[sdiy] SH-3a VCO - how does it work

[René Schmitz]

On 20.04.2020 12:14, Steve Lenham wrote:

>> Most odd is the arrangement of Tr130 and Tr131, which appears to 
>> either protect the capacitor from overvoltage, or is used as some very 
>> crude voltage reference.

Actually, most likely the cap over these zeners is there to short out 
the noise.

> It's a temperature-compensated Zener diode built out of discrete 
> components (and they do it again with Tr107 and Tr108).
> 
> The forward-biassed B-E junction has a negative temperature coefficient.
> The breakdown voltage of the reverse-biassed B-E junction has a positive 
> temperature coefficient. They cancel each other out to some extent - not 
> totally, but the result is better than a Zener alone.

The stability is usually best around 5-6V, where the diode TC cancels 
the Zeners'. (6.2V compensated zeners are common, because they're easy 
to make. 5.6V+0.6V)
> 
> You can buy TC Zeners as a single device now but perhaps you couldn't 
> back in the day. Or perhaps the transistor version was slightly cheaper. 
> I would speculate that the temperature stability was more important to 
> Roland than the exact Zener voltage, because that wouldn't have been 
> very well controlled.
>

Yes. Here the avalanche voltage of the transistors is not only a not 
well specified property, but typically is more 9V than 5V. (I could be 
wrong and the particular NPN has a low breakthrough voltage.) As you 
say, at the least its better than a pure BE-junction.

In other places there are zeners used (expo, keyboard circuit).

Best,
  René

--
synth at schmitzbits.de
http://schmitzbits.de