Greg Mayman wrote in a message to Roy J. Tellason:

 -=> Roy J. Tellason said to Greg Mayman
 -=> about "POWER SUPPLY" on 12-11-03  04:06.....

 RJT> One would think so,  only this one didn't show up with an ohmmeter.  I
 RJT> suspect that something on there is breaking down under voltage.  I need
 RJT> to be able to apply the two voltages separately,  and see what
 RJT> happens... 

 GM> Fortunately I only ever had one like that.

 GM> I finally isolated it by wiring a 100w lamp in series with the
 GM> mains input (so it wouldn't blow the fuse) and feeding the unit
 GM> from a Variac. Then I brought it up to where it was just on the
 GM> point of running away and measured lots of voltages around the
 GM> amplifier.

 GM> It turned out that one of the output transistors must have
 GM> developed a soft zener diode type action in the base-collector
 GM> junction. I've never seen any other semiconductor do it exactly
 GM> like that.

 RJT> remember where a signal went in the order FET -> optoisolator ->
 RJT> transistor -> SCR -> bridge (!) -> triac, which I thought was a bit
 RJT> much... 

 GM> I think I know what they were doing. OTOH it could probably have
 GM> been simpler.

 RJT> What do you think they were doing?

 GM> IMO the triac was controlling an AC line, with the gate drive from
 GM> the AC supply.

 GM> Obviously this won't suit the optocoupler which needs a unipolar
 GM> feed, so the bridge was added. And to boost the opto's output, the
 GM> SCR was added as well.

The bridge was after the SCR,  actually.  The stuff was in the order I gave it.

 GM> Action would be as follows:

 GM> 1. FET turns on, driving the LED of the opto.
 GM> 2. The opto's output connects between anode and gate of the SCR,   
 GM> turning it on when the opto is driven.
 GM> 3. The SCR shorts the DC side of the bridge, giving a near short   
 GM> circuit on the AC side.
 GM> 4. The AC side of the bridge is connected between anode 1 and   
 GM> gate of the Triac maybe with a series limiting resistor. When   
 GM> current flows in this path the Triac is turned on.

Something like that...  It still struck me as a little bit overly
complicated for what they were trying to do.

 GM> The current through the SCR is uni-polar due to the bridge,
 GM> although the current through the bridge to the gate of the Triac is
 GM> bipolar.

Yeah,  but you're supposed to be able to trigger a triac with either
polarity of voltage,  for either polarity of MT2 voltage.  There are
differences in sensitivity in the four "quadrants" of operation, 
nontrivial ones,  but with sufficient drive I don't see this as being a
problem.

 GM> It could have been done more simply with a relay.

Yes.

 GM> OTOH this circuit is fast so it might have been used in a phase 
 GM> controlled dimmer -- obviously a relay can't do that.

I don't recall now what the application was,  I'll have to look...   Found
it,  under "Solid State Relay"!  That's the caption the author
had on the page,  and it also appears on the schematic diagram.

Looking a little closer,  the FET has the gate tied to source,  so they're
using it to limit current through the LED of the opto,  which is why the
input can be labeled "5-25V".  There's also a diode in series
with that,  and that leg of the circuit is marked "5 mA" -- I
guess that depends on the specific FET you're using.  The output side of
the opto has a resistor tying the base of the phototransistor to ground
(why?),  and an NPN transistor that fires the gate of the SCR that in turn
fires the triac through the bridge.

I guess the power for the SCR, transistor, and output side of the opto are
all supplied through the bridge,  which accounts for that...

I've a couple of TRIACs here too that are in stud-mount packages,  though
they don't have all that hefty a rating -- something like 6A,  I think? 
Their original application was in an electronic organ's rotary speaker for
motor control,  not much of a heatsink was used,  so they used a heftier
part for a fairly light load,  I guess.

That application used a couple of *very* lightweight relays to switch the
triacs,  which carried the load.  Something like a 100 ohm 2W resistor
connted between the relay contacts and the gate and MT2 terminals.  I've
wondered at times about using the output side of a MOC3010 (DIAC output)
opto for a similar purpose,  and what resistor I might need in there.

I also did a similar mod,  in Leslie tone cabinets.  Those things used to
burn up relays,  which were used to switch motors directly.  If I could get
the okay on it before the contacts were too burned,  I'd stick a couple of
triacs in there and let the relay switch _them_ instead of the motors
directly,  saving them from further deterioration.  Too bad I don't get the
opportunity for little tweaks like that these days...

---