Dual polarity supply using a regulator module

[jay]

I needed +5V and -5V supply for a project and decided to take the easiest way I knew: a dual output DC-DC converter module from MeanWell:
http://www.meanwell.com/search/dcw05/dcw05-spec.pdf
I had used it before but now I have more load on the +5V rail than on the -5V. So when +5V is loaded it drops to 4V and -5V falls to -6V So the regulator isn't COM / GND referenced. Can you recommend an alternative that would regulate + and - rails independently? Or is this how these usually work?

And before someone suggests.. sure I could implement an inverting charge pump to get a negative supply and regulate it but I would rather use my time on the other parts of the projects, not on the power supply.

[richard.cs]

That isn't how they usually work, it's a bit worrying that there don't seem to be any clues in the datasheet that the outputs need to be loaded symmetrically. Is it possible you just had a dud module or a wiring fault? I assume you're staying within it's output rating?

[Zero999]

No the COM terminal won't be ground referenced because it's an isolated supply. By design, the secondary isn't connected to the primary.

If the output isn't regulated or is only poorly regulated on the +5V and not regulated on the -5V then this is what you should expect to happen.

[jay]

I have tried to find replacement DC-DC converters, that would behave so that +5V and -5V can be unequally loaded, but I need it to have the same pinout, because I don't want to spend more money and time to order new boards. The pin out seems to be somewhat common, but the datasheets are not very clear about the COM / Common pin and what happens when loading the rails unequally against the Common.
Traco TEN5 looks like a suitable and relatively easily available replacement:
http://www.tracopower.com/products/ten5.pdf
Has anyone used dual polarity models from the TEN5 series?

If the output isn't regulated or is only poorly regulated on the +5V and not regulated on the -5V then this is what you should expect to happen.

They promise +/- 0.5% regulation with 2% voltage accuracy so it doesn't count as "poorly regulated", right?

I did more measurements with DCW05B-05 and DCW05A-05.

Voltage between +Vout and -Vout is stable 10V (measured variation only 20mV). It does not change with asymmetric loading of +Vout and -Vout.

I put an adjustable load between COM and +Vout and another adjustable load (well, manually adjusted by swapping 5W resistors) between COM and -Vout.
When there is equal load on both rails then +Vout = +5.0V and -Vout = -5.0V.
Uneven loading causes the COM level to shift towards the more loaded rail. The effect is strongest one one rail is completely without load and the other is heavily loaded. When both rails have significant load (90mA on negative and 400mA on positive) I measured voltages -5.2V and 4.8V.

I tried connecting COM to -Vin, but that did not influence the behaviour at all.

I'm curious.. Can some power supply expert tell based on this how they implemented the module? I don't get how they produce the COM voltage.

Pin names (+Vin, -Vin, +Vout, -Vout, COM) as in the datasheet, that I provided a link to in the original post.

[jay]

Is it possible you just had a dud module or a wiring fault? I assume you're staying within it's output rating?

I've tested 2 modules on the board and 1 on a breadboard.. No difference. I've kept the loads below 500mA.

[jay]

  I have a solution..

It's nice to have additional heating in winter time . Simply putting 33 Ohm resistors on both rails prevents too asymmetric loading of the rails. I have 300mA max real load (on the positive rail) so 5V/33Ohm=150mA + 300mA=450mA < the max 500mA allowed by the module. It's generating a total of 1.5W heat so good bye efficiency.. but it works well enough: I measured max 80mV deviation from 5V, when there's only the dummy 33Ohm load on the negative rail and full 450mA on the positive. The design could tolerate even +/- 250mV variation.

I think I'm not going to use a black box DC-DC converter in the next project, because it would feel like . Inverting charge pump + linear regulator is less trouble for the negative rail (and insanely more energy efficient).