[sdiy] korg ms02 antilog amp and log amp

[Ian Fritz]

Hi Harry --

Yes, I think the MAT02 is an overkill for a VCO expo.  Here is my 
reasoning.  For a batch of CA3083's I recently tested, 18 of 20 were 
matched to better than 0.2 mV and 12 of 20 were under 0.1 mV.  This is way 
better than the 1 mV we used to shoot for using discrete transistors.  In 
an expo converter, remember, one member of the pair is just used for 
temperature compensation.  This means that matching is not all that 
critical  (especially given that most people don't bother to carefully 
compensate other drift sources).

Additionally, I measured the emitter series resistance of the CA3083 to be 
quite low -- about 1.3 Ohms.  This compares to 1.1 Ohms for the LM394, 5.0 
Ohms for the CA3046 and 2.4 Ohms for one of the Japanese pairs.  So log 
conformance is quite good!  Not as good as the MAT02, but remember that in 
a VCO there is always some reset-time error at high frequencies, so there 
is a point of diminishing returns as far as improving log conformance.  The 
CA3083 is significantly better than the CA3046 as far as hf tracking goes, 
but I don't see much improvement going to a better transistor, since some 
hf-tracking adjustment is always needed anyway, and the usual schemes are 
quite accurate in compensating the Re error.

Now when you look at matched pairs for making a roll-your-own OTA, well, 
that's a different ball game.  To avoid excessive variation in offset with 
varying bias current you need very good matching -- better than .05 mV, and 
it makes sense to use the MATs.  But here I think you get better bang for 
your buck with a good IC solution (CA3280 or better).

So that's my current take on matched pairs.  Comments welcome, of course.

   Ian


At 07:03 PM 3/27/06, harrybissell wrote:
>Yikes... above ALL of those ???
>
>(the MAT02 is pretty damn good...)
>
>(or did you mean considering price/performance ratio ???)
>
>H^) harry
>
>Ian Fritz wrote:
> >
> > Or my favorite -- above all those mentioned  -- the CA3083 array.
> >
> > At 01:02 PM 3/27/06, Harry Bissell Jr wrote:
> > >Analog Devices MAT01 or MAT02  (dual)
> > >or MAT04 (quad)
> > >
> > >National Semi LM394
> > >
> > >CA3046 or CA3086 (five transistors)
> > >
> > >The 2SC1583 is not THAT hard to get, yet...
> > >
> > >H^) harry