LM13600 VCA, latest news

[jorgen.bergfors at idg.se]

I changed the "Serge" circuit according to JH's recommendations. It now has a 
12k resistor from pin 2 to +15V. From the inputs there are 22k resistors and a 
2k2 trimmer at the bottom. Wiper is connected to -15V.
The input resistor is reduced to 15k. The control input resistor is reduced to 
22k. That gives a maximum Iabc of 0,64 mA. The load resistor is 22k. The 
Darlington transistor is replaced by a TL072 op-amp. The gain is still a little 
less than unity at maximum control voltage.

The linearity of the output is still clearly not as good as the "normal" 
circuit, from the NS datasheet. 
I changed that a little too, to drive it harder. It now has a 15k input 
resistor, 22k control current resistor and 22k load resistor on the output. The 
resistor on pin 2 is 12k as before. This circuit has slightly over unity gain 
at "full tilt".
I can still drive this circuit with as large input signals as an op-amp can 
deliver (25V p-p), without significant unlinearity! The control voltage 
bleedthrough is 20 mV, when properly trimmed.

I measured the voltage on the inputs and on pin 2/15. In the "Serge" circuit 
they are -0,27 and +0,39 volts. In the "normal" they are 0,45 and 1,12 volts. 
The difference is 0,66 V and 0,67 V. So in reality it is the same.

It seems the unlinearity in the "Serge" is not very dependent on input 
amplitude. Even at moderate input levels the unlinearity is there. Conclusion 
is that the circuit from the NS datasheet really is as good as this chip will 
get. Next step is to compare it with other VCA circuits.

/Jorgen