Fi
g
ure 1. THAT4316 block dia
g
ram.
FEATURES
• Pre-trimmed VCA and RMS detector
• Very low supply voltage: 2.7V - 5.5V
• Low supply current: 1.2mA typ. (3.3V)
• Internal Vcc/2 divider and buffer
• Wide dynamic range: 115dB as
compander
APPLICATIONS
• Companding noise reduction
• Wireless microphones
• Wireless instrument packs
• Wireless in-ear monitors
• Battery operated dynamics processors
• Compressors
• Limiters
• Noise Gates
• AGCs
THAT 4316
The THAT4316 is a single-chip Analog Engine®
optimized for very low-voltage, low-power operation.
It incorporates a high-performance class-AB voltage-
controlled amplifier (VCA) and true-RMS-responding
level detector. The 16-pin QSOP part is aimed at
battery-operated audio applications including com-
panding systems for wireless microphones, wireless
instruments, and in-ear monitors, as well as dynam-
ics processors of all types. The 4316 operates from a
single supply voltage down to 2.7V, drawing only
1.2mA at 3.3V.
The 4316's true-RMS-level detector improves the
sound of the part over averaging or peak detectors in
companding applications as well as sound
mod-ifiers. This makes the 4316 ideal for many low-
power dynamics processors including compressors,
limiters, and gates.
The part was developed as a versatile analog
engine, drawing from THAT’s long history and expe-
rience with such applications. Because both VCA
control ports and the RMS level detector output are
independently available, the part is extremely
flexible. It can be configured for a wide range of
applications including single and multi-band com-
panders with a wide range of companding ratios,
plus compressors, expanders, limiters, AGCs, de-
essers, and the like.
Description
Pre-Trimmed, Very Low-Voltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation; Document 600177 Rev 00
15
2
14
3
13
4
12
5
11
6
10
7
9
8
THAT
4316
EC-EC+
IN OUT
VCA
IN
CT OUT
RMS
16
1
VCA
IN
RMS
IN
VCA
OUT
CT
NC
NC
NC
NC
EC-
RMS
OUT
EC+
VREF
NC
FILT
VCC
RA
RB
GND
Document 600177 Rev 00 Page 2 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Positive Supply Voltage (VCC) +6.0V
Supply Current (ICC)10 mA
I/O Pin Voltage Supply Voltage
Ec+, Ec- to Vref Voltage ± 1V
Vref Output Short-Circuit Duration 30 sec
Operating Temperature Range (TOP) -40 to +85 ºC
Junction Temperature (TJ) -40 to +125 ºC
Storage Temperature Range (TST) -40 to +125 ºC
Absolute Maximum Ratings2
Parameter Symbol Conditions Min Typ Max Units
Power Supply
Positive Supply Voltage VCC Referenced to GND +2.7 3.3 +5.5 V
Supply Current ICC No Signal
VCC=+3.3 V —1.2 1.8 mA
VCC=+5 V —1.3 2.0 mA
Voltage Controlled Amplifier (VCA)3
Max. I/O Signal Current iIN(VCA) + iOUT(VCA) VCC = +3.3 V —1200 —µApeak
VCC = +5 V —1600 —µApeak
VCA Gain Range EC+ or EC- used singly -50 —+50 dB
Gain at 0V Control G0EC+ = EC- = VREF -1.5 0 +1.5 dB
Gain-Control Constant ΔEC /ΔGain (dB) -40 dB to +40 dB —6.1 —mV/dB
Gain-Control Tempco ΔEC/ΔTCHIP Ref TCHIP=27ºC —+0.33 —%/ºC
Output Offset Voltage Change4 |Δ VOFF(OUT)|R
2 = 4.7kΩ
0 dB gain —315 mV
+15 dB gain —630 mV
Output Noise eN(OUT) 0 dB gain
22Hz~22kHz, R1=R2=4.7 kΩ—-95 -93 dBV
Total Harmonic Distortion THD 1kHz
0dB (VIN = -10dBV), EC+ = EC- = Vcc/2 —0.03 0.15 %
Maximum VCA Control Voltage Ec-,Ec+ Ref: VREF -500 —500 mV
VCA Control Port Input Impedance EC+, EC- 400 500 600 Ω
Source Impedance at VCA Input 5 Frequency > 320 kHz ——2.5 kΩ
Electrical Characteristics3, 4
1. All specifications are subject to change without notice.
2. If the device is subjected to stress above the Absolute Maximum Ratings, permanent damage may result. Sustained operation at or near
the Absolute Maximum Ratings conditions is not recommended. In particular, like all semiconductor devices, device reliability declines
as operating temperature increases.
3. Unless otherwise noted, TA=25ºC, VCC=+3.3V.
4. See Figure 13 for component references.
4. Reference is to output offset with approximately -40 dB VCA gain.
5. Refer to the text in item 4 of the 4316 and 2182 comparison on page 6.
SPECIFICATIONS1
Document 600177 Rev 00 Page 3 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Parameter Symbol Conditions Min Typ Max Units
RMS Level Detector
RMS reference input current iin0 — 7.2 — µArms
Output Voltage at Reference iIN eO(0) iIN = iin0 = 7.2 µA RMS, Ref:VREF -13 0 +13 mV
Output Error at Input Extremes eO(RMS)error iIN = 200 nA RMS -3 ±1 3 dB
iIN = 150 µA RMS -3 ±1 3 dB
Output scale factor Δ eO(RMS)/Δiin (dB) 0.72 µA< iIN(RMS) < 72 µ A —6.1 —mV/dB
Scale Factor Match to VCA -20 dB < VCA gain < +20 dB
0.72 µA< iIN(RMS) < 72 µ A 0.92 1 1.08
Rectifier Balance Iin =±Iin0 DCIN -1 0 +1 dB
Timing Current IT—7.2 —µA
Filtering Time Constant τTCHIP = 27 ºC 3611 X CTs
Output Tempco ΔeO(RMS)/ΔTCHIP Ref TCHIP = 27 ºC —+0.33 —%/ºC
Load Resistance RL-400mV < VOUTRMS< +230mV, Ref:VREF 400 —— Ω
Capacitive Load CL——100 pF
Vcc/2 Reference Generator 3
VREF Output Current IOUT(VREF) -1.25 —+1.25 mA
VREF Load Capacitance CL(VREF) ——100 pF
VREF Output Voltage VREF No load on VREF Vcc/2-12 Vcc/2 Vcc/2+12 mV
Voltage Divider Resistors RA, RB—48 —kΩ
Performance as a Compander (through an encode-decode cycle)
Dynamic Range (max signal level)-(no signal A-weighted output noise) —115 —dB
Distortion THD f = 1 kHz —0.15 —%
Frequency response -20 dB re: Max Signal 20 Hz ~ 20 kHz —± 1.5 —dB
Electrical Characteristics (con’t)3
Document 600177 Rev 00 Page 4 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
1
0.1
0.01 10.10.01 Vrms
THD+N [%]
REPRESENTATIVE DATA6,7
Figure 3. VCA THD+N vs. Input Level at -15 dB gain 8.
Figure 4. VCA THD+N vs. Input Level at +15 dB gain 8. Figure 8. VCA Offset (at VCA Out in Fig. 13) vs. Gain.
Figure 9. VCA Frequency Response for various Gains10.
Figure 6. VCA Gain vs. Control Voltage (VEc+-VEc-).Figure 2. VCA THD+N vs. Level at 0 dB gain 8.
Figure 5. VCA THD+N vs. Frequency at 0dB gain 9.
Figure 7. VCA Noise vs. Gain 8.
-100
-80
-60
-40
-20
0
20
40
60
-0.6 -0.4 -0.2 0 0.2 0.4
V
dB
1
0.1
0.02 210.10.01 Vrms
THD+N [%]
-120
-110
-100
-90
-80
-70
-60
-50
-100 -80 -60 -40 -20 0 20 40dB
dBV
1
0.1
0.02 0.30.10.010.003
THD+N [%]
Vrms
1
0.1
0.01 20k2k20020 Hz
THD+N [%]
Input:0.5Vrms,1kHz
-
10
-5
0
5
10
15
20
-80 -60 -40 -20 0 20 dB
mV
Sample 1
Sample 2
Sample 3
-2
-1.5
-1
-0.5
0
0.5
100k10k1k100
dBr
Hz
+20dB
0dB
-40dB
Document 600177 Rev 00 Page 5 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 10. RMS Output vs. Input Current iIN.
6. Unless otherwise noted, TA=25ºC, VCC=+3.3
V
, f=1kHz
7. The test circuit is shown in Figure 13.
8. Measured with an Audio Precision System One with 22 kHz bandwidth.
9. Measured with an Audio Precision System One with 80 kHz bandwidth.
10. Measured with an Audio Precision System One with >500 kHz bandwidth.
Figure 11. RMS Frequency Response vs. Level 9.
Figure 12. Supply Current vs. Supply Voltage.
Figure 13. The 4316 VCA and RMS detector test circuit.
IN
EC+EC-OUT
OUT
RMS
IN CT
Vcc
U2
5534
4k99
R1
4k99
R2
C2
10u
C1
47p
CT
10u
C6
22u
VCA
VCA
Input
VCA
Output
C8
100n
R3
4k99
C3
100p
THAT
4316
4123 5678
916 15 14 13 12 11 10
4k99
R4
10u
C4
RMS
Input
Vcc
C7
4u7
RMS Output
Ec- Input
Ec+ Input
RA
RB
-400
-300
-200
-100
0
100
200
20002002020.20.020.002 uARMS
mV
1
1.1
1.2
1.3
1.4
1.5
2.5 3 3.5 4 4.5 5 5.5 V
mA
-300
-200
-100
0
100
200 38.5 dB
27 dB
15.5 dB
4 dB
-7.5 dB
-19 dB
-30.5 dB
20k2k 200 20 Hz
mV
The THAT 4316 Analog Engine combines an
exponentially controlled Voltage-Controlled Amplifier
(VCA) with a true-RMS-responding level detector to
produce a versatile dynamics processor. The part is
implemented in a low-voltage Bi-CMOS process. It
delivers wide bandwidth and excellent audio per-
formance while consuming less than 4mW when run-
ning from 3.3V.
For details of the theory of operation of the VCA
and RMS Detector building blocks, we refer inter-
ested readers to THAT Corporation’s data sheets on
the 2180-Series VCAs and the 2252 RMS Level
Detector.
The VCA — in Brief
The VCA in THAT 4316 is based on THAT Corpo-
ration’s highly successful complementary log-antilog
gain cell topology — The Blackmer™ VCA — as used
in THAT 2180-Series IC VCAs. We modified the tra-
ditional design so that the VCA works in a power-
efficient class-AB mode under supply voltages as low
as 2.7V using a Bi-CMOS process. The VCA symme-
try is trimmed during wafer probe for minimum dis-
tortion. No external adjustment is allowed.
Input signals are currents in the VCA IN pin (pin
15). This pin is a virtual ground with dc level
approximately equal to VREF; in normal operation, an
input voltage is converted to a current via an appro-
priately sized resistor. Referencing Figure 13, the
VCA input voltage is converted to a current based on
the value of R1. Because any current associated with
dc offsets present at the input pin (for instance, any
dc offset from VREF in the preceding stages) will be
modulated by gain changes (thereby becoming audi-
ble as thumps), the input pin is normally ac-coupled
(C1).
The VCA output signal (at pin 13) is also a cur-
rent, in phase with respect to the input current. In
normal applications, the output current is converted
to a voltage via an external op-amp (U2 in Figure 13),
where the ratio of the conversion is determined by
the feedback resistor R2 connected between U2’s out-
put and its inverting input. The signal path through
the VCA and op-amp (from "VCA Input" to "VCA
Output" in Figure 13) is inverting. Note that this is
in contrast to other THAT Corporation ICs featur-
ing a Blackmer™ VCA (e.g., THAT 4315 or 2180
series), which have a non-inverting signal path.
The gain of the VCA is controlled by the voltage
applied between EC+ (pin 11) and EC- (pin 12). Note
that any unused control port should be connected to
VREF. The gain (in decibels) is proportional to (VEC+ –
VEC-) (see Figure 6). The constant of proportionality is
typically 6.1mV/dB. Note the limits to the control
voltages at EC+ and EC- in the specifications section.
The VCA’s noise performance varies with gain in
a predictable way as shown in Figure 7. At large
attenuation (<-50dB), the noise floor is limited to
about -114dBV by the input noise of the output op-
amp U2 (a 5534 type) and its feedback resistor. At
0dB gain, the noise floor is ~ -95 dBV as specified.
In the vicinity of 0dB gain, the noise increases almost
linearly with the gain. This applies to the whole posi-
tive gain region. As gain drops below -20dB, the
noise floor decreases more slowly than the gain and
tends to saturate below -40dB.
While the 4316’s VCA circuitry behaves similarly
to that of the THAT 2180-Series, there are several
important differences, as follows:
1. At +3.3 V VCC, approximately 1.2 mA is avail-
able from the 4316 for the sum of VCA input and
output signal currents. This increases to ~1.6mA at
+5V VCC.
2. A SYM control port (similar to that on the
2180 VCA) exists, but is driven from an internally
trimmed current generator. This current flows into
either the positive or negative control port, depend-
ing on the (internal) trimming direction, and must be
supplied by whatever circuitry drives this port.
3. Each of the 4316 VCA control ports is con-
nected to an internal 2:1 resistive voltage divider
(internally terminating at VREF). These scale the VCA
gain control constant from the internal ~3mV/dB to
match the RMS detector output characteristic. The
control port input impedance is 500Ω ±100Ω, so the
driving circuitry must be capable of supplying the
required current into this load.
4. To maintain stability over the wide range of
possible VCA gains, the 4316 VCA’s internal CMOS
transconductance amplifier requires that the source
impedance at the VCA input pin must be kept under
2.5kΩ above 320kHz. R3 and C3 in Figure 13 are pro-
vided to accomplish this. See the Applications sec-
tion for more ideas on how best to address this
issue.
The RMS Detector — in Brief
The 4316’s detector computes RMS level by recti-
fying the input current signals, converting the recti-
fied current to a logarithmic voltage, and applying
that voltage to an internal log-domain filter. The out-
put signal is a dc voltage proportional to the decibel-
level of the RMS value of the input signal current.
Some ac component (at twice the input frequency,
2fin) remains superimposed on the dc output. The ac
signal is attenuated by the internal log-domain filter,
which constitutes a single-pole rolloff with cutoff
determined by an external capacitor.
As in the VCA, the detector’s input signals are
currents to the RMS IN pin (pin 2). This pin is a vir-
tual ground with dc level equal to VREF, so a resistor
is normally used to convert input voltages to the
desired current. The level detector is capable of accu-
rately resolving signals well below 100nA (see
Figure 10). However, if the detector is to accurately
track such low-level signals, ac coupling is required.
Document 600177 Rev 00 Page 6 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
T
heory of Operation
Note also that small, low-volta
g
e electrolytic capaci-
tors used for this purpose may create significant
leakage if they support half the supply voltage, as is
the case when the source is dc-referenced to ground.
To ensure good detector tracking to low levels, the
input coupling capacitor's leakage (given the voltage
across it in the application) should be insignificant
compared to the lowest signal current to be resolved.
The internal log-domain filter cutoff frequency is
usually placed well below the frequency range of
interest. For an audio-band detector, a typical value
would be 5Hz, or a 32ms time constant. The filter’s
time constant is determined by an external timing
capacitor, CT, attached to the CT pin (pin 4), and an
internal current source (IT) connected between GND
and the CT pin. This current source is fixed at
~7.2μA with a tolerance of ~±20%. The resulting
time constant in seconds is approximately equal to
3611 times the value of CT (in farads). Note that, as a
result of the mathematics of RMS detection, the
attack and release time constants are fixed in their
relationship to each other.
The RMS detector is capable of driving large
spikes of current into CT when the audio signal at the
RMS detector's input increases suddenly. This cur-
rent is drawn from VCC (pin 9), fed into CT at pin 4,
and returns to the power supply through the ground
end of CT. If not handled properly through layout and
bypassing, these currents can mix with the audio,
producing unpredictable and undesirable results. As
shown in Figure 13, a local bypassing capacitor, like
C6, from the VCC pin to the ground end of the timing
capacitor CT, is strongly recommended to keep these
currents out of the ground structure of the circuit.
The dc output of the detector is scaled with the
same constant of proportionality as the VCA gain
control, ~6.1mV/dB. The detector’s 0dB reference
current (iin0, the RMS input current which causes the
detector’s output to equal VREF), is approximately
equal to 7.2μA, the same value as IT. The RMS detec-
tor output stage is capable of directly driving either
of the 500Ω VCA control ports to the limits of the
detector output voltage. It is also capable of driving
up to 100pF of capacitance.
Frequency response of the detector (see
Figure 11) extends across the audio band for a wide
range of input signal levels. Note, however, that it
does fall off at high frequencies at low signal levels.
Differences between the 4316’s RMS level detec-
tor and that of the THAT 2252 include the following:
1. The rectifier in the 4316 RMS detector is inter-
nally balanced by design, and cannot be adjusted
externally. The residual mismatch in the 4316 will
not significantly increase ripple-induced distortions
in dynamics processors over that caused by the sig-
nal ripple alone.
2. The time constant of the 4316’s RMS detector
is determined by the combination of the external
timin
g
capacitor and the internal current source, IT.
A resistor is not normally connected directly to the
CT pin on the 4316.
3. The 0dB reference input current, or level
match, is equal to approximately IT. However, as in
the 2252, the level match will be affected by any addi-
tional currents drawn from the CT pin.
Reference Voltage
The 4316 input and output signals, as well as the
VCA control voltages, must be biased to a reference
voltage between VCC and ground. For optimal per-
formance, the reference must have low AC imped-
ance and noise. The 4316 contains an internal volt-
age divider (RA, RB) and buffer amplifier (OA1) for
this function, as shown in Figure 14.
Capacitor C7 is required from the FILT pin (pin
7) to ground. It serves to minimize the influence of
the thermal noise of the resistive divider on the rest
of the circuity, as well as to filter out any supply-
related noise. A 4.7μF capacitor results in a lowpass
pole of ~1.4Hz with the internal divider impedance
of 24kΩ. This is sufficient for most applications.
Larger values provide additional filtering at the
expense of longer settling times after power is
applied.
The FILT pin is internally connected to the input
of unity gain buffer OA1. The output of OA1 is avail-
able at the VREF pin (pin 6). The buffer also drives
the requisite internal nodes, including one end of
each of the control-port voltage dividers. Because
most of OA1's output current is required to drive the
low-impedance dividers at the VCA control ports,
designers should take care not to draw too much
current externally from this pin. Limit the external
current to within +/-1.25mA.
Pins 1, 3, 14 and 16, are not connected
internally; we suggest they be connected to VREF in a
PCB layout so that they provide shielding to the VCA
and RMS input pins.
Document 600177 Rev 00 Page 7 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 14. Internal voltage reference generator.
8
7
6
9
10
11
Rb Ra
Vcc Vcc
NC
EC+
GND
FILT
VREF
OA1
4u7
to other VREF
connections to other internal
circuitry
The 4316 provides the basic building blocks for
a wide variety of dynamics processing applications:
an exponentially controlled VCA and a logarithmic
RMS detector. These elements are especially versatile
because the audio performance of designs using
these blocks is determined primarily by the control
loop (or "side chain") from the detector to the VCA
control port. Theory of the interconnection of expo-
nentially controlled VCAs and log-responding level
detectors is covered in THAT Corporation’s design
note DN01A, "The Mathematics of Log-Based
Dynamic Processors".
Perhaps the most important application for the
4316 is wireless audio companding systems. In this
data sheet, we cover this application in some detail.
However, many other configurations are possible,
including all those covered within THAT's collection
of application notes for dynamics processors (though
shown with previous VCA/detector parts or Analog
Engines). For assistance with these and any other
applications, please contact our applications engi-
neers at apps_support@thatcorp.com.
Noise Reduction (Compander)
Configurations
A primary use of the 4316 is for noise reduction
systems, particularly within battery-operated devices.
In these applications, one 4316 is configured for use
as a compressor (or encoder) to condition audio sig-
nals before feeding them into a noisy channel such as
a radio-frequency (RF) link. A second 4316, config-
ured as an expander (or decoder), is located at the
receiver end of the noisy channel.
The compressor reduces the dynamic range of
the audio signals so that it can fit better through a
channel with limited dynamic range. The expander
works in an opposite, complementary fashion to
restore the dynamic range of the original audio sig-
nal (as present at the input of the compressor).
As shown in Figure 17, during low-level audio
passages, the compressor increases signal levels,
bringing them up above the noise floor of the
channel. At the receiving end, the expander reduces
the signal back to its original level, in the process
attenuating the channel noise.
During high-level audio passages, the compressor
decreases signal levels, reducing them to fit within
the headroom limits of the channel. The expander
then increases the signal back to its original level.
While the channel noise may be increased by this
action, in a well-designed compander, the noise floor
will be masked by the high-level audio signal.
Advantages of True-RMS-Level Detection
The 4316's RMS detector has the property that it
responds faster to large increases in signal level than
to small ones. This is because it responds to the
square of the input signal, instead of the signal itself.
Essentially, its attack time varies, becoming shorter
for large level changes than that for small ones. This
mimics the behavior of the human ear, resulting in
more "musical" response to audio signals than for
average or peak responding detectors.
In companding applications, the "variable" attack
time ensures that overloads are kept short in dura-
tion, because the compressor responds quickly in
cases where a low-level audio signal (causing high
VCA gain) is followed suddenly by a much higher
level signal (which reduces the VCA gain over time as
the detector acquires the new level). This minimizes
the duration of overloads for a given time constant
when compared to those using average responding
detectors.
Another advantage of RMS detection over average
or peak detection is that it is relatively insensitive to
phase shifts in the signal being measured. This is
particularly helpful in companding applications
because low- and high-frequency phase shifts com-
mon in a bandlimited transmission channel cause
less difference between the compressor’s detector
reading and that of the expander. This ensures better
tracking between the expander and detector in real-
world applications.
The combination of insensitivity to phase shift
and variable attack behavior causes companders
based on true-RMS detection to sound better than
those based on either average- or peak-responding
detectors.
Versatility in Compander Design
The 4316 was designed to facilitate the design of
a wide variety of companding noise reduction sys-
tems. The RMS detector responds accurately over a
wide range of input current (Figure 10), while the
VCA responds accurately to a wide range of gain
commands (Figure 6). The RMS output and the VCA
control inputs are fully configurable, which makes it
easy to configure the 4316 for companding ratios
different from the traditional 2:1. (See the section
"3:1 Compander" below for one such example.)
The 4316 supports a wide range of compander
designs (and more), including simple 2:1 wide range
(level-independent) systems, level-dependent systems
with thresholds and varying companding slopes, sys-
tems including noise gating and/or limiting, and sys-
tems with varying degrees of pre-emphasis and filter-
ing in both the signal and control paths. Generally,
these variations can be accomplished by conditioning
the detector side chain rather than the audio signal
itself. The audio signal passes through as little as two
VCAs and two opamps, and still supports multiple
ratios, thresholds, and time constants.
In this datasheet, we show the part used in three
example designs. First is a simple 2:1 companding
noise reduction system. Next, we show a high-
performance 2:1 compander with pre- and
Document 600177 Rev 00 Page 8 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Applications
de-emphasis networks in the si
g
nal path and pre-
emphasis in the detector path. Finally, we present a
3:1 compander with pre- and de- emphasis in the
signal path and pre-emphasis in the detector path.
One other minor point is that companders
designed using the 4316 are generally compatible
with those using other THAT Analog Engine and dis-
crete VCA or RMS detector ICs. For example, a 4316
may be configured as a low-voltage, low-power-
consumption compressor for the battery-powered
transmitter in a wireless microphone or instrument
belt pack, and paired with a higher-voltage, higher-
power-consumption 4301 or 4305 as the comple-
mentary expander in the companion AC-powered
receiver.
Simple Compander Design
Basic 2:1 Encoder
The encoder in a wireless companding system is
generally a feedback compressor located in the trans-
mitter, operating from a battery supply. Figure 15
shows a basic 2:1 encoder. The blocks within the
bold outline are the three functional circuits in the
4316, i.e., a VCA, an RMS detector and a reference
generator. Following the mathematical simplifica-
tions taught in DN01A, the steady-state transfer func-
tion of this circuit is :
(1)
OUTC=1
(KC+1)(INC+GC+KC$RMS0C)
where,
, (2)
OUTC=20log10(VoutC)
, (3)
INC=20log10(VinC)
, and (4)
GC=20log10 R2
R1
. (5)
RMS0C=20log10(V0C)
INC and OUTC are the compressor’s voltage input VinC
and output VoutC in dBV, respectively. GC is the com-
pressor's signal path gain in dB; in Figure 15, GC is
determined by the ratio of R2 to R1 as in Equation 4
and is 0dB. KC is the linear gain (in V/V) between the
compressor RMS detector output and VCA control
port. (KC is 1 in Figure 15 since the detector output is
connected directly to the VCA EC- port.) Finally,
RMS0C is the dBV value of the detector reference volt-
age, V0C, which causes iin0 (the RMS input reference
current), to flow in the compressor detector’s input.
,(6)
V0c=iin0R3
where R3 is the detector’s input resistance (4.99kΩ)
in Figure 15. Hence, V0C is 35.9mVrms and RMS0C is
-28.9dBV. For this example, the compressor output
OUTC is always half of INC plus a fixed offset of
(GC+RMS0C)/2, yielding a compression ratio of 2:1.
The compression ratio (CR), is generally defined in
Equation 7:
.(7)
CR =KC+1
At the 4316 VCA input, R4 (4.99kΩ) and C5
(100pF) comprise the compensation network
required to keep the VCA’s internal amplifier stable
for all gains. (C2 performs a similar function for U2,
neutralizing the VCA's output capacitance plus any
stray layout capacitance appearing at the inverting
input of U2.)
The RMS detector output is tied directly to the
VCA’s negative control port, EC-. (This is what makes
Document 600177 Rev 00 Page 9 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
RMS
VCC
U2
Op-Amp
R1
4k99
R2
R3
4k99
4k99
1u
C1
C2
47p
1u
C3
C4
10u C8
22u
VCA
Input Output
VCC
C6
4u7
C7
100n
R4
4k99
C5
100p
U1
4316
IN
15 EC+
EC-
11
12
OUT 13
OUT
IN
25
CT
4
VCC FILT
GND VREF
8
97
6
Figure 15. Basic 2:1 Compressor using 4316.
KC=1, and sets CR at 2:1.) Because the RMS outpu
t
connects to the negative-sense control port, EC-, this
circuit acts as a compressor. C4 (10μF) sets the RMS
detector time constant to approximately 36msec.
As described in the Theory of Operation section
“The RMS Detector - In Brief”, the RMS detector is
capable of driving large spikes of current into C4 in
Figure 15. To prevent these currents from upsetting
circuit grounds, VCC should be bypassed to a point
very near the grounded end of C4 with a capacitor
equal to or greater than the value of C4. 22μF C8 in
Figure 15 serves this purpose. The grounded ends of
these two capacitors should be connected together
before being tied to the rest of the ground system.
This will ensure that the current spikes flow within
the local loop consisting of the two capacitors, and
stay out of the ground system. This requirement
applies to the decoder and other applications of the
THAT4316 as well.
Basic 2:1 Decoder
Figure 16 shows the THAT4316 configured as a
2:1 decoder. This is a feedforward expander
intended to complement the encoder of Figure 15. It
is optimized for low-voltage operation, as might be
the case for a decoder in an in-ear monitoring system
which runs from a battery power. The expander
steady-state transfer function is:
, (8)
OUTE=(KE+1)INE+GE−KE$RMS0E
where
, (9)
OUTE=20log10(VoutE)
, (10)
INE=20log10(VinE)
, and (11)
GE=20log10 R11
R10
. (12)
RMS0E=20log10(V0E)
As in Equation 1, INE and OUTE are the dBV values of
the expander’s voltage input VinE and output VoutE,
respectively. GE is the expander signal path gain in
dB, which is 0 dB here as well. KE is the gain in lin-
ear terms (V/V) between the expander detector out-
put and VCA control port; it is unity here. RMS0E is
the dBV value of the expander detector’s reference
voltage V0E, which is calculated using Equation 6 with
the input resistor R12 (4.99kΩ). As in the encoder, it
is also -28.9dBV.
Because, in Figure 16, the detector's output is
connected directly to the VCA positive control input,
the expander’s output OUTE will always double its
input INE, except for a fixed offset (GE-RMS0E). The
expansion ratio is thus 2:1, and given generally by
Equation 13:
. (13)
ER =KE+1
Since the 4316 VCA is not stable unless it sees a
source impedance of 2.5kΩ or less at high frequen-
cies, another compensation network (R13 & C14) is
provided to maintain stability. 47pF C11 maintains
stability in U2, just as C2 does Figure 15.
In this instance, the RMS detector output is con-
nected to EC+; this reverses the polarity of the control
signal relative to the encoder, and makes this circuit
an expander rather than a compressor.
System Performance
The encoder and decoder in Figure 15 and 16
form a compander system. To a first approximation,
Document 600177 Rev 00 Page 10 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
RMS
U2
Op-Amp
R10
4k99
R11
R12
4k99
4k99
1u
C10
C11
47p
1u
C12
C13
10u
VCA
Input Output
U1
4316
VCC
C15
4u7
C16
100n
VCC
C17
22u
R13
4k99
C14
100p
IN
15 EC+
EC-
11
12
OUT 13
OUT
IN
2
5
CT
VCC FILT
GND VREF
8
97
6
4
Figure 16. Bas
i
c 2:1 E
x
pander using 4316.
the output of Fi
g
ure 15 will be connected to the inpu
t
of the Figure 16 expander. (The two are usually con-
nected by an RF link, which should be relatively
transparent within the audio band, except for noise.)
Static Performance
Assuming that both VCA and RMS detectors
match well, as the detectors have identical input
resistor values, the reference voltage terms, i.e.,
KC•RMS0C and KE•RMS0E, in Equations 1 and 8 can-
cel each other. So the overall compander system
transfer function becomes:
(14)
OUTE=INC+GC+GE
With zero dB gain in both the encoder and
decoder, the compressor input is fully recovered at
the expander output. The behavior of this compand-
ing system is shown in Table 1. The columns labeled
Encoder Out and Decoder Out use the previous
equations to generate signal and gain values. The
Encoder VCA Gain is the difference between Encoder
Out and Encoder In; The Decoder VCA Gain is calcu-
lated similarly using Decoder Out and Decoder In.
These two gains have the same absolute value but
opposite polarity. The values in the column labeled
iin_RMS, which is the detector’s RMS input current, are
derived using the equation:
, (15)
iin_RMS =10 EncoderOut
20
Rin_RMS
where Rin_RMS is the detector input resistance (4.99kΩ
in Figure 15 and 16).
Figure 17 presents this data in the form of a
"butterfly diagram" for the 2:1 compander. Signal lev-
els are shown from the encoder input, through its
output, to the decoder output. The encoder com-
presses its input dynamic range by a factor of 2, its
CR, while the decoder reverses the process and
restores the signal back to its original at the decoder
output. Hence, only half of the signal dynamic range
is required for the transmission channel between the
encoder and decoder.
The encoder VCA gain varies from -14dB to
+36dB, which covers half of the input dynamic range
as well, while the decoder VCA's gain varies from
-36dB to +14dB. Both these ranges easily fit within
the capabilities of the 4316 VCA. The RMS input cur-
rent range is also easily accommodated.
Dynamic Performance
While the VCA gains in both the compressor and
expander change with signal levels, of course the
changes are not instantaneous. As noted earlier, the
RMS detector used in THAT's Analog Engines,
including the 4316, behaves favorably when faced
with changing signal levels. Its quick response to
sudden overloads ensures that the compressor
reacts appropriately to minimize transient overloads
in the compressor and the subsequent channel. And,
its insensitivity to phase changes in the signal means
that the expander and compressor detectors will
deliver consistently similar level readings despite the
band-limiting in the transmission channel.
Nonetheless, to ensure good dynamic tracking,
the time constants of both the compressor and
expander RMS detectors must be the same. The time
constants are controlled by the internal timing cur-
rent IT, and the external timing capacitor (C4 and C13
in the two schematics). The internal timing current is
controlled to within ~±20% of its nominal value.
This tolerance adds to that of the timing capacitors.
For the best possible tracking, THAT recommends
using tight-tolerance capacitors.
Another consideration is distortion. At low fre-
quencies, the compressor RMS detector output con-
tains significant ripple at twice of the input frequency
(2fin). The amount of this ripple increases as fre-
quency decreases. The ripple adds a time-varying
component to the steady-state VCA gain. The ripple
amplitude modulates the signal in the VCA, resulting
in third harmonic distortion (3fin) in the output of the
compressor. This amounts to a "squashing" of the
tops and bottoms of the input sine wave, since the
detector output is the highest during those portions
of the input signal.
The expander RMS detector output generally
contains the same (2fin) ripple in the same phase
relationship to the fundamental as that of the com-
pressor. And, if the distortion components (at 3fin)
Document 600177 Rev 00 Page 11 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
-100-360.120-6436-100 -90-310.214-5931-90 -80-260.380-5426-80 -70-210.676-4921-70 -60-161.20-4416-60 -50-112.14-3911-50 -40-63.80-346-40 -30-16.75-291-30 -20412.00-24-4-20 -10921.40-19-9-10 01438.00-14-140
(
dBV
)
(
In dB
)
(
μ
A
)
(
dBV
)
(
In dB
)
(
dBV
)
Decoder
Out
Decoder
VCA
Gain
iin_RMS
Encoder
Out/
Decoder
In
Encoder
VCA
Gain
Encoder
In
Table 1. 2:1 compander transfer characteristics.
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
-100
0
Encoder
Input
(dBV) (dBV)
Decoder
Output
Decoder
Input
Encoder
Output
-10
-14
-19
-24
-29
-34
-39
-44
-49
-54
-59
-64
-20
-30
-40
-50
-60
-70
-80
-90
-100
Headroom Limit
Noise Floor
Figure 17. 2:1 compander butterfly diagram.
are not phase-shifted with respect to the fundamental
(fin), then the ripple in the expander RMS detector
output will reverse the dynamic effect of that in the
compressor, and the distortion in the compressor
output will be reduced or even canceled in the
expander output. But, to make this work, the low-
frequency phase shift of the channel must be very
small indeed. System designers should bear this in
mind if low distortion is important at low
frequencies.
High-Performance 2:1 Compander
While the compander in Figure 15 and Figure 16
performs adequately in some applications, a few
minor changes can result in substantially improved
overall performance. The following compander
implementation adds pre- and de- emphasis to the
signal path and pre-emphasis to the detector path.
Pre-emphasis in the encoder signal path helps over-
come the “acqua noise” characteristic of the FM RF
channel by raising the level of higher frequency por-
tions of the signal while it is in the transmission
channel. The de-emphasis in the decoder brings the
frequency response of the signal back to flat while
simultaneously lowering the noise floor of the chan-
nel. The pre-emphasis in the detector paths alleviates
high-frequency overload due to the signal path
pre-emphasis.
High-Performance 2:1 Encoder
The encoder shown in Figure 18 implements pre-
emphasis in the signal path by means of a non-
inverting stage with op-amp U3, R6, R7, and C10.
Equation 1 from the basic encoder discussion is still
valid, but both the signal path gain GC and the detec-
tor reference level RMS0C become frequency
dependent due to the associated pre-emphasis ne
t
-
works. GC is expressed in the equation below,
. (16)
GC=20log10 R2
R1
sC10(R6+R7)+1
sC10R6+1
The 2nd term inside the log is introduced by the
signal-path pre-emphasis network. Its bode plot is
shown in Figure 19. The gain (in dB) shown is the
ratio of the signal at the output of U3 to its non-
inverting input. The zero frequency f1 and pole fre-
quency f2 are calculated using the equations:
, (17)
f1=1
2(R6+R7)C10
. (18)
f2=1
2R6C10
At frequencies well below f1 (391Hz), the gain is
0dB due to the effect of C10. As frequency increases
beyond f1, the gain starts to increase at 6dB/octave,
then flattens out at f2 (2.34kHz).
So, the low-frequency (<<f1) compressor gain
GC_LF is:
, (19)
GC_LF =20log2010 R2
R1{10dB
while the high-frequency (>>f2) gain, GC_HF, is
approximately:
. (20)
GC_HF =20log10 R2
R1
R6+R7
R6{26dB
The extra 16dB gain at high-frequency is a result of
the input pre-emphasis network.
In the circuit of Figure 18, we implemented the
signal-path pre-emphasis with an additional opamp
in order to minimize noise, rather than with a series
R-C network in parallel with the VCA input resistor
Document 600177 Rev 00 Page 12 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Input U3
Op-Amp
4k99
C10
68n
R7
R6
1k
U2
Op-Amp
R1
4k99
R2
R3
3k4
R5
2k21
15k8
4u7
C1
C2
47p
4u7
C3
47n
C5
C4
10u
VCA Output
VCC
C6
4u7
C7
100n
VCC
C8
22u
R4
4k99 C5
100p
U1
A
4316
RMS
IN
15 EC+
EC-
11
12
OUT 13
OUT
IN
2
5
CT
VCC FILT
GND VREF
8
97
6
4
Figure 18. High-performance 2:1 Encoder circuit.
R1. This is because, for gains of unity and above, the
4316 VCA’s dominant noise source is its input noise
voltage, so reducing the current-to-voltage conversion
impedance at the VCA input results in a proportional
increase in the output noise. This is undesirable.
However, if the pre-emphasis network is placed in a
low-noise buffer stage in front of the VCA, there will
be less noise at the output of the compressor.
The 16dB high-frequency gain added by the
signal-path pre-emphasis increases the compressor's
output level at high frequencies. This can cause pre-
mature overload in the transmission channel. This
undesirable effect is offset by the pre-emphasis in the
detection path shown in Figure 18.
With the addition of the pre-emphasis, the RMS
detector’s reference voltage also becomes frequency
dependent as in Equation 21:
. (21)
RMS0C=20log10 iin0$R31+sC5R5
1+sC5(R3+R5)
The 2nd term inside the lo
g
arithm is the RMS
detector’s input pre-emphasis network impedance.
The RMS input current is proportional to the recip-
rocal of the impedance. The bode plot for this cur-
rent (the detector path pre-emphasis) is also drawn
in Figure 19. Its corner frequencies, f3 and f4, are
expressed in Equation 22 and 23, respectively.
(22)
f3=1
2(R3+R5)C5
(23)
f4=1
2R5C5
For the case in Figure 18, at frequencies substan-
tially under 604 Hz (f3), the detector’s input net-
work’s impedance is R3. Hence, RMS0C in that region
is -32.2dBV. At frequencies substantially above
1.53kHz (f4), the impedance is approximately the
parallel combination of R3 and R5, i.e., 1.34 kΩ. So
RMS0C reduces to -40.3dBV, making the detector
more sensitive at high frequencies. Note that the geo-
metric center frequencies for the signal-path and
RMS detector pre-emphasis networks are about the
same, which is 0.96kHz, i.e., . The
f1f2=f3f4
detector pre-emphasis gain is 8dB, about half that of
the signal path.
Increasing the detector's sensitivity at high fre-
quencies through pre-emphasis causes it to weight
high frequencies more heavily, hence, reducing gain
more strongly to high frequencies than to low ones.
The right mix of signal-path and detector pre-
emphasis avoids high-frequency overload which
would otherwise occur.
High-Performance 2:1 Decoder
The decoder shown in Figure 20 matches the
encoder of Figure 18. It includes a signal-path
Document 600177 Rev 00 Page 13 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 19. Bode plot of th
e
signal-path and detector
-
path pre-emphasis of the Fig. 18 Encoder.
Figure 20. High-performance 2:1 Decoder circuit.
U2
Op-Amp
R10
15k8
R12
R14
3k4
R11
2k21
4k99
1u
C10
C11
47p
4u7
C13
47n
C19
C14
10u
VCA
Input Output
VCC
C16
4u7
C18
100n
VCC
C17
22u
R15
2k49 C15
220p
R13C12
68n 1k
U1A
4316
RMS
IN
15 EC+
EC-
11
12
OUT 13
OUT
IN
2
5
CT
VCC FILT
GND VREF
8
97
6
4
f(Hz)
(dB)
0391 2.34k
604 1.53k
The signal path pre-emphasis
The detector path pre-emphasis
16dB
8dB
6dB/Oct
de-emphasis network that has inverse frequency
response to that of the encoder's signal-path pre-
emphasis network as in Figure 21. Equation 8 in the
previous section is still applicable. But the signal-
path gain GE becomes frequency dependent and is
shown in Equation 24.
(24)
GE=20log10 1
R10
R12(sC12R13+1)
sC12(R12+R13)+1
The same values of R1and R7 in Fi
g
ure 18 makes
it possible to reuse the signal pre-emphasis network
as part of the de-emphasis one at the VCA output in
Figure 20. The resulting in de-emphasis network
ensures that GE is complementary to the encoder GC
over frequency. Because an identical pre-emphasis
network is employed in the decoder detector path,
the expander detector reference RMS0E is also fre-
quency dependent and cancels out the RMS0C in the
compander system.
Table 2 and Figure 22 show the low-frequency
transfer characteristics of the high-performance 2:1
companding system. The encoder VCA gain varies
between -21dB and +29dB for an input dynamic
range of 100dB. At the same time, the RMS detector
current levels iin_RMS varies from 0.26μA to 82μA and
Rin_RMS=R3=3.4kΩ.
The high-frequency transfer characteristics are
shown in Table 3 and Figure 23. Because of the
detector pre-emphasis, the high frequency signal
level at the compressor's output is only 4dB higher
than that at the lower frequencies, even with 16dB
signal pre-emphasis. The VCA gain range changes to
between -33dB and +17dB, and Rin_RMS=1.34kΩ, and
iin_RMS shifts up to between 1μA and 319.2μA. All are
well within the reach of the 4316.
Document 600177 Rev 00 Page 14 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 22. Butterfly diagram of the High-performance
2:1 compander (f << 391Hz).
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
-100
0
Encoder
Input
(dBV) (dBV)
Decoder
Output
Decoder
Input
Encoder
Output
-10
-11
-16
-21
-26
-31
-36
-41
-46
-51
-56
-61
-20
-30
-40
-50
-60
-70
-80
-90
-100
Headroom Limit
Noise Floor
-100-290.260-6129-100 -90-240.460-5624-90 -80-190.820-5119-80 -70-141.460-4614-70 -60-92.590-419-60 -50-44.600-364-50 -4018.18-31-1-40 -30614.55-26-6-30 -201125.88-21-11-20 -101646.02-16-16-10 02181.83-11-210
(
dBV
)
(
In dB
)
(
μ
A
)
(
dBV
)
(
In dB
)
(
dBV
)
Decoder
Out
Decoder
VCA
Gain
iin_RMS
Encoder
Out/
Decoder
In
Encoder
VCA
Gain
Encoder
In
Table 2. High-performance 2:1 compander transfe
r
characteristics (f << 391Hz).
Figure 21. Bode plots of the Fig. 20 Decode
r
de-emphasis gain and the Fig. 18 Encoder
pre-emphasis gain.
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
-100
0
Encoder
Input
(dBV) (dBV)
Decoder
Output
Decoder
Input
Encoder
Output
-10
-7
-12
-17
-22
-27
-32
-37
-42
-47
-52
-57
-20
-30
-40
-50
-60
-70
-80
-90
-100
Headroom Limit
Noise Floor
Figure 23. Butterfly diagram of the Hig
h
-performance
2:1 compander (f >> 1.53kHz).
Table 3. High-performance 2:1 compander transfe
r
characteristics (f >> 2.34kHz).
-100-171.010-5717-100 -90-121.790-5212-90 -80-73.190-477-80 -70-25.680-422-70 -60310.09-37-3-60 -50817.95-32-8-50 -401331.92-27-13-40 -301856.76-22-18-30 -2023100.9-17-23-20 -1028179.5-12-28-10 033319.2-7-330
(
dBV
)
(
In dB
)
(
μ
A
)
(
dBV
)
(
In dB
)
(
dBV
)
Decoder
Out
Decoder
VCA
Gain
iin_RMS
Encoder
Out/
Decoder
In
Encoder
VCA
Gain
Encoder
In
f(Hz)
(dB)
391 2.34k
16dB 16dB
Encoder sign al-path pre -emphasi s
Decoder signal-path de -emphasis
0
In this desi
g
n, we set the maximum encoder ou
t
-
put level to -7dBV (or 0.63Vpeak). This level is well
within the voltage capabilities of most 3.3 or 5 V
opamps used for U2 in Figure 18 and 20. Addition-
ally, if the designer wishes to limit the voltage swing
at the VCA's output to prevent over-modulation in the
transmission channel, a pair of back-to-back silicon
diodes across R12 will accomplish this quite easily,
limiting peak swings to about ±0.7V.
Compared to the basic 2:1 compander responses
shown in Figure 17, the signal levels at the encoder
output are approximately 3 to 7 dB higher over the
audio frequency range. This is well predicted by
Equation 1. For instance, at low frequencies, the
10dB higher gain and over 3dB lower RMS reference
voltage level of the high-performance encoder result
in the overall ~3dB higher level at the encoder out-
put. The selection of the gain and RMS reference
level is application dependent. It depends on the
dynamic range of source signals, transmission chan-
nel characteristic and supply level, etc.
In the high-performance compander, any noise in
the channel is attenuated at high frequencies by the
~16dB high-frequency attenuation of the decoder
signal-path de-emphasis network. This makes a dra-
matic difference in the perception of the channel
noise, and improves masking of the channel noise by
the signal.
As a result, we observe about 5.3dB (A-weighted)
improvement in the noise floor of the high perform-
ance 2:1 compander compared to the basic 2:1 com-
pander. In fact, an even better improvement is
expected in reality as the channel noise is not
included in the simulation. But, perhaps more
importantly, the signal path pre- and de-emphasis
combination helps low-frequency signals better mask
the acqua noise of a typical FM transmission
channel.
A
nd, the pre-emphasis in the detector mini-
mizes transient overload that might result from the
signal-path pre-emphasis.
3:1 Compander
The flexible configuration of THAT Corporation’s
Analog Engine® ICs allows compression and expan-
sion ratios of other than 2:1. This feature can be par-
ticularly advantageous in situations where RF
bandwidth and power are at a premium. The circuits
in Figure 24 and 25 demonstrate a 3:1 companding
system with pre- and de-emphasis in the signal path,
and pre-emphasis in the detector path.
The topology of this system is similar to the pre-
vious examples. The transfer functions for a compan-
der system in Equations 1, 8, and 14 still apply, but
because of amplifier U3 (with a gain of 2) between the
RMS detector output and the VCA control ports (in
both the encoder and decoder), the gain factor
KC=KE=2, and thus CR=ER=3:1.
We chose an inverting topology for the gain stage
U3 is to minimize the loading to the on-chip VREF
generator. Because this inverts the polarity of the
control voltage, we swapped the VCA control ports
(using EC+ for the encoder and EC- for the decoder).
For the 3:1 compressor, pre-emphasis in the
detector is even more important than for a 2:1 sys-
tem. This is because the higher compression ratio
leads to more aggressive VCA gain variations as input
signal levels change. Making the detector more sensi-
tive at high frequencies helps mitigate potential tran-
sient overload at the compressor output. when the
input goes from very low to very high quickly.
Besides that, the detector pre-emphasis also results
in a flatter swept sine-wave response from input to
output of the compressor.
Document 600177 Rev 00 Page 15 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 24. 3:1 Encoder circuit.
RMS
VCC
U2A
Op-Amp
R1
10k2
R2
R3
6k19
205k
2u2
C1
C2
15p
3u3
C3
C4
10u C8
22u
VCA
Input Output
VCC
C6
4u7
C7
100n
R4
3k24
C5
220p
U2B
Op-Amp
10k2
C10
47n
R7
R5
1k13
U1
4316
R8
R9
2k
R10
4k02
C11
C12
1n
U3
Op-Amp
2k87
33n
IN
15 EC+
EC-
11
12
13
OUT
IN
25
CT
4
VCC FILT
GND VREF
8
97
6
OUT
The transfer functions for the frequency-
dependent GC and GE in Equations 16 and 20 still
apply. In Figure 24, the signal path pre-emphasis
starts at f1=299Hz and flattens at f2=3kHz. The
encoder low-frequency signal gain is 26dB. The pre-
emphasis boosts GC by 20dB. Hence, at f >> f2, the
signal gain GC increases to 46dB.
Equation 21 for the frequency-dependent detec-
tor reference voltage also applies here. The RMS
detector pre-emphasis starts at 532Hz and ends at
1.68kHz. And the high frequency boost is about 10
dB, half that of the signal path pre-emphasis. The
center frequencies in the signal and detector paths
are also set to be the same, ~0.95kHz. The identical
detector pre-emphasis network is employed in the
complimentary 3:1 decoder circuit as in Figure 25.
Table 4 and 5 list the transfer characteristics of
the 3:1 compander at low and high frequencies,
respectively. The amount of pre-emphasis chosen for
the RMS detector path makes the encoder output lev-
els stay the same over frequency. So the over-
frequency compander level transfer is represented in
one single butterfly plot, Figure 26.
As in the previous two compander examples, for
a 100dB input dynamic range, the channel dynamic
range requirement here is also scaled by CR, but
here is equal to 33dB: one-third of the input signal
dynamic range. Since the encoder compresses the
input signal more than that in the 2:1 systems, the
VCA gain changes over a wider range, i.e., -19dB to
+47dB at low frequencies and -35dB to +32dB at
high frequencies, which is two-thirds of the input
dynamic range.
The RMS detectors input current varies over a
narrower range, which follows the channel dynamic
range. At frequencies well below 532 Hz,
Rin_RMS=6.19kΩ and iin_RMS varies from 1.19μA to
55.1μA; for frequencies above 1.68kHz, Rin_RMS
decreases to 1.96kΩ, hence iin_RMS shifts up accord-
ingly to between 3.75μA and 174.1μA. The detector-
path pre-emphasis yields a flat overall frequency
response at the encoder output, so the encoder out-
puts at low and high frequencies are the same.
Finally, we set the decoder's maximum output
level to -9dB (0.5Vpeak) to make it easy to use a diode
clipper at the VCA output for over-modulation
protection.
Other Dynamics Processor
Configurations
The same distinguishing features that make the
4316 so applicable to companding noise reduction
systems also qualify it for dynamics processors of
many other types. Because of its low-voltage supply
rails and micro power demand, the 4316 is espe-
cially applicable to dynamic processors that run
from battery power. The 4316 is versatile enough to
be used as the heart of a compressor, expander,
noise gate, AGC, de-esser, frequency-sensitive com-
pressor, and many other dynamics processors. It is
beyond the scope of this data sheet to provide spe-
cific advice about these many functional classes. But,
we refer interested readers to THAT’s many Design
Notes covering compressors, limiters, and other
dynamic processors. With minor modifications, most
of the teachings of those notes apply directly to the
4316.
Document 600177 Rev 00 Page 16 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Figure 25. 3:1 Decoder circuit.
RMS
U2
Op-Amp
R10
205k
R12
R14
6k19
10k2
100n
C10
C11
47p
3u3
C13
C14
10u
VCA
Input Output
U1
4316
VCC
C18
100n
C16
4u7
VCC
C17
22u
R15
2k49
C15
220p
R13C12
47n 1k13
R16
R18
R19
2k
C20
C19
U3
Op-Amp
33n 2k87
4k02
1n
IN
15 EC+
EC-
11
12
OUT 13
OUT
IN
25
CT
VCC FILT
GND VREF
8
97
6
4
Please check with THAT’s applications engineer-
ing department to see if your application has been
covered yet, and for personalized assistance with
specific designs.
Where to go from here
The design of compander systems and dynamics
processors is a very intricate art: witness the prolif-
eration of companding systems, and the many differ-
ent dynamics processors available in the market
today. In the applications section of this data sheet,
we offer a few examples of companders as a starting
point only. THAT Corporation’s applications engi-
neering department is ready to assist customers with
suggestions for tailoring and extending these basic
circuits to meet specific needs.
Document 600177 Rev 00 Page 17 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
-100
0
Encoder
Input
(dBV) (dBV)
Decoder
Output
Decoder
Input
Encoder
Output
-10
-9
-16
-23
-29
-36
-43
-20
-30
-40
-50
-60
-70
-80
-90
-100
Headroom Limit
Noise Floor
Figure 26. Butterfly diagram of the 3:1 compander.
-100-471.190-4347-100 -90-411.740-3941-90 -80-342.560-3634-80 -70-273.750-3327-70 -60-215.510-2921-60 -50-148.080-2614-50 -40-711.870-237-40 -30-117.42-191-30 -20625.57-16-6-20 -101337.53-13-13-10 01955.08-9-190
(
dBV
)
(
In dB
)
(
μ
A
)
(
dBV
)
(
In dB
)
(
dBV
)
Decoder
Out
Decoder
VCA
Gain
iin_RMS
Encoder
Out/
Decoder
In
Encoder
VCA
Gain
Encoder
In
Table 4. 3:1 compander transfer characteristic
s
a
t
f << 299Hz.
-100-323.750-4332-100 -90-255.500-3925-90 -80-188.080-3618-80 -70-1211.860-3312-70 -60-517.410-295-60 -50225.55-26-2-50 -40837.50-23-8-40 -301555.05-19-15-30 -202280.80-16-22-20 -1028118.59-13-28-10 035174.07-9-350
(
dBV
)
(
In dB
)
(
μ
A
)
(
dBV
)
(
In dB
)
(
dBV
)
Decoder
Out
Decoder
VCA
Gain
iin_RMS
Encoder
Out/
Decoder
In
Encoder
VCA
Gain
Encoder
In
Table 5. 3:1 compander transfer characteristics at
f >> 3kHz.
Package Information
The THAT 4316 pins are listed in Table 6. The
part is available in a 16-pin QSOP package as shown
in Figure 27.
Document 600177 Rev 00 Page 18 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
A
D
1
B
C
E
I
J
0-8º
G
H
ITEM MILLIMETERS INCHES
A4.80 - 4.98 0.189 - 0.196
B3.81 - 3.99 0.150 - 0.157
C5.79 - 6.20 0.228 - 0.244
D0.20 - 0.30 0.008 - 0.012
E0.635 BSC 0.025 BSC
G1.35 - 1.75 0.0532 - 0.0688
H0.10 - 0.25 0.004 - 0.010
I0.40 - 1.27 0.016 - 0.050
J0.19 - 0.25 0.0075 - 0.0098
Figure 27. Surface mount package QSOP-16.
Parameter Symbol Conditions Typ Units
Surface Mount Package
Type See below for pinout and dimensions 16 pin QSOP
Thermal Resistance
θ
JA QSOP package soldered to board 150 ºC/W
Soldering Reflow Profile JEDEC JESD22-A113-D (260 ºC)
Package Characteristics
Table 6. THAT 4316 pin assignments.
16No Internal Connection
15VCA IN
14No Internal Connection
13VCA OUT
12EC-
11EC+
10No Internal Connection
9VCC
8GND
7FILTER
6VREF
5RMS OUT
4CT
3No Internal Connection
2RMS IN
1No Internal Connection
Pin NumberPin Name
Document 600177 Rev 00 Page 19 of 20 THAT4316 Pre-Trimmed, Ver
y
Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; USA
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Revision History
—Initial release10/12/1200
PageChangesDateRevision
Document 600177 Rev 00 Page 20 of 20 THAT4316 Pre-Trimmed, Very Low-
V
oltage
Low-Power Analog Engine® IC
THAT Corporation; 45 Sumner Street; Milford, MA 01757-1656; US
A
Tel: +1 508 478 9200; Fax: +1 508 478 0990; Email: info@thatcorp.com; Web: www.thatcorp.com
Copyright © 2012, THAT Corporation
Notes
