Bipolar Supply vs Voltage Doubler

Started by POTL, August 27, 2020, 07:59:09 AM

Previous topic - Next topic

POTL

Hello
I often see circuits with voltages greater than 9, 2 classic solutions are voltage doubling 18 volts and bipolar power + 9 / -9 volts.
Both methods can be implemented with a single microcircuit.
I notice that many modern components, in addition to operating on lower voltage (3-5 volts), are unipolar. Increasingly, bipolar-powered chips can be found marked "not recommended for new designs" or "obsolete".
It seems to me that in the technical and audio aspects there is no difference between + 9 / -9 and +18/0
But maybe I am missing something.


Are there any reasons for using bipolar nutrition rather than doubling? With the proviso that both versions are made on a special chip.
Or is it just another outdated design decision?

ElectricDruid

There's no difference in terms of *voltage* between +9/0/-9V and +18/+9/0V. Those are the same thing, and which you see depends where you stick your probes and what you regard as the lowest voltage.

But that's a minor point about voltage being relative. You're really asking about 18V unipolar supply versus +/-9V bipolar supply. In that case, there definite *is* a difference, because one has a true low impedance path to ground down the 0V connection, and the other one has a floating bias voltage at 9V which might or might not stay where it is supposed to be under varying loads. Big difference.

In practice of course in pedal-world we make unipolar supply circuits work perfectly reasonably all the time, even with only 9V to play with and a 4.5V bias. So it's possible. But you'll also find dozens of threads on here complaining about noise in circuits caused by voltage-doublers or dodgy Vrefs. Using a genuine bipolar supply still makes working with op-amps a lot simpler / more foolproof.

It's true that modern chips are heading more and more towards lower voltages. I think that's a drive towards lower power devices from manufacturers. Previous generations wanted the maximum headroom to improve signal/noise ratios - many of the older op-amps could run on +/-18V supplies, or even more. There are studio mixing consoles that ran the op-amps at higher-than-spec supply voltages to gain just a little extra headroom and try and keep further above the noise floor. That kind of approach is unnecessary in a world of mostly digital audio, so voltage and power levels are going down and down. I expect that will continue. As devices get smaller, heat dissipation becomes more of an issue, so lower power has benefits in terms of reducing the heat you need to lose from a tiny package.


antonis

For audio implementation, bipolar supply is more convenient due to coupling caps need elimination (at least, for signals swinging near ground..) and also no need for extra Vref & bias items configuration..

Anything else, already told by Tom.. :icon_wink:

"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

PRR

Quote from: POTL on August 27, 2020, 07:59:09 AM....Increasingly, bipolar-powered chips can be found marked "not recommended for new designs" or "obsolete"........

What chips are marked "bi-polar"?

Yes, many chips are specified in terms of a bipolar supply. Few actually NEED bi-polar, or are made-for bi-polar-only.

Bipolar was a fad in the 1940s-1960s, with long after-life, but getting pretty outdated now. The primary point was DC computations, where "zero" was a solid thing. Most such stuff has moved off to logic chips. Bipolar in large audio had some advantage segregating audio and power (though often foiled by insufficiently obsessed bypassing). In small audio it is usually moot.
  • SUPPORTER

amptramp

If you design with a Vcc/2 Vref and you have more than one stage that is biased from it, there is a possibility that there will be feedback from one stage to another via the Vref pin.  Audio voltage going to the second stage usually goes through a large resistor but if it is something like an absolute value circuit where the bias resistors are small, unless you have bypassing bordering on the magical, the output of the first stage may feed into the Vref and cause unwanted feedback.  A regulator or buffer may stiffen the Vref but then you are getting into more complication.