[sdiy] Lowest distortion neede for VCA - linearizing the LM13600/13700 & other dual VCAs

[ASSI]

Am 17.09.2013 11:04, schrieb Neil Johnson:
> I think it's doing two things:
> - converting the input voltage to a current,

The I/V conversion you care about is done by the input resistor.  The 
OTA produces a matching current to compensate the input current via 
feedback.

> minimising the differential input to the second VCA.

Sort of.  It will be a smallish voltage, but you'd certainly be in a 
region where the tanh distortion is noticeable.

>  Note that BOTH VCAs have almost
> zero differential voltage -- as per the datasheet you want to keep the
> differential input voltage to below a few mV for lowest distortion.

As long as the feedback works and the matching is perfect, the amplifier 
could have any nonlinearity and it would not show up at the output at 
all.  This curcuit, in a nutshell, implements a current mirror (with 
adjustable mirror ratio) built from two matched OTA.  This is also what 
keeps the distortion low: the amplifier nonlinearity never enters the 
equation when perfect matching can be assumed.  A translinear circuit 
built from a translinear circuit... :-)

> - pre-distorting the input signal, further improving the overall performance.

Only the first OTA ever sees the input signal and one would hope that 
the input and output resistors don't distort it.

> Note that Sims doesn't use the linearizing diodes in his scheme.
> Given the differential input is almost zero the data sheet suggests
> they won't help, but it might be worth experimenting just to see.

That would lower the input impedance and is not useful for this circuit. 
  Remember that by using the linearizing diodes the OTA really gets an 
I/I characteristic rather than the I/V that is needed here.


-- 
Achim.

(on the road :-)