13600 (was VCA chips)

[jorgen.bergfors at idg.se]

Hello Juergen and all others on the list.
I built the LM13600 dual VCA last night, with one VCA using the "normal" 
circuit as in Barry Klein's book and the other using Serge's idea (see below). 

The "normal" circuit use a 1k trimmer between the inputs. The wiper is 
connected to ground. Signal is input via a 33 k resistor to the non-inverting 
input. The linearization input has a 12k resistor to +12V. The load resistor on 
the output was 33k but had to be increased to 60k4 to get unity gain. The 
control circuit is copied from the Formant CA3080 VCA (it's the logarithmic 
one).

The "Serge" circuit has a 30k resistor to -15V on each input. The signal is fed 
to the non-inverting input via a 33k resistor. The linearization input has a 
15k resistor to +15V. Offset adjustment is from a 100k trimmer between + and - 
15 V via a 1M resistor to the invering input (I didn't put the trimmer at the 
bottom as JH suggest, because I got his mail this morning).Output and control 
stage is the same as above.

Looking at the output on an oscilloscope it is clear that the "normal" circuit 
actually has noticably better linearity than the "Serge". This wasn't what I 
had expected!
If I put the 'scope in differential mode and monitor the input and the output, 
the curve looks a bit like a bridge. It has  peaks where the sawtooth wave's 
steep edge is, and sags between. 
The "normal" circuit can accept signals up to 25V p-p vithout much linearity 
error. The "Serge" sags noticeably even at 10V p-p input.

Have I done anything wrong with the Serge circuit? Any ideas?

Another problem is that the output from the Darlington transistor has a -1,2V 
offset. This is regardless of the control voltage and offset adjustment on the 
input. Is there any way to fix this without adding an op-amp?
If replace the Darlington with a jfet op-amp the offset problem goes away. 
Maybe it is best to forget about the Darlington transistor that is built into 
the 13600?

I'm going to do proper distortion measurements on an Audio Precision System Two 
later, but it hasn't arrived yet.

/Jorgen






MIME:jhaible at metronet.de on 98-09-03 03.56.36
To:  Jorgen Bergfors at IDGSE
cc:  
Subject: Re: 13600 (was VCA chips) 


>On it you have a 1k trimmer between the inputs (pin 3 and 4 or 14 and 
13).
The 
>wiper is connected to ground. The trimmer is for offset adjustment, I
suppose. 
>The bias resistor (to pin 2 and 15) is 12k. 
>Should I remove the trimmer and use 30k resistors to -15V instead? Or 
can
I use 
>a 50 k trimmer and tie the wiper to -15V? Or maybe a smaller trimmer 
(say
5 k) 
>and a 27k resistor to -15V from the wiper? What do you think is best?

IMO it's best to leave the diodes alone for audio applications. A 
gradually
increasing
distortion is more pleasant than a sudden one, unless you get a *much*
higher SNR
with linearisation, which is not the case for the 13700.

But this is only 2nd hand information (though an article from Serge is 
not
exactly
hearsay ...), so measurements would be nice.
With 12k at pin 2 I'd suggest two 24k resistors and a 1k trimmer at the
bottom,
for a start.

>Because the 13600 is a dual OTA, I can build one VCA using the standard
circuit 
>and another one using the circuit mentioned here. Then I can use the 
Audio

>Precision to measure distortion etc, and see if there is a difference.

Good idea!

JH.