[sdiy] Discrete OTAs - (was - Best CA3280 projects?)

Phil Macphail phil.macphail at liivatera.com
Mon Jul 2 22:45:02 CEST 2012

That is true for transistors used in linear applications like mirrors and
the linearising diodes discussed, but in MOS design the only true diodes
available are usually the parasitic one between the MOS drain and
substrate. With modern processes the resistors available are probably
comparable with metal film resistors for absolute and relative matching
for reasonable size (but a worse temperature coefficient), so aren't
really that good. This is why it is normal to use circuits that are
dependent on transistor size ratios, as these will match far better, and
when biased from current sources they can be temperature stabilised with a
bandgap reference.

As you say, it's a different thought-process from discrete design,


On 02/07/2012 20:58, "Lanterman, Aaron" <lanterma at ece.gatech.edu> wrote:

>On Jul 2, 2012, at 3:34 PM, ASSI wrote:
>>> I'd be curious to see how you did this, whether you actually used
>>> or half a transistor (don't laugh - that's what I've read) and what
>>> advantage it got you - if any.
>To my knowledge, all diodes you find on chips are "diode-connected"
>transistors. (Both BJT and MOSFETs).
>It also takes a while to get used to seeing some transistors more-or-less
>intended to act as resistors. Resistors on chips are expensive
>real-estate-wise and difficult to set precisely.
>The chip design space really turns your discrete-on-breadboard intuition
>upside down.
>- Aaron
>Synth-diy mailing list
>Synth-diy at dropmix.xs4all.nl

More information about the Synth-diy mailing list