AW: LM13600 VCO - latest report

[jhaible]

>I checked the negative supply on your request. It has less than 1 mV
ripple, as 
>seen on the 'scope. So I don't think that is the problem. No, my guess is
that 
>the linearizing diodes actually overcompensate in the "Serge" circuit. If
you 
>compare the output waveform with the output from a CA3080 VCA, the
waveform 
>distortion looks a bit like the opposite. Also if you put the 'scope in 
>differential mode, the input/output difference looks very different. The 
>"Serge" produces a shape that looks a bit like a bridge, whereas the
CA3080 
>produces something that looks similar to a weak sine wave.

Hi Jorgen, and diy list,

I have finally measured the transfer courve of the VCAs in my Morphing
Programmer, and it is just as you described. So there is no reason
to build a different input scheme than that from the NS application notes.
Sorry for the confusion.

After all, it is not surprising that the original application notes show
the
right way to connect the 13700, but its strange that in the explanation
on their data sheet they operate with current sources, AND I still don't
know HOW the compensation is optimized with finite (even low) resistor
values. I mean, the usual Gilbert method is quite clear, a log function
compensating an expo function, but here ?? I have a feeling that NS
don't tell us the whole story in the data sheet.

>P.S. Yesterday I worked on a CA3046 VCA that looks very promising.
Linearity 
>looks extremely good on the 'scope. It uses one transistor in the array as

>linearizing diode, I suppose.

Which VCA circuit is that ? Are there schemos anywhere ?

JH.

----------