Computer PSU tripping as Bench Power Supply

Question

I am currently attempting to construct a bench power supply using an old PC PSU (350W) and having some issues. I have managed to set up a basic standby and power on led's, an on/off switch and a dummy load on the 5v rail using a 5w 10ohm resistor. So far it has been able to power my arduino and some small standalone circuits with LEDs and such. All of these circuits have been possible to do using my macbooks USB ports, so have pulled < 2A.

Recently I purchased myself a 120 ws2812b led strip. It has been powered by my USB port for now, but cannot pass serial data and dims LEDs at full brightness as my laptop cannot supply enough power.

Hence leads to my issue...

I connected my led strip to the PSU via 5+ and ground lines. I then connected my Arduino to the data pin via port 6 and a small resistor (same as when using USB port). The LED strip is then grounded to the Arduino as well (to complete the data circuit). The Arduino gets its 5v power using my laptops USB port where it sends serial data to change lighting effects. Up until around 70% brightness, the PSU holds, before it then trips itself. On the very rare occasion, it makes it to full brightness for a short while (2-3s) before tripping. The Arduino continues to run correctly after these trips (as expected on a separate voltage supply). I have a voltmeter that measures up to 10A, so have not been able to use it to test the 5v lines Amps (rated for 32A on side of box). I have tested the voltage however, which significantly drops to ~4.6V on the 5v rail before tripping.

I was wondering if anyone knows why this is the case. The PSU was fully functional in a pc two weeks ago and all I have done is cut off the connectors and organise the cabling. Any help is appreciated.

Thanks in advance!

BFB

Note, have jumped the terminals as not all cables fit in one socket. The terminals are 10A rated. I don't mind if one of the jumper cables burns out, though they should be fine (correct me if I'm wrong though ;D).

Answer 1

As you have seen, ATX power supplies are not good for using them as generic bench power supplies, and there is no reason to expect them to be good generic bench supplies.

As long as the power supplies are used reasonably with a typical or expected load such as a computer, and when they are used within specification, they generally work fine without issues. Typically, in a computer, the power supply would see at least some load on each of the output voltages, and there are no sudden jumps in the load current, and even if there is, there is a maximum rate of 1A per 1 microsecond change required for the current.

As there are many manufacturers and models of ATX supplies, and they are manufactured since 1995, there will be differences in them how they work internally. Early computers tended to use a lot of current on the 5V output to power the CPU, while later on CPUs are powered from the 12V output, so different power supplies can use a different focus which output is the main rail for large currents and which output is uses for global feedback for regulation.

The problem is using only one output, as multiple voltages may come from singe output transformer, and only use one of them for regulation. Early units typically used 5V and 12V on the same transformer, so using only 5V current would make the unused 12V supply voltage rise too much so that the 12V overvoltage protectiom triggers and shuts down the power supply.

Since you only load a single output rail, most likely something similar happens in this case too.

So even if some ATX power supplies may work just fine when used as a bench power supply, there is no general specification how each and every ATX supply could be used.

That is why there are many different opinions what component values such as capacitors and dummy load resistors should be used on which rails to make it work, but there is no guarantee which combination will work with your power supply the best, and even then it would waste a lot of watts as heat with dummy load resistors to keep the power supply running in a stable fashion.

And still the supply might perform poorly under dynamic load. It might shut down when something is suddenly connected to it, such as a high load, or even a small load if it takes a large surge current to charge capacitors. It might also shut down when a load is suddenly removed. That is because the power supply is not required to work outside specifications, which do have limits for both current step magnitude and the speed at what rate the current is allowed to change.

You can try adding high power dummy load resistors to 3.3V, 5V and 12V outputs, and perhaps add 470 to 2200 uF capacitors on each output help the power supply to stay better within regulation limits, but still, don't expect it to work perfectly as a bench supply.