unknown (solid?) caps on MSI "military class" graphics card?

[CapPun]

considering buying an MSI 1070 ITX format card but first need to know what caps are used :

https://i.ibb.co/hffFVKf/1070.jpg



?

on the product page MSI claims their card's made with "military class IV" (whatever the hell that means) yet those caps have no brand & that alone is suspicious hence my question anyone can tell without logo what the brand is?
(mostly wanna know the country of origin tbh)

[momaka]

Re: unknown (solid?) caps on MSI "military class" graphics card?

Those blue polymers are United Chemicon PSE series, as indicated by the "E" letter.
Datasheet here:


Not that it should matter too much. Solid capacitors even from cheap Chinese/Taiwanese brands like CapXon, seem to be OK, unlike their electrolytics.

As for the MSI "Military Class" thing - most likely just fancy advertising terms. It's a standard PCB like any other. Funny how Gigabyte, Asus, and MSI have these fancy names for their PCBs to make the customer think the product will never fail. The reality is, the GPU chip is the one that will fail - especially if you don't keep the temperature to below 55-60C. And these being a GTX 1070 for ITX format... I can only wonder how poorly sized of a cooler this thing will have... unless they used a mobile GTX 1070 chip - in which case, the TDP should be a lot lower and there may be a chance the card may not run so hot.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

ok thx but so you sure about it? (how did you figure the brand when there's no logo???)


btw I've never seen or read about a GPU failing (or an intel CPU for that matter) hope that dont happen :/

from 2011 to 2017 I used a GTX 580 these things heat up like crazy like above 80°C in gaming (even in well-cooled giant Cooler Master HAF case with 4 fans including 20mm top fan)
plus it was a single-fan with a tiny fan, brand was EVGA but the card looked like a basic generic card yet it worked without a problem for all those years till I finally sold it after upgrade

here for the MSI 1070 Aero (ITX version) reviews show temps go up to the mid-70's in Furmark tops & they say those are good temps for a 1070

[ChaosLegionnaire]

Re: unknown (solid?) caps on MSI "military class" graphics card?

ok thx but so you sure about it? (how did you figure the brand when there's no logo???)

according to the datasheet, these are low-profile, small footprint 6.3mm caps, so because they are so small, there's no space to print the logo. if u look at the datasheet pictures of the printing and marking on the cap, u'll see it matches, particularly the size and position of the voltage "v" on the cap. so rest assured, its quite certain these are chemicon npcap pse series polymer caps.

btw I've never seen or read about a GPU failing (or an intel CPU for that matter) hope that dont happen :/

i'd bet u are one of those that go through and change your hardware so frequently u dont see it happen or u are one of those "lucky ones". i could probably say the same but with polymer caps. i've never seen or read about a polymer cap failing, junk brand or otherwise...

also, the caps shown in review photos may not be the ones u get in retail aka "bait and switch". there are many cap manufacturers out there with similar cap specs so what u see in reviews may not be what u get "in the field" or in retail. sometimes they run out of one cap brand series in the factory so they substitute it with another brand series with similar specs.

here for the MSI 1070 Aero (ITX version) reviews show temps go up to the mid-70's in Furmark tops & they say those are good temps for a 1070

looking up the specs of the gtx 1070, it has a base clock tdp of 150w with up to 250w tdp with boost clocks. a review by guru3d attempted to calculate the tdp of the msi 1070 itx aero 8gb and it seems to have a tdp of 163w. looking at the pics of the msi 1070 itx aero 8gb, the cooler only has a single 92mm fan but it is a dual slot cooler with three heatpipes.

so the problem with these reviews is that either they dont state the test setup whether they test it inside the case or they test it in an open-air test bench. hence u dont truly know how well the cooler actually performs. what was the ambient temperature during testing? how much the ambient temperature actually fluctuated during testing? all these affect the temperature readings obtained during the review. if the "furmark mid-70's" was obtained in an open-air air-conditioned test bench, then when put inside a cramped itx case, u can expect the temperatures to be significantly higher in the region of 80-90°C.

another thing to note is the law of physics whereby tin in solder whiskers at an exponentially accelerated rate when run above 70°C, particularly lead-free solder. so whatever anyone in a forum says about the temperature being fine or whatever the gpu manufacturers say about the temperature being fine will not change this fact and law of physics. therefore, if a gpu runs above 70°C, i wont expect it to last long. any cooler that runs a gpu above 70°C is considered inadequate.

so in conclusion, the OP is looking at the wrong weakest link with regards to video card hardware failure and longevity. the weakest link is the cooling of the gpu NOT the polymer caps. polymer caps will more than outlive the gpu chip.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

just one thing - if brand of polymer cap doesnt matter...then how come in mid-range boards, the caps around the VRM's tend to be japanese brand (like FPCAP or something) whereas the other caps tend to be lesser quality like APAQ?

if brand doesn't matter for solids, then why dont they put APAQ solids around the VRM's also to save costs?
I mean it can't be for the customers cause most of them won't even look at the caps let alone know about the brand so it can't be marketing so it gotta be another reason right??

i'd bet u are one of those that go through and change your hardware so frequently u dont see it happen or u are one of those "lucky ones".

in the past not that often I'd a GTX580 "founder edition" (single fan) lasted 7 years till I sold it & that thing heat up like crazy I must be very lucky then

and if GPU rules apply to CPU's there's people who still got their overclocked air cooled 2600k going strong today

so the problem with these reviews is that either they dont state the test setup whether they test it inside the case or they test it in an open-air test bench. hence u dont truly know how well the cooler actually performs. what was the ambient temperature during testing? how much the ambient temperature actually fluctuated during testing? all these affect the temperature readings obtained during the review. if the "furmark mid-70's" was obtained in an open-air air-conditioned test bench, then when put inside a cramped itx case, u can expect the temperatures to be significantly higher in the region of 80-90°C.

one weird thing I noticed is that sometimes when I open the side door of my case the temps during stress test will go UP a degree or two during test +_-

any cooler that runs a gpu above 70°C is considered inadequate.

even if those coolers were tested on high-end GPU's ? (those that end with 70 or 80 in their series)

and does that also apply to coolers for CPU's then? (if over 70°C in Prime95)

[ChaosLegionnaire]

Re: unknown (solid?) caps on MSI "military class" graphics card?

how come in mid-range boards, the caps around the VRM's tend to be japanese brand (like FPCAP or something) whereas the other caps tend to be lesser quality like APAQ?

if brand doesn't matter for solids, then why dont they put APAQ solids around the VRM's also to save costs?

its more for reasons of planned obsolescence again. its to ensure the board outlasts the warranty. and since its a mid-end board, they are cutting corners in a smart way. u dont need costly japcaps for general filtering of the usb, pci-e, ram etc. as they arent electrically stressful areas, so u can just use cheaper/lower quality polymers or electrolytics there. they just put japcaps in the cpu vrm where its necessary because its a high stress area electrically there.

and if GPU rules apply to CPU's there's people who still got their overclocked air cooled 2600k going strong today

and does that also apply to coolers for CPU's then? (if over 70°C in Prime95)

u cant compare cpus and gpus that way even though it might seem that they are the same. firstly, gpus have a significantly higher transistor count and thus density than cpus, so this makes them more fragile both thermally and electrically than cpus.

secondly, gpus are soldered to the board while cpus are socketed. being bga soldered to the board puts the gpu through thermal stress from the soldering process and mechanical stress from the pcb being flexed when the video card is inserted/removed from the slot, the mounting and weight of the gpu heatsink among other things.

one weird thing I noticed is that sometimes when I open the side door of my case the temps during stress test will go UP a degree or two during test +_-

thats a good thing actually! haha! it shows that u have optimised the case convection airflow with the correct and proper placement of the case fans, so it acts as a proper airflow wind tunnel sucking in cold air and exhausting the hot air out. removing the side panel actually upsets/disrupts this "wind tunnel" airflow phenomenon so u get higher temperatures!

even if those coolers were tested on high-end GPU's ? (those that end with 70 or 80 in their series)

like i said about tin whiskering, its a physics phenomenon related to the tin in solder and its irrelevant if the gpu is high-end or not. at the end of the day, its just how long the manufacturers want or calculated their stuff to last. if its warrantied to last 3 years then they just slap on a cooler at the right cost for them to ensure the hardware can survive the 3 years.

read what i said above again about the usage of branded polymers vs cheaper polymers. otherwise, why not just slap that gigantic titan x cooler on all the same cards of the series (from the gt 1030 to the gtx 1080 ti), space constraints notwithstanding? that gt 1030 would certainly run a lot cooler if it had the titan x's cooler now, wont it? (of course not, the gt 1030 isnt as thermally stressful as the titan x...)

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

its more for reasons of planned obsolescence again. its to ensure the board outlasts the warranty. and since its a mid-end board, they are cutting corners in a smart way. u dont need costly japcaps for general filtering of the usb, pci-e, ram etc. as they arent electrically stressful areas, so u can just use cheaper/lower quality polymers or electrolytics there. they just put japcaps in the cpu vrm where its necessary because its a high stress area electrically there.

so that's what I meant in the end brand does matter even for solid caps right?

cause a lot of reviewers (like Jonyguru) when they see non-jap caps (even chinese!) they say something like "if those were electrolytics I'd pull points, but those are polymer so I'll let it pass" which they shouldn't

that's why I started this thread but if that MSI does have jap solids after all then maybe I'll buy it especially since it only has 1 pcie connector (so less headroom for overclocking so at least I know the seller didn't overclock it like a pig)




anyway - ultimately if GPU's are so fragile aint the best thing to do (for those who can) modify the card's BIOS so that it throttles the GPU when temps hit 70°C?

[Uranium-235]

Re: unknown (solid?) caps on MSI "military class" graphics card?

FPCAP are "Functional Polymer". They're kind of like hybrid caps

apaq are cheap ones, watch for them being used on vrms

[momaka]

Re: unknown (solid?) caps on MSI "military class" graphics card?

so that's what I meant in the end brand does matter even for solid caps right?

Well, it matters to the manufacturer / board designer. Most likely cheap polymer caps don't offer the same low ESR specs as the Japanese ones, which could matter in very stressful places, like the CPU VRM. So they went with Japanese polymers probably because they had no other choice due to specs. Sure, maybe they could have used more cheaper poly caps in place of fewer but more expensive Japanese ones... but maybe it still doesn't make sense from a cost perspective.

In any case, I haven't seen any alarming number of reports of cheap polies failing, so I'd still consider them alright.

anyway - ultimately if GPU's are so fragile aint the best thing to do (for those who can) modify the card's BIOS so that it throttles the GPU when temps hit 70°C?

They aren't "fragile", but their life can greatly vary once they go above 60-70C... and some cards will still last while others will fail in short order. It all has to do with manufacturing imperfections: even two identical cards from the same week may not last the same amount of time.

Generally, most GPU manufacturers know that people tend to upgrade their computers around ever 3-5 years, so they try to design their hardware to last just as long. But again, due to variations in manufacturing, some GPUs will be made almost "perfectly" (calibrated equipment/machinery at the beginning of production vs. "worn" machinery at end of production, new vs. aging materials/consumables, and etc.), while others may not be at all (late afternoon Friday shift? ) Thus, the majority of the GPUs will likely meet the "design target" of 3-5 years, with some going much longer than that, and almost as many falling quite short. It's pretty much a lottery how long one particular video card will last. But generally, running a GPU cooler will more than likely increase its chance of surviving longer.

So to answer your question: YES, it would help quite a bit if people mod their BIOSes to run the fans faster to keep the GPU cooler. But higher fan speeds also means more noise and quicker dust accumulation. More noise will obviously get more bad reviews (especially nowadays, where a PC is expected to be nearly silent.) And knowing how some people don't take any care of their stuff, it would only be a matter of time before the heatsink clogs with dust and is rendered ineffective (a good number of video cards die due to this, actually.) So that's why manufacturer's don't do it. And most people simply don't know any better. Then there is also the fact that many low and mid-range video cards simply don't come with adequate coolers. So no matter how much you turn up the fans (if they aren't running at 100% all the time already), the cooling may not improve.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

Well, it matters to the manufacturer / board designer. Most likely cheap polymer caps don't offer the same low ESR specs as the Japanese ones, which could matter in very stressful places, like the CPU VRM. So they went with Japanese polymers probably because they had no other choice due to specs. Sure, maybe they could have used more cheaper poly caps in place of fewer but more expensive Japanese ones... but maybe it still doesn't make sense from a cost perspective.

In any case, I haven't seen any alarming number of reports of cheap polies failing, so I'd still consider them alright.

I didn't know that - so you saying that the difference between japanese & non-japanese polymers is NOT just durability but ALSO performance? (jap ones do their job better resulting in "cleaner" current?)

Generally, most GPU manufacturers know that people tend to upgrade their computers around ever 3-5 years, so they try to design their hardware to last just as long. But again, due to variations in manufacturing, some GPUs will be made almost "perfectly" (calibrated equipment/machinery at the beginning of production vs. "worn" machinery at end of production, new vs. aging materials/consumables, and etc.), while others may not be at all (late afternoon Friday shift? ) Thus, the majority of the GPUs will likely meet the "design target" of 3-5 years, with some going much longer than that, and almost as many falling quite short. It's pretty much a lottery how long one particular video card will last. But generally, running a GPU cooler will more than likely increase its chance of surviving longer.

bummer - and I don't suppose when getting a new card, there's any way to "predict" how long it will last? say some subtle signals in a furmark stress test or maybe subtle voltage variations or something...



also instead of the fan is it possible to modify Bios so that it throttles CLOCK speed? like Intel's Speedstep (so that past a certain temperature the card down-clocks)

[momaka]

Re: unknown (solid?) caps on MSI "military class" graphics card?

so you saying that the difference between japanese & non-japanese polymers is NOT just durability but ALSO performance?

Well, I don't know for sure, but I'd say most likely they do.
Like you said, there has to be a reason why manufacturers use Japanese polymers in the CPU VRM. But it may not be because they are more reliable.

bummer - and I don't suppose when getting a new card, there's any way to "predict" how long it will last? say some subtle signals in a furmark stress test or maybe subtle voltage variations or something...

No, unfortunately, not.
How far the video card overclocks at stock voltage will only tell you about the ASIC quality of the actual chip (i.e. the little silicon core.) It will not tell you if the "fusing" of the GPU core to the substrate was done "better" or "late Friday shift" -bad.

Thus, you could have a really good overclocking chip fail very fast... or not.
And you cal also have a really crappy "hot-running" chip last a very long time... or not.

So in short, it's not possible to determine.
But the cooler (or within reasonable temperature range) you keep a graphics card (or any electronic device for that matter), the more likely it will last longer.

also instead of the fan is it possible to modify Bios so that it throttles CLOCK speed? like Intel's Speedstep (so that past a certain temperature the card down-clocks)

Maybe. Possibly.
I think most newer nVidia video cards (GTX 700 series and above?) have a power limit feature. ATI too possibly, though not sure at all after which series (probably the R7/R9 and onwards??) But from what I've briefly seen/heard, the power limit feature simply clocks the card only as high as it can go before hitting the power limit. So if for example you have a video card with 1 GHz core clock and 150W TDP, and you set the power limit to, say 100 Watts... then the core clock will run at whatever max frequency it can before hitting that power limit (thus, likely less than 1 GHz.)

That said, if anyone knows more about the power limits on newer cards, please correct me if I'm wrong somewhere. I don't have many newer (working) video cards, and thus have not looked into the issue. My current best (working) cards are a GTX 560 Ti and Radeon HD6850. (Yeah, I know... I'm so outdated. Whatever. )

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

alrite then I've another question it's about the cooling system itself (of that card)

I notice that not only the GPU but also the VRM & the memory chips are covered by the heatsink

HOWEVER I also noticed that the mem chips are only PARTIALLY covered by the heatsink:




looks like stupid design right?
notice how the heatsink & thermal pads only touch HALF the memory chips (and for one of them, only 1/4 of it!)


so my question is, will this create a dangerous "thermal imbalance" between the half of the memory chips that's in contact the heatsink & the other half that's not in contact?

basically, will half the chip still run super hot while the other half is colder?

or instead will this still cool the entire chips even if only half is in contact with the heatsink?

[brethin]

Re: unknown (solid?) caps on MSI "military class" graphics card?

alrite then I've another question it's about the cooling system itself (of that card)

I notice that not only the GPU but also the VRM & the memory chips are covered by the heatsink

HOWEVER I also noticed that the mem chips are only PARTIALLY covered by the heatsink:




looks like stupid design right?
notice how the heatsink & thermal pads only touch HALF the memory chips (and for one of them, only 1/4 of it!)


so my question is, will this create a dangerous "thermal imbalance" between the half of the memory chips that's in contact the heatsink & the other half that's not in contact?

basically, will half the chip still run super hot while the other half is colder?

or instead will this still cool the entire chips even if only half is in contact with the heatsink?

Looks good to me, nothing your pointing out makes any difference in the cooling off that card.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

Looks good to me, nothing your pointing out makes any difference in the cooling off that card.

I meant the memory chips

look at the VRAM chips each is only half covered by the thermal pad & heatsink - what about the other half it will be a lot hotter than the covered half right?
like a thermal imbalance or something


and what about that 1 VRAM chips which is only 1/4 covered what's the use of covering it at all

[retardware]

Re: unknown (solid?) caps on MSI "military class" graphics card?

look at the VRAM chips each is only half covered by the thermal pad & heatsink

Do they get hot at all?
Not all RAMs actually get hot.

To me this pad looks more like a spacer.
If it is actually intended to cool hot RAM chips, i think it could be an epic failure.

But that depends on perspective.
Maybe the calculations/experiments showed that the onset of stress damage (thermal, expansion, die, balls and bonds cracking etc) shortly after end of warranty period makes sure that demand for new boards will be present then?
Wouldn't this be clever?

[momaka]

Re: unknown (solid?) caps on MSI "military class" graphics card?

HOWEVER I also noticed that the mem chips are only PARTIALLY covered by the heatsink
...
so my question is, will this create a dangerous "thermal imbalance" between the half of the memory chips that's in contact the heatsink & the other half that's not in contact?

No, it would be fine.

BGA memory chips like these are generally designed to cool through the solder balls on the under side. Thus, with a properly-designed PCB with large ground and Vcc planes, the memory chips should cool themselves through the PCB a lot better than through a thick thermal pad.

With that said, I find often that if the GPU chip is itself cooled well, then this will also allow the whole card's PCB to run cooler, which in turn will cool the memory chips better. In fact, some of these "combined" coolers for both GPU chip and memory sometimes do more harm than good in terms of cooling, as they tend to spread the heat from the GPU chip to the rest of the card rather than actually remove it in the air. But generally, this mostly applies to high-end and hot-running cards with under-sized coolers.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

Do they get hot at all?
Not all RAMs actually get hot.

To me this pad looks more like a spacer.

"spacer"?

No, it would be fine.

BGA memory chips like these are generally designed to cool through the solder balls on the under side. Thus, with a properly-designed PCB with large ground and Vcc planes, the memory chips should cool themselves through the PCB a lot better than through a thick thermal pad.

With that said, I find often that if the GPU chip is itself cooled well, then this will also allow the whole card's PCB to run cooler, which in turn will cool the memory chips better. In fact, some of these "combined" coolers for both GPU chip and memory sometimes do more harm than good in terms of cooling, as they tend to spread the heat from the GPU chip to the rest of the card rather than actually remove it in the air. But generally, this mostly applies to high-end and hot-running cards with under-sized coolers.

ok what about the VRM's then - is it better for those to be cooled by the main cooler or by a small separate heatsink?

[momaka]

Re: unknown (solid?) caps on MSI "military class" graphics card?

ok what about the VRM's then - is it better for those to be cooled by the main cooler or by a small separate heatsink?

None.

They should cool through the PCB only. That's because MOSFETs have the lowest thermal resistance from their Drain tabs, which are often soldered to the board. The epoxy case has rather lousy thermal conduction, and that's why it doesn't really matter much if the MOSFETs have a heatsink on top of them or not. If those MOSFETs really need extra cooling, then small heatsinks should be soldered to the PCB where the Drain on each MOSFET is soldered. This will give the best cooling performance. Anything else is mostly just a gimmick.

I have a reflown HD4850 here that I fitted/modded with a large Xbox 360 CPU heatsink for the GPU chip. The RAM chips and MOSFETs are not covered with anything, like they normally would be with the stock cooler. Nevertheless, both the RAM and the MOSFETs in the VRM run cooler despite not having anything on them. Why? Simply because the GPU chip is dissipating all of the heat through its heatsink and not dumping it back on the RAM and VRM, which is what the stock cooler does when its tiny fan (and silly fan "cooling" profile) can't deal with removing the heat from the stock cooler.

[eccerr0r]

Re: unknown (solid?) caps on MSI "military class" graphics card?

Yes "spacer" -- to reduce the amount of torque applied to the oversized heatsink and heat spreader of the GPU chip. If there was a gap around the center chip instead, you could squeeze one side and bend the board, applying stress to the spreader and the die underneath. The RAM chips already provide some of the height needed and reduce the amount of "stuff" needed to keep the board from flexing when the heatsink is depressed on an edge.

[CapPun]

Re: unknown (solid?) caps on MSI "military class" graphics card?

If those MOSFETs really need extra cooling, then small heatsinks should be soldered to the PCB where the Drain on each MOSFET is soldered.

is there any known video cards so far that have done that? which brand? (preferably good brands like MSI / Asus / Gigabyte)

and in the end does this mean the VRM's heat up less then the GPU? (under full load like furmark)

between VRM & GPU which is designed for higher heat tolerance?

I have a reflown HD4850 here that I fitted/modded with a large Xbox 360 CPU heatsink for the GPU chip. The RAM chips and MOSFETs are not covered with anything, like they normally would be with the stock cooler. Nevertheless, both the RAM and the MOSFETs in the VRM run cooler despite not having anything on them. Why? Simply because the GPU chip is dissipating all of the heat through its heatsink and not dumping it back on the RAM and VRM, which is what the stock cooler does when its tiny fan (and silly fan "cooling" profile) can't deal with removing the heat from the stock cooler.

that's another question I had actually how do you measure VRM temps? so far I never seen a video card that shows VRM temp they only show GPU temp (so no sensors on VRM) why is that?

and how do you measure VRM temps on a video card?