45

I have a computer power supply that I'm hacking together as a bench supply. For this particular model to power on, I need a minimum load across both +5V and +12V.

"Easy," I thought, "I'll just connect both +5 and +12 to my power resistor!" And it worked, but then I started thinking, what does it mean to have these two different voltages connected in parallel? If the voltages were the same, then I would be increasing current. But what about different voltages?

Also, what if I connected +5 and +12 in series, and then put a load on that? The equivalent voltage would be +17V; what would be the difference between that and parallel?

Or am I going about this the wrong way; should I put a separate resistor on each rail? It seems like I can do better than that.

Danny Beckett
  • 359
  • 3
  • 15
rascher
  • 953
  • 2
  • 11
  • 15
  • 1
    Very interesting and a bit dangerous question. Good thing computer PSU-ts are overcurrent and overtemperature protected, so they tolerate quite a lot of mistakes. – Vorac Aug 22 '13 at 13:13
  • 2
    My electronics lecturer told us that when connecting two different voltage sources in parallel will cause the magic smoke that all electronics run on to come out of of the one voltage source. – Keagan Ladds May 21 '14 at 14:01

11 Answers11

35

First some theory:

In general, a PC power supply isn't expected to operate in a redundant mode (i.e. with outputs tied together).

In industry parlance, this function is called OR-ing (not O-ring). If a power supply is designed with OR-ing in mind, there will be several additions to the circuitry:

  1. Some means of isolation (diodes or MOSFETs)
  2. Some means of maintaining regulation at absolute zero load (anti-rollback)
  3. Some means of load balancing (forced or droop)

These factors allow you to connect identical voltage rails together to provide load current beyond what a single supply can do, and allow for the rail to stay up (if the load can be delivered by N-1 units) if a single unit goes down. It also gives you some measure of protection if you accidentally connect a higher voltage to a lower voltage.

Also, most PC power supply returns are all tied to each other. There typically isn't an isolated output (independent return) available.

Now, the practical ramifications of your experiments:

  1. Connecting the +12V and +5V rails together is only "safe" if there is some means of OR-ing on the +5V that can withstand the +12 applied to it. The +5V will not be delivering any current to the load, as it will be blocked by the OR-ing device.

Most likely you've back-biased the +5V and put 12V on some electrolytic capacitors that are probably only rated at 10V.

  1. Connecting the +12V and +5V in series is only "safe" if one of those rails has a return independent of the other. If the returns are common, all you're doing is short-circuiting the rail that's on the "bottom" (the rail that has a return connected to the high side).

Batteries != power supplies. Energy going into a battery charges it. Energy going into the output of a power supply usually smokes it.

Adam Lawrence
  • 32,921
  • 3
  • 58
  • 110
  • 1
    It's not good to connect random batteries in parallel either (as shown in the original question's link), because if they're not already at the same voltage and state of charge (fullness), one may charge the other with an unlimited current. The batteries & light bulbs example seems intended to teach about voltage, current, and resistance, but batteries are not ideal voltage or current sources. Rechargeable batteries connected in parallel without due care may be destroyed. The experiment is better done with cheap disposable batteries (to lower the cost) if it must be done at all. – Matt B. Dec 20 '11 at 04:56
27

To add to what Majenko said, in the series configuration it doesn't work the way you expected either. That is because both supplies have a common ground. Series would only work to make 17V is both were independently floating, which they are not. It is not possible to connect two supplies with a common ground in series.

Whether series or parallel, either way is a REALLY BAD IDEA.

Connecting two power supplies of different voltages together:

Any questions?

Olin Lathrop
  • 310,974
  • 36
  • 428
  • 915
16

In parallel, your overall voltage would be +12V.

BUT

Your +5V feed will be being over-powered by +7V.

This is bad

You will probably be doing damage to the internals of your power supply.

Especially if the +5V section is using capacitors that are rated at less than 12V... pop

Community
  • 1
Majenko
  • 55,955
  • 9
  • 105
  • 187
  • 1
    Would the overall voltage actually be 12v? I figured the two would both attempt to pull to their respective voltage resulting in some voltage in the middle until one or the other or both have a fatal issue. – Kellenjb Dec 19 '11 at 20:46
  • @Kellenjb - It would be somewhere in between as noted in my answer. The only way it could be 12V for both in parallel is if the 5V source had infinite internal resistance. – Oli Glaser Dec 19 '11 at 22:24
  • 1
    It would probably be 12V since a flyback converter doesn't have a way of dissipating excessive voltage. (It's why there's a load resistor to begin with) – W5VO Dec 19 '11 at 22:34
  • @W5VO Thanks for that... I am no expert on big PSUs like this and wasn't sure if they have feedback circuity that attempted to lower the output voltage or not. – Kellenjb Dec 20 '11 at 14:00
  • @Oli you might want to see what W5VO has to say as it might mean your answer is actually wrong. – Kellenjb Dec 20 '11 at 14:01
  • @Kellenjb - noted and agreed with, most likely it will hover around 12V. I just double checked a couple of typical ATX schematics and both had a load resistor on the +/- 5V lines (100 Ohms) and one had one between the +/-12V (270 Ohms) lines also. – Oli Glaser Dec 20 '11 at 20:56
  • The resultant voltage is irrelevant. What is important is that it is a Really Bad Idea™ and should never be done. – Majenko Dec 20 '11 at 22:30
12

If you put the 5V and 12V in parallel, the voltage would be somewhere in between depending on the internal resistance of each source.
If both sources have equal internal resistance then the resultant voltage would be 8.5V. This would apply for e.g. batteries or similar simple voltage source.

With two switching supplies though, as W5V0 has noted, the resultant voltage will likely be the higher of the two, as the lower rail cannot sink current (due to the diode) and will effectively look high impedance to the 12V rail. So all that should happen (see below) is the lower rail will rise to the potential of the higher rail.

It is not a good idea to connect two different supply rails directly due to the problems that can be caused by low impedance sources opposing each other and circuitry of the lower rail may not rated to handle the voltage from the higher rail.
However in the switchers case it's probable no magic smoke will appear due to the inability to sink current mentioned above. However it's possible that the lower rails diode won't like being reverse biased so much and any capacitors may not be rated for the higher voltage (definitely a possibility given the extremely competitive price these things aim for - every cent makes a difference)
If a mid point voltage source is needed then a regulator of some sort can be used to provide a low impedance source.

The link you provide is for connecting batteries of the same voltage, which can be regarded as completely separate sources. The rails in your PSU will share a common ground (like two batteries with their negative terminals connected together) If you try and connect them in series it will effectively short one of the rails to ground which is not good.

It's not too clear what you are trying to do with the outputs without a schematic or some more info as to what voltages and control system (e.g. protection, voltage/current adjust, etc) you want to end up with. For the minimum load on each rail you just need to use two separate resistors to ground.

Oli Glaser
  • 54,990
  • 3
  • 76
  • 147
  • 1
    "It's not too clear what you are trying to do with out a schematic"... He is trying to add a load to the PSU so the PSU can turn on, how can he be any more clear then that? – Kellenjb Dec 19 '11 at 20:50
  • I was referring to what he is wanting/planning to do with the outputs exactly. For example whether he wants +5V and +12V separate, or +17V, or something else. The minimum load part is clear, I'll edit to clarify. I'm a big fan of a small diagram/schematic for even the simplest stuff, as pictures generally reinforce/convey information well in these situations, particularly if the native languages of people involved in asking/answering are different. – Oli Glaser Dec 19 '11 at 22:15
  • 1
    I know you qualified your statement, but it's worth mentioning that switch mode converters usually don't have a way of sinking current. The only way the 5V circuit is contributing anything is if there's an explosion :) – W5VO Dec 20 '11 at 17:02
  • Yes, I agree with you about the switchers, I wasn't thinking of switching converters when I wrote that (silly, I know, as it's a PC supply so what else would it be..) I'll edit my answer to mention this. – Oli Glaser Dec 20 '11 at 20:36
4

A separate load resistor on each supply voltage is the only correct way to get both the +5 and the +12 loaded. However, depending on the supply, it might suffice to load just the +5, since the +12 is usually reserved for disk drive motors.

JustJeff
  • 19,163
  • 3
  • 48
  • 75
4

I don't know what your electrical bench looks like, but in my situation I could find an old 4GB hard drive pretty easily and put it on as a dummy load. It might not be as portable as you're wanting, but it has served me in the past for a dummy load.

Bryan Boettcher
  • 1,627
  • 1
  • 14
  • 20
2

enter image description here

This is a circuit diagram of a computer power supply I added it to give you some help about your experiment:

1- If you connect a 12v with 5v in parallel you got the difference which is 7v

2- You cannot make the 12v and 5v in series because they have not the same ground

m salim
  • 402
  • 2
  • 15
0

Putting two power supplies of equal voltage together is not likely to result in a good distribution of current. Putting two power supplies of different voltage together is an even more terrible idea. Either the lower voltage supply shuts off (and is useless) or it sinks current. If it is designed to sink current, it will reduce the available current from the higher voltage supply. If it is not designed to sink current (and most will not), any number of bad things may happen.

David
  • 81
  • 1
0

It appears that you are willing to take the high risk of "blowing" your power supply, just to save a "pre-loading" resistor! Yes, you should use a preloading resistor, as required, for each rail. Yes, you can not do better than that.

Guill
  • 2,430
  • 10
  • 6
-1

The straightforward way is that (from basic electrical engineering circuit analysis by David Irwin) "a series connection of current sources or a parallel connection of voltage sources is forbidden unless the sources are pointing in the same direction and have exactly the same values."

Karlo
  • 183
  • 1
  • 6
mbilal
  • 9
  • 1
-5

This is undefined. We can't connect two voltage sources in parallel until these volts are same. Similarly we can't connect current sources in series until these are same value.

Karlo
  • 183
  • 1
  • 6
Indira Samanta
  • 1
  • 4
    OP's question is not about theoretical ideal voltage sources, it's about "a computer power supply that I'm hacking together as a bench supply". – The Photon Aug 25 '12 at 18:46