Important protection mechanisms?

[andy572]

It's system upgrade time again.

I'm putting a new box together with a mix of currently used, second-hand and new parts.

Looking into power supplies, the specs and review scores don't always tie up with clear statements about protection within those reviews. Some of the PSUs that are popular on hardware sites and forums even appear lacking in this respect when compared to somewhat forgotten alternatives. They've been awarded very high scores despite having had a passing spotlight on, for example, the lack of over-temperature protection (OTP) that their reviews stated a unit should include.

It's easy to think that a supply should have all of the safety features but which are the ones that you guys consider necessary? Or if you like, what can a power supply do without?!

[kalniel]

Can you link to some examples where reviews have applauded a unit despite missing safety features?

Basic fusing is the most important thing to me! Add in over volt/current protection and you've got the most important things to me. Temperature is less vital personally and I guess might be covered by the other protections before it'd be dangerous (ie a good unit shouldn't overheat unless it's failed somewhere else.) But OTP is a given on most computer components so why not a PSU as well I guess.

[peterb]

IMHO, over voltage and over current protection are essential. Over temperature is a tricky one, because the cause of over temperature will be over current (normally) so the over current protection should cut in before the PSU overheats.

The exception is if the airflow is impeded and causes over heating. But I suspect most PSU regulators have some thermal protection built in, it's just not considered sexy enough marketing to shout about.

[andy572]

I'll try to return with links and excerpts but it was at least a couple of reviews from Tom's Hardware, TechPowerUp and/or JonnyGURU that stuck in mind recently, kalniel.

I don't recall if they were all for an EVGA G2 but remember that the first page for the 550W model on Tom's referenced the missing "important Over-Temperature Protection (OTP)."

[kalniel]

Fair enough. It wouldn't bother me, given the other protections listed (over volt, over current, over power, short circuit etc.)

[andy572]

I still wonder about many of the mechanisms and how they potentially relate to one another but that's just the sort of context that is nice to have, Peter. Likewise to the previous post. Thank you both.

In case it's of interest, here's the OTP-related comment from Page 2 of TechPowerUp's 650W G2 review:

The unit doesn't feature Over Temperature Protection (OTP), which is unusual for a Leadex platform. OTP is crucial, especially if the PSU features a semi-passive operation, which can dramatically increase internal temperatures in cases were the PSU has, for example, been installed with its top exhaust grill facing downward. Absolutely make sure to avoid installing semi-passive and passive PSUs in such a way. Should a PSU's fan fail, only OTP can save the day. In short, every PSU platform should include OTP.

[kalniel]

Yet they don't list it as a negative in the conclusion? I'm not sure they believe their own statement.

[andy572]

Yeah, that's what seemed so odd to me when reading the reviews. Particularly with the TechPowerUp one, given the ways it asserts or underlines the importance of OTP in that paragraph alone.

[andy572]

[New post to keep this easy to follow!]

After trying to remember for since I started the thread, hold-up time has just returned to mind.

It seems to be Pass or Fail and besides comparing numbers in mere milliseconds, some of the numbers created new questions. For instance, why would a lower-rated supply that appears to use the same components as its larger capacity brother have a shorter hold-up time? Where's it 'leaking'?! One might think it could be a sign of variation in production of the capacitors, though the small sample of two reviews and a relevant comment in a review for another PSU, each noted low hold-up time. It happens to be an EVGA G2 again. Two 550W units failed the test and the comment in a 650W G2 review described the reduced hold-up of a 550W box (that wasn't reviewed).

After seeing my present system somehow run through rare events in the supply of electricity that were far beyond mere milliseconds, I'd love to understand what was 'luck' and what might have been down to hold-up time.

Like before, if there's more to it, I'd be very pleased if I could gain a better grasp of what the numbers mean in the real world beyond 'longer is better' too. Everything I've found on-line so far has only included obvious descriptions or in the case of reviews, included Pass or Fail where supplies didn't meet the 16ms 'minimum' ATX spec.

[peterb]

Hold up time will be related to load and the sensitivity of the connected components, so difficult to define precisely. If the motherboard has an energy storage component incorporated that may affect system hold up time.

Hold up time could be specified at a specific load, but for a domestic PC is a largely meaningless parameter.

It may be of use if there is a separate detection of power failure and the hold up time is sufficient to complete current processor operations in (say) a real time transaction based process, but generally a UPS will perform that function and allow a controlled shutdown of the system.

[andy572]

Are you referring to motherboards where the power side of things is well-specced?

Your closing part relates to one piece of text that I read in relation to completion. Not sure if I could find it again but it did refer to an example like processor operations – something that could run in a fraction of a second and would have just long enough to finish.

That did leave me pondering the chasm between that length of time and a system managing to keep going through a longer loss earlier! A UPS immediately came to mind as well as, for a different example, how long surge protectors might be able to 'buffer' the cut of supply.

[peterb]

Surge protects don't buffer the supply, they just switch on to absorb the spike. They have a limited life and I personally think they are a waste of time.

For home use a UPS will fulfill all the requirements and if connected to the system it is connected to to initiate automatic shutdown, it only needs a hold up time of around 5 minutes.