List of Dangerous ATX Power Suplies

I was debating whether or not to title this post as it did, but after the 3rd non-compliant PSU, either I method is not the best or there is something very wrong with certification.

I am not an electrical engineer but took some courses and have dabbled with hardware for more than two decades now. I have used all the type of bad PSUs one can think of but soon realized that is not the best idea. I have posted my list of good PSUs a while back.

The List

I never recommend you use these unattended, though the ones with 12v short-circuit protection (SCP) can probably be used for testing hardware wtihout major issues.

I have used some of these for many years until problems propped up but cannot rely on them as I keep some machines on 24/7. The danger of a fire hazard is not worth $100. I am now actively testing all PSUs before using them, given the failures I have witnessed and damages I had in hardware.

For some of these units, I have contacted the manufacturer and asked for information. I will update with that information if a recall is issued and which Serial Numbers are affected.

Complete lack of SCP:

• Seasonic M12-II Evo 520W (2 samples, no OCP/OPP either)
• Seasonic S12-II (and no OCP/OPP)
• XFX Pro Core 550W (based on M12-II Evo)

Only 12v rail protected:

• FSP/Fortron FSP400-60HLN (dead after 5V short)
• LC Power LC6550 PN.5 - 550W ATX 12V
• Nox Urano 450W (2 samples, 1 dead after 5V short)
• Nox Urano SX 500W

Some Background

Most basic problems you have with cheap PSUs is little reliability or power spike protection. This means that it is sometimes cheaper to buy a 3x the cost PSU than 4-5 cheap ones over 10 years - never mind other problems it may cause or labour to replace it.

Some of the most problematic missing features have been the lack of a Metal Oxide Varistor (MOV), which protects against power spikes. The result is usually damaged hardware and even lost data. This is easily solved through the use of a surge protector but I have also come to realize that other types of protections are critical, like short-circuit protection (SCP).

As per ATX spec, SCP is required on all rails, even 5VSB, so this should not be an issue. Usually, the PSU should power off when a short is detected but most can (and should) turn back on when the short is removed.
I have a Seasonic S12-II PSU which I have tested and does this to ATX specs, so I don't think the methodology was at fault in some way.

What happens when these protections are not in place is fire hazards and damaged hardware:

Failure of SCP on 12v rail

As you can see by the marking, the traces on the motherboard were burning up before I managed to switch off the PSU. This was caused by a defective GPU with a MOSFET short.

Before powering down, the PCB material was bubbling up from the high temperatures and I think it is almost a miracle that the traces are still there and the motherboard didn't develop a short itself. Fortunately, I was around testing this hardware.

Further testing with two other PSUs resulted in both not even powering up, which is expected behavior in this situation!

These types of shorts can happen for environmental reasons, like humidity causing corrosion and shorts, or due to failure of power components like MOSFETS - which can fail shorted. Surface mount capacitors can also fail shorted, so it is critical that PCs that will be left unattended have these features working.