[sdiy] TB303 / Polysix VCO current mirror thingy.

[Gordonjcp]

On Fri, Apr 06, 2018 at 09:33:45AM +0100, rburnett at richieburnett.co.uk wrote:
> >Is there any downside to this cheap-n-cheerful alternative, Richie?
> 
> Two that I can think of:
> 
> 1. The trigger threshold of the thyristor has a temperature
> coefficient.  So the trip point will vary significantly with
> temperature, whereas a decent comparator combined with a precision
> voltage reference should do much better in this regard.  (However,
> the AN299 paper that Neil mentioned explains how this temperature
> coefficient was purposely used to cancel out the approximately equal
> and opposite temp coef of a particular type of capacitor used for
> the VCO integrator.  So I guess the temp coef of the SCR is not
> always a downside!)   I suspect the additional diode-connected
> transistor Q24 in the TB-303 schematic is an attempt to compensate
> for the temperature coefficient of the Q25/Q27 thyristor in some
> way.

I did wonder about that.  The Polysix has no "auto tune" function of any
sort as far as I can see , it just does its thing.  Mine has never been
particularly well tuned, but it seems to sit pretty much in tune
regardless of temperature.  I wonder if there's any special matching
required?

If you hit a note as you power it on, it'll drift up to pitch over about
ten or fifteen seconds, going sharp and then settling back down to
"about right".  I don't know how much of this is the oscillators warming
up, the CV generator warming up or just the general DC conditions
settling.

Interesting that it's essentially a relaxation oscillator with a current
mirror to keep it nice and linear.  It reminds me of a weird issue that
someone I know had with their house a good few years ago - after the
switched to compact florries (LEDs weren't cheap yet) when they switched
off the hall lights with the switch downstairs it was okay, but when
they switched them off with the switch upstairs when they went to bed,
the hall light would - after a few minutes - begin to flash about every
30 seconds.  Blink... Blink... Blink...

It turns out when they removed the upstairs switch there was a bit of
damp in the brickwork behind it, condensation had formed in the switch,
and when it was "off" it really had a resistance of a few thousand ohms.
This is nothing for an incandescent lamp, but for a compact florry it
would allow the smoothing capacitor for the ballast to charge slowly
until it eventually hit the tube's striking voltage :-D

-- 
Gordonjcp