diyAudio (
-   Parts (
-   -   paralleling computer power supply outputs: can they be connected in parallel... (

jarthel 3rd March 2009 12:53 AM

paralleling computer power supply outputs: can they be connected in parallel...
to get more current?

I've done it before in linear supplies but I'm not so sure with smps (I assume computer supplies uses smps).

Is it as simple as connecting "red wire" from 1st supply to the "red wire" of the 2nd supply?

thank you

unclejed613 3rd March 2009 01:14 AM

the SMPS supplies sometimes have feedback connections to the outputs (as do most linear supplies). it's best to use small value high wattage resistors to balance the load between supplies. you might want to fuse the outputs as well, to protect the other supplies in case you have a rectifier diode or cap short on one of them. now that i think of it, the resistance of the wires (usually #18 or #20) will probably suffice for balancing the loads. fuse the supplies for about 90% of their rated current or less. although SMPS's have overcurrent shutdown, two or more supplies operating in pulsed shutdown might react unpredictably in event of a short.

Scott Novak 4th March 2009 04:25 PM

The biggest problem you will have trying to parallel switching power supplies is getting the load to balance between the 2 power supplies. This is pain in the butt even when you are designing the system yourself. Using series resistors between each power supply and the load may help to some degree, but then you will be affecting the power supply regulation.

I wouldn't recommend trying to parallel switching power supplies unless the power supplies were designed to be operated in parallel.

Scott Novak

unclejed613 4th March 2009 10:47 PM

another way to parallel SMPS's is to parallel through high current schottky diodes to isolate the supplies from each other. using schottky diodes gives only a 300mV forward drop across the diodes. if you want a "no-drop" diode, this is done with MOSFETs. there's a few ways of doing this. look up "lossless diode MOSFET" or "ideal diode MOSFET" in a search engine. if you can diode isolate the supplies, you will avoid some of the squirrely things that can happen with the interaction between two or more SMPSs.

one of the tricks that worked with linear open frame supplies, but is more difficult with SMPSs unless you know where the feedback is tapped off, is to run the sense line to the other side of the isolation diode, so that the supply will compensate automatically for the diode drop.

coloradosound 4th March 2009 11:03 PM

There are dual redundant computer power supplies on the
market. Have you checked into those. I picked up some
off the bay of E place. They share a common chassis and
have the load balncing resistors and the schottky diodes
(or MOSFET equiv.) which blocks the supplies as unclejed613
mentions and be cause of that they are hot swappable!
(If one supply goes down, you can usually flip some type
of locking lever and pull out the bad unit while powered up
and plug in replacement later, again while powered up!)

stu1958 5th December 2012 03:37 PM

Re: PSU's Switching Output Connections Together
Can switching power supply outputs be connected , no not really is my take on what the consensus here is saying, so :::::

For the sake of argument, lets just say NAY to hooking up switching output supplies together. OK? But is there another way, off course there is.

Here's another idea, Based on UPS technology they switch from Line to Battery within the time necessary to keep the computer from crashing. They throw around the term, "hold up time" as it may be a useful trick to switch from one PSU to another in a FLASH. The flash being some way to detect any failure quick enough to switch all outputs from one supply to another. This brings to mind TRI-STATE digital outputs due to a similiar issue with NOT being able to tie outputs together. If you accept the previous idea then heres the issues.

Several questions being the following design issues :::
1) How fast can you bring up a regular switching power supply, fast enough not to let computer to crash ?
2) IF NOT THEN can a switching supply be running with OPEN outputs OR need a LOAD BANK to idle ?
3) What parameters to sense sufficient enough to
INITIATE AUTOMATIC SWITCHOVER within approx 15 msec ???

IMHO Any decent redundant supply less than 10x cost of two supplies alone is worth the cost of not having unscheduled downtime. And reinvening the wheel.
Just remember, everything looks good on paper until you tinker with it and you smoke a few parts,it happens. the end...
ps please dont reply saying im crazy and you dont understand anything I say.
I dont want to discuss it. TY.
Just do the math and remember to use capacitors,inductors and resisters wisely. Like on a house. Try not to burn it down.

freax 12th December 2012 05:19 AM

a discrepancy of 0.5v between two 70 amp power supplies means you would have to dissipate 35 watts!

It would also do nast things to the voltage regulator circuits without any diode in place.

stu1958 12th December 2012 09:13 PM

Re: Tinker Tinker Ozzy Osbourne
Ok, 0.5 just a mere one have volt difference and two 70w power supplies is mentioned to be 35w of energy that needs to go somewhere.

Ok, doesn't sound to bad on paper until you multipy the power supply X10 and then theres a 350w difference.

The problems with these supplies is what the big UPS companies do with their enterprise equip that are built in stackable expansion size blocks of
many kilowatts. The use a one wire approach to sync' ing in unison all
the supplies to the same switching frequency.

Unfortunately I did not go to work for them to see how its done. The are
extremely sticky about protecting their proprietary schematics.

Somewhere on ebay I almost bought a behemouth enterprise UPS just to
tinker with. I know sounds crazy. They would not appreciate any backward engineering as if I hacked their mothers mainframe.

Smart Grid technology and older means of automatic switchover has been
going on for quite awhile.

I'm an obsolete dinosaur, I don't know nothin.

All times are GMT. The time now is 02:38 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2