9631EX–AT42–12/13
Features
Number of keys:
Up to 24
Technology:
Patented charge-transfer (transverse mode), with frequency hopping
Key outline sizes:
6 mm × 6 mm or larger (panel thickness dependent); different sizes and
shapes possible
Key spacings:
8 mm or wider, center to center (panel thickness dependent)
Electrode design:
Two-part electrode shapes (drive-receive); wide variety of possible layouts
Layers required:
One layer (with jumpers), two layers (no jumpers)
Electrode materials:
PCB, FPCB, silver or carbon on film, ITO on film
Panel materials:
Plastic, glass, composites, painted surfaces (low particle density metallic
paints possible)
Panel thickness:
Up to 50 mm glass, 20 mm plastic (key size dependent)
Key sensitivity:
Individually settable via simple commands over communication interface
Interface:
I2C slave mode (up to 400 kHz data transfer speed)
Debug output
Signal processing:
Self-calibration, auto drift compensation, noise filtering, Adjacent Key
Suppression® (AKS®) technology
FMEA compliant design features
IEC/EN/UL60730 compliant design features
UL approval
VDE compliance
For use in both class B and class C safety-critical products
Detects and Reports Key Failure
Power:
+3 V to 5V
Package:
32-pin 7 × 7 mm TQFP RoHS compliant
32-pin 5 × 5 mm QFN RoHS compliant
Atmel AT42QT1244
24-key QMatrix FMEA IEC/EN/UL60730 Touch Sensor
PRELIMINARY DATASHEET
2
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
1. Pinout and Schematic
1.1 Pinout Configuration
1.2 Pin Descriptions
Y2A
Y1A
Y0A
RST
SCL
SDA
SMP
Y2B
DBG_CLK
VREF
S_SYNC/DBG_DATA
X0
X1
X2
X3
X4
M_SYNC
CHANGE
VSS
VDD
VSS
VDD
X6
X7 X5
VDD
A1
VDD
VSS
A0
Y0B
Y1B
1
2
3
4
5
6
7
817
18
19
20
21
22
23
24
32 31 30 29 28 27 26 25
910 11 16
15
14
13
12
QT1244
Table 1-1. Pin Descriptions
Pin Name Type Comments
If Unused, connect
To...
1M_SYNC IMains Sync input Vdd
2CHANGE OD State change notification, active low. Has internal
20 k–50k pull-up resistor Leave open
3VSS PGround –
4VDD PPower –
5VSS PGround –
6VDD PPower –
7X6 OX matrix drive line Leave open
8X7 OX matrix drive line Leave open
3
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
I Input only O Output only, push-pull I/O Input and output
OD Open-drain output P Ground or power
9DBG_CLK ODebug clock Leave open
10 VREF IConnect to Vss –
11 S_SYNC/
DBG_DATA OOscilloscope sync / Debug data Leave open
12 X0 OX matrix drive line Leave open
13 X1 OX matrix drive line Leave open
14 X2 OX matrix drive line Leave open
15 X3 OX matrix drive line Leave open
16 X4 OX matrix drive line Leave open
17 X5 OX matrix drive line Leave open
18 VDD PPower –
19 A1 I I2C address 1, read after reset –
20 VDD PPower –
21 VSS PGround –
22 A0 I I2C address 0, read after reset –
23 Y0B I/O Y line connection Leave open
24 Y1B I/O Y line connection Leave open
25 Y2B I/O Y line connection Leave open
26 SMP O Sample output –
27 SDA OD Serial data line –
28 SCL OD Serial clock line –
29 RST IReset low; has internal 30 k – 60 k pull-up
resistor. This pin should be controlled by the host. Leave open or Vdd
30 Y0A I/O Y line connection Leave open
31 Y1A I/O Y line connection Leave open
32 Y2A I/O Y line connection Leave open
Table 1-1. Pin Descriptions (Continued)
Pin Name Type Comments
If Unused, connect
To...
4
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
1.3 Schematic
Figure 1-1. Typical Circuit
For component values in Figure 1-1 check the following sections:
Section 3.3 on page 6: Cs capacitors (Cs0 – Cs2)
Section 3.5 on page 8: Sample resistors (Rs0 – Rs2)
Section 3.7 on page 8: Matrix resistors (Rx0 – Rx7, Ry0 – Ry2)
Section 3.10 on page 11: Voltage levels
Rp2
VDD
MATRIX Y SCAN IN MATRIX X DRIVE
SCL
SDA
Rp1
VREG
Cs2
Cs1
Cs0
*Rx6
*Rx7
**Ry2
**Ry1
**Ry0
*Rx1
*Rx4
Creg
VDD
Rs2 Rs0
*Rx3
*Rx2
*Rx0
*Rx5
Vunreg
MAINS SYNC
CHANGE
SCOPE SYNC
I2C
QT1244
Creg
Rs1
Note:
1) It is important to place all Cs, Rs, Rx and Ry
components physically near to the chip.
2) Leave YnA, YnB unconnected if not used.
* optional - for emission suppression
** optional - for RF susceptibility improvement
Follow regulator manufacturer's recommended values
for input and output bypass capacitors Creg.
Keep the output capacitor close to pins 4 and 6.
Add a 100 nF capacitor close to pin 18.
5
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
2. Overview
The AT42QT1244 (QT1244) device is a digital burst mode charge-transfer (QT™) sensor, designed specifically for
matrix layout touch controls. It includes all signal processing functions necessary to provide stable sensing under a
wide variety of changing conditions. Only a few external parts are required for operation. The entire circuit can be
built within a few square centimeters of single-sided PCB area. CEM-1 and FR1 punched, single-sided materials can
be used for the lowest possible cost. The rear of the PCB can be mounted flush on the back of a glass or plastic
panel using a conventional adhesive, such as 3M VHB two-sided adhesive acrylic film.
The QT1244 employs transverse charge-transfer (QT) sensing, a technology that senses changes in electrical
charge forced across two electrode elements by a pulse edge (see Figure 3.).
Figure 2-1. Field Flow Between X and Y Elements
The QT1244 allows a wide range of key sizes and shapes to be mixed together in a single touch panel (see Section
3.8 on page 10).
The device uses an I2C interface to allow key data to be extracted and to permit individual key parameter setup. This
interface uses a memory mapped structure, designed to minimize the amount of data traffic while maximizing the
amount of information conveyed.
In addition to normal operating and setup functions the device can also report back actual signal strengths.
QmBtn software for the PC can be used to program the operation of the device as well as read back key status and
signal levels in real time.
The device also includes a Debug output interface, which can be used to monitor many operating variables during
product development.
overlying panel
CMOS
driver
X element Y element
6
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
3. Hardware and Functional
3.1 Matrix Scan Sequence
The circuit operates by scanning each key sequentially. The keys are numbered from 0 – 23. Key scanning begins
with location X = 0, Y = 0 (key 0). All keys on Y0 are scanned first, then Y1 and finishing with all keys on Y2.
For example, the sequence is:
X0Y0, X1Y0 to X7Y0
X0Y1, X1Y1 to X7Y1
X0Y2, X1Y2 to X7Y2
Table 3-1 shows the key numbering.
Table 3-1. Key Numbers
Each key is sampled in a burst of acquisition pulses whose length is determined by the setups parameter BL
(Section 7. on page 45). This can be set on a per-key basis. A burst is completed entirely before the next key is
sampled. At the end of each burst the resulting signal is converted to digital form and processed. The burst length
directly impacts key gain. Each key can have a unique burst length in order to allow tailoring of key sensitivity on a
key-by-key basis.
3.2 Burst Paring
Keys that are disabled by setting NDIL to zero (Section 6.2.1 on page 30) are eliminated from the scan sequence to
save scan time and thus power. As a consequence, the fewer keys that are used the faster the device can respond.
All calibration times are reduced when keys are disabled.
3.3 Cs Sample Capacitor Operation
Cs capacitors absorb charge from the key electrodes on the rising edge of each X pulse. On each falling edge of X,
the Y matrix line is clamped to ground to allow the electrode and wiring charges to neutralize in preparation for the
next pulse. With each X pulse charge accumulates on Cs causing a staircase increase in its differential voltage.
After the burst completes, the device clamps the Y line to ground causing the opposite terminal to go negative. The
charge on Cs is then measured using an external resistor to ramp the negative terminal upwards until a zero
crossing is achieved. The time required to zero cross becomes the measurement result.
The Cs capacitors should be connected as shown in Figure 1-1 on page 4. They should be NP0 (preferred), X7R
ceramics or PPS film. NPO offers the best stability. The value of these capacitors is not critical but 4.7 nF is
recommended for most cases. If the transverse capacitive coupling from X to Y is large enough the voltage on a Cs
capacitor can saturate, destroying gain. In such cases the burst length should be reduced and/or the Cs value
increased. See Section 3.4.
If a Y line is not used its corresponding Cs capacitor may be omitted and the pins left floating.
X7 X6 X5 X4 X3 X2 X1 X0
Y0 76543210Key
numbers
Y1 15 14 13 12 11 10 9 8
Y2 23 22 21 20 19 18 17 16
7
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
3.4 Sample Capacitor; Saturation Effects
Cs voltage saturation at a pin YnB is shown in Figure 3-1. Saturation begins to occur when the voltage at a YnB pin
becomes more negative than –0.25 V at the end of the burst. This nonlinearity is caused by excessive voltage
accumulation on Cs inducing conduction in the pin protection diodes. This badly saturated signal destroys key gain
and introduces a strong thermal coefficient which can cause phantom detection.
Figure 3-1. VCs – Nonlinear During Burst
(Burst too long, or Cs too small, or X-Y transcapacitance too large)
The cause of this is either from the burst length being too long, the Cs value being too small, or the X-Y transfer
coupling being too large. Solutions include loosening up the key structure interleaving, greater separation of the X
and Y lines on the PCB, increasing Cs, and decreasing the burst length.
Increasing Cs makes the part slower, decreasing burst length makes it less sensitive. A better PCB layout and a
looser key structure (up to a point) have no negative effects.
Cs voltages should be observed on an oscilloscope with the matrix layer bonded to the panel material. If the Rs side
of any Cs ramps is more negative than –0.25 V during any burst (not counting overshoot spikes which are probe
artifacts), there is a potential saturation problem.
Figure 3-2 shows a defective waveform similar to that of Figure 3-1, but in this case the distortion is caused by
excessive stray capacitance coupling from the Y line to AC ground (for example, from running too near and too far
alongside a ground trace, ground plane, or other traces). The excess coupling causes the charge-transfer effect to
dissipate a significant portion of the received charge from a key into the stray capacitance.
Figure 3-2. VCs – Poor Gain, Nonlinear During Burst
(Excess capacitance from Y line to Gnd)
This phenomenon is more subtle. It can be best detected by increasing BL to a high count and watching what the
waveform does as it descends towards and below –0.25 V. The waveform appears deceptively straight, but it slowly
starts to flatten even before the –0.25 V level is reached.
A correct waveform is shown in Figure 3-3. Note that the bottom edge of the bottom trace is substantially straight
(ignoring the downward spikes).
X Drive
YnB
X Drive
YnB
8
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Figure 3-3. VCs – Correct
Unlike other QT circuits, the Cs capacitor values on QT1244 have no effect on conversion gain. However, they do
affect conversion time.
Unused Y lines should be left open.
3.5 Sample Resistors
The sample resistors (Rs0 – Rs2) are used to perform single-slope ADC conversion of the acquired charge on each
Cs capacitor. These resistors directly control acquisition gain. Larger values of Rs proportionately increase signal
gain. For most applications Rs should be 1 M. Unused Y lines do not require an Rs resistor.
3.6 Signal Levels
Using Atmel QmBtn software it is easy to observe the absolute level of signal received by the sensor on each key.
The signal values should normally be in the range of 200 – 750 counts with properly designed key shapes (see the
Touch Sensors Design Guide, available on the Atmel website) and values of Rs. However, long adjacent runs of X
and Y lines can also artificially boost the signal values, and induce signal saturation: this is to be avoided. The X-to-
Y coupling should come mostly from intra-key electrode coupling, not from stray X-to-Y trace coupling.
QmBtn software is available free of charge on the Atmel web site.
The signal swing from the smallest finger touch should preferably exceed 8 counts, with 12 being a reasonable
target. The signal threshold setting (NTHR) should be set to a value guaranteed to be less than the signal swing
caused by the smallest touch.
Increasing the burst length (BL) parameter increases the signal strengths as does increasing the sampling resistor
(Rs) values.
3.7 Matrix Series Resistors
The X and Y matrix scan lines can use series resistors (Rx0 – Rx7 and Ry0 – Ry2 respectively) for improved EMC
performance (Figure 1-1 on page 4).
X drive lines require Rx in most cases to reduce edge rates and thus reduce RF emissions. Values range from
1k– 100 k, typically 1 k.
Y lines need Ry to reduce EMC susceptibility problems and in some extreme cases, ESD. Values range from 1 k–
100 k, typically 1 k.Y resistors act to reduce noise susceptibility problems by forming a natural low-pass filter with
the Cs capacitors.
It is essential that the Rx and Ry resistors and Cs capacitors be placed very close to the chip. Placing these parts
more than a few millimeters away opens the circuit up to high frequency interference problems (above 20 MHz) as
the trace lengths between the components and the chip start to act as RF antennae.
The upper limits of Rx and Ry are reached when the signal level and hence key sensitivity are clearly reduced. The
limits of Rx and Ry depend on key geometry and stray capacitance, and thus an oscilloscope is required to
determine optimum values of both.
X Drive
YnB
9
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Dwell time is the duration in which charge coupled from X to Y is captured. Increasing the dwell time recovers some
of the signal levels lost to higher values of Rx and Ry, as shown in Figure 3-4. Too short a dwell time causes charge
to be lost if there is too much rising edge roll-off. Lengthening the dwell time causes this lost charge to be recaptured,
thereby restoring key sensitivity. Dwell time is adjustable (see Section 6.10.1 on page 39) to one of 8 values.
Figure 3-4. Drive Pulse Roll-off and Dwell Time
Dwell time problems can also be solved by either reducing the stray capacitance on the X line(s) (by a layout
change, for example by reducing X line exposure to nearby ground planes or traces), or, the Rx resistor needs to be
reduced in value (or a combination of both approaches).
One way to determine X settling time is to monitor the fields using a patch of metal foil or a small coin over the key
(Figure 3-5). Only one key along a particular X line needs to be observed, as each of the keys along that X line will
be identical. The chosen dwell time should exceed the observed 95% settling of the X-pulse by 25% or more.
Figure 3-5. Probing X-Drive Waveforms With a Coin
X drive
Y gate
Dwell time Lost charge due to
inadequate settling
before end of dwell time
10
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
3.8 Key Design
For information about key design refer to the Touch Sensors Design Guide on the Atmel website.
3.9 PCB Layout, Construction
3.9.1 Overview
The chip should be placed near the touch keys on the same PCB so as to reduce X and Y trace lengths, thereby
reducing the chances for EMC problems. Long connection traces act as RF antennae. The Y (receive) lines are
much more susceptible to noise pickup than the X (drive) lines.
Even more importantly, all signal related discrete parts (resistors and capacitors) should be very close to the body of
the chip. Wiring between the chip and the various resistors and capacitors should be as short and direct as possible
to suppress noise pickup.
Note: Ground planes and traces should NOT be used around the keys and the Y lines from the keys. Ground
areas, traces, and other adjacent signal conductors that act as AC ground (such as Vdd and LED drive
lines) absorb the received key signals and reduce signal-to-noise ratio (SNR) and thus are
counterproductive. Ground planes around keys also make water film effects worse.
Ground planes, if used, should be placed under or around the chip itself and the associated resistors and capacitors
in the circuit, under or around the power supply, and back to a connector, but nowhere else.
3.9.2 LED Traces and Other Switching Signals
Digital switching signals near the Y lines induces transients into the acquired signals, deteriorating the SNR
performance of the device. Such signals should be routed away from the Y lines, or the design should be such that
these lines are not switched during the course of signal acquisition (bursts).
LED terminals which are multiplexed or switched into a floating state and which are within or physically very near a
key structure (even if on another nearby PCB) should be bypassed to either Vss or Vdd with at least a 10 nF
capacitor to suppress capacitive coupling effects which can induce false signal shifts. The bypass capacitor does not
need to be next to the LED, in fact it can be quite distant. The bypass capacitor is noncritical and can be of any type.
LED terminals which are constantly connected to Vss or Vdd do not need further bypassing.
3.9.3 PCB Cleanliness
Modern no-clean flux is generally compatible with capacitive sensing circuits.
If a PCB is reworked in any way, clean it thoroughly to remove all traces of the flux residue around the capacitive
sensor components. Dry it thoroughly before any further testing is conducted.
CAUTION: If a PCB is reworked in any way, it is highly likely that the behavior of the
no-clean flux changes. This can mean that the flux changes from an inert material to
one that can absorb moisture and dramatically affect capacitive measurements due
to additional leakage currents. If so, the circuit can become erratic and exhibit poor
environmental stability.
11
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
3.10 Power Supply Considerations
For the power supply range see Section 8.2 on page 47. If the power supply fluctuates slowly with temperature, the
device tracks and compensates for these changes automatically with only minor changes in sensitivity. If the supply
voltage drifts or shifts quickly, the drift compensation mechanism is not able to keep up, causing sensitivity
anomalies or false detections.
As this device uses the power supply itself as an analog reference, the power should be very clean and come from a
separate regulator. A standard inexpensive Low Dropout (LDO) type regulator should be used that is not also used
to power other loads such as relays or other high current devices. Load shifts on the output of the LDO can cause
Vdd to fluctuate enough to cause false detection or sensitivity shifts.
Caution: A regulator IC shared with other logic can result in erratic operation and is not advised.
A regulator can be shared among two or more devices on one board.
A single ceramic 0.1 µF bypass capacitor, with short traces, should be placed very close to each supply pin of the IC.
Failure to do so can result in device oscillation, high current consumption and erratic operation.
3.11 Startup/Calibration Times
The device employs a rigorous initialization and self-check sequence for EN 60730 compliance. The last step in this
sequence enables the serial communication interface, but only if the self-checks pass. The communication interface
is not enabled if a safety critical fault is detected during the startup sequence. A maximum of 95 ms is required from
reset before the device is ready to communicate.
The device determines a reference level for each key by calibrating all the keys immediately after initialization. Each
key is calibrated independently and in parallel with all other enabled keys. Calibration takes between 11 and 62
keyscan cycles, each cycle being made up of one sample from each enabled key. The device ends calibration for a
key if its reference has converged with the signal DC level. The calibration time is shortest when the keys signals are
stable, typically increasing with increasing noise levels to the maximum of 62 keyscan cycles.
An error is reported for each key where calibration continues for the maximum number of keyscan cycles and the key
reference does not appear to have converged with the signals DC level. Noise levels can vary from key to key such
that some keys may take longer to calibrate than others. However, the device can report during this interval that the
key(s) affected are still in calibration via status function bits. Table 3-2 shows keyscan cycles times and calibration
times per key versus dwell time and burst length for all 24 keys enabled. The values given assume that
MSYNC = off, SLEEP = off, FHM = off and DEBUG = off .
12
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Keyscan cycle and thus calibration time varies with different setups as well as keypad design. Increasing capacitive
loading at each key increases the signal levels and hence increases the keyscan cycle time. Conversely, disabled
keys are subtracted from the burst sequence and thus the keyscan cycle time is shortened. The keyscan cycle time
should be measured on an oscilloscope for the specific setups in use.
The initialization time and calibration time must be added together to determine the total time from reset to a keys
ability to report detection.
3.12 Reset Input
The RST pin can be used to reset the device to simulate a power-down cycle, in order to then bring the device up
into a known state should communications with the device be lost. The pin is active low, and a low pulse lasting at
least 10 µs must be applied to this pin to cause a reset.
The reset pin has an internal 30 k–60k resistor. A 2.2 µF capacitor plus a diode to Vdd can be connected to this
pin as a traditional reset circuit, but this is not essential. Where the QT1245 has detected a failure of one of the
internal EN60730 checks and has subsequently locked up in an infinite loop, only a power cycle or an external
hardware reset can restore normal operation. It is strongly recommended that the host has control over the RST pin.
If an external hardware reset is not used, this pin may be connected to Vdd or left floating.
3.13 Frequency Hopping
The QT1244 supports frequency hopping to avoid a clash between the sampling frequency and noise at specific
frequencies elsewhere in products or product-operating environments. It tries to hop away from the noise.
During the acquisition bursts, a sequence of pulses are emitted with a particular spacing, which equates to a
particular sampling frequency. If the latter should coincide with significant noise generated elsewhere, touch sensing
may be seriously impaired or false detections may occur.
To help combat such noise, the burst frequency can either be preset to one specific frequency (frequency hopping
disabled) away from the noisy frequency, or frequency hopping can be enabled and set to switch dynamically
between three specific configured frequencies or even set to sweep a configured range of frequencies.
3.14 Detection Integrators
See also Section 6.2.1 on page 30.
The device features a detection integration mechanism, which acts to confirm a detection in a robust fashion. A per-
key counter is incremented each time the key has exceeded its threshold and is decremented each time the key
does not exceed its threshold. When this counter reaches a preset limit the key is finally declared to be touched.
For example, if the limit value is 10, then the device has to exceed its threshold and stay there for a minimum of 10
acquisitions before the key is declared to be touched.
Table 3-2. Keyscan Cycle and Calibration Times
Setups Keyscan Cycle Time Calibration Time (min) Calibration Time (max)
BL = 0 (16 pulses),
DWELL = 0 (125 ns)
FREQ0 = 0
Signal level = 100 counts
7ms 77 ms (11 × 7) 434 ms (62 × 7)
BL = 3 (64 pulses),
DWELL = 7 (4.5 µs)
FREQ0 = 63
Signal level = 4000 counts
105 ms 1155 ms (11 × 105) 6510 ms (62 × 105)
13
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
The QT1244 uses a two-tier confirmation mechanism having two such counters for each key. These can be thought
of as inner loop and outer loop confirmation counters.
The inner counter is referred to as the fast DI. This acts to attempt to confirm a detection via rapid successive
acquisition bursts, at the expense of delaying the sampling of the next key. Each key has its own fast DI counter and
limit value. These limits can be changed via the setups block on a per-key basis.
The outer counter is referred to as the normal DI. This DI counter increments whenever the fast DI counter has
reached its limit value. The normal DI counter also has a limit value which is settable on a per-key basis.
If a normal DI counter reaches its terminal count, the corresponding key is declared to be touched and becomes
active. Note that the normal DI can only be incremented once per complete keyscan cycle (that is, more slowly,
whereas the fast DI is incremented on the spot without interruption).
The net effect of this mechanism is a multiplication of the inner and outer counters and hence a highly noise-resistant
sensing method. If the inner limit is set to 5, and the outer to 3, the net effect is a minimum of 5 × 3 = 15 threshold
crossings to declare a key as active.
3.15 Sleep
If the sleep feature is enabled (see Section 7. on page 45), the device sleeps whenever possible to conserve power.
Periodically, it wakes automatically, scans the matrix, and returns to sleep unless there is activity which demands
further attention. The device returns to sleep automatically once all activity has ceased. The time for which it sleeps
before automatically awakening can be configured.
A new communication with the device while it is asleep causes it to wake up, service the communication and scan
the matrix. At least one full matrix scan is always performed after waking up and before returning to sleep.
At the end of each matrix scan, the part returns to sleep unless recent activity demands further attention. If there has
been recent activity, the part performs another complete matrix scan and then attempts to sleep once again. This
process is repeated indefinitely until the activity stops and the part returns to sleep.
Key touch activity prevents the part from sleeping. The part will not sleep while any key is calibrating or if any touch
events were detected at any key in the most recent scan of the key matrix, or while a serial communication is in
progress.
If the sleep feature is disabled in the setups, the device never sleeps. Sleep should be disabled if the device is being
used in an FMEA or EN 60730 compliant design because all operations are stopped within the device while the part
is asleep and the host is not able to distinguish between faulty operation and EN 60730 counters appearing to run
slow because the part is intermittently sleeping. It should also be noted that the drift compensation interval will
effectively be stretched each time the device sleeps because the device is fully halted during sleep and cannot
perform drift compensation. If sleep is enabled, a shorter drift compensation interval may be required.
3.16 FMEA Tests
Failure Modes and Effects Analysis (FMEA) is a tool used to determine critical failure problems in control systems.
FMEA analysis is being applied increasingly to a wide variety of applications including domestic appliances. To
survive FMEA testing the control board must survive any single problem in a way that the overall product can either
continue to operate in a safe way, or shut down.
The most common FMEA requirements regard opens and shorts analysis of adjacent pins on components and
connectors. However, other criteria must usually be taken into account, for example complete device failure.
The device incorporates special self-test features which allow products to pass such FMEA tests easily, and enable
key failure to be detected. These tests are performed in an extra burst slot after the last enabled key.
The FMEA testing is done on all enabled keys in the matrix, and results are reported via the serial interface. Disabled
keys are not tested.
All FMEA tests are repeated every few seconds or faster during normal run operation, if the sleep feature is disabled.
Sometimes FMEA errors can occur intermittently (for example, due to momentary power fluctuations). It is advisable
to confirm a true FMEA fault condition by making sure the error flags persist for several seconds.
14
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Also, since the device only communicates in slave mode, the host can determine immediately if the device has
suffered a catastrophic failure.
The FMEA tests performed allow detection of fault conditions including the following:
X drive line shorts to Vdd and Vss
X drive line shorts to other pins
X drive signal deviation
Y line shorts to Vdd and Vss
Y line shorts to other pins
X to Y line shorts
Cs capacitor shorts and open circuits
Vref
Key gain (see Section 6.9.2 on page 39 (Key Gain Test Threshold, KGTT))
Other tests incorporated into the device include:
A test for signal levels against a preset minimum value (LSL setup, see Section 7. on page 45). If any signal
level falls below this level, an error flag is generated.
CRC communications checks on all data read from the device.
Some very small key designs have very low X-Y coupling. In these cases, the amount of signal is very small, and the
key gain is low. As a result, small keys can fail the LSL test see (Section 6.9.1 on page 38) or the FMEA key gain
test. In such cases, the burst length of the key should be increased so that the key gain increases. Failing that, a
small ceramic capacitor, for example 3 pF, can be added between the X and Y lines serving the key to artificially
boost signal strength. This capacitor must be located at the key's site to ensure the FMEA tests are not
compromised.
3.17 IEC/EN 60730 Compliance
The device also incorporates special test features which, together with the FMEA tests, allow products to achieve
IEC/EN60730 compliance with ease.
IEC/EN60730 compliance demands dynamic verification of all safety related components and sub-components
within a product. The QT1244 is able to verify some sub-components internally, but others require verification by a
separate, independent processing unit with another timing source. To this end the QT1244 exposes a number of
internal operating parameters through its serial communications interface and requires the cooperation of a host to
check and verify these parameters regularly. It is also necessary for the host to verify the communications
themselves by checking and validating the communications CRC, which the QT1244computes during each read
transfer sequence (see Section 5.5 on page 24).
If a CRC check should fail, the host should not rely on the data but retry the transmission. Occasional CRC failures
might be anticipated as a result of noise spikes. Repeated CRC failures might indicate a safety-critical failure. Where
the QT1244 is able to verify sub-components internally, but any such verification fails, the device disables serial
communication and locks up in an infinite loop. The host can detect this condition as repeated serial communication
CRC failures.
During normal operation the host must perform regular reads of the IEC/EN60730 counters (see Section 5.4 on page
23) to verify correct operation of the QT1244. The host must also perform regular reads of the device status (see
Section 5.5 on page 24) and verify there are no errors reported. The FMEA error flag, LSL error flag and setups CRC
error flag must all be considered as part of an IEC/EN60730 compliant design.
The host can try to recover from any safety critical failure by resetting the QT1244 using its RST pin. The host should
allow a grace period in consideration of the startup and initialization time the QT1244 requires after reset to
communicate (see Section 3.11 on page 11).
15
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
The sub-components that the QT1244 is able to verify internally are tested repeatedly during the normal running of
the device, with the various tests run in parallel. As each test ends the result is recorded and the test is restarted.
The real time that elapses from the start of each test to the start of the next iteration of the same test is called the
failure detect time, or hazard time, the maximum time for which an error could be undetected.
Each test is broken down into a number of smaller parts, each of which is processed in turn during each matrix scan.
Each test is therefore completed either after a number of matrix scans, as shown in Table 3-3.
Table 3-4 shows matrix scan times for Setups that yield the shortest matrix scan time and a much longer scan time
resulting from the use of long dwell and low frequency settings.
Longer matrix scan times are possible than those shown in Table 3-4 by using even higher values for FREQ0 (lower
burst frequencies), but these are considered extreme settings.
Table 3-3. Test run times (expressed in matrix scans)
Test
Required Matrix Scans
to complete test
FMEA 24
Other 18
Variable Memory 768
Firmware CRC 342
Setups CRC 5
Table 3-4. Matrix Scan Times
Setups Conditions Matrix Scan Time (ms)
BL = 0 (16 pulses),
DWELL = 0 (0.13 µs),
FREQ0 = 1,
All keys enabled,
FHM = 0,
MSYNC = 0 (off),
SLEEP = 0 (sleep disabled),
DEBUG = 0 (off).
7.5
BL = 3 (64 pulses),
DWELL=7(4.5µs),
FREQ0 = 25,
All keys enabled,
FHM = 0,
MSYNC = 0 (off),
SLEEP = 0 (sleep disabled),
DEBUG = 0(off).
38
16
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Table 3-5 shows the failure detect times for the internal tests assuming a matrix scan time of 9 ms, which is valid for
typical Setups.
Longer failure detect times are possible than those shown in Table 3-5 where the matrix scan time is longer. The
failure detect times are proportional to the matrix scan time. The failure detect time for other setups can therefore be
determined by observing the matrix scan time using an oscilloscope and scaling the times given in Table 3-5
accordingly. Alternatively, the failure detect times can be calculated by taking the numbers from Table 3-5 and
multiplying them by the matrix scan time.
Unnecessarily long settings of dwell and low burst frequencies should be avoided because these will also result in
undesirably long failure detect times.
3.17.1 UL approval / VDE compliance
The QT1244 has been given a compliance test report by VDE and is approved by UL as a component suitable for
use in both class B and class C safety critical products. By using this device and following the safety critical
information throughout this datasheet, manufacturers can easily add a touch sense interface to their product, and be
confident it can also readily pass UL or VDE testing.
Table 3-5. Failure Detect Time
Test Failure Detect Time (ms)
FMEA 216
Other 162
Variable Memory 6912
Firmware CRC 3078
Setups CRC 45
Conditions: Matrix scan time = 9 ms. QT1244 does not
sleep for duration of tests.
17
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
4. Serial Communications
4.1 Introduction
The device uses an I2C interface for communications with a host, and also includes a Debug output interface, which
can be used to monitor many operating variables during product development.
4.2 I2C Serial Communication Bus
The device communicates over an I2C bus, only in slave mode.
Pins A0, A1 are used to configure the I2C addresses (see Table 4-1) and should be pulled up to Vdd or pulled down
to Vss using 10 k resistors. These pins are read after reset, during initialisation.
The QT1244 only responds to the correct address match. I2C operating parameters are as follows:
Max Clock Frequency: 400 kHz
Address: 7-bit
The QT1244 allows multiple byte transmissions to provide a more efficient communication. This is particularly useful
to retrieve several information bytes at once. Every time the host retrieves data from the QT1244, an internal
address pointer is incremented. Therefore, the host needs only to write the initial address pointer of interest (the
lowest address) and the number of bytes to read, followed by read cycles for as many bytes as required.
The device also calculates a 16-bit CRC on the data read, thus allowing the host to validate the data read. Although
the CRC is a requirement for EN 60730 compliance, and is recommended for robust communications, reading the
CRC is optional.
4.3 Debug Output Interface
The QT1244 includes a debug interface which may be used for observing many internal operating variables, in real
time, even while the part is actively communicating with a host over the I2C serial communications bus. The Debug
interface provides a useful aid during product development (see Appendix C. on page 55).
4.4 CHANGE Pin
The CHANGE pin can be used to alert the host to key touches or key releases, thus reducing the need for
unnecessary communications. Normally, the host can simply not bother to communicate with the device, except
when the CHANGE pin becomes active.
CHANGE becomes active after reset and when there is a change in key state (either touch or touch release) and
becomes inactive again only when the host performs a read from address 6, the detect status register for all keys on
Y0. CHANGE does not self-clear. Only an I2C read from location 6, or a device reset, clears it.
It is important to read all three key state addresses to ensure the host has a complete picture of which keys have
changed.
Table 4-1. Interface Details
A1 A0
I2C Address
(decimal)
Vss Vss 57
Vss Vdd 7
Vdd Vss 17
Vdd Vdd 117
18
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
After the device is reset it performs internal initialisation and then sets CHANGE active (low) to signal the host that it
is ready to communicate.
CHANGE is an open-drain output with an internal 20 k–50k pull-up resistor. This allows multiple devices to be
connected together in a single wire-OR logic connection with the host. When the CHANGE pin goes active, the host
can poll all devices to identify which one is reporting a touch change.
Every key can be individually configured to wake a host microcontroller upon a touch change. Therefore, a product
can wake from sleep when any key state changes, or only when certain desired keys change state. The
configuration is set in the setups block (see Appendix 6.2.3 on page 32) on a key-by-key basis.
IEC/EN60730 compliant products cannot rely on the CHANGE pin because its operation cannot be verified. The
CHANGE pin can still be utilized but only to optimize the key response time. The host must also poll the QT1244
Detect Status bytes (Addresses 6, 7, 8), but at a rate suitable to guarantee IEC/EN60730 compliance. A poll rate of
once every 100 ms would impose very little extra load on the QT1244.
19
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
5. Memory Map
5.1 Introduction
The device features a set of commands which are used for control and status reporting.
As well as Table 5-1 refer to Table 7-1 on page 45 for further details.
Table 5-1. Memory Map
Address Use Access
0Reserved Read
1Reserved Read
2100 ms counter (IEC/EN60730) Read
3Signal fail counter (IEC/EN60730) Read
4Matrix Scan counter (IEC/EN60730) Read
5Device Status. Collection of bit flags Read
6Detect status for keys 0 to 7, one bit per key Read
7Detect status for keys 8 to 15, one bit per key Read
8Detect status for keys 16 to 23, one bit per key Read
9Reserved Read
10 Current frequency Read
11 Current pulse spacing Read
12–15 Data for key 0. See Table 5-6 on page 26 for details. Read
16–19 Data for key 1 Read
20–23 Data for key 2 Read
24–27 Data for key 3 Read
28–31 Data for key 4 Read
32–35 Data for key 5 Read
36–39 Data for key 6 Read
40–43 Data for key 7 Read
44–47 Data for key 8 Read
48–51 Data for key 9 Read
52–55 Data for key 10 Read
56–59 Data for key 11 Read
60–63 Data for key 12 Read
64–67 Data for key 13 Read
68–71 Data for key 14 Read
72–75 Data for key 15 Read
20
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Poll rate: The host can make use of the CHANGE pin output to initiate a communication. This guarantees the
optimal polling rate.
If the host cannot make use of the CHANGE pin the poll rate in normal run operation should be no faster than once
per matrix scan (see Section 8.4 on page 48). Typically 10 to 20 ms is more than fast enough to extract the key
status. Anything faster will not provide new information and slows down the chip operation.
Run Poll Sequence: In normal run mode the host should limit traffic with a minimalist control structure. The host
should just read the IEC/EN60730 counters, the Device Status and the three detect status registers (see Figure 5-1
on page 21).
Repeated Start: Using repeated start is not allowed and can cause communication failure. The host should instead
perform a STOP condition followed by a minimum delay of 150 µs and then perform a START condition.
76–79 Data for key 16 Read
80–83 Data for key 17 Read
84–87 Data for key 18 Read
88–91 Data for key 19 Read
92–95 Data for key 20 Read
96–99 Data for key 21 Read
100 – 103 Data for key 22 Read
104 – 107 Data for key 23 Read
108 – 139 Reserved
140
Control command. Write 0xFF to calibrate all keys. Write 0xFE
immediately before writing setups. Write 0xFD to perform low level
calibration and offset for frequency hopping. Write k to calibrate key k.
Write 0x18 to reset the device.
Write
141 – 250 Setups – see Section 6. on page 28 for details Read/Write
Table 5-1. Memory Map (Continued)
Address Use Access
21
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Figure 5-1. Power-on or Hardware Reset Flow Chart
5.2 Writing Data to the Device
The sequence of events required to write data to the device is shown below:
Verify Setup
Block
Recalibrate All
Send 255 to
Address 191
Read key
status registers
Addresses: 6, 7
and 8
Recalibrate All
Send 255 to
Address 191
Host main
Process
Key Detection(s) / End of
Detection Processing
'Change'
output set
Incorrect
Setup Data
Correct Setup
Block
Power On or Hardware Reset
Stuck Key
Detected
Keys OK
Send Correct
Setup Block
Internal Host
Processes
Comms with
QT
Legend
Host to Device Device Tx to Host
SSLA+W AMemAddress A0x00 AData A P
Table 5-2. Key to Write Sequence
Name Use
SStart condition
SLA+W Slave address plus write bit
AAcknowledge bit
MemAddress Target memory address within device
0x00 Reserved byte
Data Data to be written
PStop condition
22
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
The host initiates the transfer by sending the START condition, and follows this by sending the slave address of the
device together with the Write-bit. The device then sends an ACK. The host sends the internal memory address it
wishes to write to. The device then sends an ACK. The host transmits a byte with value 0x00. The device then sends
an ACK. The host transmits one or more data bytes. Each is acknowledged by the device.
If the host sends more than one data byte, they are written to consecutive memory addresses. The device
automatically increments the target memory address after writing each data byte. After writing the last data byte, the
host should send the STOP condition.
The host should not try to write beyond address 255 because the device will not increment the internal memory
address beyond this.
5.3 Reading Data From the Device
The sequence of events required to read data from the device is shown below:
The host initiates the transfer by sending the START condition, and follows this by sending the slave address of the
device together with the Write-bit. The device then sends an ACK. The host sends the internal memory address it
wishes to read from. The device then sends an ACK. The host sends n, where n is the number of data bytes it wishes
to read.
The host sends a STOP condition, waits 150 µs and sends another START condition followed by the slave address,
but this time accompanied by the Read-bit. The device returns an ACK, followed by a data byte. The host must
return either an ACK or NACK. If the host responds with an ACK, the device subsequently transmits another byte. If
the host responds with a NACK, the device will not perform another transmission.
Host to Device Device Tx to Host
SSLA+W AMemAddress A n A P 150 µs S SLA+R A
Data1 AData 2 AData n ACRC LSB ACRC MSB /A
Table 5-3. Key to Read Sequence
Name Use
SStart condition
SLA+W Slave address plus write bit
AAcknowledge bit
MemAddress Target memory address within device
n# bytes to read. Valid range is 1 – 253.
CRC is available after reading n bytes
Data Data from device
PStop condition
150µsDelay between STOP and START conditions
SLA+R Slave address plus read bit
/A Not Acknowledge bit/indicates last byte transmission
23
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Each time a data byte is transmitted, the device automatically increments the internal address. When all n data bytes
have been returned, the device returns a 16-bit CRC, LSB first. After the device has returned the CRC LSB, the host
should respond with an ACK. After the device has returned the CRC MSB, the host responds with a NACK. The host
should terminate the transfer by issuing the STOP condition.
The device calculates the 16-bit CRC using the slave address, the internal memory start address, the number of
bytes to be read (n), and the n data bytes themselves, all in the same sequence they occur during the transmission
(see Appendix B. on page 54).
5.4 Report IEC/EN60730 Counters
These counters can be used by the host to check the correct speed and operation of the device. The host must
check these values regularly to meet the requirements of IEC/EN60730. IEC/EN60730 requires that each
component of a system be checked for correct operation. Where correct speed of operation must be confirmed and
the device has no way to perform such a cross-check internally, counters are exposed through the communication
interface to enable independent cross checking by the host.
Address 2: 100 ms counter (IEC/EN60730)
This is an 8-bit unsigned counter that is incremented once every 100 ms, counting 256 steps repeatedly from 0 to
255. When the counter has reached 255 it wraps back to 0 at the next 100 ms interval. The counter should take
between 25 and 26 seconds (256 × 100 ms = 25.6s) to count up from zero through 255 and wrap back to zero again.
The host must read this counter regularly and cross-check the counting rate against one of its own clock sources.
If the 100 ms counter is read once every second, for example, the host should find the counter has increased by 10
counts from the value returned at each previous read and should traverse one full count range (256 steps) when the
host has read the counter 25 or 26 times. The host should verify the 100 ms counter is incrementing at the expected
rate. If the counter advances faster or slower than expected, there could be a fault with the QT1244 or the host, and
the host should adopt an appropriate strategy to meet the required safety standard.
Address 3: Signal fail counter (IEC/EN60730)
This is an 8-bit unsigned counter that is incremented each time a signal capture failure occurs. Signal capture failure
can occur where keys are enabled but do not physically exist. Only keys that exist should be enabled. All other keys
should be disabled.
Signal capture failure can also occur where heavy noise spikes corrupt the signal. Occasional capture failure is to be
expected and does not unduly affect the device performance. Regular capture failure would extend the key response
time. The host must check this counter regularly. Tests should be made with a heavy noise source during
development to determine how the key response time is affected and determine a maximum acceptable count rate
for this counter.
Address 4: Matrix scan counter (IEC/EN60730)
This is an 8-bit counter that is incremented before the start of each matrix scan, or keyscan cycle, counting 256 steps
repeatedly from 0 to 255. When the counter has reached 255 it wraps back to 0 at the start of the next keyscan cycle.
The keyscan cycle time should be measured with an oscilloscope during development. The Matrix Scan count rate
can be calculated directly from this.
For example, if the keyscan cycle time is measured as 10 ms, the counter counts 256 steps in 2560 ms
(256 × 10 ms). The host must read this counter regularly to check the matrix scan is operating at the expected rate.
If the Matrix Scan counter is read once every 100 ms, for example, the host should find the counter has increased by
10 counts from the value returned at each previous read and should traverse one full count range (256 steps) when
the host has read the counter 25 or 26 times. The host should verify the counter is incrementing at the expected rate.
If the counter advances faster or slower than expected, there could be a fault with the QT1244 or the host, and the
host should adopt an appropriate strategy to meet the required safety standard.
24
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
5.5 Report Device Status
Address 5: Device status
This byte contains the general status bits. The bits report as follows:
Bit 4: Set if an FMEA error was detected during operation. See Section 3.16 on page 13.
Bit 3: Reports that an enabled key has a very low reference value, lower than the user-configurable LSL value (see
Section 7. on page 45).
Bit 2: Set if any key is in the process of calibrating.
Bit 1: Set if there was a mains sync error, for example there was no Sync signal detected within the allotted 100 ms.
See Section 7. on page 45. This condition is not necessarily fatal to operation, however the device operates very
slowly and may suffer from noise problems if the sync feature was required for noise reasons. Reset the device to
clear this bit after the MSYNC setup has been cleared (OFF).
Bit 0: Set if the setups CRC does not match the CRC uploaded by the host (HCRC). The CRC is computed
repeatedly and checked against the value uploaded by the host. This bit is set if the two values do not match.
A host in an IEC/EN60730 compliant product must check bits 0, 2, 3, and 4 and handle persistent errors
appropriately to maintain safety.
Bit Description
7Reserved
6Reserved
5Reserved
41 = FMEA failure detected
31 = LSL failure detected
21 = any key in calibration
11 = Mains sync. error
01 = Setups CRC does not match HCRC
25
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
5.6 Report Detections for All Keys
Address 6: detect status for keys 0 – 7
Address 7: detect status for keys 8 – 15
Address 8: detect status for keys 16 – 23
Each location indicates all keys in detection, if any, as a bitfield. Touched keys report as 1, untouched or disabled
keys report as 0.
Note: The CHANGE pin becomes inactive on reading address 6.
Note: The device should be reset after disabling keys because, if a key was in detect when it was disabled, it
could incorrectly report detect.
5.7 Frequency Hopping
The following locations in the memory map report runtime variables used by the frequency hopping module.
Address 10: Current Frequency
When FHM = 1 or FHM = 2, the frequency hopping module switches between three configurable frequencies. This
byte indicates which of the three frequency selections is currently in use.
Address 11: Current Pulse-Spacing
The burst sampling frequency is changed by using different idle times, or pulse spacing, between pulses in the burst.
This byte reflects that idle time, with each increment representing an increase of approximately 375 ns.
Table 5-4. Bits for Key Reporting and Numbering
Address Bit Number
76543210
676543210
715 14 13 12 11 10 9 8
823 22 21 20 19 18 17 16
Table 5-5. Frequency in Use
Value Description
0Configured frequency – 0 is in use
1Configured frequency – 1 is in use
2Configured frequency – 2 is in use
26
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
5.8 Data Set for Keys
Addresses 12 – 107 allow data to be read for each key. There are a total of 24 keys and 4 bytes of data per key,
yielding a total of 96 addresses. These addresses are read-only.
The data for the keys is mapped in sequence, starting with key 0 at addresses 12 – 15. The data for key 15 is located
at addresses 72 – 75, and that for key 23 is located at addresses 104 – 107. Table 5-6 summarizes this.
Table 5-6. Data Set for Keys
Address Variable Key # # Bits Bit-mask
Valid
Range Description
12–13 Signal
DI count 012
4
0x0FFF
0xF000
0 – 4095
0–15
Signal
Normal Detector Integrator
14–15
Reference
LSL failure
FMEA failure
0
12
1
1
0x0FFF
0x1000
0x2000
0 – 4095
0, 1
0, 1
Reference
LSL 1 = Reference level < LSL (low
signal level)
FMEA 1 = FMEA failure
16–17 Signal
DI count 1
18–19
Reference
LSL failure
FMEA failure
1
72–73 Signal
DI count 15
74–75
Reference
LSL failure
FMEA failure
15
104 – 105 Signal
DI count 23 12
4
0x0FFF
0xF000
0 – 4095
0–15
Signal
Normal Detector Integrator
106 – 107
Reference
LSL failure
FMEA failure
23
12
1
1
0x0FFF
0x1000
0x2000
0 – 4095
0, 1
0, 1
Reference
LSL 1 = Reference level < LSL (low
signal level)
FMEA 1 = FMEA failure
27
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
5.9 Command Address – 140
The Command Address (140) is a write-only location. Reading from this address will not cause any side-effects but
returns undefined values. The value written defines the action taken.
(0xFF) Calibrate All:
Shortly after the QT1244 receives a value of 0xFF at the Command Address it recalibrates all keys and
restarts operation.
The host can monitor the progress of the calibration by checking the device status byte, and the data set for
each key.
(0xFE) Setups Write-enable:
Writing a value of 0xFE to the Command Address write-enables the setups block of the device. Normally the
setups are write-protected. The write-protection is engaged as soon as a read operation is performed at any
address. By writing a value of 0xFE to the Command Address, the write-protection is disengaged. The Com-
mand Address is located conveniently immediately before the setups so that the write protection may be
disengaged and the setups written in a single I2C communication sequence.
(0xFD) Low Level Cal and Offset:
Shortly after receiving this command the QT1244 performs a calibration and offset procedure across all keys
and restarts operation. If a previous command 0xFD is still being processed, the new request will be ignored.
This command takes up to 3 seconds to complete. The host can monitor the progress of the calibration by
checking the QT1244 Device Status at address 5. The calibration bit will be set throughout the process.
The low level calibration and offset procedure involves the device calibrating each key in turn at each of the
operating frequencies selected with FREQ0, FREQ1 and FREQ2, calculating the difference between the sig-
nals at those frequencies and storing the results as offsets into CFO_1 and CFO_2 for each key. When the
procedure is complete, the host can read back the setups and record CFO_1 and CFO_2 into its own copy of
the setups block. The QT1244 does not change the setups CRC, so there will be a mismatch in the setups
CRC after this command completes. The onus is on the host to compute the CRC and upload a definitive set-
ups block to the QT1244.
(0x18) Force Reset:
Shortly after the QT1244 receives a value of 0x18 at the Command Address it performs a reset. After any
reset, the device automatically performs a full key calibration on all keys.
(k) Calibrate Key:
Writing a value k in the range 0 – 23 to the Command Address requests the QT1244 to recalibrate key k.
The operation is the same as if 0xFF were written except only one key k is affected where k is from 0 – 23.
The chosen key k is recalibrated in its native timeslot. Normal running of the part is not interrupted and all
other keys operate correctly throughout. This command is for use only during normal operation to try to
recover a single key that has failed calibration or is not calibrated correctly.
28
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6. Setups
Addresses 141 – 250 provide read/write access to the setups. The host may write the entire setups block in one
operation or may write to individual setups.
The setups must be write-enabled by writing 0xFE to the Command Address immediately before writing the setups
themselves. Normally the setups are write-protected and the write-protection is engaged as soon as a read operation
is performed from any address.
When the host is writing new setups, the values are recorded into EEPROM as they arrive from the host. Setup
changes become effective immediately. However, the device should be reset because if a key was in detect when it
was disabled, it could incorrectly continue to report detect. Calibration is typically required after setups change.
Many setups employ lookup-table (LUT) value translation. The setups block summary (Table 7-1 on page 45) shows
all translation values. Default values shown are factory defaults.
6.1 Address 141 –164: NTHR, PTHR, NDRIFT, BL
6.1.1 Threshold – NTHR, PTHR
The negative threshold value is established relative to a key signal reference value. The threshold is used to
determine key touch when crossed by a negative-going signal swing after having been filtered by the detection
integrator. Larger absolute values of threshold desensitize keys since the signal must travel further in order to cross
the threshold level. Conversely, lower thresholds make keys more sensitive.
As Cx and Cs drift, the reference point drift-compensates for these changes at a user-settable rate. The threshold
level is recomputed whenever the reference point moves, and thus it also is drift compensated.
The amount of NTHR required depends on the amount of signal swing that occurs when a key is touched. Thicker
panels or smaller key geometries reduce key gain (that is, signal swing from touch, thus requiring smaller NTHR
values to detect touch).
The positive threshold is used to provide a mechanism for recalibration of the reference point when a key's signal
moves abruptly to the positive. This condition is not normal, and usually occurs only after a recalibration when an
object is touching the key and is subsequently removed.
Positive hysteresis: PHYST is fixed at 12.5% of the positive threshold value and cannot be altered.
NTHR and PTHR are programmed on a per-key basis (see Table 7-1 on page 45). They share the same space in the
address map and hence share the same value for each key. See also Section 6.10.3 on page 40.
NTHR, PTHR Typical values: 2 – 4 (8 – 12 counts of threshold; translation via LUT).
NTHR, PTHR Default value: 3 (10 counts of threshold)
Table 6-1. NTHR
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
141 – 164 BL NDRIFT NTHR, PTHR
29
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.1.2 Drift Compensation – NDRIFT, PDRIFT
Signals can drift because of changes in Cx and Cs over time and temperature. It is crucial that such drift be
compensated, else false detections and sensitivity shifts can occur.
Drift compensation (Figure 6-1) is performed by making the reference level track the raw signal at a slow rate, but
only while there is no detection in effect. The rate of adjustment must be performed slowly, otherwise legitimate
detections could be ignored. The device drift compensates using a slew-rate limited change to the reference level.
The threshold and hysteresis values are slaved to this reference.
Figure 6-1. Thresholds and Drift Compensation
When a finger is sensed, the signal falls since the human body acts to absorb charge from the cross-coupling
between X and Y lines. An isolated, untouched foreign object (a coin, or a water film) cause the signal to rise very
slightly due to an enhancement of coupling. This is contrary to the way most capacitive sensors operate.
Once a finger is sensed, the drift compensation mechanism ceases since the signal is legitimately detecting an
object. Drift compensation only works when the signal in question has not crossed the negative threshold level.
The drift compensation mechanism can be made asymmetric if desired. The drift-compensation can be made to
occur in one direction faster than it does in the other simply by changing the NDRIFT and PDRIFT setups
parameters. NDRIFT is set on a per-key basis, whereas PDRIFT is set using one global value which is applied
across all keys.
Specifically, drift compensation should be set to compensate faster for increasing signals than for decreasing
signals. Decreasing signals should not be compensated quickly, since an approaching finger could be compensated
for partially or entirely before even touching the touch pad. However, an obstruction over the sense pad, for which
the sensor has already made full allowance for, could suddenly be removed leaving the sensor with an artificially
suppressed reference level and thus become insensitive to touch. In this latter case, the sensor should compensate
for the object's removal by raising the reference level relatively quickly.
Drift compensation and the detection time-outs work together to provide for robust, adaptive sensing. The time-outs
provide abrupt changes in reference calibration depending on the duration of the signal event.
This function is programmed on a per-key basis. See Table 7-1 on page 45.
NDRIFT Typical values: 4–5
(2 – 3.3 s / count of drift compensation)
translation via LUT, page 45)
Table 6-2. NDRIFT
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
141 – 164 BL NDRIFT NTHR
Threshold
Signal
Hysteresis
Reference
Output
DRIFTST
30
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
NDRIFT Default value: 4
(2 s / count of drift compensation)
PDRIFT Typical values: 4–5
(2 – 3.3 s / count of drift compensation;
translation via LUT, page 45)
PDRIFT Default value: 4
(2 s / count of drift compensation)
6.1.3 Burst Length – BL
The signal gain for each key is controlled by circuit parameters as well as the burst length.
The burst length is simply the number of times the charge-transfer (QT) process is performed on a given key. Each
QT process is simply the pulsing of an X line once, with a corresponding Y line enabled to capture the resulting
charge passed through the keys capacitance Cx.
QT1244 uses a fixed number of QT cycles which are executed in burst mode. There can be up to 64 QT cycles in a
burst, in accordance with the list of permitted values shown in Table 7-1 on page 45.
Increasing burst length directly affects key sensitivity. This occurs because the accumulation of charge in the charge
integrator is directly linked to the burst length. The burst length of each key can be set individually, allowing for direct
digital control over the signal gains of each key individually.
Apparent touch sensitivity is also controlled by the Negative Threshold level (NTHR). Burst length and NTHR
interact. Normally burst lengths should be kept as short as possible to limit RF emissions, but NTHR should be kept
above 6 to reduce false detections due to external noise. The detection integrator mechanism also helps to prevent
false detections.
This function is programmed on a per-key basis. See Table 7-1 on page 45.
BL Typical values: 1, 2 (32, 48 pulses / burst)
BL Default value: 2 (48 pulses / burst)
BL Possible values: 0, 1, 2, 3 (16, 32, 48, 64 pulses)
6.2 Address 165 –188: NDIL, FDIL, AKS, WAKE
6.2.1 Detect Integrators – NDIL, FDIL
The NDIL parameter is used to enable and disable keys in the matrix and to provide signal filtering. To enable a key,
its NDIL parameter should be nonzero (that is, NDIL = 0 disables a key).
Table 6-3. BL
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
141 – 164 BL NDRIFT NTHR
Table 6-4. NDIL, FDIL
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
165 – 188 WAKE AKS FDIL NDIL
31
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
To suppress false detections caused by spurious events like electrical noise, the device incorporates a detection
integrator or DI counter mechanism. A per-key counter is incremented each time the key has exceeded its threshold
and is decremented each time the key does not exceed its threshold. When this counter reaches a preset limit the
key is finally declared to be touched.
The DI mechanism uses two counters. The first is the fast DI counter FDIL. When a key signal is first noted to be
below the negative threshold, the key enters fast burst mode. In this mode the burst is rapidly repeated for up to the
specified limit count of the fast DI counter. Each key has its own counter and its own specified fast-DI limit (FDIL),
which can range from 1 – 15. When fast burst is entered the QT device locks onto the key and repeats the acquire
burst until the fast-DI counter reaches FDIL or drops back to zero. After this the device resumes normal keyscanning
and goes on to the next key.
The Normal DI counter counts the number of times the fast-DI counter reached its FDIL value. The Normal DI
counter can only increment once per complete scan of all keys. Only when the Normal DI counter reaches NDIL
does the key become formally active.
The net effect of this is that the sensor can rapidly lock onto and confirm a detection with many confirmations, while
still scanning other keys. The ratio of fast to normal counts is completely user settable via the setups process. The
total number of required confirmations is equal to FDIL times NDIL.
If FDIL = 6 and NDIL = 2, the total detection confirmations required is 12, even though the device only
scanned through all keys twice.
The DI is extremely effective at reducing false detections at the expense of slower reaction times. In some
applications a slow reaction time is desirable. The DI can be used to intentionally slow down touch response in
order to require the user to touch longer to operate the key.
If FDIL = 1, the device functions conventionally. Each channel acquires once in rotation, and the normal detect
integrator counter (NDIL) operates to confirm a detection. Fast-DI is in essence not operational.
If FDIL 2, then the fast-DI counter also operates in addition to the NDIL counter.
If Signal NTHR: The fast-DI counter is incremented towards FDIL due to touch.
If Signal > NTHR then the fast-DI counter is decremented due to lack of touch.
Disabling a key: If NDIL = 0, the key becomes disabled. Keys disabled in this way are pared from the burst
sequence in order to improve sampling rates and thus response time. See Section 3. on page 6.
This function is programmed on a per-key basis. See Table 7-1 on page 45.
NDIL Typical values:1, 2
(2 or 4 confirmations; translation via LUT)
NDIL Default value: 2
(4 confirmations)
FDIL Typical values:2, 3
(4 or 6 confirmations; translation via LUT)
FDIL Default value: 3
(6 confirmations)
32
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.2.2 Adjacent Key Suppression Technology – AKS
This device incorporates adjacent key suppression (AKS) technology that can be selected on a per-key basis. AKS
technology permits the suppression of multiple key presses based on relative signal strength. This feature assists in
solving the problem of surface moisture which can bridge a key touch to an adjacent key, causing multiple key
presses. This feature is also useful for panels with tightly spaced keys, where a fingertip might inadvertently activate
an adjacent key.
AKS technology works for keys that are AKS-enabled anywhere in the matrix and is not restricted to physically
adjacent keys. The device has no knowledge of which keys are actually physically adjacent. When enabled for a key,
adjacent key suppression causes detections on that key to be suppressed if any other AKS-enabled key in the panel
has a more negative signal deviation from its reference during the DI process. Once a key reaches detect it stays in
detect as long as the touch remains, regardless of the signal strength on any other AKS enabled keys.
This feature does not account for varying key gains (burst length) but ignores the actual negative detection threshold
setting for the key. If AKS-enabled keys have different sizes, it may be necessary to reduce the gains of larger keys,
relative to smaller ones, to equalize the effects of AKS technology. The signal threshold of the larger keys can be
altered to compensate for this without causing problems with key suppression.
Adjacent key suppression works to augment the natural moisture suppression of narrow gated transfer switches
creating a more robust sensing method.
This function is programmed on a per-key basis. See Table 7-1 on page 45.
AKS Default value: 0 (Off)
6.2.3 Wake on Touch – WAKE
The device can be configured for full time wake-up from Sleep mode when specific keys are touched or released
using this feature, in order to improve response time after each key state change. This feature makes the CHANGE
pin go active on a key touch or key release (see Section 4.4 on page 17) enabling a host to sleep until a suitable
touch event occurs.
Each key has its own WAKE configuration bit so that any combination of keys can be configured for this function.
The time the part remains awake after the last key is released can also be configured in the setup block (see the
AWAKE feature, Section 7. on page 45).
Any key, even one where the WAKE feature is not enabled, prolongs the time the part remains awake once the part
is awake.
This function is programmed on a per-key basis. See Table 7-1 on page 45.
WAKE Default value: 1 (On)
WAKE Possible range: 0, 1 (Off, On)
Table 6-5. AKS
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
165 – 188 WAKE AKS FDIL NDIL
Table 6-6. WAKE
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
165 – 188 WAKE AKS FDIL NDIL
33
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.3 Calibrated Frequency Offset – CFO_1 and CFO_2
CFO_1 and CFO_2 are only used if FHM = 2.
If frequency hopping is enabled with FHM = 2, the QT1244 adjusts each key's reference at each frequency hop. The
amount of the adjustment must first be configured using the setups CFO_1 and CFO_2. These values are the
amounts by which reference levels need to be adjusted to track the signal changes when changing from the
frequency set by FREQ0 to the frequency set by FREQ1 or FREQ2. Each key uses its individual settings of CFO_1
and CFO_2.
Assuming FREQ0 defines the highest of the three selected frequencies, CFO_1 indicates the offset to be applied
when the frequency is changed between those defined by FREQ0 and FREQ1. CFO_2 indicates the offset to be
applied when the frequency is changed between those defined by FREQ0 and FREQ2. The QT1244 automatically
uses combinations of these offset values, as necessary, when switching between the three different frequencies.
The Calibrated Frequency Offset (CFO) values are used to adjust each key reference whenever a frequency hop
occurs, taking into account any differences in calibrated signal at the different frequencies. The device uses the
highest selected frequency as a point of reference and calculates offsets in calibrated signals at the other two
frequencies relative to the highest one.
The following explanation assumes that FREQ0 defines the highest frequency, for convenience of this discussion,
but that does not need to be the case.
Note: Configure FREQ0, FREQ1 and FREQ2 before trying to configure CFO_1 and CFO_2.
Command 0xFD (Low level calibration and offset) can be written to the Command Address (140) to automatically fill
out the CFO_1 and CFO_2 values.
Alternatively, CFO_1 and CFO_2 can be manually configured.
Once FREQ0, FREQ1 and FREQ2 have been configured for the three chosen frequencies, CFO_1 and CFO_2
should be loaded with the signal offsets appropriate for the signal shifts observed when switching between the three
different frequencies.
Follow these steps to determine the different signal levels for each key at each frequency and to determine
appropriate values to load into CFO_1 and CFO_2:
1. Configure FHM = 0 to disable frequency hopping temporarily.
2. Configure FREQ0 to select the highest of the chosen frequencies (FREQ0 should be set to a lower value than
FREQ1 and FREQ2).
3. Recalibrate all keys.
4. Make a note of each key reference level, Ref(k, f0).
5. Configure FREQ0 to select the second of the chosen frequencies.
6. Recalibrate all keys.
7. Make a note of each key reference level, Ref(k, f1).
8. Configure FREQ0 to select the last of the chosen frequencies.
9. Recalibrate all keys.
10. Make a note of each key reference level, Ref(k, f2).
Table 6-7. CFO_1 and CFO_2
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
189 – 212 CFO_1
213 – 236 CFO_2
34
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
11. Determine the difference in signal for each key in turn when the frequency changes from FREQ0 to FREQ1,
using the following equation:
Diff(k, f0, f1) = Ref(k, f0) – Ref(k, f1)
(this value will nearly always be positive. If it is negative, set Diff(k, f0, f1) to zero).
12. Store Diff(k, f0, f1) into the corresponding CFO_1 for each key k.
13. Determine the difference in signal at each key when the frequency changes from FREQ0 to FREQ2, using the
following equation:
Diff(k, f0, f2) = Ref(k, f0) – Ref(k, f2)
(this value will nearly always be positive. If it is negative, set Diff(k, f0, f1) to zero).
14. Store Diff(k, f0, f2) into the corresponding CFO_2 for each key k.
15. Configure FHM = 2.
These functions are programmed on a per-key basis. See Table 7-1 on page 45.
CFO_1/2 Default value: 0
CFO_1/2 Possible range: 0 – 255
6.4 Address 237: Negative Recal Delay – NRD
If an object unintentionally contacts a key, resulting in a detection for a prolonged interval, it is usually desirable to
recalibrate the key in order to restore its function, perhaps after a time delay of some seconds.
The Negative Recal Delay timer monitors detections. If a detection event exceeds the timer's setting, the key is
automatically recalibrated. After a recalibration has taken place, the affected key once again functions normally even
if it is still being contacted by the foreign object. This feature is set on a global basis using the NRD setup parameter.
NRD can be disabled by setting it to zero (infinite timeout) in which case the key never auto-recalibrates during a
continuous detection (but the host could still command it).
NRD above 0 is expressed in 0.5 s increments. Thus if NRD = 120, the timeout value is actually 60 seconds.
This function is programmed on a global basis. See Table 7-1 on page 45.
NRD Typical values: 20 – 60 (10 s – 30 s)
NRD Default value: 20 (10 s)
NRD Range: 0 – 255 (, 0.5 s – 127.5 s)
NRD Accuracy: to within ± 250 ms
Table 6-8. NRD
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
237 NRD
35
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.5 Address 238: SLEEP, MSYNC, NHYST, DEBUG
6.5.1 Sleep Duration – SLEEP
With the sleep feature enabled, the QT1244 is designed to sleep as much as possible to conserve power.
Periodically, the part wakes automatically, scans the keyboard matrix and then returns to sleep. The length of time
the part sleeps before automatically waking up can be configured to one of eight different values, via a look-up table.
As with most other setups, the look-up table index must be written to the setups (see Table 7-1 on page 45).
The QT1244 will not sleep while keys are in detect or while keys are calibrating.
If the sleep feature is disabled, the device never sleeps. Sleep might need to be disabled if the device is being used
in an FMEA or EN 60730 compliant design because all operations are stopped within the device while the part is
asleep and the host is not able to distinguish between faulty operation and EN 60730 counters appearing to run slow
because the part is intermittently sleeping.
It should also be noted that the configured drift compensation interval is valid only while the device is not sleeping,
and will effectively be stretched each time the device sleeps because the device is fully halted during sleep and
cannot perform drift compensation. If sleep is enabled, a shorter drift compensation interval may be required.
Note that when a key changes state, the CHANGE pin can be made to go active and the device can go into 'fast
mode' automatically if the WAKE feature is enabled on that key (see Section 6.2.3 on page 32).
This function is programmed on a global basis. See Table 7-1 on page 45.
SLEEP default value: 0 (sleep disabled)
SLEEP range: 0 – 7 (Off, 16 ms – 1s; translation via LUT)
6.5.2 Mains Sync – MSYNC
The MSYNC feature uses the M_SYNC pin.
External fields can cause interference leading to false detections or sensitivity shifts. Most fields come from AC
power sources. RFI noise sources are heavily suppressed by the low impedance nature of the QT circuitry itself.
Noise such as from 50Hz or 60Hz fields becomes a problem if it is uncorrelated with acquisition signal sampling.
Uncorrelated noise can cause aliasing effects in the key signals. To suppress this problem the M_SYNC input allows
bursts to synchronize to the noise source.
The noise sync operating mode is set by parameter MSYNC in setups.
The synchronization only occurs at the burst for the lowest numbered enabled key in the matrix. The device waits for
the synchronization signal for up to 100 ms after the end of a preceding full matrix scan, then when a negative sync
edge is received, the matrix is scanned in its entirety again.
Table 6-9. SLEEP
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
238 DEBUG NHYST MSYNC SLEEP
Table 6-10. MSYNC
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
238 DEBUG NHYST MSYNC SLEEP
36
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
The sync signal drive should be a buffered logic signal, or perhaps a diode-clamped signal, but never a raw AC
signal from the mains. The device synchronizes to the falling edge.
Since noise synchronization is highly effective and inexpensive to implement, it is strongly advised to take advantage
of it anywhere there is a possibility of encountering low frequency (50/60 Hz) electric fields. Atmel QmBtn software
can show such noise effects on signals, and hence assist in determining the need to make use of this feature.
If the synchronization feature is enabled but no synchronization signal exists, the sensor continues to operate but
with a delay of 100 ms before the start of each matrix scan, and hence has a slow response time.
This function is programmed on a global basis. See Table 7-1 on page 45.
MSYNC Default value: 0 (Off)
MSYNC Possible range: 0, 1 (Off, On)
6.5.3 Negative Hysteresis – NHYST
The device employs programmable hysteresis levels of 6.25%, 12.5%, 25%, or 50% The hysteresis is a percentage
of the distance from the threshold level back towards the reference, and defines the point at which a touch detection
drops out. A 12.5% hysteresis point is closer to the threshold level than to the signal reference level.
Hysteresis prevents chatter and works to make key detection more robust. Hysteresis is only used after the key has
been declared to be in detection, in order to determine when the key should drop out.
Excessive amounts of hysteresis can result in stuck keys that do not release. Conversely, low amounts of hysteresis
can cause key chatter due to noise or minor amounts of finger motion.
NHYST is set using one global value which is applied across all keys.
See Table 7-1 on page 45.
NHYST Typical values: 0, 1 (6.25%, 12.5%)
NHYST Default value: 1 (12.5%)
NHYST Possible range: 0, 1, 2, 3 (6.25%, 12.5%, 25%, 50%)
6.5.4 DEBUG Output
The QT1244 supports a Debug output interface which is useful during product development. (See Appendix C. on
page 55). If DEBUG is enabled, the device streams one frame of data out of the Debug port after each keyscan
cycle. The value set for DEBUG dictates the number of bytes in the data frame.
When DEBUG is enabled it impacts the key response time because the next keyscan cycle is delayed until the
debug frame has been fully transmitted. The oscilloscope sync and the debug data output share an I/O pin, and will
interfere with each other if both are enabled at the same time. Ensure SSYNC and DEBUG are not enabled
simultaneously.
Table 6-11. NHYST
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
238 DEBUG NHYST MSYNC SLEEP
Table 6-12. DEBUG
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
238 DEBUG NHYST MSYNC SLEEP
37
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
This function is programmed on a global basis. See Table 7-1 on page 45.
DEBUG Default value: 0 (off)
DEBUG Possible range: 0,1, 2, 3 (off, 32, 64, 128 bytes)
6.6 Address 239: Awake Timeout – AWAKE
The AWAKE timeout feature determines the period of time that elapses from the last key release before the part tries
to sleep.
Once the part has been awakened by a touch change on a WAKE enabled key, key response is fast while the
keyboard is in use, until sleep mode is resumed. Sleep mode will not resume unless there is a period of key
inactivity. This period is defined by the AWAKE setup. Sleep mode is inhibited while keys are in detect.
Note: If the sleep feature has been disabled, the device never sleeps and the AWAKE setup has no effect.
The AWAKE period can be configured to a value between 100 ms and 25.5 s, in increments of 100 ms.
This function is programmed on a global basis. See Table 7-1 on page 45.
AWAKE default value: 25 (2.5 s)
AWAKE range: 1 – 255 (100 ms – 25.5 s)
AWAKE Timeout accuracy: to within ±50 ms
6.7 Address 240: Drift Hold Time – DHT
Drift Hold Time (DHT) is used to inhibit drift on all keys while one or more keys are activated. DHT defines the length
of time the drift is halted after a key detection.
This feature is particularly useful in cases of high-density keypads where touching a key or hovering a finger over the
keypad would cause untouched keys to drift, and therefore create a sensitivity shift, and ultimately inhibit any touch
detection.
DHT can be configured to a value of between 100 ms and 25.5 s, in increments of 100 ms. Setting this parameter to
0 disables this feature and the drift compensation on any key is not dependent on the state of other keys.
This function is programmed on a global basis. See Table 7-1 on page 45.
DHT default value: 0 (Off)
DHT range: 0 – 255 (Off, 100 ms – 25.5 s)
Table 6-13. AWAKE
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
239 AWAKE
Table 6-14. DHT
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
240 DHT
38
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.8 Address 241: PDRIFT, SSYNC
6.8.1 Positive Drift Compensation – PDRIFT
See Section 6.1.2 on page 29.
6.8.2 Oscilloscope Sync – SSYNC
The S_SYNC pin can output a positive pulse oscilloscope sync that brackets the burst of a selected key. Only one
burst can output a sync pulse as determined by the setups parameter SSYNC.
This feature is invaluable for diagnostics. Without it, observing signals clearly on an oscilloscope for a particular burst
is very difficult. This function is supported in Atmel QmBtn PC software.
The oscilloscope sync and the debug data output share an I/O pin, and will interfere with each other if both are
enabled at the same time. Ensure SSYNC and DEBUG are not enabled simultaneously.
This function is programmed on a global basis. See Table 7-1 on page 45.
SSYNC Default value: 0 (Key 0)
SSYNC Possible range: 0 – 23 (Key n), 24 – 31 (Off)
6.9 Address 242 –243: LSL, KGTT
6.9.1 Lower Signal Limit – LSL
This setup determines the lowest acceptable value of signal level for all keys. If any key reference level falls below
this value, the device declares an error condition in the key data set (see Section 5.8 on page 26) and in the device
status byte (see Section 5.5 on page 24).
Testing is required to ensure that there are adequate margins in this determination. Key size, shape, panel material,
burst length, and dwell time all factor into the detected signal levels.
This parameter occupies 2 bytes (11 bits) of the setups table. Address 242 occupies the LSB, 243 the MSB.
Table 6-15. PDRIFT
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
241 SSYNC PDRIFT
Table 6-16. SSYNC
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
241 SSYNC PDRIFT
Table 6-17. LSL
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
242 LSL
243 KGTT LSL
39
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
This function is programmed on a global basis. See Table 7-1 on page 45.
LSL Default value: 100 (recommended value)
LSL Possible range: 0 – 2047
6.9.2 Key Gain Test Threshold – KGTT
The Key Gain test takes a special sample from a key using half the usual burst length and compares the resulting
value against the key normal signal. The test passes if the signal has decreased by the Key Gain Test Threshold
(KGTT). This means that the following equation must be valid for the test to pass. Each enabled key is tested.
Normal Signal – Test Signal > KGTT
The Key Gain Test Threshold can be configured to a value between 4 and 60, via LUT (see Table 7-1 on page 45).
KGTT occupies 4 bits only, sharing the same word as the Lower Signal Limit.
This function is programmed on a global basis. See Table 7-1 on page 45.
KGTT Default value: 6 (28)
KGTT Possible range: 0 – 15 (4 – 64)
6.10 Address 244: DWELL, RIB, THRM, FHM
6.10.1 Dwell Time – DWELL
The Dwell parameter in setups causes the acquisition pulses to have differing charge capture durations. Generally,
shorter durations provide for enhanced surface moisture suppression, while longer durations are required where the
keypanel design includes higher-resistance tracks such as silver and ITO. Longer durations are also usually more
compatible with EMC requirements. Longer dwell times permit the use of larger series resistors in the X and Y lines
to suppress RFI effects, without compromising key gain.
This function is programmed on a global basis. See Table 7-1 on page 45.
DWELL Default value: 0 (125 ns)
DWELL Possible range: 0 – 7 (125 ns – 4.5 µs)
Table 6-18. KGTT
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
242 LSL
243 KGTT –LSL
Table 6-19. DWELL
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
244 FHM FHM THRM RIB DWELL
40
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.10.2 Restart Interrupted Burst – RIB
The RIB parameter in setups allows a burst to be interrupted, and restarted, by host communications over the serial
bus. The QT1244 has limited processing resources available such that a burst and host communication cannot both
be serviced simultaneously. One must give way to the other. This setup lets the designer prioritize one over the
other.
If RIB is configured on, a burst can be interrupted by a host communication, and is automatically restarted.
If RIB is configured off, bursts will not be restarted. The interruption may introduce some signal noise.
This function is programmed on a global basis. See Table 7-1 on page 45.
RIB Default value: 0 (off)
RIB Possible range: 0, 1 (off, on)
6.10.3 Threshold Multiplier – THRM
It is sometimes useful to be able to operate the device with much higher detect thresholds than can be set with
NTHR and PTHR alone. The Threshold Multiplier (THRM) is a multiplier, set via LUT, which extends the range of
NTHR and PTHR considerably. The operating detect threshold for a key is arrived at by multiplying the threshold for
that key (set with NTHR or PTHR) by the THRM multiplier. Note that the detect threshold range is extended at the
expense of the step size. Figure 6-22 shows the extended threshold range for each value of THRM.
This function is programmed on a global basis. See Table 7-1 on page 45.
THRM Default value: 0 (x1)
THRM Possible range: 0, 1, 2, 3 (x1, x2, x4, x8)
Table 6-20. RIB
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
244 FHM FHM THRM RIB DWELL
Table 6-21. THRM
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
244 FHM FHM THRM RIB DWELL
Table 6-22. Extended Detect Threshold
THRM Multiplier Extended Threshold
0x1 4–18
1x2 8–36
2x4 16–72
3x8 32 – 144
41
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.10.4 Frequency Hopping Mode – FHM
Frequency hopping can be disabled altogether or enabled to one of three different active modes with FHM.
FHM = 0 – Frequency hopping disabled
If frequency hopping is disabled, the QT1244 always uses the same frequency, defined by FREQ0.
FHM = 1 or 2
If frequency hopping is enabled with FHM = 1 or FHM = 2, the QT1244 continually monitors the noise across
all keys and switches frequencies dynamically to try and find the frequency with the lower noise amplitude. Up
to three different frequencies can be selected using FREQ0, FREQ1 and FREQ2. If frequency hopping
between only two frequencies is desired, the same frequency should be selected for two of these functions.
The signal levels can vary slightly with frequency, and so the reference for each key might need correspond-
ing adjustment, in one way or another, during frequency hopping to compensate and to prevent false
detections and unresponsive keys. The mechanism used to adapt the references is different depending on the
frequency hop mode set.
FHM = 1 – Calibrate after hop
If frequency hopping is enabled with FHM = 1, the QT1244 compensates for the variations in signal level by
recalibrating each key immediately after each frequency hop. A negative aspect to this mode is the danger
that a key is being touched at the time of the frequency hop. If this were to happen, the touched key would be
recalibrated to the touching finger and the detect would be cancelled. However, on subsequent removal of the
touch (for a time greater than or equal to the PRD function) the key would be recalibrated once again and the
detect flagged when the touch is re-established.
FHM = 2 – Adjust each key reference during hop
If frequency hopping is enabled with FHM = 2, the QT1244 adjusts each key reference at each frequency hop.
The amount of the adjustment must first be configured using the setups CFO_1 and CFO_2.
The QT1244 will not switch the frequency during calibration if FHM is set at 1 or 2.
FHM = 3 – Frequency sweep
If frequency hopping is enabled with FHM = 3, the QT1244 repeatedly sweeps the sampling frequency
through a range bounded by two frequencies defined by FREQ0 and FREQ1. The frequency is changed after
each and every matrix scan. The QT1244 is shipped with the frequency hopping mode preset at FHM = 3.
In this hop mode, the reference for each key is not adjusted as the frequency is changed, so the signal levels
must not vary significantly over the selected frequency range otherwise false detects or unintentional touch
dropouts could occur. Variations in signal levels can be limited by restricting the frequency sweep range. If the
signal variation for any particular key is found to be too great, the range of frequencies should be narrowed by
decreasing the difference in values set at FREQ0 and FREQ1.
This function is programmed on a global basis. See Table 7-1 on page 45.
FHM Default value: 3 = Frequency sweep
FHM Possible range: 0 – 3 (0 = off
1 = Calibrate all keys after hop
2 = Adjust each key's reference during hop
3 = Frequency sweep)
Table 6-23. FHM
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
244 FHM THRM RIB DWELL
42
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
6.11 Address 245: Frequency 0 – FREQ0
FREQ0 is used in all frequency hopping modes and even if frequency hopping is disabled. In all modes, it defines the
idle time between pulses in the burst. Larger values yield longer times between pulses and thus a lower fundamental
frequency.
FHM = 0 – Frequency hopping disabled
If frequency hopping is disabled, the QT1244 always uses the same frequency, defined by FREQ0. Frequency
hopping might not be desirable in all applications and it might be more appropriate to preselect a burst fre-
quency at the factory which is known not to coincide with other operating frequencies within the end product or
other frequencies in the operating environment. In such cases, FHM can be set at zero, and the burst fre-
quency set with FREQ0.
With DWELL set at the default (minimum), the fundamental frequency can be set in the range 10.5 kHz – 590
kHz. The frequency for a specific DWELL and FREQ0 combination should be measured using an oscilloscope
(temporarily disable frequency hopping to make the measurement easier).
FHM = 1 or FHM = 2 – Frequency hopping between three frequencies
If frequency hopping is enabled with FHM = 1 or 2, the QT1244 can hop between three frequencies configured
using FREQ0, FREQ1 and FREQ2.
FHM = 3 – Frequency sweep
With FHM = 3, the QT1244 sweeps a range of frequencies, with the upper frequency boundary (shortest idle
time) defined by FREQ0. FREQ1 must be set to a value greater than FREQ0. The behavior is otherwise
undefined.
FREQ0 = 0 is invalid and must be avoided when FHM = 3.
This function is programmed on a global basis.
FREQ0 Default value: 1 (delay cycle)
FREQ0 Possible range: 1 – 256 (Highest frequency to Lowest frequency, 256 is obtained with setting 0)
6.12 Address 246: Frequency1 – FREQ1
FREQ1 is only used if FHM is set to a non-zero value. It is not used if frequency hopping is disabled with FHM = 0.
With FHM = 1 or FHM = 2, FREQ1 allows configuration of one of three operating frequencies by defining the idle
time between pulses in the burst. Larger values yield longer times between pulses and thus a lower fundamental
frequency.
With FHM = 3, the QT1244 sweeps a range of frequencies, with the lower frequency boundary (longest idle time)
defined by FREQ1. FREQ1 must be set to a value greater than FREQ0. The behavior is otherwise undefined.
This function is programmed on a global basis. See Table 7-1 on page 45.
Table 6-24. FREQ0
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
245 FREQ0
Table 6-25. FREQ1
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
246 FREQ1
43
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
FREQ1 Default value: 6 (delay cycles)
FREQ1 Possible range: 1 – 256 (Highest frequency to Lowest frequency, 256 is obtained with setting 0)
6.13 Address 247: Frequency2 – FREQ2
FREQ2 is only used if FHM = 1 or FHM = 2. It is not used if FHM is set to zero or 3.
With FHM = 1 or FHM = 2, FREQ2 allows configuration of one of three operating frequencies by defining the idle
time between pulses in the burst. Larger values yield longer times between pulses and thus a lower fundamental
frequency.
This function is programmed on a global basis. See Table 7-1 on page 45.
FREQ2 Default value: 63 (delay cycles)
FREQ2 Possible range: 1 – 256 (Highest frequency to Lowest frequency, 256 is obtained with setting 0)
6.14 Address 248: NSTHR, NIL
6.14.1 Noise Threshold – NSTHR
NSTHR is only used if FHM = 1 or FHM = 2. It is not used if FHM is set to 0 or 3.
When FHM = 1 or 2, the QT1244 considers a hop to one of the other frequencies when the noise at the current
frequency consistently exceeds the threshold configured with NSTHR. NSTHR is used by the frequency hopping
algorithms to determine if a signal delta should be considered as noise. A delta, of either polarity, greater than or
equal to NSTHR is considered as possible noise and forces the Noise Integrator counter to be incremented.
This function is programmed on a global basis. See Table 7-1 on page 45.
NSTHR Default value: 2 (11 counts)
NSTHR Possible range: 0 – 15 (5 – 50 counts)
6.14.2 Noise Integrator Limit – NIL
NIL is only used if FHM = 1 or FHM = 2. It is not used if FHM is set to zero or 3.
Table 6-26. FREQ2
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
247 FREQ2
Table 6-27. NSTHR
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
248 NIL NSTHR
Table 0-1. NIL
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
248 NIL NSTHR
44
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
The QT1244 considers a hop to one of the other frequencies when the noise at the current frequency consistently
exceeds the threshold configured with NSTHR. To prevent true touch events and other brief signal anomalies being
considered as noise, the QT1244 employs counters, termed noise integrators, to track the number of signal deltas
that exceed the noise threshold. It maintains two such counters, one for positive deltas and one for negative deltas.
The QT1244 considers a frequency hop only if both these counters reach the noise integrator limit (NIL). The
counters are reset to zero at the end of each matrix scan.
This mechanism provides a robust way of detecting strong noise while suppressing unnecessary frequency hopping.
This function is programmed on a global basis. See Table 7-1 on page 45.
NIL Default value: 3
NIL Possible range: 0 – 15 (0 – 2 = factory use only)
6.15 Address 249 – 250: Host CRC – HCRC
The setups block terminates with a 16-bit CRC (HCRC) of the entire setups block, which the host must calculate and
upload. The setups CRC is computed repeatedly and checked against this value. An error is reported in the Device
Status (see Section 5.5 on page 24) if the two values do not match. The formula for calculating this CRC is shown in
Appendix B. The LSB occupies address 249 and the MSB occupies address 250.
6.16 Positive Recalibration Delay – PRD
A recalibration can occur automatically if the signal swings more positive than the positive threshold level. This
condition can occur if there is positive drift but insufficient positive drift compensation, or, if the reference moved
negative due to a NRD autorecalibration, and thereafter the signal rapidly returned to normal (positive excursion).
As an example of the latter, if a foreign object or a finger contacts a key for period longer than the Negative Recal
Delay (NRD), the key is by recalibrated to a new lower reference level. Then, when the condition causing the
negative swing ceases to exist (for example, the object is removed) the signal can suddenly swing back positive to
near its normal reference.
It is almost always desirable in these cases to cause the key to recalibrate quickly so as to restore normal touch
operation. The time required to do this is governed by PRD. In order for this to work, the signal must rise through the
positive threshold level PTHR continuously for the PRD period.
After the PRD interval has expired and the autorecalibration has taken place, the affected key once again functions
normally.
PRD Accuracy: to within ± 50 ms
Delay: PRD is fixed at 1000 ms for all keys, and cannot be altered.
Table 6-28. HCRC
Address Bit 7 Bit 6 Bit 5 Bit 4 Bit 3 Bit 2 Bit 1 Bit 0
249 – 250 HCRC
45
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
7. Lookup Table
Typical values: For most touch applications, use the values shown in the boxed cells. Bold text items indicate default settings. The number to send to the
device is the index number in the leftmost column (0 – 15), not numbers from the table. The device uses a lookup table internally to translate the indices 0 – 15
to the parameters for each function.
NRD is an exception: It can range from 0 – 255 which is translated from 1 = 0.5 s to 255 = 127.5 s, in increments of 0.5 s, with zero = infinity.
AWAKE is an exception: it can range from 1 – 255 which is translated from 1 = 0.1 s to 255 = 25.5 s, in increments of 0.1 s. Zero is illegal.
DHT is an exception: it can range from 0 – 255 which is translated from 1 = 0.1 s to 255 = 25.5 s, in increments of 0.1 s. Zero disables DHT.
CFO_1 and CFO_2 are exceptions. They can range from 0 – 255.
Table 7-1. Lookup Table
Parameter
Index
NTHR/PTHR
counts
NDRIFT
secs
NDIL
counts
FDIL
counts
WAKE BL
pulses
AKS CFO_1 CFO_2 PDRIFT
secs
NRD
secs
SSYNC SLEEP
ms
NHYST MSYNC AWAKE
secs
DHT
secs
Per Key Global
0 4 0.2 Key off 1Off 16
Off 0 0
0.2 0 (infinite)
0 (Key 0) Off
6.25%
Off unused Off
1 6 0.4 2- 2
On
32 On 0
–
255 0
–
255 0.4 0.5
–
127 Key 0 – 23 16
12.5%
On 0.1
–
25.5 0.1
–
25.5
2 8 0.8
4
4
48
0.8
20 (10 s)
32 25%
25 (2.5 s)
3
10
1.2 6
6
64 1.2 64 50%
412
2
8 8
2
125
514 3.3 10 10 3.3 250
616 612 12 6500
718 10 15 15 10 1,000
8
9
10
11
12
13
14
15
46
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
LSL is an exception: it can range from 0 – 2047.
FREQ0, FREQ1, FREQ2 are exceptions. They can range from 0 – 255, with zero = 256 counts.
Table 7-2. Lookup Table
Index LSL KGTT DWELL
ns
RIB DEBUG
bytes
FHM
FREQ0
delay
cycles
FREQ1
delay
cycles
FREQ2
delay
cycles
NSTHR
counts
NIL
counts
THRM
multiplier
00–2047 4
125 Off Off
Off 256 (lowest
frequency)
256 (lowest
frequency)
256 (lowest
frequency) 5
0
(factory use
only)
1
1
100
8375 On 32 Recal on hop 1 – 255 1 – 255 1 – 255 8
1
(factory use
only)
2
212 750 64 Adjust Ref
on hop
1 6 63 11
2
(factory use
only) 4
316 1125 128
Sweep
14
3
420 1500 17 4 8
524 2250 20 5
6
28
3375 23 6
732 4500 26 7
836 29 8
940 32 9
10 44 35 10
11 48 38 11
12 52 41 12
13 56 44 13
14 60 47 14
15 64 50 15
47
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8. Specifications
8.1 Absolute Maximum Specifications
8.2 Recommended Operating Conditions
8.3 DC Specifications
Parameter Specification
Operating temp –40oC to +105oC
Storage temp –55oC to +125oC
Vdd 6V
Max continuous pin current, any control or drive pin ±10 mA
Short circuit duration to ground, any pin infinite
Short circuit duration to Vdd, any pin infinite
Voltage forced onto any pin –0.5 V to (Vdd + 0.5) V
EEPROM setups maximum writes 100,000 write cycles
CAUTION: Stresses beyond those listed may cause permanent damage to the device. This is a stress rating only and
functional operation of the device at these or other conditions beyond those indicated in the operational sections of this
specification is not implied. Exposure to absolute maximum specification conditions for extended periods may affect
device reliability.
Parameter Specification
Vdd +3 V to +5 V
Supply ripple + noise 20 mV p-p max
Cx transverse load capacitance per key 2–20pF
Cs = 4.7 nF, Rs = 470 k, Ta (Ambient Temperature)= recommended range, unless otherwise noted
Parameter Description Min Typ Max Units Notes
Iddr Average supply current,
running –6.6 –mA Vdd = 5.0 V
Idds Average supply current,
sleeping –6.00 –µA Vdd = 5.0 V
Vil Low input logic level – – 0.3 × Vdd V
Vih High input logic level 0.6 × Vdd – – V
Vol Low output voltage – – 0.6 VVdd = 3.0 V
Voh High output voltage 2.3 – – V Vdd = 3.0 V
Iil Input leakage current ––1µA
48
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8.4 Timing Specifications
Note: The values quoted in the above table are valid with the default settings, except where noted, and in the absence
of serial communications.
Ar Acquisition resolution –10 –bits
Rp Internal pull-up resistors 20 –50 kCHANGE pin
Rrst Internal RST pull-up resistor 30 –60 k
Parameter Description Min Typ Max Units Notes
TBS Burst spacing –
290
320
350
390
–µs
BL = 16
BL = 32
BL = 48
BL = 64
All with FHM = 0
Fc Burst center frequency 9.7 –590 kHz
Cs = 4.7 nF, Rs = 470 k, Ta (Ambient Temperature)= recommended range, unless otherwise noted
Parameter Description Min Typ Max Units Notes
49
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8.5 Mechanical Dimensions
8.5.1 QFN
$
50
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8.5.2 TQFP
51
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8.6 Part Marking
8.6.1 QFN
8.6.2 TQFP
1
32
Pin 1 ID
MU 3
QT1244
ATMEL
YYWW
Atmel Logo
Shortened
Part Number
‘3’=
Code
Revision 3
YY = Last two digits of the mark year
WW = Test workweek
Date code Description
1
32
Pin 1 ID
AU 3
QT1244
ATMEL
YYWW
Atmel Logo
Shortened
Part Number
‘3’=
Code
Revision 3
YY = Last two digits of the mark year
WW = Test workweek
Date code Description
52
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
8.7 Part Number
The part number comprises:
AT = Atmel
42 = Touch Business Unit
QT = Charge-transfer technology
1244 = (1) Keys (24) number of channels (4) variant number
AU = TQFP chip
MU = QFN chip
R = Tape and reel
8.8 Moisture Sensitivity Level (MSL)
Part Number Description
AT42QT1244-AU 32-pin 7 × 7 mm TQFP RoHS compliant IC
AT42QT1244-AUR 32-pin 7× 7 mm TQFP RoHS compliant IC - Tape and reel
AT42QT1244-MU 32-pad 5 × 5 mm QFN RoHS compliant IC
AT42QT1244-MUR 32-pad 5 × 5 mm QFN RoHS compliant IC - Tape and reel
MSL Rating Peak Body Temperature Specifications
MSL3 260oCIPC/JEDEC J-STD-020
53
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Appendix A. I2C Operation
A.1 Clock Stretching
The device supports clock stretching in accordance with the I2C specification. It may also instigate a clock stretch if a
communications event happens during a period when the device is busy internally. The maximum clock stretch is
4ms.
The device has an internal bus monitor that can reset the internal I2C hardware if SDA or SCL is stuck low for more
than 100 – 200 ms. If a clock stretch of more than 100 – 200 ms is seen by the device, any ongoing transfers may be
corrupted.
54
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Appendix B. 16-bit CRC Software C Algorithm
// 16 bits crc calculation. Initial crc entry value must be 0.
// The message is not augmented with 'zero' bits.
// polynomial = X16 + X15 + X2 + 1
// data is an 8 bit number, unsigned
// crc is a 16 bit number, unsigned
// repeat this function for each data block byte, folding the result
// back into the call parameter crc
unsigned long sixteen_bit_crc(unsigned long crc, unsigned char data)
{ unsigned char index;// shift counter
crc ^= (unsigned long)(data) << 8;
index = 8;
do // loop 8 times
{ if(crc & 0x8000)
{ crc= (crc << 1) ^ 0x1021;
}
else
{ crc= crc << 1;
}
} while(--index);
return crc;
}
A CRC calculator for Windows® is available free of charge from Atmel.
55
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Appendix C. DEBUG Output
The QT1244 includes a debug interface which may be used for observing many internal operating variables, in real
time, even while the part is actively communicating with a host over the I2C serial communications bus. The Debug
interface provides a useful aid during product development and uses two pins, one for clock and one for data, to
stream data out of the part.
If DEBUG is enabled in the setups (see Section 6.5.4 on page 36), the QT1244 streams a 32-byte, 64-byte or
128-byte frame of data out of the two Debug pins after each keyscan cycle. The transmission format is compatible
with Atmel's Plug-in USB card (Part Number 9206) and the data can be viewed using Atmel's Hawkeye PC software
(contact Atmel for information). Table 8-1 shows the Debug interface details.
The meaning of each byte in the frame is described in Table 8-2.
Table 8-1. Debug Interface
Debug Interface
Debug Clock output Pin 9
Debug Data output Pin 11
Data valid Clock high
Data changing Clock low
Clock frequency 400 kHz
Blank time between byte transmissions 6µs
Frame transmission time (min) 3.3 ms
Byte transmission order Most significant bit first
Table 8-2. Debug Output Data Frame
Frame Byte # Description Access
0Reserved Read
1Reserved Read
2100 ms counter (IEC/EN60730) Read
3Signal fail counter (IEC/EN60730) Read
4Matrix Scan counter (IEC/EN60730) Read
5Device Status. Collection of bit flags Read
6Detect status for keys 0 – 7, one bit per key Read
7Detect status for keys 8 – 15, one bit per key Read
8Detect status for keys 16 – 23, one bit per key Read
9Reserved Read
10 Current frequency Read
11 Current pulse spacing Read
12–15 Data for key 0 See Table 5-6 on page 26 for details Read
16–19 Data for key 1 Read
56
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Bytes 0 – 107 in Table 8-2 are identical to bytes 0 – 107 in the memory map (Table 5-1 on page 19) which may be
accessed through the serial communication bussee Section 5.1 on page 19 for more details). Bytes 108 – 127 are
reserved and should be ignored.
20–23 Data for key 2 Read
24–27 Data for key 3 Read
28–31 Data for key 4 Read
32–35 Data for key 5 Read
36–39 Data for key 6 Read
40–43 Data for key 7 Read
44–47 Data for key 8 Read
48–51 Data for key 9 Read
52–55 Data for key 10 Read
56–59 Data for key 11 Read
60–63 Data for key 12 Read
64–67 Data for key 13 Read
68–71 Data for key 14 Read
72–75 Data for key 15 Read
76–79 Data for key 16 Read
80–83 Data for key 17 Read
84–87 Data for key 18 Read
88–91 Data for key 19 Read
92–95 Data for key 20 Read
96–99 Data for key 21 Read
100 – 103 Data for key 22 Read
104 – 107 Data for key 23 Read
108 – 127 Reserved
Table 8-2. Debug Output Data Frame (Continued)
Frame Byte # Description Access
57
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Appendix D. Conducted Noise Immunity
Electrically conducted noise can increase the noise on the touch signals considerably and can lead to both false
detects and missed touches. There is a specific test for conducted immunity, as part of typical EMC testing, which
injects noise into the device under test across a broad range of frequencies and with significant amplitude. This test
is designed to test the immunity of devices and products against environmental noise that is generated by
commercial radio transmitters and other sources. Passing this test can be a challenge and might appear daunting,
but with good design practice from the outset coupled with fine tuning of the QT1244 frequency hopping Setups and,
designs based on the QT1244 show excellent noise immunity and can pass the conducted noise test by some
margin.
From the outset, the design must target achieving the best possible touch sensitivity while minimizing noise. A clean
design can achieve excellent signal delta on touch, but its easy to destroy this quality by pushing the overall product
requirements too far and poor attention to detail during the design. The more imperfections that exist in a design, the
more the sensitivity will be eroded and the conducted immunity performance with it.
A good target for signal delta on touch is 100 or even 150 counts.
Best touch performance is achieved with thinner overlay panels, optimum key electrode design, clean tracking
between the QT1244 and the keys with low stray capacitance between X and Y traces and low stray capacitance
from X and Y traces to ground, coupled with appropriate selection of the external matrix components. See the
documents referred to in “Associated Documents” on page 59 for further details on best practice for designing a
touch sensor interface.
The following paragraphs make some recommendations for initial values or type for the additional matrix
components to accompany a QT1244 design together with some Setups the have been found to produce good noise
immunity results.
Cs - NPO/COG
Use COG type for the charge sample capacitors (Cs0 - Cs7). This type of capacitor exhibits excellent stability,
albeit at a higher cost, and is preferred over X7R and other types. Lower values can be used to reduce the
cost or help with availability provided the charge transfer is not too high such as to saturate Cs. An excellent
layout with low stray capacitance will typically allow Cs to be reduced as low as 1 nF.
Rs - 1 M
Increasing the digital conversion ramp resistors (Rs0 – Rs7) increases sensitivity. The optimum selection for
Rs is one that balances highest achievable sensitivity against conversion time and other undesirable side
effects such as increased noise and possibly some reduction in temperature stability. Any increase in noise is
typically insignificant compared to conducted noise observed during conducted immunity EMC testing.
Rs can be increased as high as 5 M, although this is probably extreme in most cases. A good initial value is
1M. This value increases sensitivity to a good level with acceptable additional signal noise and often
achieves the best compromise in Signal-to-Noise Ratio (SNR). The increased sensitivity allows a higher
detect threshold to be employed, preventing noise spikes exceeding the threshold and thus preventing false
detects.
Ry - 47 k
Immunity to conducted electrical noise can be increased considerably by increasing the series resistors in the
matrix Y lines (Ry0 - Ry7) at the expense of moisture tolerance. There is a practical limit to how far their value
can be increased because higher values must be accompanied by longer dwell times, possibly in excess of
the longest settings available in the device. In any case excessively long dwell times would slow the response
time intolerably.
DWELL
Increasing the Ry values alone results in a reduction in charge transfer and sensitivity. This can be recovered
by increasing the charge transfer time, or dwell time, through higher DWELL settings. To achieve the optimum
DWELL setting, start with the maximum value and observe the reference values. Then reduce the dwell, con-
58
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
tinue to observe the reference values, and choose the lowest DWELL setting where the reference values are
not significantly reduced from their maximum.
Longer dwell times also result in reduced moisture tolerance, and so a careful balance might be necessary
between these conflicting requirements.
NTHR & THRM
Together NTHR (and PTHR) and THRM allow the threshold to be set. The threshold should be set as high as
possible to prevent false detects but should not be set so high that true touches are not detected. The Atmel
QmBtn software can be used to observe the signals and signal delta on touch, and used to determine the opti-
mum threshold settings for a specific design.
The threshold will typically need setting at a lower value to accommodate noise than is evident when testing in
the absence of noise. This may initially seem counter intuitive until consideration is given to the fact that heavy
noise can cause undesirable detect dropouts as well as missed touch events. The threshold should be set as
high as possible, but not so high that the noise causes detect dropouts while the key is still touched.
Frequency Hopping - FHM = 3 (sweep)
The immunity of QT1244-based designs to higher noise amplitude can be increased by configuring the fre-
quency hopping with 3 different frequencies and using FHM = 2. Choosing the 3 frequencies is a process of
trial and error and varies from design to design, but a considerable increase in tolerable noise amplitude can
be achieved once 3 appropriate frequencies have been identified.
However, frequency hopping using sweep mode (FHM = 3) is much easier to configure and may provide suffi-
cient immunity for many applications and products.
59
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Associated Documents
The following documents are a good source of general information regarding design and development of Atmel
touch sensors and should be studied before starting the development of a new touch interface.
QTAN0079 – Buttons, Sliders and Wheels Sensor Design Guide
Refer to this application note for details of different possible X/Y electrode designs and patterns, and the
optimum geometry to match the panel thickness.
QTAN0062 – Qtouch and Qmatrix Sensitivity Tuning for Keys, Sliders and Wheels
AVR3000 Qtouch Conducted Immunity
Refer to this guide for further information on immunity from conducted noise.
These documents are available on the Atmel website (www.atmel.com).
60
QT1244 [PRELIMINARY DATASHEET]
9631EX–AT42–12/13
Revision History
Revision No. History
Revision AX and BX – 2011 Initial release for customer
Revision CX – August 2011 Firmware version 1. Changes to datasheet clarifying text
Revision DX – March 2013 Firmware version 3.0
Revision EX – December 2013 Updated part marking diagram and part numbering section
Atmel Corporation
1600 Technology Drive
San Jose, CA 95110
USA
Tel: (+1) (408) 441-0311
Fax: (+1) (408) 487-2600
www.atmel.com
Atmel Asia Limited
Unit 01-5 & 16, 19F
BEA Tower, Millennium City 5
418 Kwun Tong Roa
Kwun Tong, Kowloon
HONG KONG
Tel: (+852) 2245-6100
Fax: (+852) 2722-1369
Atmel München GmbH
Business Campus
Parkring 4
D-85748 Garching bei München
GERMANY
Tel: (+49) 89-31970-0
Fax: (+49) 89-3194621
Atmel Japan G.K.
16F Shin-Osaki Kangyo Bldg
1-6-4 Osaki, Shinagawa-ku
Tokyo 141-0032
JAPAN
Tel: (+81) (3) 6417-0300
Fax: (+81) (3) 6417-0370
© 2009 – 2013 Atmel Corporation. All rights reserved. / Rev.: 9631EX–AT42–12/13
Disclaimer: The information in this document is provided in connection with Atmel products. No license, express or implied, by estoppel or otherwise, to any intellectual property right is granted by this
document or in connection with the sale of Atmel products. EXCEPT AS SET FORTH IN THE ATMEL TERMS AND CONDITIONS OF SALES LOCATED ON THE ATMEL WEBSITE, ATMEL
ASSUMES NO LIABILITY WHATSOEVER AND DISCLAIMS ANY EXPRESS, IMPLIED OR STATUTORY WARRANTY RELATING TO ITS PRODUCTS INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. IN NO EVENT SHALL ATMEL BE LIABLE FOR ANY DIRECT, INDIRECT,
CONSEQUENTIAL, PUNITIVE, SPECIAL OR INCIDENTAL DAMAGES (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS AND PROFITS, BUSINESS INTERRUPTION, OR LOSS OF
INFORMATION) ARISING OUT OF THE USE OR INABILITY TO USE THIS DOCUMENT, EVEN IF ATMEL HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Atmel makes no
representations or warranties with respect to the accuracy or completeness of the contents of this document and reserves the right to make changes to specifications and products descriptions at any
time without notice. Atmel does not make any commitment to update the information contained herein. Unless specifically provided otherwise, Atmel products are not suitable for, and shall not be
used in, automotive applications. Atmel products are not intended, authorized, or warranted for use as components in applications intended to support or sustain life.
Atmel®, Atmel logo and combinations thereof, Adjacent Key Suppression®, AKS®, and others are registered trademarks, QT™ and others are trademarks of
Atmel Corporation or its subsidiaries. Windows® is a registered trademark of Microsoft Corporation in U.S. and or other countries. Other terms and product
names may be registered trademarks or trademarks of others.
