CA3046 VCA

Using discrete transistors from a transistor array, this circuit avoids an OTA altogether. It uses one transistor as a Gilbert multiplier to predistort the signal, so that a larger signal can be fed into the circuit. The circuit is based on the one described in Modulus issue 5, that was provided by Chris Crosskey.
I have modfied it with a trimmer to adjust the DC offset and adjusted the input sensitivity and output gain to give some headroom and unity gain. This VCA has a linear response. A diode has been added on the control input, to block out negative voltages, which cause DC on the output. Because of the diode, the control caracreristics is unlinear below 1 volt.
The predistortion really works in this circuit. The distortion stays low up to a point where it suddenly increases dramatically. With the chosen resistor values, that point is well above normal signal levels.
Noise figures for this circuit is comparable to the SSM2024 and the LM13600 but signal bleedthrough is not as good. On the other hand, CV bleedthrough is lower than the LM13600, with proper trimming.

Noise & signal attenuation

Red = signal bleedthtrough at 0V CV. Blue = 10V CV, no signal. Green = 0V CV, no signal.

Distortion (THD+N) vs. input level

Frequency response

Test results

Dynamic range 10 V CV, no signal  82 dBr A
  0 V CV, no signal 107 dBr A
  0 V CV, 1kHz 10 V p-p in 86 dBr A
  0 V CV, 2 kHz 10 V p-pin  80 dBr A
  0 V CV, 10 kHz 10 V p-pin  63 dBr A
  Headroom (over 10V p-p)  6 dB
CV bleedthrough with careful trimming 5 mV

 

Summary

A clever design but has no real advantages over the LM13600.

+

Low CV bleedthrough

Fairly low distortion over a wide range of signal levels

Fairly high signal bleedthrough