assymetric to symetrical power supply question

[Rob Strand]

Can't you achieve that with using a regulator to isolate a second +9v (ish) and run it like that?

As antonis is coyly pointing out, probably not. The issue is not the voltages, it's where the ground wire connects.

You could take the +9V, then created a second isolated 9V output with an isolated converter, then wire it in reverse to get a negative rail.  An additional regulator would be optional.  The fact the output is isolated means the polarity mentioned in "A second +9V" is a little superfluous.

If you bought an ready made isolated converter then that solution is a possibility.   However if you were designing your own converter making it isolated is just creating a second set of problems.  You still need to get a quiet switching converter working and a non-isolated design is going to easier.    And taking that further ... a charge pump design with a good layout should work.

[mzy12]

Any schematic available..??

Here is something I cooked up. Please excuse me if this is, like, an abomination against mankind.

[Rob Strand]

Here is something I cooked up. Please excuse me if this is, like, an abomination against mankind.

You can't connect the outputs of the regulators in series when the input are fed from the same power source.   You can only do that if the regulators are fed by two supplies which are isolated.

Check out your -ve rail regulator.   The ground for the *input* power is the same wire as the -15V output rail.

[mzy12]

Check out your -ve rail regulator.   The ground for the *input* power is the same wire as the -15V output rail.

I guessed that would be a problem. It was worth a shot anyways. There is no way around that without an DC - DC isolator, correct? It's such a shame they are so expensive. I suppose a charge pump is really the best solution here.

[Rob Strand]

There is no way around that without an DC - DC isolator, correct?

Correct.

[mzy12]

Okay, what would be the problem with something like this 

[Rob Strand]

Okay, what would be the problem with something like this

It's got the same/similar problem to the previous one.

GNDREF doesn't go anywhere.   For the +15V regulator to get power from the +20V converter GNDREF must connect to GNDA, which is what you had before.   For the -15V regulator to get power from +20V, GNDA needs to connect to 15V_B (the negative rail).   Whatever you do to get one polarity working it will kill the other one.

If you look at what feeds each regulator in isolation you can see the "lower rail" of each regulator has to connect to the 0V for the 20V rail (ie. GNDA).   So the outputs of the both regulators can only be +15V.

[ElectricDruid]

Rob said it;

You can't feed a LM317 with *just* 20V. It needs two wires going in. Voltages are relative, so 20V only makes sense at all when referenced to something, and your regulators have got no ground connection going to them. They've got one going *out*, but that's not going to help...

[antonis]

Let's say it once more..

+20V on D1 & D4 Anodes MUST come from "isolated" DC supplies..
(no common ground shared to refer on..)
Those +20V must be added in series (resulting in +40V across their outer points) and their junction point be taken as GND for the downstream circuit..

[Rob Strand]

Let's say it once more..

[PRR]

[antonis]

@mzy12: Pay good attention to PRR's acoustic equivalent of floating ground..!!