Epox 9NDA3+ with exsessive coil noise

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

That is a coil wrapped in heatshrink tube I think, the heatshrink tube has 125°C written on it so if it indeed had elongated on your PSU I think your caps where in pretty bad shape too :P

Look in this thread (the linked pictures) to know why it is dead...
https://www.badcaps.net/forum/showthread.php?t=735

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Trodas: yet another pic, this of my DFI Expert that had coil noise from this coil that I fixed with hot glue...

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Damn, it just struck me; why the Nexus NX3000 failed... It is because the voltage controller for the fan failed!!!

After some futher investigation (refer to this pic) it is giving out 2.8v on the fan output, regardless of the temp of the thermistor... I guess the most likley candidate is the thermistor? Or is it more likley that something else is dead?

With a fan connected the temp of the large resistor close to the resistor gets very very hot, upto 100°C, but this is not very strange since all current is going into it instead of the fan...

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Desoldered the thermistor, at ca 25°C it is showing 0.8kΩ on my multimeter, it should be 10kΩ...

However this is still strange because this is a Negative Temperature Coefficient (NTC) thermistor so the lower the resistance the more current it should let through... So if anything the fan should have run at 12v instead of 2.8v...

So put back the thermistor and measured the voltage with no load present: 11.6v...

The plot thickens... What could be wrong?

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Trodas: yet another pic, this of my DFI Expert that had coil noise from this coil that I fixed with hot glue...

LOL Per.
With that attack-of-the-silicone, I think Trodas won't be speaking to you ever again!

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Desoldered the thermistor, at ca 25°C it is showing 0.8kΩ on my multimeter, it should be 10kΩ...

However this is still strange because this is a Negative Temperature Coefficient (NTC) thermistor so the lower the resistance the more current it should let through... So if anything the fan should have run at 12v instead of 2.8v...

So put back the thermistor and measured the voltage with no load present: 11.6v...

The plot thickens... What could be wrong?

Would it make your troubleshooting easier if I were to remove the fan boards from one of the Fortrons here and compare measurements?

Let me know.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Would it make your troubleshooting easier if I were to remove the fan boards from one of the Fortrons here and compare measurements?

Let me know.

Nahh, don't bother, I'm just thinking out load, plus, it increases my postcount :P

Tested the thing with another fan, rated 12v 0.09Ah=1.09w only...

It spun and got ca 7v, but when I placed the soldertip against the thermistor the RPM did not increase... So I don't know what to think anymore

EDIT; shorted the thermistor with a needle and the fan spun at full RPM, so I was atleast correct in that lower resistance increases voltage... I think this pretty much prooves that the thermistor is dead right?

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Wohoom progress; the mainboard now gives the C1 memory error tiresias was experiencing

It went like this, computer became more and more unstable, did not pass Prime95 torture test for even a few minutes and would hard lock up but still restart with a press of the reset button... So I lowered my OC from 240x11 to 230x11 and it was Prime95 stable for 5 hours before an error was detected...

So I went up early in the morning before work today just to lower the CPU OC further so I could resume Prime95 testing, then it failed to boot at stock speed 200x11, hardlock... Reset gave C1 error message. Power down and up and it hung at Port80H stage 52: "Test all memory"

So I relaxed the memory timings from 1T to 2T, computer started and hardlocked in Windows, reset=C1... Had to leave for work now...

Got home, start computer, C1 again, so I moved the DIMMs over from slot 1&2 to 3&4, computer booted to Windows and hardlocked, reset=C1... I kinda see a pattern now :P

So I went into the BIOS and watched the mem voltage, I have it set at default, the board seems to think that is 2.7v (should be 2.5v but anyway) it is fluctuating wildly, even idle in the BIOS, between 2.7v to 2.83v!!!

So I shutdown and remove a DIMM, back into BIOS mem voltage is now between 2.67 to 2.73v much better but still pretty poor, I resume my overclock at 11x240, Prime95 fails in Windows after a few minutes, so now I have relaxed it to 230x11 and been running Prime95 for over an hour...

Now here comes my theory; when tiresias sent the board to me the capacitors electrolyte had time to stabilize, and go back to a good enough state for the board to work with two DIMMs, remember the shipping took a few days, and then I had the board for about 2 weeks before I even started it... And now these errors have started after about 2 weeks of usage, I think my theory might be right so I'm ordering some Sanyo MV-WX caps tonight and we will see who is the wiser at the end of the week

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Got home, start computer, C1 again, so I moved the DIMMs over from slot 1&2 to 3&4, computer booted to Windows and hardlocked, reset=C1... I kinda see a pattern now :P

That's the one! I feel no longer "misunderstood"!

Interesting how the fluctuating Vdimm is associated with two DIMMs installed more than with one - this would correlate with the problem I was, and you are , experiencing.

Now here comes my theory; when tiresias sent the board to me the capacitors electrolyte had time to stabilize, and go back to a good enough state for the board to work with two DIMMs, remember the shipping took a few days, and then I had the board for about 2 weeks before I even started it...

Perhaps I forgot to mention this, Per - the board had not been used for at least 3 months, ever since replacing it with the Tyan I'm now using, which was in November 2005.

...

PS. So you think it could be the Sanyo near the Vdimm FETs that's failed? Interesting - since I had always suspected the Hermeis to have been the problem, which you very skillfully replaced with MBZ. Let us know when you find out more.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

PS. So you think it could be the Sanyo near the Vdimm FETs that's failed? Interesting - since I had always suspected the Hermeis to have been the problem, which you very skillfully replaced with MBZ. Let us know when you find out more.

No, I should have clarified, I do think it is the Hermei caps that have failed... I only swapped the 3300uF 6.3v Hermei caps with Rubycons, I left the smaller 1000uF 6.3v Hermei caps, mainly because I did not have proper caps to replace them with at home, I think using 3300uF everywhere where there is now 1000uF would be fatal for my powersupply; because the inrush current would be extermley large when the caps charge up for the first time...

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Swapped all 16x 6.3v 1000uF Hermei caps with Sanyo MV-WX 10v 1000uF, also one small 470uF by the ATX connector, had to place the new cap way too far away for it to be safe though, since the original cap was 8mm and mine was 10 and there was no chance to get it to fit any other way

No difference what so ever, if I run two DIMMs now it lockups if I try to run memtest86+... (With C1 error upon reset)

The coil noise did not go away either, was hoping that 470uF cap would have done it...

And measured with my multimeter at the coil at the very right on the board that I presume supplies the memory with current I read 2.735-2.743v with one DIMM... And the max goes up to 2.746v with two DIMMs, so it's not a power issue it seems afterall...

Right now there are 8x 100uF 16v caps and one 470uF 16v by the AGP slot left, I was meaning to replace those too but bought only one 470uF cap (did not see the one behind the ATX connector...) and the job was so difficult so I decided to leave the 100uF caps...

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Per - sorry to hear that recapping didn't help. I must say though, I'm not entirely surprised - I also had found the "2 DIMM issue" a beastly problem to troubleshoot (as I had mentioned in the 'for sale' thread). At least you seem to be able to reliably reproduce the problem with memtest, which can help in letting you know if any particular "maneuver" happens to fix it.

If I were to make a suggestion at this point, it would involve trying to assess the state of the BGA connections under the chipset. While it may be coincidence, this annoying problem did start soon after having removed the stock chipset cooler. Epox had used double-sided tape which very strongly adhered to the chip's surface, and although I used great care in removing it, I wonder whether the procedure may somehow have disturbed any (previously already weak or defective) connections under the chip.

Perhaps you could check whether pushing on the heatsink a little helps? You could see if freeze-spray exacerbates (or even improves!) the problem...

Let us know.

...

PS. I was thinking - the way the problem manifests itself is definitely weird (ie. it wasn't working with me, but it then worked with 2 DIMMs seemingly perfectly when you received it, while now it C1's on you almost immediately. Almost sounds like a loose connection... in any case, see if that MCP chip looks suspicious in any way.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Per - sorry to hear that recapping didn't help. I must say though, I'm not entirely surprised - I also had found the "2 DIMM issue" a beastly problem to troubleshoot (as I had mentioned in the 'for sale' thread). At least you seem to be able to reliably reproduce the problem with memtest, which can help in letting you know if any particular "maneuver" happens to fix it.

Yea, you know what, it is a bit odd because before I was not able to reproduce the error, it would happen after a while just like for you... But now I fire up memtest and it hardlocks and I get C1 upon reset...

If I were to make a suggestion at this point, it would involve trying to assess the state of the BGA connections under the chipset. While it may be coincidence, this annoying problem did start soon after having removed the stock chipset cooler. Epox had used double-sided tape which very strongly adhered to the chip's surface, and although I used great care in removing it, I wonder whether the procedure may somehow have disturbed any (previously already weak or defective) connections under the chip.

Perhaps you could check whether pushing on the heatsink a little helps? You could see if freeze-spray exacerbates (or even improves!) the problem...

Well, it felt like Murphy was playing me one but after first trying to restart the computer with the C1 error 10 times in a row I took your advice and pressed down on the heatsink; and it booted right up into memtest... I tried it a few times with the same success, of course I was unable to hold an even pressure but it looked like as soon as I removed pressure it would lockup...

So I tried to fix it by screwing down the heatsink really hard against the chipset, however this does strangley enough not help at all... Go figure... Strange problem indeed :P

[Rainbow]

Re: Epox 9NDA3+ with exsessive coil noise

It bends the PCB while pushing the heatsink by hand does not.
I have some boards with bad joints under BGA chips. It can be probably fixed by using heatgun - but I don't have any.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

It bends the PCB while pushing the heatsink by hand does not.
I have some boards with bad joints under BGA chips. It can be probably fixed by using heatgun - but I don't have any.

You where right, after I removed that thing the HSF fell off, I had broken the epoxy or whatever it was that held it in place...

However then I got an idea and tried it with a G-Clamp, tightened really hard, C1 memory error...

Murphy!!!

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

trodas: is this the type of caps you recommend using to filter the vcore, in parallell with the bulkcaps that are already there? Should I put them on both the 12v and vcore (1.5v) side or only one of them?

http://www.elfa.se/elfa-bin/setpage....htm?_65_752_29

???

[hkivan]

Re: Epox 9NDA3+ with exsessive coil noise

In some Hi Fi DIY forums, I have learnt of by soldering a small cap, just a few
pF or 0.xx pF, parallel to the original large cap to filter out high frequency noise. Maybe, this is what trodas want to say.

Further, I usually find some gue inside PSUs. Not only on coils, they are on the bulk
caps, heatsinks and plastic sheets... too.

[Zorro]

Re: Epox 9NDA3+ with exsessive coil noise

I've an Epox 8RDA3+Pro with coil noise. One of the four coils next to the CPU squeals during some 3D operations, including F@H when the graphic updates. During the course of my troubleshooting, I discovered that reducing my FSB from 200 to 166 opened up the option for AGP Spread Spectrum in the BIOS. An AGP-SS of .50% greatly reduced the coil noise. So, you may want to try a different video card and ensure that AGP-SS is enabled.

Of note, I've read on the Rage3D forums that a lot of people are experiencing coil noise in conjunction with their video cards, and it's gotten so bad that even Nvidia owners were asking for help in the Radeon forums. It has been asserted that it's not the quality of the board parts, so much as it is the combination of hardware, so replacing the caps may be in vain.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

I've an Epox 8RDA3+Pro with coil noise. One of the four coils next to the CPU squeals during some 3D operations, including F@H when the graphic updates. During the course of my troubleshooting, I discovered that reducing my FSB from 200 to 166 opened up the option for AGP Spread Spectrum in the BIOS. An AGP-SS of .50% greatly reduced the coil noise. So, you may want to try a different video card and ensure that AGP-SS is enabled.

In my course of troubleshooting I have no idea if I have tried another graphicscard or not anymore
However since running F@H makes the coils quiet I'm letting it run like this now... My brother is using the computer from now on...
I guess I kinda feel a bit defeated by not actually being able to fix either issue with the board, it's still a great board though and I do not regret buying it...

Of note, I've read on the Rage3D forums that a lot of people are experiencing coil noise in conjunction with their video cards, and it's gotten so bad that even Nvidia owners were asking for help in the Radeon forums. It has been asserted that it's not the quality of the board parts, so much as it is the combination of hardware, so replacing the caps may be in vain.

Yea, tell me about it, even my nVidia (XFX) 7900GT squeals now... Got damn coil noise
However I think I will not try to make it stop, it is not at all loud and I do not want to put glue on the coils on it because they are varm as they are...
Damn memory v-reg IC's on the back get over 70°C, it's like this on all 7900GT cards with the same v-reg layout...
http://www.xtremesystems.org/forums/...&postcount=417

[tazwegion]

Re: Epox 9NDA3+ with exsessive coil noise

I was wondering Per Hansson, have you, or do you know of anyone who has successfully pulled a coil out and rewound it firmly by hand?

The reason I ask this is... many moons ago I required a Uniden UHF transciever to be repaired, being of a mature age parts (the coil) was unavailable, so the technician prefabricated one from scratch... for what it's worth, that very same transciever is still in service today