News PSU Expert Aris Mpitziopoulos Responds to Gigabyte's Exploding PSU Problem

2Be_or_Not2Be

Distinguished
Aug 23, 2013
985
57
19,440
115
Congrats to Aris for calling out these poor, shoddy products released by Gigabyte! Props also to Gamers Nexus for highlighting them as well. It's a shame that it took some lurid headlines ("Exploding PSU" was a favorite of mine) to finally get Gigabyte to issue a recall.

I wish mfgs would just focus on quality and stop replacing quality parts w/cheaper parts (sadly, too many examples here). If you have to wait for the quality component to come back into stock, it's okay - you don't need to produce an inferior product. Especially when the market you're in is already saturated.
 
Apr 1, 2020
274
120
860
0
And stuff like this is why I will -only- buy SeaSonic PSUs given a choice, and will only consider Enermax and Superflower if SeaSonic isn't available, even though they are #2 and #3 in top quality .

The PSU is the heart of the machine (literally), it's not a place where you want to cut corners or have even a question of quality and safety.

And yes, I learned that lesson the hard way years ago...
 

hasten

Distinguished
Oct 25, 2007
159
3
18,695
3
Congrats to Aris for calling out these poor, shoddy products released by Gigabyte! Props also to Gamers Nexus for highlighting them as well. It's a shame that it took some lurid headlines ("Exploding PSU" was a favorite of mine) to finally get Gigabyte to issue a recall.

I wish mfgs would just focus on quality and stop replacing quality parts w/cheaper parts (sadly, too many examples here). If you have to wait for the quality component to come back into stock, it's okay - you don't need to produce an inferior product. Especially when the market you're in is already saturated.
Unfortunately managerial and financial accounting provide simple formulas that calculate the (tolerable) defect rate that will generate the highest profit based on materials provisioned. So in engineering the goal is to generate a product at the lowest cost, while meeting the standards of performance needed for (whatever) certifications to base their msrp and accept the returns which are covered by allowance. Much more goes into it but you get jest.

When profit drives ALL C-level decisions you get Gigabyte and the above very thorough calculations. The reviews across their product lines are often, "it works great, if it works." I've had both a 3070 and 80 Gaming OC and thought both were junk in comparison to all other nvidia brands. Heck, the 70 gaming oc performed notably worse than an xc black in all the benchmarks I ran - "even" with the big ol gigabyte bios PL and the SC's nothing (not even a backplate, cmon EVGA).

That said, I do have an aorus master z490 i bought for a great price off a gentlemen who got newegg shuffled and the second one is great. The first was DOA.
 

Math Geek

Polypheme
Ambassador
what GB fails to understand is that something like this will haunt them for a long time to come.

average users here and in many other places on the web won't fully understand the issue or what units it applies to. instead they will only recall that GB psu's WILL EXPLODE and every model should be avoided forever.

we see it all the time with advice here from folks who don't truly know. but when they say it, others believe it for a long time and pass on the "advice" over and over.

they need to step up, mass recall and save themselves from years of folks saying to avoid any and all GB units. frankly i won't even bother to correct folks with this one since GB is so obviously avoiding the issue and asking for this result :)
 
Reactions: Soaptrail

Spielwurfel

Distinguished
Dec 7, 2005
195
2
18,685
0
Gigabyte should be ashamed of such poor response and what I consider a clear attempt to downplay a serious issue with their products. I has a bad experience with Gigabyte motherboards, annoiances with one of their graphics card and now seeing their position to this whole situation made me want to stay even further away from their products.

Congratulations to everyone from the media for providing such good and well informative content.
 

escksu

Respectable
Aug 8, 2019
450
157
1,860
0
My POV has always been to buy PSUs thats made by reputable OEMs. The brand doesnt matter, its the OEM that matters.

Those gigabyte psu are made by oem i have never heard of. So, i wont even bother. I would only get those made by the likes of seasonic, superflower, delta, flextronics.

Btw, i am not a fan of cwt (most corsair psus are made by them, except flagship AXi). They are decent but those 4 i mentioend are better.
 
Reactions: 2Be_or_Not2Be

InvalidError

Titan
Moderator
Current measurement shunts are available with thermal coefficients under 100ppm, so you can definitely achieve analog current limits with better than 1% accuracy across the operating temperature range if you are willing to pay for it.
 

pjmelect

Distinguished
The problem is that 110%-120% OPP is impossible with analog controllers, which use resistors to adjust OPP. These resistors drift with temperature.
As an electronic engineer I would disagree with this statement, it should be possible to set the current limit to within 1% using resistors over the operating temperature range, and you do not need to use very expensive resistors to do it.
 
Aug 15, 2021
5
5
15
0
Congrats to Aris for calling out these poor, shoddy products released by Gigabyte! Props also to Gamers Nexus for highlighting them as well. It's a shame that it took some lurid headlines ("Exploding PSU" was a favorite of mine) to finally get Gigabyte to issue a recall.

I wish mfgs would just focus on quality and stop replacing quality parts w/cheaper parts (sadly, too many examples here). If you have to wait for the quality component to come back into stock, it's okay - you don't need to produce an inferior product. Especially when the market you're in is already saturated.
Yeah, except it’s a limited voluntary recall with a bunch of stipulations attached, when it should probably be a mandatory recall with stop-use guidance. I agree with Aris, they have a fundamental design flaw, most likely they are missing appropriate resistance from gate to source and/or have mismatched components, leading to unpredictable switching behavior, which is why the PSU, in his instance, popped right after he was load testing. Meaning on power down followed by power up, there was some kind of drain to source occurring, or mismatched timing, which caused the driver IC to switch to the wrong state and, subsequently, the fireworks of exploding fets occurred.

None of this is reliant on overload stress testing, it could be a number of different power events and environmental factors that could lead to this condition, e.g. machine hang followed by rapid power on/off cycle, load cycling of components, issues on the AC input side (fluctuations in AC voltage), could all be part of exciting the problem in the PSU. What shouldn’t happen is the exploding part - not ever.

Significant reported incidents of things exploding isn’t indicative of a manufacturing defect, it’s a strong indicator of a design defect, because the design should prevent this. This is why violently exploding PSU’s almost never happen. Even the cheapest old Antec ketchup & mustard PSU’s (which used to fail pretty darn regularly) generally failed in a safe manner (e.g. non explosively). Steve’s tear-down and comparison of different components, being used across different examples of these PSU’s, would indicate that this isn’t a specific part issue that’s relegated to a specific production run. I’m going to guess that Gigabyte finally tested a few units from a specific run, got a failure result, and decided to recall from that run (because, legally, if they did otherwise, then they’d really be in hot water upon discovery - and believe me, subpoena’s are coming for them eventually).

So again, Aris is most likely right: Gigabyte is facing a fundamental design-flaw with this line of PSU’s, due to having gone with an inexperienced contract vendor that appears to have since disappeared. Gigabyte, being fully unprepared to deal with this problem (due to not having proper crisis handling know-how), is going about this all wrong. They continue to obfuscate the issue with claims that the testing regimen was the cause of exploding PSU’s, that they’re fine under normal conditions, and they effectively are trying to blame misuse (and customers) as the cause.

Why? Because they suspect they have an increasingly serious liability problem and they’d have to backtrack significantly, thus now increasing their liability. Whereas, if they’d just worked earlier to get ahead of this issue, starting with examining Aris’s sample, they’d have been acting responsibly, thus decreasing their exposure (e.g. the Johnson & Johnson Tylenol model).

Ultimately, they need to pull the product from the market, issue a full-line recall and stop-use guidance, before one of these things starts a serious fire. This is going to be expensive for them, as it appears they have no contract OEM with which to share or shunt costs, since the actual manufacturer appears to have gone out of business. But Gigabyte is a two billion dollar company, and if they wish to remain as such, they need to suck it up and start dealing with this problem properly.
 

InvalidError

Titan
Moderator
I agree with Aris, they have a fundamental design flaw, most likely they are missing appropriate resistance from gate to source and/or have mismatched components, leading to unpredictable switching behavior, which is why the PSU, in his instance, popped right after he was load testing.
With the HN video showing at least two of their PSUs with "APFC failure" stickers on them, I'm going to guess there was a design flaw in the APFC boost circuit that causes it to exceed the FETs' SOA and ends with a spectacular failure.
 

ASK THE COMMUNITY