2.2 Cs Sample Capacitors
Cs1, Cs2 and Cs3 are the charge sensing sample capacitors;
normally they are identical in nominal value. They should be
of type X7R dielectric.
The optimal Cs values depend on the thickness of the panel
and its dielectric constant. Lower coupling to a finger caused
by a low dielectric constant and/or thicker panel will cause the
position result to become granular and more subject to
position errors. The ideal panel is made of thin glass. The
worst panel is thick plastic. Granularity due to poor coupling
can be compensated for by the use of larger values of sample
capacitors.
A table of suggested values for no missing position values is
shown in Table 1-2. Values of Cs smaller than those shown in
the table can cause skipping of position codes. Code skipping
may be acceptable in many applications where fine position
data is not required. Smaller Cs capacitors have the
advantage of requiring shorter acquisition bursts and hence
lower power drain.
Larger values of Cs improve granularity at the expense of
longer burst lengths and hence more average power.
Cs1, Cs2 and Cs3 should be X7R type, matched to within
10% of each other (ie, 5% tolerance) for best accuracy. The
PCB reference layout (Figure 1-3) is highly recommended. If
the Cs capacitors are poorly matched, the wheel accuracy will
be affected and there could also be missing codes.
2.3 Rs Resistors
Rs1, Rs2, and Rs3 are low value (typically 4.7K) resistors
used to suppress the effects of ESD and assist with EMC
compliance. They are optional in many cases.
2.4 Power Supply
The usual power supply considerations with QT parts applies
also to the QT511. The power should be very clean and come
from a separate regulator if possible. This is particularly
critical with the QT511 which reports continuous position as
opposed to just an on/off output.
A ceramic 0.1µF bypass capacitor should be placed very
close to the power pins of the IC.
Regulator stability: Most low power LDO regulators have
very poor transient stability, especially when the load
transitions from zero current to full operating current in a few
microseconds. With the QT511 this happens when the device
comes out of sleep mode. The regulator output can suffer
from hundreds of microseconds of instability at this time,
which will have a negative effect on acquisition accuracy.
To assist with this problem, the QT511 waits 500µs after the
400µs taken to come out of sleep mode before acquiring to
allow power to fully stabilize. This delay is not present before
an acquisition burst if there is no preceding sleep state.
Use an oscilloscope to verify that Vdd has stabilized to within
5mV or better of final settled voltage before a burst begins.
The QT511 has specially enhanced power supply rejection
built in. This means that it is often possible to share the
regulator with other circuits. However, it is always advised to
be sure that Vdd is free from spikes and transients, and is
filtered sufficiently to prevent detection problems.
During development it is wise to first design a regulator onto
the PCB just for (and next to) the QT511, but allow for it to be
‘jumpered out’. If in development it is clear that there are no
problems with false detection or ‘angle noise’ even without a
regulator just for the QT511, then the regulator can be safely
omitted.
2.5 PCB Layout and Mounting
The E510 PCB layout (Figure 1-3) should be followed if
possible. This is a 1-sided board; the blank side is simply
adhered to the inside of a 2mm thick (or less) control panel.
Thicker panels can be tolerated with additional position error
due to capacitive ‘hand shadow’ effects and will also have
poorer EMC performance.
This layout uses 18 copper pads connected with intervening
series resistors in a circle. The finger interpolates between
the copper pads (if the pads are narrow enough) to make a
smooth, 0..127 step output with no apparent stair-casing. The
lateral dimension along the centre of each electrode should
be no wider than the expected smallest diameter of finger
touch, to prevent stair-casing of the position response (if that
matters).
Other geometries are possible, for example triangles and
squares. The wheel can be made in various diameters up to
at least 80mm. The electrode width should be about 12mm
wide or more, as a rule.
The SMT components should be oriented perpendicular to
the direction of bending so that they do not fracture when the
PCB is flexed during bonding to the panel.
Additional ground area or a ground plane on the PCB will
compromise signal strength and is to be avoided. A single
sided PCB can be made of FR-2 or CEM-1 for low cost.
‘Handshadow’ effects: With thicker and wider panels an
effect known as ‘handshadow’ can become noticeable. If the
capacitive coupling from finger to electrode element is weak,
for example due to a narrow electrode width or a thick, low
dielectric constant panel, the remaining portion of the human
hand can contribute a significant portion of the total
detectable capacitive load. This will induce an offset error,
which will depend on the proximity and orientation of the hand
to the remainder of the element. Thinner panels and those
with a smaller diameter will reduce this effect since the finger
contact surface will strongly domina te the total signal, and the
remaining handshadow capacitance will not contribute
significantly to create an error offset.
PCB Cleanliness: All capacitive sensors should be treated
as highly sensitive circuits which can be influenced by stray
conductive leakage paths. QT devices have a basic
resolution in the femtofarad range; in this region, there is no
such thing as ‘no clean flux’. Flux absorbs moisture and
becomes conductive between solder joints, causing signal
drift and resultant false detections or temporary loss of
sensitivity. Conformal coatings will trap in existing amounts of
moisture which will then become highly temperature
sensitive.
The designer should specify ultrasonic cleaning as part of the
manufacturing process, and in extreme cases, the use of
conformal coatings after cleaning.
lQ
5 QT511-ISSG R6.01/1005