1
\$\begingroup\$

Following the well known instructions I've built the PSU from a spare 500W ATX. I've used 10 Ohm resistor for artificial load on 5V line.

When I connect 1A load to 12V line, the ATX shuts off. I need to wait a bit and restart it.

This only happens if I connect the large load rapidly (i.e. turn on the LED driver with around 1A load). When gradually increasing the load, I can reach 2.5A with no issues.

Say I connect a 0.5A load and it works. Let it run for a few minutes, then load it with 1A, let it run again and eventually I can have my 2.5A LED light connected. I can turn off the ATX and turn it on again and it still works.

If I however keep it off for a few minutes and turn on again, connecting anything above ~0.8A causes immediate shutdown.

According to the sticker, 12V lines are rated 15A / 17A. I've tried increasing the load using 3.2 Ohm (10R and 4R7 in parallel) but that didn't help.

Is the ATX faulty or have I overlooked something? The only possible explanation that came to my mind is that the LEDs driver I'm using has some extreme power peak when warming up. So maybe it's not about turning off the PSU but about turning off the LED driver... will test it against some 12V motor

\$\endgroup\$
6
  • 3
    \$\begingroup\$ Are your LEDs just LEDs or do they have an electronic power-supply built in. Add the details into your question (rather than bury them in the comments). \$\endgroup\$
    – Transistor
    Commented Oct 14, 2020 at 21:27
  • 1
    \$\begingroup\$ It appears that the converter's feedback/control network has a "response-to rapid-load-changes" problem. Rapid load change causes a short-time overload because the converter's feedback loop acts slowly. Thus the converter's overload protection kicks in. If I however keep it off for a few minutes and turn on again, connecting anything above ~0.8A causes immediate shutdown. that's because the output capacitors discharge in a few minutes. And, at turn-on, some high peak currents (due to empty capacitors - act like nearly zero-Ohm) flow. And the converter's overload protection kicks in. \$\endgroup\$ Commented Oct 14, 2020 at 21:34
  • 1
    \$\begingroup\$ What "well known" instructions, who knows if they are good or bad? ATX power supplies are not good general purpose power supplies. They do not work well with random loads in random situations. Care must be taken not to exceed rated load step, and that the step happens at the rated time. They are especially bad working properly, if all other outputs are unloaded except the 12V, and you disconnect the load so there is suddenly no load at all on any output, the voltage will overshoot and most likely cause the overvoltage protection to shut it down or at least the control loop to go haywire. \$\endgroup\$
    – Justme
    Commented Oct 14, 2020 at 21:40
  • 1
    \$\begingroup\$ Is the ATX PSU in good condition? I've had one with leaky/burst capacitors that would work OK under no load / light load but fall over if you plugged too much into it as there was not enough capacitance to maintain good power, triggering the PSU to shut down. \$\endgroup\$
    – John U
    Commented Oct 15, 2020 at 8:35
  • \$\begingroup\$ In the ATX specification there is a limit to the rise rate of current. Someone mentioned the value in a comment here about two weeks ago. If your inrush is too large, you may simply be out of specification. \$\endgroup\$
    – winny
    Commented Oct 15, 2020 at 9:42

2 Answers 2

2
\$\begingroup\$

Load transients are limited to allow for more efficient power supplies.

The electrical specifications for your PSU should list what they are, and system builders need to verify that the PSU capabilities meet the requirements of the mainboard, just as they do for maximum current.

For example, Intel's power supply design guide lists transient slew rates of 1.0 A/μs, and a step size of 40% on the +12V1DC rail as required from a compatible PSU.

If your PSU can deliver power to a mainboard as specified, but go into protection mode when connected to your load, then it is likely that you have more stringent requirements on the PSU than a mainboard would have, so either you need a PSU that meets these requirements, or you redesign your application to relax them, e.g. with capacitors up to 3,300μF that are charged during the inrush current limited phase while the PSU turns on.

\$\endgroup\$
0
\$\begingroup\$

If you want to power heavy stuff, the PSU will automatically shut down unless you bypass the security as explained in this video: https://youtu.be/HbS6ZK1fDsg

It works for all models I tried; you just need to find the chip that deals with the safety find the right pin (eg. fault protection latch, FPOB, Fault Protection Output...) and short it to ground, you may want to add a switch to only disable it when needed).

(PS: for your safety avoid disabling this protection)

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.