Beware of Exploding Gigabyte PSU's being Dumped by Newegg in Forced Bundles

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Not at all. A properly designed PSU will not have a catastrophic failure under those conditions. This has been standard procedure for PSU testing pretty much since anyone started to take PSU testing seriously. It's only a problem when there is a critical fault with the unit itself, whether a design flaw or a faulty part in the unit.

A power supply with protections in place should basically never fail the way these units did.

OPP testing has no standard procedure for testing or spec definition so it has not been standard procedure for PSU testing pretty much since anyone started to take PSU testing seriously.

Since there is no specification definition there is no way to apply a fair test of units across the board since each design and implementation can/is different.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Then why are you defending Gigabyte? They're clearly in the wrong here and haven't taken reasonable actions, which should be to recall these products. These PSU's are a fire hazard waiting to get much worse. Instead Gigabyte wants buyers to contact them for a replacement which is not going to be most people who own these products, and they know that. All it takes is for someone to buy a RTX 3090 and use it with these PSU's and failure is inevitable.
Pointing poor test methodology that is being misinterpreted by people who don't know what they are looking it is not defending Gigabyte.We reviewed one of these units. It was not good unit when we reviewed. On the flipside Aris sold them a Cybenetics award.
 

GoodBoy

2[H]4U
Joined
Nov 29, 2004
Messages
2,188
OPP testing has no standard procedure for testing or spec definition so it has not been standard procedure for PSU testing pretty much since anyone started to take PSU testing seriously.

Since there is no specification definition there is no way to apply a fair test of units across the board since each design and implementation can/is different.
OPP was put into the PSU design by gigabyte.

Testing procedure: Put the unit into an 'over power' situation?? lol, how the hell else is that 'feature' supposed to be tested, please tell us. You put the unit into an OP scenario. If it dies, then Gigabytes' OPP isn't worth shit.

No one cares about Cybenetics. If Gigabyte wanted that certification, you would think they would have listened to his test results instead of ignoring it. The one thing it points out is that his 'certification' should be as stringent as his testing. Beyond that, has no bearing on Gigabytes' failures. Failure to design. Failure to act swiftly. Failure to reimburse for damages done. Failure to step up. Failure to impress. Failure to accept responsibility. Failure to admit they messed up.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
OPP was put into the PSU design by gigabyte.

Testing procedure: Put the unit into an 'over power' situation?? lol, how the hell else is that 'feature' supposed to be tested, please tell us. You put the unit into an OP scenario. If it dies, then Gigabytes' OPP isn't worth shit.

No one cares about Cybenetics. If Gigabyte wanted that certification, you would think they would have listened to his test results instead of ignoring it. The one thing it points out is that his 'certification' should be as stringent as his testing. Beyond that, has no bearing on Gigabytes' failures. Failure to design. Failure to act swiftly. Failure to reimburse for damages done. Failure to step up. Failure to impress. Failure to accept responsibility. Failure to admit they messed up.

Already explained it: https://hardforum.com/threads/bewar...forced-bundles.2012841/page-5#post-1045111986

And OPP wasn't put in by Gigabyte. It was done by MEIC.
 

hititnquitit

[H]ard|Gawd
Joined
Apr 16, 2018
Messages
1,386

cybereality

[H]F Junkie
Joined
Mar 22, 2008
Messages
8,492
Fact is, these were dying under normal use conditions. The high failure rate is what prompted GN to do the video.

It's not just that they die under some unrealistic test condition, which is the false narrative Gigabyte is trying to push.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095

emphy

Limp Gawd
Joined
Aug 31, 2016
Messages
298
OPP testing has no standard procedure for testing or spec definition so it has not been standard procedure for PSU testing pretty much since anyone started to take PSU testing seriously.

Since there is no specification definition there is no way to apply a fair test of units across the board since each design and implementation can/is different.

Why would you want a "fair test"?

Psu's were breaking down in the field, reported to be at much higher than normal rates. One would want tests to detect why they are doing so, not some test designed to compare them with other psu's for a review. Gigabyte had their "fair test" when costumers built the psu's into their pc's that stopped functioning due to the psu breaking down.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Why would you want a "fair test"?

Psu's were breaking down in the field, reported to be at much higher than normal rates. One would want tests to detect why they are doing so, not some test designed to compare them with other psu's for a review. Gigabyte had their "fair test" when costumers built the psu's into their pc's that stopped functioning due to the psu breaking down.

Because I can make any unfair test fail any unit I want. That is not a legitimate or ethical way to do my job. It also would not allow me to ever draw a legitimate conclusion or compare units. Some of us have a job to do and to do it correctly.
 

Nafensoriel

Limp Gawd
Joined
Nov 23, 2015
Messages
382
So are you saying that if tech jesus took a high quality PSU (like a Seasonic) and did the same test that it would explode?
Not wanting to put words in mouths but I'd guess he's saying that all data can be manipulated when dealing with something as complex as a psu.
Steve also is not an unbiased source as much as he claims to be. He is in entertainment news after all.

Hell from my own experience with industrial level PSUs I can tell you the PC space ones are not exactly built to what everyone expects as "robust" standards. PSU space goes "Ripple is 8 YAY!" and I look at the tech sheet for a part and... 3. Of course, there is a big difference between a consumer who gives a damn PC parts list and a 150k inline test setup.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
OVP, OPP, OCP, SCP, OTP and BOP Definitions: These protections are only designed to protect the unit itself, not the whole system.
https://landing.coolermaster.com/faq/what-are-the-definitions-of-ovp-opp-ocp-scp-otp-and-bop/

OPP, over power protection should protect the power supply from damage, not explode, catch fire etc. after it trips once on just a normal restart.

A bad batch of anything really means very poor quality control. Been in manufacturing for over 20 years, several different companies. Taking a sample and testing on every batch routinely is the norm, taking samples when a new batch of material is used is the norm (a.k.a. a different manufactured component or material). Taking a sample and testing for changes in the formula. Any design change is not only sampled and tested but engineering review mandatory. During a run routine samples are also conducted. Last company had ZERO issues destroying (because of testing) $36000/day on one line while having way more than 1 line of production for routine samples. Toyota routinely takes one whole car off the line to test and destroy afterwards, routinely, a complete car value lost.

I see zero excuse for Gigabyte in this, this points to a very shady company who really do not care about their customers, extremely poor QC for their power supplies. Even if OEM made them for Gigabyte it is still falls on Gigabyte to ensure what they are selling is good quality and not fail. For example FDOT of Florida would do their own separate QC checks for pipe we manufactured to them, if failed the company gets to buy back the whole batch. Zero tolerance.

Combined with Gigabyte poor QC or really give a shit about their customer in this, they then pretty much ignored it when plenty of evidence pointed a rather big issue. You can't tell me the many return samples of failed power supplies did not show a big issue to them. This is really gross negligence overall.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
So are you saying that if tech jesus took a high quality PSU (like a Seasonic) and did the same test that it would explode?

HIm, probably not. He doesn't know what he is doing. Could, I or Lee, or Jeremy, etc? Sure, setting units up for failure is easy.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
HIm, probably not. He doesn't know what he is doing. Could, I or Lee, or Jeremy, etc? Sure, setting units up for failure is easy.
The problem I had with his report was that it did not, except for one if I remember right, test each PS on normal loads and a torture test. If it failed during normal design loads, maybe up to 80% power then that would indicate much better than taking it out of the box and putting 120%+ over load on each one.

I am thinking part of PS testing, the protection circuits or abilities should be tested, that can be very important to users having $2000 plus video card or cards, $4000 CPUs etc. As for a method or protocol to test them -> the designer and builders of power supplies should have this already down pat or at least how they approve a given design before manufacturing. That same protocol or methodology could be adapted by competent reviewers, the issue is the reviewer finding one bad power supply, as in one failure could be very misleading and not represent well the power supply in general.

Reading Amazon.com feedback, the 8% that gave a 1 Star had fires, destroyed video cards, SSDs, melted cords, DOA and so on. If a Power Supply after protecting itself then destroys, catches fire etc. is then no real protection at all for the user when abnormal conditions exists or a user who overloads the PS mistakenly or ignorantly.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
Sure, find the OPP protection circuitry definitions and testing procedures.
Ask the manufacturers how they test, what QC is done on batches for these features. That is if they will be forth coming.

UL should have good tests as well except I cannot dig deeper since I am not a member:
https://www.ul.com/news/power-supply-standards-compliance

These over protection features need to work, they are listed, promoted etc. which users depend on. I hope reviewers actually will test these in the future.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Ask the manufacturers how they test, what QC is done on batches for these features. That is if they will be forth coming.

UL should have good tests as well except I cannot dig deeper since I am not a member:
https://www.ul.com/news/power-supply-standards-compliance

These over protection features need to work, they are listed, promoted etc. which users depend on. I hope reviewers actually will test these in the future.

The answer is that, as I have said before, there is no OPP defintion or testing procedure. So, I can make up whatever I want to fail a unit. We can't test what is not a standard becuase I can manipulate the test any way I want to fail a unit. And that is not ethical even though you guys want some pitchforks to raise.
 

hititnquitit

[H]ard|Gawd
Joined
Apr 16, 2018
Messages
1,386
You keep saying that you can manipulate the testing anyway you want to make a psu fail. We get it. You know how to kill them. That's not what were trying to understand (at least not myself). If OPP is an industry standard. Why is it so difficult to test properly?

Edit- is it because reviewers cant figure out what the OPP setting is prior to testing?
 
Last edited:

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
The answer is that, as I have said before, there is no OPP defintion or testing procedure. So, I can make up whatever I want to fail a unit. We can't test what is not a standard becuase I can manipulate the test any way I want to fail a unit. And that is not ethical even though you guys want some pitchforks to raise.
So basically you commit in not testing these features on a power supply. That is your choice and decision as a reviewer.


Reviewers I would expect to come up with a logical approach to testing using what ever methods are available and be creative as well using sound reasoning for these features. Plus refine them with good feedback as needed: I made up this below as only a thought, crude at best maybe laughable:

OTP, Over Temperature Protection
  • Likely scenario - failed fan, possible blocked or clogged fan
    • How to test:
      • Block fan or disconnect fan
      • Use OScope to record power supply output voltage for abnormal spikes
      • Run a load on the PS on Electronic or dynamic load capable of handling PS range for protection
      • Monitor temperature with IR camera to see hot spots and what component reaches over temperature
      • Power Supply shuts down -> Did it protect itself without damage
      • Look at components indicated hot by IR and visually check for damage
      • Check if power supply will operate normally after test
    • Purpose of test was if a fan failure of the power supply, will it protect itself from damage and not damage other components
OCP/OPP, Over Current/Power protection
  • These are similar, what rails to test if even if all are protected, I would just test the 12v rail(s). Anyways this is not rocket science dealing with power. P = VI in a DC circuit
    • How to test
      • Separate rails if multiple 12v rails are used would each need to be tested
      • Use OScope to record power supply output voltage for abnormal spikes
      • Hook up the rail or each rail one at a time to Electronic or dynamic load capable of handling PS range for protection: voltage and current
        • If the OCP/OPP is known, let say 130% (actual trip point would be in a range and depending upon circuit design could be affected like by temperature)
          • Rapid increase in Current/Power test for the rail to the 130% to see if it trips
            • If it trips it passes
          • If 130% it does not pass go until it is like150% max
            • If it trips or protects, measure the trip point that it did protect at
            • If it did not trip at 150% or whatever max value is reasonable then it fails
            • If fuse blows, some power supplies have external access fuses while others is on the board -> pass if replaceable fail if not
      • After all rails have been tested, check for normal operation, voltage under a normal load, temperatures, any component that visually looks stressed/damaged etc.
      • Note/reasoning: A slow over power condition but below the trip setpoint maybe tested but I would not think that would be a normal condition for a power supply to see, computer loads are quick to load and unload. The user who incorrectly configures a system with an insufficient power rated power supply could damage the power supply for something it was not designed for. These protections are for normal range of operations and for example an extra video card is added that is like 350w, this would be a sudden increase in power when used where the OCP/OPP should protect itself and hopefully the system as well
SCP, Short Circuit Protection, this one saved my ass a number of times from quality power supplies
  • How to test
    • Not hard, short out the positive and negative 12v rail (I would think any rail for a multiple rail PS would be enough for this check)
      • Power supply should not even turn on or very briefly otherwise it would trip on OCP or OPP instead which has a time delay most likely not fast enough to protect the PS
    • See if the PS resets and can operate normally after clearing the short
OVP, Over Voltage Protection, this one is the hardest one to test, probably one that should not be tested except by the most skilled and knowledgeable person
  • Only test if the voltage regulator circuit can be used to give a higher voltage on the output, for example on some regulators a shunting resistor is used going to ground. Decreasing the resistance, which drops the feedback voltage fools the power supply thinking its output is less than actual which can be measured. So voltage can be controlled up to test the setpoint for this trip.
    • Except voltage regulators can also be digital, meaning not so easy to test and can be vastly different from one design to the next
BOP, Brown Out Protection, here in Florida my PS routinely have to go through voltage spikes and drops in a fraction of a second. Worst power grid I've seen!
  • How to test:
    • The wide variance of voltage from your outlet to your power supply from sudden and even erratic drops in voltage or increases, times etc. is almost infinite
    • Use OScope to record power supply output voltage for abnormal spikes
    • With normal load on the power supply
      • Perform sudden drops on voltage going into the power supply and back to normal voltage:
        • 10%, 30%, 60%, 100% this may trip the PS or not
      • Perform longer term drops of input voltage to the power supply
        • If the power supply protects itself, as in shutting down and can operate normally afterwards then it effectively protected itself from Brown Outs
        • If it can protect itself at all possible voltage drops and raises I think would be impossible to test for
 
Last edited:

GoodBoy

2[H]4U
Joined
Nov 29, 2004
Messages
2,188
Because I can make any unfair test fail any unit I want. That is not a legitimate or ethical way to do my job. It also would not allow me to ever draw a legitimate conclusion or compare units. Some of us have a job to do and to do it correctly.
Selective hearing much?

This guy is trolling us...
 

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,875
Pointing poor test methodology that is being misinterpreted by people who don't know what they are looking it is not defending Gigabyte.We reviewed one of these units. It was not good unit when we reviewed. On the flipside Aris sold them a Cybenetics award.
Ok so I wanna know what GN did wrong? They pushed the PSU beyond it's designed limit? What did they do that you consider wrong?
HIm, probably not. He doesn't know what he is doing. Could, I or Lee, or Jeremy, etc? Sure, setting units up for failure is easy.
So wait, if GN did the same thing to a SeaSonic unit then it wouldn't fail or it would? "Probably not" could mean many things to me. If it's easy to make a unit fail then that just tells me that PSU manufacturers aren't making good quality products in general. Keep in mind these tests were not done because GN felt like doing it but because an alarming amount of reviews on NewEgg were reporting their PSU's dying.

You want to defend Gigabyte then fine, you get your test setup and you do the tests and include some SeaSonics for good measure. Make sure to put a video of it on YouTube. You said you reviewed it but did you go so far as to see how easy these units fail? Does your review match what people are saying on NewEgg posts?
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Ok so I wanna know what GN did wrong? They pushed the PSU beyond it's designed limit? What did they do that you consider wrong?
Tested something that has no specification to test. With no specification to test, no test can be proper.

So wait, if GN did the same thing to a SeaSonic unit then it wouldn't fail or it would? "Probably not" could mean many things to me.

Probably not means many things to you, ok. But I answered your question.

"Him, probably not. He doesn't know what he is doing. Could, I or Lee, or Jeremy, etc? Sure, setting units up for failure is easy."

If it's easy to make a unit fail then that just tells me that PSU manufacturers aren't making good quality products in general.
Ok, and? You want a $100 power supply don't expect it to be the quality of industrial space stuff.

Nafensoriel already told you as much https://hardforum.com/threads/bewar...forced-bundles.2012841/page-6#post-1045114443

You want better quality start getting ready to pay an order of magnitude more.

Keep in mind these tests were not done because GN felt like doing it but because an alarming amount of reviews on NewEgg were reporting their PSU's dying.
And now we take Newegg reviews seriously? No. I have seen completely defective designs get great reviews on Newegg and great designs get terrible reviews. I have a power supply in the publishing queue right now that has 4/5 rating on Newegg. It is far worse than the Gigabyte. But, Newegg reviews yo!

You want to defend Gigabyte then fine, you get your test setup and you do the tests and include some SeaSonics for good measure. Make sure to put a video of it on YouTube. You said you reviewed it but did you go so far as to see how easy these units fail? Does your review match what people are saying on NewEgg posts?
I am not defending Gigabyte, I gave the unit a poor review when I reviewed it. Youtube is not the be all end all of anything. Most of us who know how to do this work still mostly work in print because we aren't trying to generate clickbait youtube videos. Make sure to like, follow, and subscribe for more on how to test powers supplies. Oops. My bad.
 

Derangel

Fully [H]
Joined
Jan 31, 2008
Messages
19,785
I am not defending Gigabyte, I gave the unit a poor review when I reviewed it. Youtube is not the be all end all of anything. Most of us who know how to do this work still mostly work in print because we aren't trying to generate clickbait youtube videos. Make sure to like, follow, and subscribe for more on how to test powers supplies. Oops. My bad.

And you want to accuse them of acting holier than thou? Christ, this just makes it look like you're blindly dismissing everything solely because you don't like the source. It is an exceptionally bad look for someone trying to claim superiority.
 

hititnquitit

[H]ard|Gawd
Joined
Apr 16, 2018
Messages
1,386
Tested something that has no specification to test. With no specification to test, no test can be proper.
Im admittedly a novice when it comes to psu testing specifics. Im not trying to argue or accuse. Or raise a pitchfork 😆 Im just trying to understand how an industry standard cant be tested? It makes no sense to me.
Gigabyte states that OPP is an industry standard in their press release regarding these units. Every major company has OPP included as part of the protections. Most go on to explain what it is and how it works. So im going to assume that OPP is in fact an industry standard that's accepted by most if not all of the major OEMs and psu companies.
From Seasonics website

Over Power Protection

If the output rating (maximum load limit) of the power supply is insufficient for the system’s requirements, this protection function is activated to switch off the power supply.

That seems simple enough. You already know the output rating/ maximum load limit(MLL) of every psu you test. You run the same tests. The same way everytime for every unit. To each units MLL.
The only way for the test to be biased would be for the reviewer to lie about the MLL prior to testing. Or intentionally try to overload the unit right?
Why would an experienced reviewer such as yourself need specifications to run such a test?
Am i missing something? If so, please explain.
 

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,875
Tested something that has no specification to test. With no specification to test, no test can be proper.
So there needs to be a specification to "load goes up and PSU goes boom"?
"Him, probably not. He doesn't know what he is doing. Could, I or Lee, or Jeremy, etc? Sure, setting units up for failure is easy."
You think this why?
Ok, and? You want a $100 power supply don't expect it to be the quality of industrial space stuff.

Nafensoriel already told you as much https://hardforum.com/threads/bewar...forced-bundles.2012841/page-6#post-1045114443

You want better quality start getting ready to pay an order of magnitude more.
The least I expect a PSU to do is not blow up and take parts of my PC with it. If pushing a PSU to it's limit will do that then these are junk. It's not uncommon for the industry to collaborate to make products at a certain price range within a certain quality. If that's the case then a lot of PSU manufacturers need to be thrown under a bus. This is crap I would find in late 90's or early 2000's PSU's, not something I should expect to see in 2021.
And now we take Newegg reviews seriously? No.
If a lot of people are saying the same thing as GN is saying then yes. Two positive sources don't make a positive product but two negative sources do make a negative product. If I'm in the market to buy a PSU I want a reviewer who takes apart the PSU and actually load tests it and explain the quality of the components inside it. If you aren't doing this then you're wasting everyone's time. You might as well say, "This Power Supply has a little bit for everyone".
I have seen completely defective designs get great reviews on Newegg and great designs get terrible reviews. I have a power supply in the publishing queue right now that has 4/5 rating on Newegg. It is far worse than the Gigabyte. But, Newegg reviews yo!
What makes you think your review is better? If you found a PSU that's worse than Gigabytes then I imagine it must explode the moment you plug it in.
I am not defending Gigabyte, I gave the unit a poor review when I reviewed it.
I need to see this review of yours to see how poor it was.
Youtube is not the be all end all of anything. Most of us who know how to do this work still mostly work in print because we aren't trying to generate clickbait youtube videos. Make sure to like, follow, and subscribe for more on how to test powers supplies. Oops. My bad.
It's a lot more impactful when I can see the product explode on camera than for you to articulate it in print. Seeing it go boom is far more damaging to Gigabytes brand. Speaking of which.

 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
Im admittedly a novice when it comes to psu testing specifics. Im not trying to argue or accuse. Or raise a pitchfork 😆 Im just trying to understand how an industry standard cant be tested? It makes no sense to me.
Gigabyte states that OPP is an industry standard in their press release regarding these units. Every major company has OPP included as part of the protections. Most go on to explain what it is and how it works. So im going to assume that OPP is in fact an industry standard that's accepted by most if not all of the major OEMs and psu companies.
From Seasonics website

Over Power Protection

If the output rating (maximum load limit) of the power supply is insufficient for the system’s requirements, this protection function is activated to switch off the power supply.

That seems simple enough. You already know the output rating/ maximum load limit(MLL) of every psu you test. You run the same tests. The same way everytime for every unit. To each units MLL.
The only way for the test to be biased would be for the reviewer to lie about the MLL prior to testing. Or intentionally try to overload the unit right?
Why would an experienced reviewer such as yourself need specifications to run such a test?
Am i missing something? If so, please explain.
Agree there on OPP testing, it is an overload shutdown, you have to OL the power supply. What I want to know if it will shut itself down, not destroy my computer and if reusable after the overload. Not destroying my computer is most important followed by that the PS actually protected itself and not become a bomb afterwards. All a reviewer has to do is say we are testing a rapid overpower condition, like adding an extra video card for mining that will add 300w to an already highly loaded power supply, will it protect itself and not destroy your system? In general, all protective features are based if the item like a power supply is properly rated for the system and an abnormal condition occurs. Spectre has a point with how tested on Gamers Nexus. While I applaud Gamers Nexus for actually testing a feature that is critical for many reasons, they did a test with the power supply operating at a higher power level, beyond specification for a period of time. 2 min may not sound significant but that is not how protective mechanisms work. They are meant for the most part for normal operation and in spec applications that suffers an abnormal condition. Gamers Nexus if they took the power supply from 70% and then gave it a sudden OL then I would think that would be a better test. Even 2 min or longer beyond specification will allow components to reach much higher temperatures, maybe not the ones looked at for over temperature. Since reviewers have failed in general to test these protective features, which can not only save your power supply, computer but maybe your life, I see no excuse not to test these and be scientific about it and refine the testing when good feedback comes at you like from manufacturers, engineers, readers and so on.

The Short Circuit Protection and Over temperature shutdown has to be the easiest to test. No one needs to hold hands and give permission, show how to do, to test a PS if it will shutdown on Overtemperature or melt or catch fire or destroy your computer if the fan fails at high loads. This may even protect your house from burning down. Yet we see ripple graphs after ripple graphs where as it makes zero difference for giving 1mhz of OC and may or may not indicate if the engineering and components used were of high quality. Short Circuit Protection is so easy to test and yet so vital not only to the power supply but also protecting motherboard and anything plugged into it from burning up.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
And you want to accuse them of acting holier than thou? Christ, this just makes it look like you're blindly dismissing everything solely because you don't like the source. It is an exceptionally bad look for someone trying to claim superiority.

Well, Youtube and the internet in general are not the sum knowledge nor keyholders to all or even correct information.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Short Circuit Protection is so easy to test and yet so vital not only to the power supply but also protecting motherboard and anything plugged into it from burning up.
The one issue with testing it is that it comes with a lot of risk on the testing end. When it doesn't work, I am out at least $2K or so repairing, calibrating, and certifying test equipment.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
The one issue with testing it is that it comes with a lot of risk on the testing end. When it doesn't work, I am out at least $2K or so repairing, calibrating, and certifying test equipment.
Maybe a do or die test without the test equipment exposed to a resulting deranged power supply can be done if test equipment does not have shunting circuits combined with OV SSR. Short out the 12v rail or one of them, turn on, if power supply does not turn on or turns off quickly and then operates normally afterwards -> a pass for the test. Sometimes using just old fashion voltage meters (which would not really react for quick spikes, but recorded by video could show a longer term abnormal voltage high or low). I would recommend taking a stab at a reasonable methodology and see over time how to improve with good feedback. These safety features can be one of the most important feature for a power supply for many reasons and for the most part are not tested.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Maybe a do or die test without the test equipment exposed to a resulting deranged power supply can be done if test equipment does not have shunting circuits combined with OV SSR. Short out the 12v rail or one of them, turn on, if power supply does not turn on or turns off quickly and then operates normally afterwards -> a pass for the test. Sometimes using just old fashion voltage meters (which would not really react for quick spikes, but recorded by video could show a longer term abnormal voltage high or low). I would recommend taking a stab at a reasonable methodology and see over time how to improve with good feedback. These safety features can be one of the most important feature for a power supply for many reasons and for the most part are not tested.

They do, but they are designed for the ATX12v standards which limits a unit to 240VA per rail. Do that SCP test on a single 12v rail say 850W unit that has 72A on the 12v rail and...........well that is a whole lot more than the spec calls for and not guaranteed to not fry everything down stream.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
They do, but they are designed for the ATX12v standards which limits a unit to 240VA per rail. Do that SCP test on a single 12v rail say 850W unit that has 72A on the 12v rail and...........well that is a whole lot more than the spec calls for and not guaranteed to not fry everything down stream.
Ok, I would then just skip the OScope analysis (which would be very interesting) and do the do or die test. A secondary coil separating the test equipment and PS power cable shorted could also be used to measure the current like a clamp on amp meter, there are ways to protect the test equipment but it needs to be accurate/calibrated I would think if used to validate results. The main point of test is will this power supply protect itself and still be operational afterwards with a short. If it shuts itself down or never starts to begin with then one can say it is likely any component or item causing the short has not been damaged further. Seeing if excessive amperage occurred before the shutdown is a great point. I worked with load centers with thousands of amps and sudden changes and there are monitoring meters and automatic recorders that capture the transients, I see nothing in any computer power supply that in that case could not be tested. Maybe just an added test gear is needed.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Ok, I would then just skip the OScope analysis (which would be very interesting) and do the do or die test. A secondary coil separating the test equipment and PS power cable shorted could also be used to measure the current like a clamp on amp meter, there are ways to protect the test equipment but it needs to be accurate/calibrated I would think if used to validate results. The main point of test is will this power supply protect itself and still be operational afterwards with a short. If it shuts itself down or never starts to begin with then one can say it is likely any component or item causing the short has not been damaged further. Seeing if excessive amperage occurred before the shutdown is a great point. I worked with load centers with thousands of amps and sudden changes and there are monitoring meters and automatic recorders that capture the transients, I see nothing in any computer power supply that in that case could not be tested. Maybe just an added test gear is needed.

Yeah, I get the idea. The issue is I have already gone down the road of setting up custom test gear when we built the Transient Load Tester. It's, honestly, a pain in the ass if there is not an off the shelf plug and play solution. Plus, the first time you get a result a vendor doesn't like your custom built test gear gets thrown under the bus. It really would not be an issue though if units were spec compliant and limited to 240VA per rail.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
Yeah, I get the idea. The issue is I have already gone down the road of setting up custom test gear when we built the Transient Load Tester. It's, honestly, a pain in the ass if there is not an off the shelf plug and play solution. Plus, the first time you get a result a vendor doesn't like your custom built test gear gets thrown under the bus. It really would not be an issue though if units were spec compliant and limited to 240VA per rail.
Maybe common sense applies as in conditions or scenarios that are very much possible.
  • Fan fails at high loads, will the over temperature protection work and protect the power supply?
  • The 12v cable is shorted, will the power supply shut down and not burn up the cable (if it burns up the cable before it shuts down -> it will likely burn up anything in the computer that was shorted).
    • Short the 12v cable with a very thin piece of wire or fuse (if a .5amp fuse does not blow for an example -> great! PASS!, a 100a fuse blows -> HOLY SMOKES -> FAIL!)
  • OPP/OCP -> Will power supply configured in spec for a system but operator mistakenly or ignorantly puts a 350w extra video card when loaded, an abnormal new condition, will this power supply protect against that when loaded?
No one expects every possible scenario needs to be tested but logical tests that weed out the power supplies that fails these simple tests would be most helpful and could be very important for a purchase decision. Who cares what a vendor doesn't like? Let them explain better exactly what it will protect against, if it can't protect against a bad fan then what the hell good is the Over Temperature protection? Make them reveal how tested and what is tested, since they don't now then I conclude there is nothing they want to expose on these which look good for them.
 
Last edited:

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
18,095
Who cares what a vendor doesn't like?

Those of us who get sued when we they don't like things. That is why we have to mitigate things by using off the shelf equipment and test methodology that is supported by the specifications.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
6,652
Those of us who get sued when we they don't like things. That is why we have to mitigate things by using off the shelf equipment and test methodology that is supported by the specifications.
How can a disconnecting a fan to test if the OT protection could ever be considered in a law suite. You put a condition which is possible and this is the result. What shelf equipment do you need for that?

Shorting the 12v rail, there is no test equipment needed except maybe fuses to see if high amperage occurs before the shutdown. I've done this more than once accidently and I am sure others have too.

Reviewers are also journalist with certain protections, if you are clear on how tested, what means, give ample time for feedback from the vendor they I would say have zero chance of a lawsuit. Just them threatening a law suite would be very bad for them.

Reviewers all the time bash products, that is their job if they found it to be poor quality and most of those reviews have very little to do with standard equipment or test methodology. What makes Power Supplies a non touchable piece of hardware for reviews? I don't see it.
 

Starfalcon

Gawd
Joined
Jan 7, 2020
Messages
758
Those of us who get sued when we they don't like things. That is why we have to mitigate things by using off the shelf equipment and test methodology that is supported by the specifications.

That seems rather strange....as it isnt much of a review if you are afraid to say anything bad about the unit. At that point you more or less doing a sponsored press release for the company. I personally would rather know if my PSU can blow up, versus a company having hurt feelings over your "review" and making lawsuits over it.
 

D-EJ915

[H]ard|Gawd
Joined
Jan 31, 2003
Messages
1,514
Got an email from Newegg about it.
This email is being sent to you concerning a recent public statement from Gigabyte https://www.gigabyte.com/Press/News/1930 relating to a product which our records indicate you have previously purchased from Newegg. Please see the order details below:
Order #​
Item #: 17-233-032​
Item Description: PSU GIGABYTE 850W GP-P850GM R​
Gigabyte is offering return and exchange service to customers who purchased the below power supplies within a defined range of Serial Numbers. The data table below sets forth the products in question and their respective serial number ranges:​
Model name:GP-P850GM
S/N From SN20343G031011 to SN20513G022635
Model name:GP-P750GM
S/N From SN20243G001301 to SN20453G025430
If your item’s serial number falls within the range of the corresponding model number, please reply back to this email so we can assist with setting up a return for a refund or replacement.​
To be honest I haven't even looked at it since I got it with my 3090, I guess I should see what the serial number is.To be honest I haven't even looked at it since I got it with my 3090, I guess I should see what the serial number is.
 
Last edited:
Top