Which caps to replace: ECS P4M800Pro-M V1.0A

[jonny124]

Found this comment from someone on VOGONS forum:

I recently found out that 775 VIA chipsets are unstable with a 1066 bus. This mainly concerns chipsets released before 2007 +/-.
I found out this on P5VDC-MX V2.0 motherboard with P4M800PRO chipset, that was released at the end of 2004 and it was modified version of P4M800 chipset that released in 2003.
If you set for example Core 2 Quad Q6600 (or maybe Core 2 Duo E6600) on P5VDC-MX V2.0, that supports Core 2 CPUs, an unpredictable reboot may occur at any moment. But possibility of this reload decreases with lower bus frequency and later chipset's release date (somewhere around 2007). This problem maybe comes from 2003, when Gallatin 1066 bus was supported by Intel for exapmle 865 chipset, but VIA not aimed on that frequency and this problem was fixed in later chipsets but not completely (mainly fixed by newer production technology)
P4M890 and P4M900 chipsets have the same problem, but I have never seen motherboard on these chipsets with changeable bus frequency to the downside. Reload problem with 1066 can may occur after weeks or months of use, but more often appears under load. This makes VIA chipsets on 1066 MHz unreliable.
So latest absolutely stable bus frequency on VIA is 800 (but I saw posts with ECS P4M900T-M2 mobo that suddenly stopped working with Core 2 Quad CPU's and only started with 533 bus, but this is not beacause of caps, power, memory, BIOS or something like that, thats very interesting)

So it appears my theory is correct, some early revision of P4M800Pro chipset are unstable, and later manufactured model fix the issues with 1066FSB, i also found an article in fact were it's mentioned in the Chipset diagram / datasheet that it supports 1066FSB, but in original VIA announcement on their website they listed max 800FSB, so definitely they did an oopsie in some earlier silicon and instead of wasting stock and using fixed ICs they dediced to hide this issue and maybe hope that owners of early revision P4M800PRO will never try 1066FSB CPU

[stj]

try getting the chipset datasheets

[jonny124]

well i think it looks like now if anyone on the internet has ever gotten that board to work with a pd cpu reliably and stably being the key words. if no one has ever seen that, then its just false advertising on ecs' part.

i could say my cpu is compatible with donkeys and monkeys but has anyone ever seen a donkey or a monkey use a computer? so basically, anyone can say whatever they like but whether that is true or not or will it become true or not, people will have to see it and judge for themselves.

also a pentium d doesnt have a mem controller in the cpu. for intel, that only started much later with the core i7 series. so i guess this settles it for this board. it will only boot to bios with a pd cpu but cant boot an os properly, so this is strictly a single core p4 board only. if u wanna run pd, plenty of other better boards out there that will run it like the gigabyte 945p-s3, 945p-ds3, 965p-s3 and 965p-ds3.

There were people who got it running stable with Pentium D according to decade old posts, however those same people had to return the board before getting one that ran the CPU in a stable way, why we don't know but it would also be strange of them to offer a Pentium D bundled with this board at Fry's back in 2006 (most forum posts you find talking about ECS are guys who got the board for this exact reason).
But anyways i already found another guy selling some hw bundle which include a board and 2 775 CPUs, a core 2 duo E6xxx something and a pentium dual code, i will try both in the ECS board just for a final test if not then i will put this new G35 mobo.
Maybe some specific revisions of the P4M800PRO chipset have some fault and that's why they have issues with different CPUs, otherwise it makes no sense to me that someone just RMA the board and the new one works as expected.

[ChaosLegionnaire]

well i think it looks like now if anyone on the internet has ever gotten that board to work with a pd cpu reliably and stably being the key words. if no one has ever seen that, then its just false advertising on ecs' part.

i could say my cpu is compatible with donkeys and monkeys but has anyone ever seen a donkey or a monkey use a computer? so basically, anyone can say whatever they like but whether that is true or not or will it become true or not, people will have to see it and judge for themselves.

also a pentium d doesnt have a mem controller in the cpu. for intel, that only started much later with the core i7 series. so i guess this settles it for this board. it will only boot to bios with a pd cpu but cant boot an os properly, so this is strictly a single core p4 board only. if u wanna run pd, plenty of other better boards out there that will run it like the gigabyte 945p-s3, 945p-ds3, 965p-s3 and 965p-ds3.

[jonny124]

you could scope the vrm output,
if it's weak then upgrading the fets or inductors may fix it

I dont have oscilloscope to test the phases, but i could replace the mosfets, the Rev 2.0 has all mosfets slot populated and also claims support for more Core 2 CPUs, the Rev 1.0 looks almost identical just with less mosfets (but it does have extra available spots for them), i wanted to get schematic but the only ECS Rev 1.0A i found is on Vinafix which requires payment to download

[stj]

you could scope the vrm output,
if it's weak then upgrading the fets or inductors may fix it

[jonny124]

At this point might just be faulty CPU who knows, why cant it boot anything past 200mhz or even has difficulties like locking up in the BIOS? Its not using much power to stress VRM in this state, maybe memory controller inside CPU is just damaged

[jonny124]

install Linux so it creates a crashlog.
and try running memtestx86+ on it

Honestly what can it do to help me? The system just resets randomly after some time with 200mhz (no bsod, just hard reset) anything higher cant even boot the OS, dont really want to bother with Linux if we know the issue is hardware already, RAM was tested at 533mhz with Pentium 4 and gave no errors (tm5 anta777 extreme + memtestx86 bootable cd).
Tried pentium d with 1 stick DDR2, 2 sticks, 1 stick DDR1 and 2 sticks DDR1, always same behaviour even if CPU is locked at 800mhz by setting speedstep to "minimum".

[jonny124]

Noticed that the ECS REV 1.0 page has a different CPU support list, on that only 1 pentium d is sipported (Pentium d 820), but it shows support for 6 Core 2 Duo CPUs all with 65W tdp, max one being E6700.
While my REV 1.0A doesnt shows any core 2 duo, but it shows 95W pentium D.
Mobo design beetween Rev 1.0 and Rev 1.0A is exactly the same, except a moved fan header, same exact caps hole and they even have same BIOS file.
I will try to find a Core 2 Duo mentioned on Rev 1.0 sipport page and see if that will be it to fix the problems, i also checked that my Pentium D is of D0 stepping (the one mentioned in Rev 1.0A support list for Pentium D 945) so it should work but here we are, so next on my list is to get a cheap core 2 duo and test, otherwise mobo will be swapped

[stj]

install Linux so it creates a crashlog.
and try running memtestx86+ on it

[jonny124]

After doing some tests the situation is exactly the same, the system doesnt give BSOD only with 200mhz ddr2 frequency, however there are still random reboots (even with all possible power saving settings disabled, and speedstep set to "Maximum" instead of Auto), using ddr1 instead of ddr2 seems to make no difference at all, also tried different BIOS versions

[jonny124]

Update:
So i replaced the capacitors, board still refuses to do anything with 533Mhz at stock timings with Pentium D 945, Pentium 4 3ghz works just fine as before evem after hours of stresstest with testmem5.
Now i was able to boot with 200mhz ddr2 clock and system already stayed on for 30 mins, im doing testmem5 and some other stress tests we will see how it does.
In the end if you have this motherboard i feel like its not worth it to recap VRM for maxing out the CPU, even if the previous polymers tested meh board was still working with Pentium 4, its just bad quality mobo, nothing can be done unfortunately.

[Agent24]

ALL polymers have esr lower than liquid electrolytics.
some have lower esr than others but they are all good for replacing older caps

Yea but I wouldn't put it past the Chinese to make fake polymers that are just cheap electrolytics hidden inside or something.

[stj]

ALL polymers have esr lower than liquid electrolytics.
some have lower esr than others but they are all good for replacing older caps

[jonny124]

Polymers should be very low ESR and suitable for motherboard VRMs,, but the listing will say whatever they think will get you to buy it. "100% genuine" is also thrown around on many AliExpress listings when the part is clearly priced far too low to be genuine. Like 10x lot of some IC for $2.50 that costs $5 each at DigiKey, and they claim 100% genuine selling something that would be $50 from a real distributor, but for like a 20th of the price...

As far as I know, and I don't have much experience with them to be fair, polymers should be ultra low ESR, lower than what most ESR meters can really measure, unless you have high end test equipment, that most people can't afford.

But I guess if reviews are OK then yeah they could be good. I have also read that any polymers are okay but that was some years ago and the Chinese will fake or produce low quality of literally anything if they think it will make them more money, so without the ability to test and verify them, who can really say...

The Panasonics are questionable. Capacitance reading high can indicate low quality or a capacitor that's old with high leakage, and if that's the case, ESR may not be great either. If you can get better ones from a known good shop like DigiKey that would be great.

That said ECS are known to kinda suck so your problems may be an underpowered VRM... BUT I would definitely populate ALL the capacitor positions for the VRM not just the ones that were used already. If the capacitance on the VRM input is too low or they are high ESR it could also cause the problem.

Yea i dont trust the "panasonic" ones as the listing had like 2 reviews, will try to find some decent ultra low esr 1800uf caps on Ebay Italy as i need them fast and cant bother with Ali again.
The thing with mouser / digikey is i always have to do a big order to avoid paying shipping / import duties, so its good when i have to buy a lot but i cant always rely on them.

[Agent24]

Reviews looked very good, the listing said "for motherboard vrm low esr" and many people posted pictures of capacitance test, unfortunately i dont have anything to measure ESR and i cant always order from mouser / digikey / ebay US due to high shipping and import cost to EU, so i dont have many places to purchase from, i also heard that polymers arent ultra bad on Ali so i rolled the dice and decided to risk it.
The electrolytics they sent me are probably crap so i will need to find a different replacement

Polymers should be very low ESR and suitable for motherboard VRMs,, but the listing will say whatever they think will get you to buy it. "100% genuine" is also thrown around on many AliExpress listings when the part is clearly priced far too low to be genuine. Like 10x lot of some IC for $2.50 that costs $5 each at DigiKey, and they claim 100% genuine selling something that would be $50 from a real distributor, but for like a 20th of the price...

As far as I know, and I don't have much experience with them to be fair, polymers should be ultra low ESR, lower than what most ESR meters can really measure, unless you have high end test equipment, that most people can't afford.

But I guess if reviews are OK then yeah they could be good. I have also read that any polymers are okay but that was some years ago and the Chinese will fake or produce low quality of literally anything if they think it will make them more money, so without the ability to test and verify them, who can really say...

The Panasonics are questionable. Capacitance reading high can indicate low quality or a capacitor that's old with high leakage, and if that's the case, ESR may not be great either. If you can get better ones from a known good shop like DigiKey that would be great.

That said ECS are known to kinda suck so your problems may be an underpowered VRM... BUT I would definitely populate ALL the capacitor positions for the VRM not just the ones that were used already. If the capacitance on the VRM input is too low or they are high ESR it could also cause the problem.

[jonny124]

Capacitors from AliExpress might be the problem.... You listed measured capacity not ESR. If those polymer aren't actually ultra low ESR they may be the problem...

Reviews looked very good, the listing said "for motherboard vrm low esr" and many people posted pictures of capacitance test, unfortunately i dont have anything to measure ESR and i cant always order from mouser / digikey / ebay US due to high shipping and import cost to EU, so i dont have many places to purchase from, i also heard that polymers arent ultra bad on Ali so i rolled the dice and decided to risk it.
The electrolytics they sent me are probably crap so i will need to find a different replacement

[Agent24]

Capacitors from AliExpress might be the problem.... You listed measured capacity not ESR. If those polymer aren't actually ultra low ESR they may be the problem...

[jonny124]

ChaosLegionnaire
I just installed the polymers, filled all 9 slots with 6.3V 680uF from Aliexpress (ESR measured 687uF) here is a picture of the board and also the electrolytics next to the polymers, which are 6.3V 1800uF.
i did buy some "" Panasonic"" 6.3v 1800uF on Ali but they are measuring at 2200uF so i dont know about the quality (there were 4 of these caps on the board, i desoldered one to check it), there are extra slots for these caps too so i also plan to max them out like i did for the 680uF ones.

[jonny124]

why would u trash the board if it was working fine with a pentium 4 before? obviously the board simply cant support the pentium d due to its high tdp specs since ecs skimped out on the vrm components and caps. if one isnt good with a soldering iron, then just stick with the low speed single core pentium 4 cpus from 2.8 to 3.2 ghz max. 3.4 ghz and higher have higher tdp requirements and so the board may not run stably with those high powered cpus.

so u have found out the hard way that just because its supported by the manufacturer and even if it fits in the cpu socket doesnt necessarily mean that it will work. dont trash the board just simply because one component, the cpu, doesnt work. the board uses an agp slot and has hybrid ddr1 and 2 mem support so it can support many OSes from win98 to winxp like i said above. very much a keeper if u dont need dual core cpu support and frankly, if i needed dual core cpu support, i would build another system on core 2 duo or phenom ii x2 as those are some of the best dual core cpus around. pentium d was intel's first attempt at a dual core cpu and it turned out to be quite the flop. not sure why some people like it unless for intel fanboi reasons. athlon 64 x2 was amd's first attempt at a dual core cpu and also the world's first dual core cpu and they scored a wonderful goal indeed in the annals of computing history!

so just forget the board ever supported pentium d. it was just false hope on the manufacturer's part. it was never designed to esp. since ecs made a later revision v2.0 of this board that did indeed work with up to core 2 duo allendale and conroe cpus.

Yes i understand that, i simply thought since the board listed a lot of high powered CPUs it would work, i bought some 6.3v polymers on aliexpress that had good reviews by people with esr test pictures.
As for the dual core CPU the reason i went with pentium d was just because i got this PC for free and looking online i found a Pentium D 945 for 5€ + Ati hd3650 agp for 20€, paired with 2GB of Ram and it was gonna be an okay HTPC with DVD, HD DVD, Bluray and DVB-T support.
So im assuming even the Rev 2.0 of the board has problems with Pentium D and Core 2 Duo, mine is Rev 1.0A and it has the exact same capacitor configuration as OP, 6 polymer 4 electrolytic same brand / model, thats why i didnt post any pictures.

I already have a Win 98 build with a mobo that has ISA, AGP, PCI and maxed out to 1.5gb ram, so i dont really need this board, i will replace the caps and if it works great, otherwise i will put it for sale (pentium 4 is just not enough for hd decoding)