Quad Rail ATX Power Supply

[Einric]

I am looking for a SMPS for +- 24v.
Would it be possible to take a Quad rail ATX PS and tie two rails in series for two 24v supply rails?

[eem2am]

only if they are isolated from each other before you do it.

please send schem of the quad-rail, so we can be sure wot you do

[Einric]

At this point I am just wondering theoretically.
I'm not really ready to drop any cash on a quad rail PS when I could series two dual rail PS's for the same effect.
I have two dual rail PS's in my junk pile at the house.

[!]

^ No. Or I should write, not usually. Typical quad rail ATX PSU do not have 4 separate transformer windings, they usually just isolate the output with resistance, possibly also the rectification/CLC subcircuits, and either way, as constructed the transformer has a built in, common ground which one of the two 12V windings per positive, and per negative rails, would need to be a floating ground.

Similar often applies to two dual rail PSUs, you cannot assume the dual rails can be summed to reach a higher voltage without a substantial rework. What you could do instead is abandon ATX PSU and use two, 24V SMPS, one with a floating ground as the negative power rail and its positive rail tied to the ground of the other PSU.

[Einric]

I have kind of gathered that from things I have been reading.
I am looking at the Mean-Well SP-320-27.
It has the outputs I need with the power I require.

[luka]

Originally Posted by !

Typical quad rail ATX PSU do not have 4 separate transformer windings, they usually just isolate the output with resistance, possibly also the rectification/CLC subcircuits, and either way, as constructed the transformer has a built in, common ground which one of the two 12V windings per positive, and per negative rails, would need to be a floating ground.

Similar often applies to two dual rail PSUs, you cannot assume the dual rails can be summed to reach a higher voltage without a substantial rework. What you could do instead is abandon ATX PSU and use two, 24V SMPS, one with a floating ground as the negative power rail and its positive rail tied to the ground of the other PSU.

How so? as I know, there are 4 windings, each has its own (can be the same) current rating. but all have same common point, gnd. this splits and creates more smaller output filters, better filtering, smaller size, instead of one large filter. In PC supply you can't expect split outputs, since there is only one gnd, and few voltages compared to that one.

for +/- you would need 2 or more physical units.

[!]

^ How so means what? Separate rails is more a matter of marketing with ATX PSU than a similarity to other types of SMPS that claim separate rails.

I don't claim there are "none" with separate windings, certainly there are some but most do not have separate windings for each 12V power rail claimed, especially when 4 are claimed instead of 2. It is more likely on higher priced PSU, but if it has significant value then reuse for an audio amp starts to make less sense relative to buying another SMPS more suited to the task.

To put it another way, if there were separate windings many would come closer to being capable of the sustained output number you'd get by adding all rails x voltage x current rating.

[!]

Here's an example of there not being separate transformer windings for each so-called "rail", and it's a pretty good (expensive, $199 USD) power supply not a medium to lower end PSU. It is labeled as quad rail, but notice that the label also lists a max wattage across all 4 of only 744W, while if you added them all up it should have been 1032W. Also notice that two of the 12V rails are rated for 18A and the other two for 25A.

Also notice that when they loaded all rails (page 11), the output voltage dropped quite a bit while loading individual rails (page 13) they could achieve 24A & 25A (fewer rails loaded means higher current per rail given shared transformer windings).

Also notice the picture of the bottom of the PCB. The transformer has only 3 groups of pins on the secondary side, and one of those groups isn't even for the 12V rails, it's for 5V/3V rails.

AnandTech.com - Gigabyte's New Odin GT 800W Power Supply

[luka]

rail is a rail, doesn't matter if it is the same voltage and/or it has same return path. this is still one voltage supply, from which you can't make +/-24v

[!]

It is true you can't get +-24V without substantial modification, but if the PSU has two 12V windings you could probably get +-12V by reversing the 2nd 12V winding's connections to ground and the rectifier, so long as you flip the rectifier and capacitor polarity, and sever any further electrical connection between the two (except for the common ground) including the feedback loop on the negative rail portion.