CPU at 100% after RECAP

[Gianni]

Hi all,

I have read a lot of post and I must say this is a great forum, a lot of stuff to read.

I hope my English is good enough to explain my trouble and to reply correctly.

I'm trying to repair a friend's motherboard, the PC had a lot of problem: it crashes very often, sometimes it freezes during boot, etc...
First of all I replaced all cap near the CPU (GSC) because some of them were rusty on top and were bulged.

When I finished, I have reassembled the PC and reinstalled the OS (WinXP HE) and the PC worked fine for one day without any problems.

Two days later it started to freeze again so I replaced also the power supply without success.
Then I replaced all the caps except the small ones, I mean smaller than 22uF.
Now the PC is extremely slow and when XP is ready, in the task manager I see the CPU is always at 100% even if non application is open.

I don't know what to do because checking the PC with a tool, everything seem good.

The motherboard is a VIA P4... and I have replaced the GSC caps with Rubycon.
For 2 CAPS I didn't have the right value 330uF 10V so I have replaced them with 470uF 10V; could be this difference the problem?

At the moment I can't name the right type of motherboard because I'm in office.

Any help is greatly appreciated because I don't want to give up now that I have spent a lot of time on this board and I hate to be defeated.

TIA
Gianni

[Rob]

Re: CPU at 100% after RECAP

Run some memtest86. See if you get errors.

[Gianni]

Re: CPU at 100% after RECAP

Hi Rob,

thanks for your reply.

Last night I have launched memory test with Memtest86+ V 2.01, found on Hiren's BootCD 9.5, this morning it was at pass 47% test 25% after 7 hours. I had stopped it because I prefer not to leave the PC on when I'm not at home; anyway when I stopped memtest no error were found.
I don't know if such kind of test take so long time.

Before this test, I have checked the board with PCCheck V6.05 by Euorosoft and all tests were good.
I have seen that the program said memory bandwidth is 154MB/s and when he tested the memory the results were:
memory read 20MB/s
memory write 70MB/s
memory move 12MB/s

Memory MMX speed:

read 40MB/s
write 135MB/s
move 20MB/s

I don't know what those speeds mean and if they are correct

I post now the correct name of board, CPU and memory:
MB Via P4PB 400C-FL Rev. C2
Intel Pentium 4 2.8GHz Northwood 13um
Ram Kingston KVR400X64C3A 512MB
Video card Ati Radeon 9200 128MB

The problem is independent from WinXP because working with Hiren's and KNOPPIX Live CD it takes long time before it is possible to use the PC and working is right impossible due to the very slow behavior.

I have verified the right polarity of CAP and their welding.

What can I do now?

Regards
Gianni

[robofind]

Re: CPU at 100% after RECAP

Gianni Yes, this could be the problem.
When you change capacitors, the value of the new must be in +/- 20%.
So 470 / 330 = 1.42
It is 42% bigger then the original.

And what series of Rubicon you use?

[Gianni]

Re: CPU at 100% after RECAP

Ciao Robofind

I have mounted the series YXF 105ºC, we use this kind of capacitors in SMPS for consumer application.
Unfortunately I have not other types.

Anyway today I have checked all voltages across capacitors with my Tektronix DSO TDS210 and I have not seen spikes, ripple or under/over voltage.
After this test I have reassembled the PC and run a speed test with a tool, I attach here the picture.
The benchmark result is strange, it seems that this CPU is worst than a Pentium 133MHz.
As I don't know if this tool is reliable, I have tested my PC and the result are good.
By the way, while on my PC the test take only few minutes, on the VIA board I have stopped it after 45 minutes while was running memory test.

Do you think is it possible that I have damaged the CPU? I mean it is possible that an electrostatic discharge has ruined something on the board? I tried also to use the DDR of my PC on this board but nothing changed.

I don't know what to do, I'm thinking to recap the board with the original CAPS I removed, just to see if the speed return regular.

Ciao
Gianni

[robofind]

Re: CPU at 100% after RECAP

Rubicon MBZ or MCZ
Nichicon HN or HZ
National Chemi-Con KZG
is what must be used on that place of the board.

I think the processor is OK, but the voltage regulator on the board can not supply enough current to it.
Anyway you must solder on the board ONLY with all components removed. This includes processor, memory, Cmos battery, etc.

Do not recap with the original capacitirs. First measure their capacitance with C-meter or L/C-meter. They are failed because their leakage curent has gone high, capacitance decreased or ESR increased.

You are lucky - I have a board that worked only 30 minutes after recaping with Hitano EXR capacitors. I think the board is OK, but the capacitors are not for that place.

You need Ultra Low ESR capacitors. ( I need too. )
I think to post the things about my board too.

[Gianni]

Re: CPU at 100% after RECAP

Hi robofind

I know if you want to repair something you need the right components, but I think it is very strange that the CPU works so slowly after I have changed the CAPS around memory slots.
Ok, YXF CAPS have higher ESR and lower current capability than MBZ/MCZ but I expect that the board won't start or freeze suddenly, not to work so slow.
Maybe I'm wrong thinking in this way, I'm not a guru in this field.
I'm afraid I have done something wrong the second time, when I changed the capacitors around DDR slots; today I have inspected the board with a magnifying lens to see if I have damaged tracks but I have found nothing.

Three years ago I replaced all capacitors on a colleague's motherboard with YXF capacitors and he had non problem at all; it must be underline that the system was older than the board I'm trying to repair now and probably the power needed by the CPU was lower.

Today I have tested the original caps I replaced near DDR slots, they are marked as 1000uF 10V; measuring their parameters I have seen they are all around 830uF and the ESR is 43mOhm. So I have remounted them on the board, tonight I will reassemble the PC, I want to see if it works at normal speed; in this case I can be sure YXF CAPS are the problem. I know GSC are crap, but before I replaced them, the board was working faster and only sometimes it crashed.
Only the capacitors close to the CPU were rusty and completely ruined, instead of 1500uF they were 78uF or less, I can't remember the ESR value and I can't check it now because I have thrown them all in the trash so for now I have to leave YXF type.
Yesterday I went to a computer shop to look for a motherboard for P4, so I can replace it and keep RAM, HD, DVD and the case; it costs 65,00€. I have to check "if" and "where" I can find MBZ/MCZ CAPS and how much do they cost and then ask to my friend what he wants to do.
I'm doing this work only as a favor for a friend, because he has no money to spend and if he went to a computer shop for sure they have not even considered to replace caps.
I have cooperated for 6 years with a computer shop run by a friend, he has never repaired a motherboard with bad capacitors.
Now I work in a small company, we produce SMPS for industrial and consumer application, so I can have some components for free and I can use tools to rework electronic boards but I can't order components for myself.

I was really surprised when I found this site when I was looking for information about this board, I didn't know that "bad caps" were such a big problem for a lot of people.
I hate to trhown away something that can be repaired, but today this seems the way we live

Ciao
Gianni

[robofind]

Re: CPU at 100% after RECAP

Hi Gianni.
I am not a specialist in that field too, I am an radioamateur. I help to my friends like you do.

I think its the capacitors in the processor voltage regulation. Not these near the memory.
In my case, written here the board worked about 30 minutes, and then stopped.
In your case it at last works - slowly, but works.

I have recapped many old boards - AMD Duron, Intel P3, Intel Celeron.
And they all work fine.
But their processors work at higher voltage and draw much smaller current. P4 draw bigger currents - thats the reason for Ultra Low ESR.

Its not only the price - if the device can be repaired, why not to be repaired?
Sometimes the motherboard is nonstandard and than the computer goes useless.

Its sad, but thats the case now - everyone says if its broken by new. All the industry works on that principle.

So post the results - it will be interesting.

[kc8adu]

Re: CPU at 100% after RECAP

the yxf are not the problem.nor is the subbed 470 in place of 330.
try another cpu.
i have been starting to see skt 478 chips that overheat instantly due to a problem with the thermal compound between the die and slug.
a p4 will throttle down as it overheats.

[Gianni]

Re: CPU at 100% after RECAP

My last test gave bad result: with the original CAPS the CPU is always at 100% end the PC very slow.
Here attached the pics of system properties and processes. It took me 50 minutes to get them.
The strange thing is that the audio is good, but when you move a window it takes some time to refresh the video.

I don't believe, like kc8adu wrote, that using better capacitors will solve this problem, I have read a lot of thread here and the problems due to bad caps are quite different.

I'm convinced I have done something wrong or maybe something happened when I have taken the board from home to work and vice versa.

@ kc8adu: I will try as soon as I find a P4 to test the board and, obviously, if it will be good, I will change immediately the CGS CAPS.

Thanks
Gianni

[Gianni]

Re: CPU at 100% after RECAP

@ robofind: sorry, it is very late for me and I'm so tired that I haven't seen your last message.

If the problem is the ESR to high, I expect that the CPU crash or won't start at all. I don't see any reason why it must work so slowly.
AFAIK an high ESR cause ripple problem and voltage drop when the CPU sink current, but in this case under voltage should turn off the CPU and not decrease it speed or making it to work always at 100%.
I don't know what happened, but it seems to me that for some reason the CPU and the memory don't "talk" to each other, like if the CPU is doing all the works.
This is my poor opinion and probably a big stupid thing, but this is what I feel.

I hope to find the problem because I'm loosing my life

Thanks again for your help
Gianni

[robofind]

Re: CPU at 100% after RECAP

Hi Gianni.
Clear CMOS and verify all the settings are correct. Especially these for processor, memory and disk devices. After recaping the last ECS board it got "frozen" with default settings. So I changed them and it works OK - 3 months up to now.

[gdement]

Re: CPU at 100% after RECAP

I agree with robofind - definitely worth a try to reset the CMOS. If the settings got messed up somehow, it is possible for it to become unbelievably slow. Especially if the memory caches got turned off or various other CPU features are disabled.

Even if it looks good in the menu, reset it anyway. There can be hidden parameters that you can't see, so fully resetting the CMOS is the surest way to clear it up if it's corrupt.


If the CPU is overheating, you might be able to see that in the BIOS, or else in a software utility in Windows. It would depend whether the temperature reading actually comes from the CPU or if it's coming from an external sensor.

[Gianni]

Re: CPU at 100% after RECAP

robofind and gdement what can I say......: THANKS SO MUCH

After clearing CMOS and loading Fail Safe Setup the board works perfectly!!!
Your suggestion was right!!!

Ok the BOARD is not problem free yet, I have to RECAP it with good capacitor and I have to look for the type suggested, otherwise I will stay with YXF.

I have worked with the PC for 2 hours now and I had non problem at all; since my friend has left for vacation, I have one week to use it so I can verify it is OK.

I have run some test after clening CMOS and I have seen the +12V was low; I have opened the main PSU and found one damn cap dead!! See the photo.
So I have replaced once again the PSU with a new one and run test again and this time everything was OK.

Thanks again for your help, I have never think about CMOS issue because looking in the BIOS everything was correct.

Now the nightmare of bad cap is in my head: when I will go to buy something (TV set, notebook, motherboard, etc...) do I have to ask the BOM before buying it?
Reading some post here I see that even if you buy something very expensive you can't be safe from this problem... bad caps are spread everywhere...

Ok that's all for now, I will be happy to offer you a good fresh beer but unfortunately it's not possible... but if you come to Italy don't forget to contact me, I have always beer in my fridge

Ciao
Gianni

[robofind]

Re: CPU at 100% after RECAP

Hi Gianni.
Its best to see that the problem is solved.

For the power suply - at last, change not only the bad capacitor, but all that are the same. Since they may look good but to leak after some time.
With some boards I have this issue. Replaceng only the visualy bad capacitors. After some time other capacitors, that are the same, but were not changed got bad.

[kc8adu]

Re: CPU at 100% after RECAP

i bet if you put the yxf back and replace the psu caps(all the ones on the secondary)it will be fine.
how low is the +12?
looking at your pic the burst cap is most likely on +3.3
but its hard to tell exactly.
i am just going on where it is in respect to the output wires.

[Gianni]

Re: CPU at 100% after RECAP

Hi robofind and kc8adu

for sure I will replace all damn caps in the PSU.

@ kc8adu: I have not replaced the cap as on saturday and sunday I don't work, so I have not all tools to do the work, I will do it tomorrow.
Don't know if the cap is on +12V, anyway in the test I had 2 warning because the +12 dropped to 11.3 and 10.9 two times.
I replaced the PSU with a new one and the voltage with the same test has its lower value at 11.55V.
Tomorrow I will check the PSU and I will let you know if the damaged cap is on +12V.
I have a question for you because I didn't understand one of your post:

i have been starting to see skt 478 chips that overheat instantly due to a problem with the thermal compound between the die and slug.
a p4 will throttle down as it overheats.

You mean that could be a problem with the CPU heatsink?
In this case I think the CPU will decrease the speed but I don't think it stays always at 100% even if you are doing nothing, like was happening to me.
The PC was so slow because CPU was always working at 100%, the temperature I have measured on the CPU, using a small program, was always between 38°C and 49°C. I have touched sometimes the heatsink and it was not vey hot. Before changing the caps, I have removed the heatsink, I have cleaned the CPU and HS removing the white paste and I have put the silver paste, the one specific for CPU Titan TTG-S104.

@ robofind: you have suggested "National Chemi-Con KZG" but probably you meant "Nippon Chemi-Con", am I right?

I am looking for those caps in Italy but till now I have found nothing, the usual site like RS-Components, Farnell , etc... don't have such kind of stuff.
Only Distrelec has Rubycon caps but they are ZL and ZLH series. I have to check if they are better than YXF, but for sure are not the same as you suggested.

Ciao
Gianni

[robofind]

Re: CPU at 100% after RECAP

Hi Gianni.
"National Chemi-Con" = "Nippon Chemi-Con"
Panasonic FJ/FL/FM or Samxon GC/GD are good too.

ZL and ZLH have smaller ESR than YXF. But if the board works correct - stay with YXF. Many solder/desolder operations tend to damage the board.

kc8adu means not the thermal paste / thermal grease between the processor and the heatsink. He means the thermal compaund used to "glue" the silicon chip of the processor to its metal plate. This metal plate is what contacts with the heatsink. The silicon chip is insade, and is not visible, nor the "glue" compaund.

[robofind]

Re: CPU at 100% after RECAP

Here is what i found for description:
This picture will give you an idea. Its computer generated, not real. Its for core quad. It gives an idea of the processor construction.
The processor's metal plate is made transparent and the two silicon chips (which are the processor itself) are visible. P4 is sigle chip however.
The "glue" compaund is between the chip and the plate and is not illustrated.

[Gianni]

Re: CPU at 100% after RECAP

Ciao robofind,

thanks for all explanations now I understand what kc8adu meant.

By the way kc8adu was right: the burst cap inside the PSU (1000uF 10V) was on +3.3V and not on +12V. This morning I have replaced all caps, they were all "Jun Fu"; I have measured them and only 2 were still within specs or at least they had the right value and low ESR.

For the motherboard I know it is not recommended to solder/desolder the board many times, but if I can find better caps than YXF I think it is worth to change them.
Probably I can buy Rubycon series ZL and ZHL, I know they are not like MBZ or MCZ but they are better than YXF.
I don't know where to find this kind of caps, they are too specific to be found in electronic shops where you can buy electronic kits.

Bye
Gianni