Help with virtual ground? type circuit

[armdnrdy]

I'm fairly familiar with virtual ground circuits but...I'm a little lost at the moment with what is going on in this circuit.

Usually the output of a virtual ground op amp goes to bias op amps or a component return. This output is going to ground via a 2.7K resistor.

The BBD audio path is biased with the "Delay Line Bias" (far right of the circuit snippet)
The bias on the page 2, Oscillator/LFO (not present) is taken care of by a typical voltage divider/op amp virtual ground circuit.

Neither page 1 or page 2 depicts the supply voltage for the op amps. (I figured it was +12V and GND)

If the op amps were powered with +18V and ground, wouldn't the +6V reference that I clearly see on page 2 be too low?


Update: Page 2 has a detail for the op amps mislabeled "CA3046" instead of LM324 that clearly depicts the op amps being fed from the regulated +12 volts and GND.


Edit: The Zener voltage is 6.2V.


Here is the circuit in question:


This is page 1 of my Colorsound Flanger redraw:


Here is the original drawing for reference:

[PRR]

> output is going to ground via a 2.7K resistor.

OK, this IS funky.

You must use a *single* opamp. ('741, TL071)

One input is fed Zener-regulated 6.2V, the other input is fed half of "+12V".

If the two inputs do not agree (actually even if they do), the output dumps current in the 2.7K.

Where does this current come from? Power supply (+18V).

But there is 270r in the way.

So a voltage will be dropped in the 270r. This voltage turns-on Q3. Q3 leaks power from +18V to "+12V".

When "+12V" is pulled-up to 12.2V (D2 R41 R42), Q3 will slack-off, passing just the amount of power needed to hold "+12V" at 12.2V.

While it is clever, I do think it could have been simpler. It is also not short-proof. Q3 can flow "any" current, "infinite" if things go wrong.

Since the other thing happening is a delay-line bias, which is trimmable, and the only reason for "+12V"/12.2V is to give smooth safe power for SAD and minor stuff, *today* I would throw a LM7812 at IC3 Q3 and friends. I might derive "bias" from a fraction of that 12V, or retain R44 D2.... probably depending how near-by a 6.2V Zener was.

[PRR]

On further thought: the IC3 Q3 scheme is "low dropout". "+18V" can probably come very close to 13V and it will still hold 12.2V steady. If you are sucking batteries, this may be important. If you have wall-power over 15V, it probably isn't important, and LM7812 sure is simpler and more compact.

[armdnrdy]

Thanks for that Paul!

If you think that this circuit is overcomplicated....wait until you see the CA3046 oscillator page!

There was sort of an attraction to this circuit because it is so non typical!

Even though you are completely correct and my better judgment tells me don't do it! I think I'd like to build it as designed. It's sort of a relic design with discreet components and a few op amps thrown in for good measure. 

[armdnrdy]

Once given a some direction...so I knew what to look for...look what I found.

LDO Voltage Regulator Using Discrete Semiconductors

[PRR]

> LDO Voltage Regulator Using Discrete Semiconductors

Broadly similar. Better drawn. Again I would not do it that way today without a "nostalgic" reason.

That plan requires an opamp which will swing output to a half-volt of the negative rail. Many won't.

Every opamp already has the equivalent of that plan's Q2, one way or another.

That plan has a LOT of gain. It will seek instability. C2 may have to be a boulder.

The Colorsound regulator uses the internal transistor, and stands it up on 2.7K. And note the ratio 2.7K:270r, 10:1. Colorsound's Q3 Base will want about 0.6V drop across the 270r. This will need about 6V drop in 2.7K. Which puts the opamp output around +6V, well within its rails. If Q3 needs as much as another 1mA Base current, the opamp output may rise to 8.7V, still well within the rails. It can't pass more than about 6mA to Q3 Base (but that still allows >600mA through Q3 which may not be safe).