Professional Documents
Culture Documents
THAT 4305 Datasheet
THAT 4305 Datasheet
Dynamics Processor IC
THAT 4305
FEATURES APPLICATIONS
y Pre-trimmed Blackmer® VCA & •Compressors & Limiters
RMS-level detector y Gates & Expanders
y Wide supply voltage range: y AGCs
±4.5V ~ ±16V y Line-operated dynamics processors
y De-Essers
y Low supply current:
3.5 mA typ. (±15V) y Duckers
y Mixers
y Wide dynamic range:
117 dB (VCA) y Level indicators
60 dB (RMS-level detector) y Companding noise reduction systems
Description
The THAT4305 is a single-chip Analog Engine® The 4305 was developed specifically for use in
optimized for low-cost applications. It incorporates a low-cost dynamics processors, drawing from THAT's
high-performance Blackmer® voltage-controlled long history and experience with such designs. Both
amplifier (VCA) and log-responding RMS-level VCA control ports and the detector input and output
sensor. The VCA and RMS detector are pre-trimmed are available for the designer to connect as s/he sees
at wafer stage to deliver low distortion without fit. As a result, the part is extremely flexible and can
further adjustment. be configured for a wide range of applications includ-
ing single- and multi-band companders, digital
Available in a small (QSOP) surface-mount
overload protectors, voltage-controlled faders, level
package, the 4305 is aimed at line-operated audio
indicators, etc.
applications such as compressor/limiters, gates, and
other dynamic processors. The part normally What really sets the 4305 apart from other
operates from a split supply voltage up to ±16Vdc, manufacturers’ offerings is the transparent sound of
drawing only 3.5mA at ±15V. This IC also works at its Blackmer VCA, coupled with its accurate true-
supply voltages as low as ±4.5V, making it useful in RMS level detector. This makes the IC useful in a
some battery-operated products as well. wide range of analog audio products.
SPECIFICATIONS
Electrical Characteristics 2
Parameter Symbol Conditions Min Typ Max Units
Power Supply
Gain-Control Constant EC+/Gain (dB) -60 dB < gain < +60 dB - 6.2 - mV/dB
Total Harmonic Distortion THD VIN= -5dBV, 1kHz, EC+ = EC- = 0V 0.07 0.15 %
Scale Factor Match to VCA -20 dB < VCA gain < +20 dB
1 μa< iIN(RMS) < 100 μA .95 1 1.05 -
1. If the devices are 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.
2. Unless otherwise noted, TA=25ºC, VCC=+15V, VEE= -15V.
3. Reference is to output offset with -60 dB VCA gain.
U1B
R2 THAT4305
+15V RMS In C1
C5 20k R1 RMS Out
R4 C3 RMS In
5k1 2 RMS Out
11 5
6k8 10u CT
100p Ec+ 9
Vcc 22p NPO 4 R5
VCA In C2 +15V
R3 13
CTIME 2k
VCA In VCA Out VCA Out
20k 15 C4 10u
Gnd
10u Vee
Ec- 6 U2
8 10u
12
Control Voltage
-15V U1A
THAT4305
Theory of Operation
The THAT 4305 Dynamics Processor combines VCAs and the 2252 RMS Level Detector. Theory of the
THAT Corporation's proven exponentially controlled interconnection of exponentially controlled VCAs and
Blackmer® Voltage-Controlled Amplifier (VCA) and log-responding level detectors is covered in THAT
log-responding RMS-Level Detector building blocks Corporation's application note AN101A, The Mathe-
in a small package optimized for low cost designs. matics of Log-Based Dynamic Processors.
The part is fabricated using a proprietary, fully
complementary, dielectric-isolation process. This The VCA - in Brief
process produces very high-quality bipolar transis-
tors (both NPNs and PNPs) with unusually low collec- The VCA in the 4305 is based on THAT Corpora-
tor- substrate capacitances. The 4305 takes tion's highly successful complementary log-antilog
advantage of these devices to deliver wide bandwidth gain cell topology (the Blackmer® VCA) as used in
and excellent audio performance while consuming THAT 2180-Series IC VCAs. VCA symmetry is
very low current and operating over a wide range of trimmed during wafer probe for minimum
power supply voltages. distortion. No external adjustment is allowed.
For details of the theory of operation of the VCA Input signals are currents in the VCA's VCAIN pin
and RMS Detector, we refer the interested reader to (pin 15). This pin is a virtual ground with a small dc
THAT Corporation's data sheets on the 2180-Series offset, so in normal operation an input voltage is
converted to input current via an appropriately sized 2. The SYM control port (similar to that on the
resistor (R3 in Figure 2). Because the dc current 2180 VCA) is not brought out to an external pin; it is
associated with dc offsets present at the input pin driven from an internally trimmed current generator.
plus any dc offset in the preceding stages will be
modulated by gain changes (thereby becoming
audible as thumps), the input pin is normally 3. The control-voltage constant is approximately
ac-coupled. This blocks such offset currents and 6.2 mV/dB, due primarily to the higher internal
reduces dc offset variation with gain. operating temperature of the 4305 compared to that
of the 2180 Series.
The VCA output signal, VCAOUT (pin 13), is also a
current, inverted with respect to the input current. In The RMS Detector - in Brief
normal operation, the output current is converted to
a voltage via an external op-amp, where the current- The 4305's detector computes RMS level by
to-voltage conversion ratio is determined by the rectifying input current signals, converting the recti-
feedback resistor connected between the op-amp's fied current to a logarithmic voltage, and applying
output and its inverting input (R2 in Figure 2). The that voltage to a log-domain filter. The output signal
resulting signal path through the VCA plus op-amp is is a dc voltage proportional to the decibel-level of the
noninverting. RMS value of the input signal current. Some ac
component (at twice the input frequency plus higher-
The VCA gain is controlled by the voltage applied order even harmonics) remains superimposed on the
between EC+ (pin 11) and EC- (pin 12). Note that any dc output. The ac signal is attenuated by a log
unused control port should be connected to ground domain filter, which constitutes a single-pole rolloff
(as EC+ is in Figure 2). The gain (in decibels) is with cutoff determined by an external capacitor (C4
proportional to (EC+ -EC-). The constant of propor- in Figure 2).
tionality is 6.2 mV/dB for the voltage at EC+(relative
to EC-). Note that neither EC+ or EC- should be driven The rectifier is balanced to within ±3 dB, so a
more than ±0.6 V away from ground. small amount of fundamental (and higher odd-order
harmonics) ripple can be present at the detector
The VCA's noise performance varies with gain in output. By design, this ripple contributes less total
a predictable way, but due to the way internal bias ripple than the even-order products that are
currents vary with gain, noise at the output is not naturally and inevitably present at the output of a
strictly the product of a static input noise times the perfectly balanced detector.
voltage gain commanded. At large attenuation, the
noise floor is usually limited by the input noise of the As in the VCA, input signals are currents to the
output op-amp and its feedback resistor. At 0 dB RMSIN pin (pin 2). This input is a virtual ground, so a
gain, the noise floor of ~ -97.5 dBV is the result of resistor (R1 in Figure 2) is normally used to convert
the VCA’s output noise current, converted to a input voltages to the desired current. The level detec-
voltage by the typical 20k I-V converter resistor (R2 tor is capable of accurately resolving signals well
in Figure 2). In the vicinity of 0 dB gain, the noise below 10 mV (with a 5 kΩ input resistor). However,
increases more slowly than the gain: approximately if the detector is to accurately track such low-level
5 dB noise increase for every 10 dB gain increase. signals, ac coupling (C1 in Figure 2) is required to
Finally, as gain approaches 30 dB, output noise prevent dc offsets from causing a dc current to flow
begins to increase directly with gain. in the detector’s input, which would obscure
low-level ac signal currents.
While the 4305's VCA circuitry is very similar to
that of the THAT 2180 Series VCAs, there are several The log-domain filter cutoff frequency is usually
important differences, as follows. placed well below the frequency range of interest. For
an audio-band detector, a typical value would be
1. Supply current for the 4305 VCA depends on 5 Hz, or a 32 ms time constant (τ). The filter's time
the supply voltage. At ±5 V, approximately 800 μA is constant is determined by an external timing capaci-
available for the sum of input and output signal tor (CTIME) attached to the CT pin (pin 4), and an
currents. This increases to about 1.8 mA at ±15 V. internal current source (IT) connected to CT. The
(Compare this to ~1.8 mA for a 2180 Series VCA current source is internally fixed at 7.5 μA. The
when biased as recommended.)
resulting time constant in seconds is approximately 2. The time constant of the 4305's RMS detector
equal to 3467 times the value of the timing capacitor is determined by the combination of an external
in Farads. Note that, as a result of the mathematics capacitor CTIME and an internal current source. The
of RMS detection, the attack and release time internal current source is set to about 7.5 μA. A
constants are fixed in their relationship to each resistor is not normally connected directly to the CT
other. pin on the 4305.
The RMS detector is capable of driving large 3. The 0 dB reference point, or level match, is
spikes of current into CTIME, particularly when the also set to approximately 7.5 μA. However, as in the
audio signal input to the RMS detector increases 2252, the level match will be affected by any
suddenly. This current is drawn from VCC (pin 9), additional currents drawn from the CT pin.
fed through CTIME at pin 4, and returns to the power
supply through the ground end of CTIME. If not Compressor (or Limiter) Configurations
handled properly through layout and bypassing,
these currents can mix with the audio in the circuit’s The 4305 provides the two essential building
ground structure with unpredictable and undesirable blocks required for a wide variety of dynamics
results. As noted in the Applications section, local processing applications. The part may be configured
bypassing from the VCC pin to the ground end of CTIME into practically any type of dynamics processor
is strongly recommended in order to keep these system.
currents out of the ground structure of the circuit
(see C4 in Figure 2.) Perhaps the most common application for the
4305 is as a compressor or limiter. These circuits
The dc output of the detector is scaled with the are intended to reduce gain above some determined
same constant of proportionality as the VCA gain signal level in order to prevent subsequent stages
control: 6.2 mV/dB. The detector's 0 dB reference from being overloaded by too high a signal.
(iin0, the input current which causes the detector's Compressors generally have low to moderate
output to equal 0V), is trimmed during wafer probe compression ratios, while limiters have high ratios.
to equal approximately 7.5 μA. The RMS detector In such applications, the signal path has static gain
output stage is capable of sinking or sourcing so long as the input signal remains below some
125 μA. It is also capable of driving up to 150 pF of threshold, but gain is reduced when the signal rises
capacitance. above the threshold. Compression ratio is defined as
the number of dB the input signal increases for a
Frequency response of the detector extends 1 dB increase in output signal.
across the audio band for a wide range of input
signal levels. Note, however, that it does fall off at Feedforward Topologies
high frequencies at low signal levels like THAT’s
other RMS detectors. To make a compressor or limiter with a 4305,
typically, the input signal is applied to both the VCA
Differences between the 4305's RMS level detec- and the RMS detector. The RMS output signal is fed
tor circuitry and that of the THAT 2252 RMS detec- forward to the VCA's negative control port (EC-) via a
tor include the following. dc-coupled op-amp based stage. This stage has gain
above some dc level (the threshold), and no trans-
1. The rectifier in the 4305 RMS Detector is mission below that level. This path, called the
internally balanced by design, and cannot be "sidechain," — from detector output to VCA control
balanced via an external control. The 4305 will port — determines the compression behavior of the
typically balance positive and negative halves of the circuit. As signal level rises, the dc voltage at the
input signal within 10% but in extreme cases the RMS' output rises. Once the dc level exceeds the
mismatch may reach +40% or -30% (±3dB). threshold, the rms output signal is transmitted
However, even such extreme-seeming mismatches through the sidechain and presented to the VCA
will not significantly increase ripple-induced distor- control port, lowering the gain to signals passing
tion in dynamics processors over that caused by through the VCA. As a result, the output signal level
balanced signal ripple alone. is reduced, or compressed, relative to rising input
signal levels.
Varying the threshold setting of the sidechain producing an expander. This is typically applied
will vary the point at which compression begins. below a threshold (so, the threshold detector’s polar-
Varying the gain between the RMS output and the ity is reversed from that of a compressor) to reduce
VCA control input varies the compression ratio. noise or crosstalk during pauses in program
material. This technique has long been used for
Feedforward compressor topologies are "cleaning up" individual drum tracks to reduce rever-
especially versatile because they cannot become beration, interference from microphones picking up
unstable due to oscillation in the control loop. Unity adjacent drum sounds, and alter the attack/decay
gain in the sidechain produces infinite compression characteristic of individual drum sounds.
(where the output remains constant regardless of
increases in the input signal). With feedforward, Practical gates usually require very fast attack
negative compression ratios are easily achievable. times, and carefully programmable release times. In
(Negative compression occurs when the output signal a 4305, this is best accomplished by using the RMS
decreases as the input signal increases.) This detector as a log rectifier with very short time
approximates the effect of playing music backwards, constants, and following the detector output with a
since the attack is suppressed and the release is time-constant stage that applies the desired attack
increased in volume. and release behavior. This alters the 4305 detector’s
natural response characteristics to peak, rather than
Many other variations of the feedforward rms, time constants. We intend to produce an appli-
concept are possible. These include implementing cation note showing examples of these circuits. Until
more than one threshold, different ratios, additional that is available, see DN 100, which shows a noise
time constants, ac-coupling of some (or all) of the gate application using THAT's 4301 Analog Engine®.
detector output signal, and many more. See AN101A,
The Mathematics of Log-Based Dynamic Noise Reduction (Compander)
Processors, for more details of how the sidechain
Configurations
gain determines compression ratios.
An additional application of the 4305 is for
Feedback Topologies
noise reduction systems. In these applications, one
Analog Engine is configured for use as a compressor
An alternative configuration for compressor/
to condition audio signals before feeding them into a
limiter design is to feed the output signal into the
noisy channel. A second Analog Engine, configured
RMS detector. The RMS output is fed back (dc-
as an expander, is located at the receiver end of the
coupled) to the VCA's negative control port to reduce
noisy channel. Most commonly, the
signal levels. Similarly as with a feedforward
compression/expansion ratio is modest (e.g. 2:1:2)
designs, a threshold in the sidechain serves to stop
and is linearly applied across the entire signal
the compression action at low signal levels.
dynamic range.
The feedback topology behaves somewhat differ-
During low-level audio passages, the compressor
ently from feedforward. First, reaching infinite
increases signal levels, bringing them up above the
compression requires infinite gain in the feedback
noise floor of the channel. At the receiving end, the
loop from RMS output to VCA control port. Of
expander reduces the signal back to its original level,
course, infinite gain is impossible, so practical
in the process attenuating channel noise.
feedback compressors are usually limited to ratios
no greater than 20 or so. Additionally, the gain in
During high-level audio passages, the compres-
the feedback loop alters the effective time constant of
sor decreases signal levels, reducing them to fit
the detector, shortening the attack as the ratio
within the headroom limits of the channel. The
becomes higher. This may or may not be
expander increases the signal back to its original
appropriate, depending on the desired effect.
level. While the channel noise may be increased by
this action, in a well-designed compander, at such
Expander (Gate) Configurations times the noise floor will be masked by the high-level
signal.
By changing the sign of the sidechain in a
feedforward compressor, it is possible to arrange
signal gain to decrease along with signal level, thus
expand (decode) that signal using the 4305 in a rack- can make a feedforward compressor/limiter by having
mount, line-operated receiver. the gain of the sidechain vary from zero to one.
The Mathematics of Log-Based Dynamics Note that if k>1, the compression ratio becomes
Processors negative. Negative compression results with 1<k<2.
At first, the logarithmic output of the RMS detec- The Feedback Compressor
tor and the exponential control ports of the VCA can We can use the same approach for a feedback
be intimidating for designers unfamiliar with THAT compressor. Figure 4 shows a simplified theoretical
Corporation's offerings. However, in fact, these model of a feedback compressor. By inspection,
characteristics make developing audio processors
easy once a designer understands the concepts Out dB = IndB + GdB , and
involved. As noted earlier, AN101A: The Mathemat-
ics of Log-Based Dynamics Processors, discusses GdB = −k $ Out dB . Therefore,
these concepts in some detail. The following discus-
sion draws heavily from that application note. Out dB = IndB − k $ Out dB, and
limit effectively enough that infinite gain is not In the equations so far, we have made the
required. implicit assumption that the decibel reference level
everywhere is that of the rms-level detector. This
Log-Based Expanders assumption simplifies the math, but it may not corre-
spond to real-world reference levels such as 1V rms
Similarly, for the feedforward expander shown (0 dBV) or 0.775 VRMS (0 dBu). Additionally, it is
in Figure 5, we can state that possible to offset the VCA's inherent behavior of
producing unity (0 dB) current gain at 0 mV control
voltage (EC+-EC-) by selecting asymmetrical voltage-to-
current and current-to-voltage converting resistors
(R3 and R2, respectively, in Figure 2).
IndB G
dB Out dB Figure 6 allows for a VCA voltage gain offset of
AdB, as well as an offset (LMdB) to vary the "level
RMS
match" point of the RMS detector. Using similar “log
Out dB math” from AN101A, we can state for Figure 6:
-k
Out dB = IndB + GdB + AdB, where GdB is the
VCA's control port gain in dB, and AdB is any static
gain or attenuation (in dB) applied to the signal.
Figure 4. Simplified feedback compressor,
conceptual diagram. We can also state that:
Out dB = IndB + GdB , and that GdB = −k (Out dB − L.M. dB), where L.M. is a
varying dc voltage intended to change the system’s
GdB = k $ IndB . zero dB reference point (often referred to as level
match point) from that of the detector, and k is the
Note that the sign of k makes this an expander gain of the sidechain.
in which gain GdB decreases as input signal level
IndB decreases. Combining these equations: Substituting the second equation into the first,
In dB +k $ L.M. dB +A dB
Rearranging yields OutdB = (1+k) .
Out dB
In dB = 1 + k = E.R. Similarly, for the expander shown in Figure 7,
we can state that:
This is the expansion ratio.
L.M.dB
Figure 5. Simplified feedforward expander, Figure 6. Feedback compressor with level match.
conceptual diagram
20
LM= -40
0
-40
RMS
dB Out
IndB -60
k=0
+ -80 k=1
k=2
Σ k -100
k=3
-
-120
-120 -100 -80 -60 -40 -20 0 20
L.M.dB dB In
Figure 7. Feedforward expander with level match. Figure 8. Compressor input-output transfer
characteristics with varying k.
GdB = k(IndB − L.M. dB). Thus:
k=1
Out dB = (1 + k)IndB − k $ L.M. dB − AdB 0
-20
In both Figures 6 and 7, we show two ways to
dB Out
adjust the point at which input and output levels of -40
the compressor or expander become equal. These L.M.= -10
are a) adjusting the 0 dB reference voltage level of the -60 L.M.= -20
L.M.= -30
RMS detector by setting its voltage-to current -80
L.M.= -40
20
Figure 8 shows the behavior of the compressor
-20
of Figure 6 with varying sidechain gain k for a
-60 k=0
constant AdB and LMdB. Note that as k increases, the k=1
k=2
compression ratio increases, and the dynamic range -100 k=3
of output levels decreases. Low input signal levels -140
-60 -40 -20 0
(including noise) are increased, and high input signal dB In
levels are decreased. Figure 9 shows how the same
Figure 10. Expander input-output transfer
compressor behaves with varying LMdB, but fixed AdB characteristics with varying k.
and k. Note that as LMdB is reduced, all output levels
decrease, and vice-versa.
40
20
Figures 10 and 11 show how the input-output
0
transfer characteristics of the expander of Figure 7
-20
change while varying k (Fig 10) and LM (Fig 11). In
dB Out
-40
each case, all other parameters remain fixed. -60
L.M.= -10
L.M.= -20
-80 L.M.= -30
L.M.= -40
-100
-120
-60 -50 -40 -30 -20 -10 0
dB In
Substituting yields 0
-10
-20
Out dB = In dB − k (In dB − L.M. − T.A.) + A V.
dB Out
-30
-40
Out
-50
For input signals below the level set by the In
-60
threshold setting, the signal at the output of the ideal -70
diode threshold is 0 (dB), so -80
-90
-100
G dB = A dB , thus, -100 -80 -60 -40 -20 0 20
dB In
Out dB = In dB + A dB .
Figure 13. Transfer function of feedforward compressor
0
IndB GdB Out dB
RMS
dB Out
Figure 14. Feedback compressor with threshold (T), Figure 15. Transfer function of feedback compressor
gain (A), and ratio (k) adjustments
Applications
In this datasheet, we will show detailed circuits for about 750 μA. At 0 dB gain, this will cause the same
the 4305 in a relatively simple above-threshold current at the VCA output, which is converted back
compressor, and in two simple companding systems. to 15 V by U2 and R13 (also 20 kΩ). C2 provides AC
As mentioned above, many other configurations of coupling, required to block any DC currents that
the 4305 are possible. THAT intends to publish might otherwise flow into the VCA input. This
additional circuits in forthcoming applications notes. prevents changes in gain from modulating this
Please check with THAT's applications engineering current, which could produce audible "thumps".
department to see if your application has been
covered yet, and for personalized assistance with The compensation circuit of R28 and C16 is
specific designs. required for stability. The VCA must “see” a source
impedance no greater than 5 kΩ above 1 MHz. R28
Feedforward Compressor/Limiter in parallel with R14 accomplishes this. C16 is
chosen to prevent the added noise gain of the lower
The circuit in Figure 16 shows a typical hard- source impedance from increasing noise within the
knee, feedforward compressor/limiter. In addition to audio band.
compression ratio, the sidechain includes controls
for threshold and make-up gain as well. Note that such compensation is unnecessary
when the voltage-to-current converting resistor (R14)
The Signal Path is 5 kΩ or less. For example, if the input signal were
The input of the VCA (pin 15) is a virtual limited to lower voltages, the input voltage-to-current
ground, and R14 converts the input signal into a converting resistor (R14) could be reduced in value,
current flowing into the VCA. The maximum total possibly eliminating the need for R28 and C16.
signal current, (IIN + IOUT) is 1.8 mA with ±15 V
supplies, so R14 is sized to keep the maximum U2, along with C4 and R13 forms a trans-
current at unity gain to below this level. With peak impedance amplifier that converts the VCA's output
input voltage swing limited by the ±15 V supply rails, current into a voltage. C4 prevents the VCA's output
the 20 kΩ resistor at R14 limits maximum iIN to capacitance from destabilizing the op-amp in this
configuration.
+15V R13
C16 20k
R28 11 U1A C4
9 THAT4305
100p 6k2
EC+ 22p NPO
Vcc
In C2
R14 13 Out
VCA In VCA Out
20k 15 +15V
10u Gnd
Vee U2
-
EC- 6
8
-20 dB R12
12 10k
C6 R8 R18
-15V 22p Gain
620k 5k1
+20 dB C5
Increase
U1B
THAT4305 -15V
C1 100n
R1 2 5 R17 R2 R9 R3
RMS In
33k RMS Out 5k1 10k 1k43 10k
10u CT D2 INF:1 CR
4 D1 1N4148 U4
Compression
Ratio R11
CTIME 1N4148
+15V 10u +15V 10k
C13 C15 1:1 CR
The Sidechain the threshold over ±30 dB, we can calculate the
required value of R7 as follows:
As noted earlier, for most effects compressors, it
is best not to linearly compress the entire dynamic 15V( 10k
R7 )
V = 30dB .
range of the audio signal. To this end, in the 0.0124 dB
sidechain of figure 16, we have added a threshold
amplifier with ±30 dB of threshold adjustment. This We can rearrange this to be
restricts compression to signals above the threshold,
passing those below threshold without any change.
10k
R7 = 30dB
15V $ 0.0124 dB
V
, and therefore,
Additionally, we added a compression ratio control
adjustable from 1:1 to ∞:1. Finally, we've added a R7 = 403.2k
means to apply up to ±20 dB of static, or "make-up"
gain. We chose 430 kΩ for R7.
The input signal is fed into the RMS detector U4 is a variable-gain inverter that serves to
through C1 and R1. Like the VCA, the input to the buffer the VCA's control port, ensuring a low- imped-
detector is AC coupled. This prevents any DC ance drive at that point. (High impedances, even as
current flowing into the detector's input from being little as 50 to 100 ohms, will increase VCA distortion
measured by the detector. If unchecked, such offsets at high signal levels.) Above threshold, when U4's
would limit resolution at low levels, producing an gain is -½, the net gain of the sidechain (from RMS
artificial "floor" to the detector's low-level response. output to VCA control input) is unity, and the
compression ratio is ∞:1. The network of R3, R9,
As previously discussed (in the theory section), and R11 in conjunction with R18, allows the gain of
the output of the detector is proportional to the log of U4 to vary from 0 to -½, and simultaneously shapes
the RMS level of the input voltage. The output of the the (linear) pot's response so that 50% rotation
detector will be approximately zero volts at the "zero results in 4:1 compression. 4:1 ratio at 50% rotation
dB reference level" -- the point at which the RMS is often considered a useful target.
value of the input current equals the timing current
(set at 7.5 μA for the 4305). We have chosen a value Finally, R8 and R12 provide the means for
of -10 dBu, (245 mVRMS) for the zero dB reference adding static, or "make-up" gain. The control-voltage
level. The required input resistor can be calculated sensitivity at the output of U4 is 6.2 mV/dB.
as Therefore,
245mVRMS
R= = 32.6k
l 33k
. 15V( 5.1k
R8 )
7.5A
V
0.0062 dB
= 20dB .
Inverting threshold amplifier (U3) provides gain
of approximately -2 to the detector output signal We can rearrange this to be
above threshold, and zero gain (AV=0) to signals
below threshold. The change in gain is accomplished
5.1k
R8 = 20dB
15V $ 0.0062 dB
V
, and therefore,
by D1 and D2, which allow negative-going output
signals to pass but block positive-going ones.
Because U3 is configured to invert, positive-going R8 = 20dB
5k
V
$0.0062 dB
= 625k
15V
signals at the RMS output (indicating increasing ac
input levels) are passed onwards, while negative- We've chosen 620 kΩ for R8 since it is the
going RMS outputs are blocked. By feeding variable nearest 5% value.
dc into this stage via R7 and the threshold pot R10,
we can vary the point at which RMS output signals The signals in the sidechain, and at the output of
begin to be passed through to the threshold amplifier U4, are generally relatively slow moving, so the
stage’s output (at the junction of D2 and R2.) sidechain does not usually require wide bandwidth.
Furthermore, noise on the VCA control port can
The scaling at the output of the detector is modulate the VCA signal, thus adding noise to the
+6.2 mV/dB, but because R2 is approximately twice signal path. Accordingly, we added C5 in order
R17, the threshold amplifier (U3) has a gain of -2, so reduce the noise gain of U4 at high frequencies,
the scaling at its output is -12.4 mV/dB. To swing which slightly reduces the high-frequency noise floor
of the entire circuit. C5 may be omitted for and a 5.1 kΩ resistor (R3) along with a 220 pF
non-critical applications. capacitor (C4) comprise the compensation network
is required to keep the VCA's input amplifier stable
As described in the Theory of Operation section for all gains.
("The RMS Detector - In Brief"), the RMS detector is
capable of driving large spikes of current into the Since the RMS detector output is tied directly to
averaging capacitor CTIME. To prevent these currents the VCA's EC-, the compression ratio will be 2:1.
from upsetting circuit grounds, it is necessary to Note that the use of the negative-sense control port,
bypass VCC to a point very near the grounded end of EC-, makes this circuit a compressor. The RMS
CTIME with a capacitor equal to or greater than the detector timing capacitor is set for a release rate of
value of CTIME. This is C13 in Figure 16. The -125 dB per second by using a value of 10 μF.
grounded ends of these two capacitors should be
connected together before being tied to the rest of the C5 serves to keep high currents through the
ground system. Doing so will ensure that the current timing capacitor (C3) from upsetting circuit grounds
spikes flow within the local loop consisting of the two as described in the previous section.
capacitors, and stay out of the ground system.
The output of the RMS detector is zero volts
when the RMS input current is equal to the timing
Companding Systems
current (internally set to ~7.5 μA). A voltage level of
The Encoder -28.5 dBV was chosen as the desired zero dB refer-
ence. The RMS detector's input resistance can be
Figure 17 shows the 4305 configured as a calculated as:
simple 2:1 encoder or feedback compressor. The −28.5
encoder in a companding system is positioned before R RMS In = 10 20
7.5A l 5.1k
.
the noisy channel (wireless link, storage system,
etc.). The static gain of this circuit is 1, or zero dB, This value also applies to the decoder.
C5
22p
U1A
R2
11 THAT4305
EC+ 20k
INPUT C1
R1 15 13 U2A
VCA In VCA Out OUTPUT
20k
22u
R3 EC- Op-Amp
5k1 12
C4
220p C2
R4
5 RMS In
RMS Out 2 5k1
U1B CT 22u
THAT4305 4
+15V
C3 C5
10u
22u
U3B
INPUT C12 THAT4305
R10 C15
RMS In 5
5k1 2 RMS Out
22u
CT 47p
+15V 4 U3A
C13
C14 10u THAT4305 R7
11 20k
22u
EC+ U4A
C11 R8 15 13 2 OUTPUT
VCA In VCA Out 1
20k 3
22u
R13 EC-
5k1 Op-Amp
12
C21
220p
Table 3 shows the transfer characteristics of this -30 1 -29 0.0069 -1 -30
companding system. The columns labeled Encoder -40 6 -34 0.0039 -6 -40
VCA Gain, Encoder Out, Decoder VCA Gain, and -50 11 -39 0.0022 11 -50
Decoder Out use the equtions derived previously in
-60 16 -44 0.0012 16 -60
the Theory sub-section entitled "The Mathematics of
-70 21 49 0.0007 -21 -70
Log Based Companding Systems". The values in the
column labeled IRMS In are derived using the equation: -80 26 -54 0.0004 -26 -80
-90 31 -59 0.0002 -31 -90
EncoderOut )
10 (
I RMS In =
20
-100 36 -64 0.0001 -36 -100
R RMS In
Table 3. 2:1 compander transfer characteristics
Figure 19 tracks signal levels through the given signal level, high-frequency signals are lowered
encoder and decoder of Figures 17 and 18. The in level by the VCA more than low-frequency signals.
encoder reduces the dynamic range at its input by a
factor of 2, compressing 120 dB into 60 dB. The As an additional enhancement, we have included
decoder expands this dynamic range back to track a means to truncate the RMS detector’s low-level
that of the encoder's input signal. response. This improves low level tracking between
different detectors by forcing each detector to
"bottom out" at a predetermined level, eliminating the
Compression Expansion effects of different low-level behavior from one detec-
Process Process
tor to the next.
20
System Performance
0 The compander shown in Figures 20 and 21
20
0 implements all of the aforementioned improvements.
-20
-20 Assuming no change in VCA gain (GdB), the
-40 pre-emphasis network of R3 and C7 produces
dB
-40
-60
~20 dB of signal-path pre-emphasis starting at
-80
-60 ~2 kHz and stopping at ~19 kHz.
-100
−50dBV
i R6 = RMSOut
, and VIN = 10 ( 20 )
= 3.16mVRMS . Therefore,
R6
(−23.6−(−50))$0.0062
0.00316V
= , And
i R12 = RMSOut
R12
8.87k
R6
(−23.6−(−50))$0.0062
We'll choose a point about 6 dB above the R6 = 0.00316V = 459k
l 464k
8.87k
encoder output level (-56 dBV) corresponding with
an input level of -100 dBV. The same is true for R12.
C5
Signal Path
Pre-emphasis
22p
C7 R3 U1A
THAT4305 R2
2k32 11
U2 3n3 EC+ 84k5
Encode In
C1
R1 15 13 U3 Encode Out
VCA In VCA Out
1u 21k0
Op-Amp EC- Op-Amp
12
R5 C8
R6
464k 4k02 3n3
RMS
Pre-emphasis
RMS Out
5 C2
RMS In 2 R4
Vcc U1B CT 8k87
THAT4305 1u
C4 4
22u
C3
10u
RMS Pre-emphasis
C18 R11 R12
C21
220p
Encoder
wide range of dynamics processor configurations.
Encode Decoder These include companding noise reduction systems
Encoder Out/ Decoder
VCA IRMS In VCA
In Decoder Out with ratios other (higher or lower) than 2:1:2, multi-
Gain Gain
In
band companders, etc. The 4305 is versatile enough
(dBV) (In dB) (dBV) (mA) (In dB) (dBV)
to be used as the heart of a compressor, expander,
20 -28 -4 0.1841 28 20 noise gate, AGC, de-esser, frequency-sensitive
10 -23 -1 0.1035 23 10 compressor, and many other dynamics processors.
0 -18 -6 0.0582 18 0 It is beyond the scope of this data sheet to provide
-10 -13 -11 0.0327 13 -10
specific advice about any of these functional classes.
We refer the interested reader to THAT's applications
-20 -8 -16 0.0184 8 -20
notebooks volumes 1 and 2, which contain many
-30 -3 -21 0.0104 3 -30 circuits based on THAT's other VCAs and RMS level
-40 2 -26 0.0058 -2 -40 detectors, but are largely applicable to the 4305 with
-50 7 -31 0.0033 -7 -50 only minor variations. Of course, look for more
-60 12 -36 0.0018 -12 -60 applications information aimed specifically at the
4305 in the future.
-70 17 -41 0.0010 -17 -70
-80 22 -46 0.0006 -22 -80
Compression Expansion
-90 27 -51 0.0003 -27 -90
Process Process
-100 32 -56 0.0002 -32 -100
20
Table 4. Hi-fi compander transfer characteristics
(applies to low frequencies only) 20
0
0
Table 4 shows the transfer characteristics of this -20
-20
companding system (neglecting the effects of R6 and -40
-60
dB
Figure 22 tracks signal levels through the Figure 22. Hi-fi compander transfer characteristics
encoder and decoder of Figures 20 and 21. The
compression and expansion ratios here are the same
as those of the previous circuits, but the frequency Closing Thoughts
shaping afforded by signal pre- and de-emphasis and
detector pre-emphasis make this a superior sound- The design of dynamics processors and
ing system. companding systems is a very intricate art: witness
the proliferation of dynamics processors available in
In this application, the VCA gain ranges over the market today. Many of these are based on
about ±30 dB, which is well within specification, as THAT's VCAs and level detectors, yet they all have
is the RMS detector input current. individual sonic characteristics. In the applications
section of this data sheet, we have offered a few
Other Dynamics Processor Configurations examples only as starting points. THAT Corpora-
tion's applications engineering department is ready
We have said before that the building blocks to assist customers with suggestions for tailoring and
contained within the 4305 are applicable to a very extending these basic circuits to meet specific needs.
The THAT 4305 is available in a 16-pin QSOP specified in the RoHS directive. For more informa-
package. The package dimensions are shown in tion, including MDDS forms which disclose the
Figure 23 below, while the pinout is given in Table 1 substances contained in our ICs and their packaging,
on page 1. please visit: www.thatcorp.com/RoHShome.html.
The 4305 is available in a lead-free, "green" The package has been qualified using reflow
package. The lead frame is copper, plated with temperatures as high as 260°C for 10 seconds. This
successive layers of nickel palladium, and gold. This makes them suitable for use in a 100% tin solder
approach makes it possible to solder these devices process. Furthermore, the 4305 has been qualified to
using lead-free and lead-bearing solders. The plastic a JEDEC moisture sensitivity level of MSL1. No
mold compound, and the material in which the parts special humidity precautions are required prior to
are packaged, contains no hazardous substances as flow soldering the parts.
Package Characteristics
Parameter Symbol Conditions Min Typ Max Units
Package Style See Fig. 23 for dimensions 16 Pin QSOP
Moisture Sensitivity Level MSL Above referenced JEDEC soldering profile MSL-1
D A
E
B
C G
J H
0-8º
I
Revision History
Revision ECO Date Changes Page
00 — 10/01/06 Released. —
02 2935 08/05/15 Corrected labels in Fig 15. Updated Package Characteristics. 11, 19