Very interesting failure rate statistics

[goodpsusearch]

http://linustechtips.com/main/topic/...ly-everything/

[shovenose]

Re: Very interesting failure rate statistics

huh. Is this not just under this specific e-tailer's returns? What is the time limit on this? Newegg for example has a 30 day return policy.
So these numbers could be majorly influenced by handling, packaging, audience (as in, what type of customers bought it), or even shipping carrier (for like hard drives)...

Actually long term reliability should not be reflected in those numbers...

OCZ solid state drives do suck, though.

[RJARRRPCGP]

Re: Very interesting failure rate statistics

Interesting that the ratings for AsRock are like ones I would expect with the old AsRock...

AsRock seemed to typically use poor caps during the 775 period...

But they appear to not have that problem with their socket AM3+ boards...

So I doubt that AsRock's gonna suck like those OCZ horror stories.
(Which even those may be exaggerated.)

[mariushm]

Re: Very interesting failure rate statistics

The AM3+ Asrock boards are budget models, some of the cheapest available with the 970+SB950 chipset combo.

ASRock 970 Pro3 R2.0 for example doesn't even have a heatsink on the VRM and the ASRock 970 Extreme3 has a tiny heatsink but it's still a board with relatively weak VRM.

All three (ASRock 990FX Extreme3 is the third with high failures) are boards with 4+1 or 4+2 phase VRM, which is just too weak for overclocked 125w rated processors - lots of people buy these with cheap FX-6300 (95w tdp) or FX-8320 (125w tdp) and overclock them and then act surprised when they fail.

Even worse, lots of people buy third party heatsinks because they think the stock ones are too noisy, and these third party heatsinks are often too tall and blow air straight out the case instead of blowing some air down the board and onto the vrm heatsinks or mosfets and the mosfets simply cook when they overclock the processor.

-
What I found very interesting (but i expected it) is the high number of Corsair CX returns and failures. If I remember correctly, huge problems with the primary capxon capacitor in that series, and coil whining.. these are CWT designs (again, if i remember correctly).

They also started to cheapen out on the memories,so you can see the return rates - my guess is mainly due to the large heatsinks blocking some cpu fans and due to low overclocking ability.

[RJARRRPCGP]

Re: Very interesting failure rate statistics

The AM3+ Asrock boards are budget models, some of the cheapest available with the 970+SB950 chipset combo.

ASRock 970 Pro3 R2.0 for example doesn't even have a heatsink on the VRM and the ASRock 970 Extreme3 has a tiny heatsink but it's still a board with relatively weak VRM.

All three (ASRock 990FX Extreme3 is the third with high failures) are boards with 4+1 or 4+2 phase VRM, which is just too weak for overclocked 125w rated processors - lots of people buy these with cheap FX-6300 (95w tdp) or FX-8320 (125w tdp) and overclock them and then act surprised when they fail.

Even worse, lots of people buy third party heatsinks because they think the stock ones are too noisy, and these third party heatsinks are often too tall and blow air straight out the case instead of blowing some air down the board and onto the vrm heatsinks or mosfets and the mosfets simply cook when they overclock the processor.

-
What I found very interesting (but i expected it) is the high number of Corsair CX returns and failures. If I remember correctly, huge problems with the primary capxon capacitor in that series, and coil whining.. these are CWT designs (again, if i remember correctly).

They also started to cheapen out on the memories,so you can see the return rates - my guess is mainly due to the large heatsinks blocking some cpu fans and due to low overclocking ability.

It actually looks like MSI is worse, IIRC, some MSI AM3+ boards come with a warning to basically not use more than about 50 percent CPU, because of using cheap components.
Basically caught MSI admitting that they cheaped out and that it's about as useless as tits on a bull!

It looks like some of those MSIs can't even handle a stock CPU!

[Dan81]

Re: Very interesting failure rate statistics

What about ASUS?I heard they were (and maybe still are?That's why I ask this question)as bad as ECS.However,one ASUS board I have (P4P800-VM,T-Systems OEM)has only Rubycon and Panasonic caps.(4-6 Panasonic FJ caps and the rest is just Rubycon ZL and MBZ)

[goodpsusearch]

Re: Very interesting failure rate statistics

They are definitely worse than Gigabyte, but I don't agree with the bad reputation they have here.

[shovenose]

Re: Very interesting failure rate statistics

I had an AsRock 970 Extreme3, well, it wasn't great for overclocking, but it worked fine. It's now in somebody else's PC and they have not had any problems. Running an FX-8350 so it's not a low power CPU either...

[c_hegge]

Re: Very interesting failure rate statistics

My take on boards:

Gigabyte: as good as they get. I only ever use gigabyte boards in new PCs I build for others (and in my own PCs), and have only ever had a handful fail

ASUS: Absolute junk - we tried using a few in customer PCs at work, and most are dead now.

MSI: Decent, but just not quite as good as Gigabyte

ASRock: No idea. I used to class them as being nearly as bad as they get, since they were just a cheaped out and stripped down ASUS. Now that ASUS no longer own them, I don't know.

ECS: Just as bad as ASUS.

[RJARRRPCGP]

Re: Very interesting failure rate statistics

For new builds after 2002, I only used Asus and all of the boards I still have, still work.

[delaware74b]

Re: Very interesting failure rate statistics

I have had over a dozen Asus boards over the past 10 years. One developed a bad Bios (fans only, no boot) which was RMA'd under warranty with no complaints. The replacement board (A7N8E-Deluxe) would still be alive today if it wasn't victim of Antec's F-you caps in the +5VSB 2-transistor circuit.

I have an AM3-based (M4A785TD-V EVO) that has been running 24/7 for three years as my desktop/server with Win 7. It's been shut down long enough for cleaning and 2 weeks for moving.....

[momaka]

Re: Very interesting failure rate statistics

My take on boards:

Gigabyte: as good as they get. I only ever use gigabyte boards in new PCs I build for others (and in my own PCs), and have only ever had a handful fail

ASUS: Absolute junk - we tried using a few in customer PCs at work, and most are dead now.

MSI: Decent, but just not quite as good as Gigabyte

ASRock: No idea. I used to class them as being nearly as bad as they get, since they were just a cheaped out and stripped down ASUS. Now that ASUS no longer own them, I don't know.

ECS: Just as bad as ASUS.

I agree with you on ASUS and older ASROCK board that were built by ASUS - they are either finicky or just plain straight bad. On ASUS boards, especially socket-939 and P4-era, you'd often see them cook caps with hot MOSFETs around the chipset or RAM areas. The only ASUS boards I cannot complain about is the P4SD (and their variants). Aside from bad caps (TEAPO, OST, and affected-era Nichicon HM), they are designed very well - almost every power module on the motherboard uses a buck-type regulator. I haven't had anything newer from ASUS or ASRock, so I can't comment on how their newer stuff is.

ESC and MSI, on the other hand - I love them! Why? Because they are excellent boards that just need a little love by giving them a recap. ECS loves OST and MSI, TEAPO - and we all know neither of those are good brands on a motherboard. I feel the same way about Jetway and EVA boards too - most excellent boards if it wasn't for those awful GSC/Evercon/Sacon caps.

Another brand of motherboards that deserves an honorable mention in terms of built-quality is aBit. Again, I like the way they do their power regulation. Not like ASUS, older AsRock, and some Intel boards.

[Topcat]

Re: Very interesting failure rate statistics

There's a reason my Supermicro wondersboards arent on that list.... I'm typing on one thats 4 years old running 24/7/365 the whole time. Never so much as a crash. Buy quality, get quality!

[rhomanski]

Re: Very interesting failure rate statistics

I don't know about it, but I'm still typing this with a P5Q-E. It's been going almost constantly since I bought it in 2008. I even had it plugged straight in the wall for several years until the power flickered way too much. I got tired of rebuilding the raid and put it on a UPC. The power supply died last year and I'm on my second set of drives, but the mother boards still going. Hope I don't jinx it.

[c_hegge]

Re: Very interesting failure rate statistics

ESC and MSI, on the other hand - I love them! Why? Because they are excellent boards that just need a little love by giving them a recap. ECS loves OST and MSI, TEAPO - and we all know neither of those are good brands on a motherboard. I feel the same way about Jetway and EVA boards too - most excellent boards if it wasn't for those awful GSC/Evercon/Sacon caps.

I find that ECS suffer from the same sort of issues as ASUS. Even if there's nothing obviously wrong with them, they just die.

There's a reason my Supermicro wondersboards arent on that list.... I'm typing on one thats 4 years old running 24/7/365 the whole time. Never so much as a crash. Buy quality, get quality!

That, and because they're not available through most stores. AFAIK, only two or three very expensive server boards are available at all in this country, so for my builds, Gigabyte will have to do.

[larrymoencurly]

Re: Very interesting failure rate statistics

It actually looks like MSI is worse, IIRC, some MSI AM3+ boards come with a warning to basically not use more than about 50 percent CPU, because of using cheap components.
Basically caught MSI admitting that they cheaped out and that it's about as useless as tits on a bull!

It looks like some of those MSIs can't even handle a stock CPU!

Somebody who kept a count of motherboard failures said MSI had the highest rate of CPU voltage regulator failures. Also I wouldn't be surprised if a higher percentage of motherboards returned due to unsuccessful BIOS updates are MSI because MSI's update methods have been so flakey that the moderators of MSI's forums recommend against using 2 out of 3 of the methods.

[shovenose]

Re: Very interesting failure rate statistics

Somebody who kept a count of motherboard failures said MSI had the highest rate of CPU voltage regulator failures. Also I wouldn't be surprised if a higher percentage of motherboards returned due to unsuccessful BIOS updates are MSI because MSI's update methods have been so flakey that the moderators of MSI's forums recommend against using 2 out of 3 of the methods.

They have a Windows utility for updating MSI BIOS now thankfully.

[momaka]

Re: Very interesting failure rate statistics

I find that ECS suffer from the same sort of issues as ASUS. Even if there's nothing obviously wrong with them, they just die.

I don't know, I've never had issues with them or riviving them if they broke (usually just need a recap). In fact, based on the few older P4-era boards I found from dumpster-saved computers, I'd say they are troopers.

For example, one of the boards I have, an ECS P4VXASD2+, was powered its entire life by a completely gutless CyberLink PSU. That PSU had only a 16A schottky on the 5V rail and 2 (yes TWO!) 470uF caps for filtering. Nothing more. Why this matters? Well, the P4VXASD2+ is kind of a funky P4 board - it uses the 5V rail to power the CPU instead of 12V. Moreover, it uses an ancient design where instead of a synchrnous buck regulator for the CPU VRM, it only has a 1-sided buck regulator (i.e. MOSFETs between 5V and CPU V_core and schottky diodes between CPU V_core and ground). This design is terribly inefficient (but very simple and robust)!
Coupled with a 1.7 GHz Willamate CPU (83W max), this thing had apparently been running quite some time judging by all of the dust, despite the crappy PSU. I gave it a good cleaning, a 2 GHz Northwood CPU (65W max), and a HiPro PSU, and it's been running great... OST RLS/RLX caps and all (a recap is planned for it, though). No signs of instability either. I even stuffed it with some crappy PC3200 RAM modules that would give errors on other motherboards and make them not boot. But not this one - boots every time, provided those crappy RAM modules are set to run at PC2700 speeds or lower.

The other board that surprised me was an ECS P4IBMS. That one, too, was likely a victim of a crappy power supply. It came in some cheap case with a "gutless" PSU that didn't have any signs of use whatsoever. My guess is, the original gutless PSU that came with the case blew up and did some damage (as evident by several meltdowns on the 3rd RAM slot and a few dead SMD transistors). When the owner installed the new POS PSU and noticed that the system still did not work, he threw it away. HDD was dead too when I found it (spins but does not appear in BIOS - which is why I think it was the PSU that caused this). Also appears the owner wasn't that nice to the board either. Several of the G-Luxon caps around the CPU seem to have been ripped out - quite literally. Then there's the CPU fan wire: twist-tied to the 12V line on the case's fan. Possibly some other "handywork" in there that I can't remember right now.

Juding by the number of components that blew, that old PSU must have caused quite a bit of havoc (especially to kill the HDD). The fact that I was able to save it by just changing a transistor and doing a crappy recap with used caps means that this board is actually not designed that badly after all.

ASUS boards, on the other hand, don't need a crappy PSU to die randomly - they just do.
Oh yeah, PS3 boards are also made by ASUS (at least I remember it saying so on their PCB). Let me tell you what a heap of crap those boards are! I hate troubleshooting them.

[kc8adu]

Re: Very interesting failure rate statistics

just rebuilt an amptron(another ecs brand)with that same cheap vrm.
had been run all these years on a gutless "premier" "300w"power supply shaped object.
of course the psu was recycling pile fodder.the board had 1 blown cap in the vrm and 1 near agp.
the board in the psu was black from overheating and all caps were blown.i grabbed the joke of a heatsink to pick it up and the board fell apart.
that this thing ran that long with that pos is a miracle.
and the owners grandkids never turned it off.
so even doomed to fail stuff sometimes holds on longer than expected.
this combo is from the badcaps era all around.even the video card with its skywell sht caps was ok.it got recapped too.

[momaka]

Re: Very interesting failure rate statistics

the board in the psu was black from overheating and all caps were blown.i grabbed the joke of a heatsink to pick it up and the board fell apart.

The CyberLink I found was not as bad like that but still pretty discolored under the 5V rectifier. I guess the lack of a fan controller was a good thing here . If the fan had seized, though...

this combo is from the badcaps era all around.even the video card with its skywell sht caps was ok.it got recapped too.

^ HAHA, Skywell SHIT caps . Been a while since I've seen those.

Yeah, I guess people keep their old junk so old stuff still tends to pop every now and then. Just 2 months ago, I picked up a Deer/L&C PSU that was left as junk parts in one of our labs in the university. It one of those older ones with the VIVA primary capacitors and a mix of Ruycon and CapXon caps on output. Not the oldest design with the 7805 regulator for 5VSB, but still ancient. So it had a standard 2-transistor self-resonator but with a 5.1V zener for the regulation rather than a 431 shunt (read: cheap). I put some used Nichicon HZ/HNs from Xbox 360 motherboards, and it's up and running good now. At least for testing stuff on breadboards, that is.