Getting a bipolar power supply from simple 9V dc supply

[Jdansti]

Just for theoretical purposes, if all one had was a 1044, could you regulate the input at 9v with a Zener?

[armdnrdy]

Just for theoretical purposes, if all one had was a 1044, could you regulate the input at 9v with a Zener?

I would think yes.

I've seen quite a few transformer power supply designs in older stomp boxes where they do just that.

As you know...a transformer's secondary voltage output is relative to the primary, and voltage can vary from one side of town to the next.

Thank god for the invention of voltage regulators!

[Keppy]

I just tried a transient analysis of the LT1054 as an inverter in LTspice with a 9V supply and output voltage never reaches -9V.  It only gets to about -8.6V which is within the specs for voltage loss in the datasheets.  My question is how would you get a 'perfect' +/-9V supply or even a +/-8.6V.  Would you have to then regulate the +9v supply after the LT1054 to make it exactly +8.6V.  It's probably not a huge concern, just curious.

The negative supply will vary with a varying positive supply, so regulating AFTER the charge pump is probably not the way to go. I'd more likely run the positive supply through a diode to drop the voltage into range. If you're REALLY picky, you could put a trimpot between the charge pump voltage input and the circuit input to control the exact voltage drop. Either way, you'd want another filter cap after the diode/trimmer to filter the new supply voltage.

Not that I would bother with any of it. I'm with armdnrdy, I think it's unlikely to matter.

[Wounded Paw]

Actually putting a diode in series on the +9V input to the circuit puts it pretty much exactly in range.  I just ran the spice sim again and got +/-8.3V, assuming equal loads on the + and - supplies.  Not that it matters.

[PRR]

> My question is how would you get a 'perfect' +/-9V supply or even a +/-8.6V.

My question is when would you need a 'perfect' +/-9V supply??

Usually you need enough to cover signal and bias, not too much so stuff blows.

You don't demand that both shoulders on a road be equal.

Yes, there ARE a few cases where exact equality is relied-on. "Tracking" power regulators try to do this. But this is so hard to do that you are always better off finding another way, not depending on exact-equal supplies.

All of the ordinary chips will do most of the ordinary audio work with severe un-balance, like +12V/-5V. Doesn't faze them.

[Wounded Paw]

I was just curious.  I wanted ground to be ground regardless.  Voltage dividers or rail splitters like TLE2426 make a virtual ground but is that as solid as using the ground that comes in and out of a circuit.  It's just a bit of a mental exercise, not critical in any way.

[PRR]

> Voltage dividers or rail splitters ... is that as solid as using the ground that comes in and out of a circuit.

No.

> I wanted ground to be ground regardless.

"Ground" (a dubious thing) is where you make it.

Generally your signals never touch the supplies, so do not "know" where the supplies are.

Yes, we do bump the supplies in *power* amps worked to MAX output. And a 10% un-balance means the "maximum un-distorted power" is reduced 21%. So it may be something to consider. OTOH the nature of serious power supplies tends to give well-balanced rails. If they don't, or your circuit is off-center, well if it won't hit your power target you build a bigger supply.

[Wounded Paw]

The reason I'm on this is I've had some problems with noise in a fairly extensive circuit involving several op-amps and have tracked down at least some of it to the ground being noisy.  Not the virtual ground or bias voltage, the actual ground as referenced to the in and out jacks.  I'm already going to redesign the whole circuit board to have individual grounds for different signal paths like in a mixing console but I'm wondering if going to bipolar power will help as well and would rather keep the circuit on a single 9V supply if possible.  I'm currently using the TLE2426 rail splitter to generate my bias voltage instead of a voltage divider.