SV: [sdiy] QVCO schemo up

[harrybissell]

Ian Fritz wrote: (and I replied...)

> At 07:52 PM 11/11/05, harrybissell wrote:
>
> >let me say that I have never seen an LM13600/13700 that did not
> >have huge offset problems when I activate the diode bias
>
> Maybe I'm not clear on what you are saying here.  Do you mean (a) you have
> to make a big change in offset correction when the diodes are turned on or
> (b) that the change in bias with Iabc gets much worse (after re-trimming)?
>
> For me, (a) is certainly true, but not a concern and (b) sometimes happens
> but isn't always a lot worse.  But again, these chips are a total crap
> shoot, as far as I am concerned.  I learned why making discrete
> versions.  So now I always pretest any OTA I use.  I built test circuits
> and ran dozens of chips through them, recording all the results on paper.

I'm saying "A".   I usually lay out al my OTA stuff to allow the diode bias
resistor to be added.  Adding it blows the offset WAY out.

> >I'd add that I usually use equal value bias resistors... then inject a small
> >offset voltage through a high value (megohms) resistor.
>
> You waste a resistor?!?!?  MEGA Ohms down the drain?!?!?  When you only
> need to use 100 or 200?!?!?  Gasp!!!!!

I like to use a pair of 1% metal films on the inputs, because they are more
stable
long term imho. THEN I make a really high value resistive divider and run that
from
a trimpot connected to +/- rails... probably ruining all the benefit of the 1%
resistors
in the first place.  I decouple the pot wiper to prevent noise.

I could see that the manufacturer's method has some advantages (low component
count) but if the trimpot fails in my case, the offset gets a bit untrimmed. If
the
trim fails in their case, the circuit will become totally non functional, na ???

H^) harry