PSU tier list 2.0

[Onus]

...
Now I can see the possible relevance when back in the AT days and psu switches were toggle on/off not ATX soft on/ hold off, that that might be a concern, but psu design has come a considerable way since then and I feel like it's nothing more now than an excuse to say something bad about a good psu or something good about a bad one. I personally would much rather have a quality 550G2 psu with its outstanding ripple and great everything else except for its 12ms pwr_ok, than to have spent less for a 550GS unit with mediocre everything, including protections, just because pwr_ok is higher than 16ms. And yet that is exactly what some claim, the GS is a better psu than the G2.

In a pigs eye...

Pretty much this. This is a mountain being made into a molehill. A decent UPS has a <4ms switching time. For any system that will be plugged into one, hold up time is a non-issue.

 

[turkey3_scratch]

...
Now I can see the possible relevance when back in the AT days and psu switches were toggle on/off not ATX soft on/ hold off, that that might be a concern, but psu design has come a considerable way since then and I feel like it's nothing more now than an excuse to say something bad about a good psu or something good about a bad one. I personally would much rather have a quality 550G2 psu with its outstanding ripple and great everything else except for its 12ms pwr_ok, than to have spent less for a 550GS unit with mediocre everything, including protections, just because pwr_ok is higher than 16ms. And yet that is exactly what some claim, the GS is a better psu than the G2.

In a pigs eye...

Pretty much this. This is a mountain being made into a molehill. A decent UPS has a <4ms switching time. For any system that will be plugged into one, hold up time is a non-issue.

No no you guys still have it all backwards. It's the ones with >16ms for the PWR_OK signal that are not as good because they set the voltage limit too low in order to get the value that high. The 550 G2 is safer than the 650 G2 (which has like 21ms) because the 550 G2's threshold is at a safer in-spec voltage.

The requirement for an 11ms transfer time UPS or less is another issue, but having less than 16ms isn't harmful in any way (assuming the voltage is safe) other than more likely shut down during rapid AC loss or when connected to a UPS.

 

[turkey3_scratch]

I own a UPS for almost a year and I have tried it with a shitty corsair vs650 and a high quality EVGA G2 850W and none of them shut down when the power went off.......

The G2 850 has 17.1ms of AC_LOSS to PWR_OK time, I wouldn't expect it to shut down.

@SOmeone Somewhere: Whenever there is AC_LOSS and my PC shuts down from it, right when it starts back up it boots by itself to Windows for some reason. I think that's what he meant.

 

[Someone Somewhere]

Yeah, that's an option in the BIOS.

 

[superninja12]

all this tech stuff about the g2 confuses me....
in noob language, whats wrong with it and could it have a big consequence?

 

[Onus]

I didn't recheck the spec, but I believe the ATX spec says that a PSU should be able to continue running if it loses power for up to 16ms (such as might happen during a brief line disturbance).
When the power is cut to a PSU, it should drop the PWR_OK signal before its output voltage(s) drop out of spec.
Some manufacturers are artificially keeping the PWR_OK signal active even after the output voltage(s) have dropped below the ATX specification for them. This means a PC may shut down or crash, possibly in the middle of a write-to-disk or some other "data-risky" operation. If the PWR_OK signal drops before the outputs drop out of spec, the motherboard can halt such data-risky operations before the dropping voltage to it causes such a crash.
Note that the 16ms hold-up time is required at full output. If a PSU can manage that, than the hold-up time should be even longer under less than full output load.

Question then. AC input fails. Is PWR_OK being dropped by a timer from that point, or is PWR_OK dropped by a measurement of the output? If the former, the PSU wouldn't have the problem under light loads, but if the latter, the PSU would still fail to de-assert PWR_OK before the output voltage dropped out of spec.

In either case though, this situation won't damage hardware, and only may cause data loss. Also, in either case, the risk would be 100% mitigated by use of a UPS that switches to battery in <10ms (which should be most/all of them).

 

[Darkbreeze]

Moral of the story, don't write/copy data from drive to drive or work on mission critical documents and applications during an electrical storm when it's likely or probable that there may be a power loss. Even if you HAVE a UPS, the potential for damage regardless of breakers, surge protectors and UPS devices makes it a bad idea. And if you live somewhere where the power grid is so bad that this is something likely to occur without the presence of electrical storms, you have bigger issues than the pwr_ok signal.

 

[superninja12]

browing through ibuypower psu's

https://gyazo.com/d3e2ca6dc48fd9d38954ede8f465804a

 

[turkey3_scratch]

Question then. AC input fails. Is PWR_OK being dropped by a timer from that point, or is PWR_OK dropped by a measurement of the output? If the former, the PSU wouldn't have the problem under light loads, but if the latter, the PSU would still fail to de-assert PWR_OK before the output voltage dropped out of spec.

That's a good question. If I was a betting man I'd say measurement.

@superninja: You have all rights to be confused, because there are multiple types of holdup times so it's important to not confuse AC_LOSS to PWR_OK holdup time with AC_LOSS to DC_LOSS holdup time.

 

[Nuckles_56]

browing through ibuypower psu's

https://gyazo.com/d3e2ca6dc48fd9d38954ede8f465804a

I did laugh over the fact that it cost less to put a 750W GQ in than it did to put the CX750M in

 

[Someone Somewhere]

I didn't recheck the spec, but I believe the ATX spec says that a PSU should be able to continue running if it loses power for up to 16ms (such as might happen during a brief line disturbance).
When the power is cut to a PSU, it should drop the PWR_OK signal before its output voltage(s) drop out of spec.
Some manufacturers are artificially keeping the PWR_OK signal active even after the output voltage(s) have dropped below the ATX specification for them. This means a PC may shut down or crash, possibly in the middle of a write-to-disk or some other "data-risky" operation. If the PWR_OK signal drops before the outputs drop out of spec, the motherboard can halt such data-risky operations before the dropping voltage to it causes such a crash.
Note that the 16ms hold-up time is required at full output. If a PSU can manage that, than the hold-up time should be even longer under less than full output load.

Question then. AC input fails. Is PWR_OK being dropped by a timer from that point, or is PWR_OK dropped by a measurement of the output? If the former, the PSU wouldn't have the problem under light loads, but if the latter, the PSU would still fail to de-assert PWR_OK before the output voltage dropped out of spec.

In either case though, this situation won't damage hardware, and only may cause data loss. Also, in either case, the risk would be 100% mitigated by use of a UPS that switches to battery in <10ms (which should be most/all of them).

Definitely measurement.

I'd be very surprised if the motherboard was able to prevent writes in that kind of time frame, though. Write latency on a typical SATA drive is well into the milliseconds.

 

[logainofhades]

browing through ibuypower psu's

https://gyazo.com/d3e2ca6dc48fd9d38954ede8f465804a

Holy, crap, choices, Batman!

 

[Karadjgne]

So according to my math, pretty much what that standard is saying is that any power interruption of less than 16ms should not adversely affect the psu whatsoever, it would continue normal operation. Now I may be wrong, but in all the years I've been cognizant of electricity, I can honestly say I don't think I've ever come across a power interruption of such limited time. When the power goes out at my house, its for hours, not ms.

If you are going to drop $1000+ on a good system, spend the extra $100, buy a decent UPS and that'll give you minutes to shutdown. It's not a difficult concept to grasp.

 

[Karadjgne]

@Logain,

Why do you think the name of the company is ibuypower? Because you will, repeatedly, every 6 months or so

 

[Karadjgne]

@Onus,

From what I understood about UPS, there wasn't any switching involved. 120v charges the battery, 12v from the battery is stepped to 120v to power the ups outlets. It's a constant cycle, if power goes out or spikes etc, its all absorbed by the battery, the outlets aren't run from the wall in bypass.

 

[turkey3_scratch]

@Onus,

From what I understood about UPS, there wasn't any switching involved. 120v charges the battery, 12v from the battery is stepped to 120v to power the ups outlets. It's a constant cycle, if power goes out or spikes etc, its all absorbed by the battery, the outlets aren't run from the wall in bypass.

Only thing I don't quite get is that, if the outlets aren't run from the wall in bypass, why would there be any need for transfer time in the first place?

 

[Darkbreeze]

There isn't, or shouldn't be. That's the whole reason a UPS system must have a compatible sine wave form or NORMAL operation, not power loss operation, is affected. The problems with the sine wave compatibility don't exist only in the event of a power loss, they exist, or don't, any time the unit is powered through the UPS because it's providing the power 100% of the time, as mentioned there shouldn't be any bypass. If there was, then wave form compatibility would only be an issue after a power loss.

 

[turkey3_scratch]

So according to my math, pretty much what that standard is saying is that any power interruption of less than 16ms should not adversely affect the psu whatsoever, it would continue normal operation. Now I may be wrong, but in all the years I've been cognizant of electricity, I can honestly say I don't think I've ever come across a power interruption of such limited time. When the power goes out at my house, its for hours, not ms.

I experience power interruption at my house in the ms at times. It happened about a week or two ago multiple times in one day. My Xbox and router turned off from the interruption but my PC (with an 850 G2) stayed on since it was working at minimal load and has good enough AC_LOSS to PWR_OK time.

 

[Onus]

@Onus,

From what I understood about UPS, there wasn't any switching involved. 120v charges the battery, 12v from the battery is stepped to 120v to power the ups outlets. It's a constant cycle, if power goes out or spikes etc, its all absorbed by the battery, the outlets aren't run from the wall in bypass.

That's how a true online UPS runs. Most [smaller] UPS units are actually standby units, that switch over to battery faster than a typical load will even notice, e.g. <4ms.

Turkey, I'd think dropping PWR_OK would be by measurement as well. To wait until a DC output has dropped out of spec in order to maintain an artificially high hold-up time is indeed cheating; fortunately it's not the sort that kills hardware.

 

[Onus]

So according to my math, pretty much what that standard is saying is that any power interruption of less than 16ms should not adversely affect the psu whatsoever, it would continue normal operation...

Exactly; this. It shouldn't matter at all.

...which reminds me, I need to order a replacement battery for my BE750G...
Edit: Done!

 

[Karadjgne]

I hate replacing the battery in my minuteman 700pro. If it swells even slightly you almost need a crowbar to get it out of the slot it sits in. Got @ a year or so on mine atm. Of course it's all prolly my own fault, I use U1500 instead of U1200 batteries.

 

[turkey3_scratch]

I'm looking to buy some really shady power supplies for dirt cheap. Anyone have any ideas of units I could get?

 

[Turb0Yoda]

I'm looking to buy some really shady power supplies for dirt cheap. Anyone have any ideas of units I could get?

HP OEM PSU

 

[turkey3_scratch]

Is this the weirdest PSU you've seen or what?

 

[turkey3_scratch]

I'll be venturing to Microcenter and getting 3 PSUs for $38.91.

Also, question. While we are on the topic of PWR_OK, say that they set the measurement of the PWR_OK signal to cut out at 10.8V. Does this mean normal undervoltage protection in the PSU is set to 10.8V? Are they the same circuits?