VOGONS


First post, by majestyk

User metadata
Rank Member
Rank
Member

The FIC PT-2003 is a nice board. You can add a VRM and run a Pentium MMX 233 with the latest BIOS. It can be equipped with 512K of L2 Cache either asynchronous using DIL SRAM or pipeline burst, when you populate the CELP slot. Since the chipset can address 128MB of RAM (64MB cachable) only, I doubt using 512K L2 cache is useful at all.
But I like the 430FX chipset for nostalgic reasons.

The board came in a quite dirty state. It had some brownish stains on the surface that looked like battery leakage of regular battery cells. I cleaned everything carefully but there still was some corrosion left at the pins of the clock generator and the chips nearby.

When I started testing this PT-2003 it would refuse to POST at any FSB above 50 MHz. It would run fine at 50 MHz, at 60 MHz or 66 MHz it wouldn´t POST or even beep.
I tried with different RAM sticks and CPUs with and without L2 cache to rule out that any component was limiting the maximum FSB.
Then, from time to time all of a sudden it would POST at 60 MHz perfectly, but never at 66 MHz. After a while I suspected this behaviour to be dependant on temperature.
I carefully heated up single components, then cooled them down again to narrow the cause of this issue.
Finally I found that the sentitive component is the Northbridge of the 430FX chipset. When the MCH (SB82437FX66) is heated up at it´s center, the system even POSTs at 66 MHz, but hangs later, when the heat is turned off.
Of course I carefully checked every single pin to make sure the soldering joints are still o.k. , but they are.

PT2003_1a.JPG
Filename
PT2003_1a.JPG
File size
1.56 MiB
Views
328 views
File license
Public domain

I have seen many cases when chips started acting flaky as soon as they reached a certain temperature, but I can't remember any case where a chip worked perfectly only when it´s overheated.
What do you think - can I be sure the MCH is the cuplprit? Are there any alternative explanations for this behaviour?
Replacing this chip is quite time consuming - it would be frustrating to do it without fixing the issue.

Last edited by majestyk on 2021-10-12, 14:43. Edited 4 times in total.

Reply 1 of 11, by dionb

User metadata
Rank l33t
Rank
l33t

My first thought would be bad solder joins.

If you're absolutely sure that's not the case (i.e. already reflowed stuff), the only other time I noticed similar behaviour was with a Pentium VLB system. Was solid at 50MHz, unstable at 60MHz and usually failed to even post at 66MHz - but with better chances after it had been running a while, despite exactly same RAM and CPU being fine on an identical board I have. Eventually narrowed it down to the PSU. I was using an old AT PSU and not a particularly good one, and this particular board seemed to draw more current than the other one, just enough for it to fail stability with that PSU. With a better one, it works fine.

Reply 2 of 11, by majestyk

User metadata
Rank Member
Rank
Member

I think the PSU can be ruled out here, I checked with two different AT and an ATX PSU.
My usual practice to test the soldering joints is to tick every single pin with the tip of a small cutter knife applying some force to see if it´s still firmly connected:

PT2003_2.JPG
Filename
PT2003_2.JPG
File size
238.04 KiB
Views
308 views
File license
Public domain

Everything was still perfect here. But I´ll check again...
__________________________________

Yesterday I repaired a different PT-2003. About 2/3 of the southbridge´s pins had become loose (no idea how the former owner managed to accomplish this).

I had to resolder the red marked areas:

PT-2003_B.JPG
Filename
PT-2003_B.JPG
File size
636.1 KiB
Views
308 views
File license
Public domain

That one is working perfectly now.

Reply 4 of 11, by majestyk

User metadata
Rank Member
Rank
Member

I gave it a try and resoldered all 4 angles, but nothing changed. The fact, that only heating at the very center of the chip cures the issue temporarily indicates that the chip might be the culprit here. (At least I hope so...)
I removed it, but sadly I lost one soldering pad in the process.
Depending on the PCB´s quality there´s a risk of detaching pads even during the 2nd or 3rd soldering.

PT-2003_3.JPG
Filename
PT-2003_3.JPG
File size
1023.32 KiB
Views
266 views
File license
Public domain

I´ll now have to acquire a new 430FX Northbridge.

Reply 5 of 11, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
majestyk wrote on 2021-09-21, 15:36:
I gave it a try and resoldered all 4 angles, but nothing changed. The fact, that only heating at the very center of the chip cur […]
Show full quote

I gave it a try and resoldered all 4 angles, but nothing changed. The fact, that only heating at the very center of the chip cures the issue temporarily indicates that the chip might be the culprit here. (At least I hope so...)
I removed it, but sadly I lost one soldering pad in the process.
Depending on the PCB´s quality there´s a risk of detaching pads even during the 2nd or 3rd soldering.

PT-2003_3.JPG

I´ll now have to acquire a new 430FX Northbridge.

Could it be a cracked via under the chip, expanding and making contact as it warms up? Don't know if it's worth checking continuity with the chip out of the way.

Reply 6 of 11, by tony359

User metadata
Rank Newbie
Rank
Newbie

what about that lonely electrolytic capacitor which sits just by the chip? Worth replacing in case the new 430FX does not fix?

My Youtube channel: https://www.youtube.com/channel/UCvxgGy-GkPY9RA9BcLTBaFQ

Reply 7 of 11, by majestyk

User metadata
Rank Member
Rank
Member

I have checked the traces and vias uner the removed northbridge very carefully. No corrosion, no damages or scratches to be found.

When you carefully heat up the chip on it´s upper side, it would take quite a while until the heat reaches the downside, and finally heats up the PCB. There´s also a gap between chip and PCB.
But the effect sets in after 2 seconds - so I assume there´s something happening inside the chip.

The electrolytic sits next to the southbridge and isn´t affected when applying heat to the northbridge due to the distance (about 4 inches).

While waiting for the new MCH ton arrive I will try my best to restore the missing soldering pad and check the conductivity of the traces in that area.

Reply 8 of 11, by tony359

User metadata
Rank Newbie
Rank
Newbie

oh sorry, I was watching at the wrong picture! Well, your analysis seems correct, so it is now just a matter of waiting I guess! Keep us posted!

My Youtube channel: https://www.youtube.com/channel/UCvxgGy-GkPY9RA9BcLTBaFQ

Reply 9 of 11, by majestyk

User metadata
Rank Member
Rank
Member

Hi everybody,
it took a while for the SB82437FX66 to arrive from China. I received it yesterday well packed in mint conition in it´s original tray, no legs bent.

I couldn´t wait to fit it in so first I restored the lost soldering pad with a thin wire (see blue arrow in the picture), then carefully adjusted the new chip, soldered the pins at the corners and all remaining pins afterwards.
After some cleaning and drying this was the final result:

PT-2003_4.JPG
Filename
PT-2003_4.JPG
File size
1.7 MiB
Views
155 views
File license
Public domain

Everything is working perfectly now, all FSBs including 66 MHz are running stable. Another old SS7 board saved.

Reply 11 of 11, by Roman555

User metadata
Rank Member
Rank
Member
majestyk wrote on 2021-10-12, 14:34:
Hi everybody, it took a while for the SB82437FX66 to arrive from China. I received it yesterday well packed in mint conition in […]
Show full quote

Hi everybody,
it took a while for the SB82437FX66 to arrive from China. I received it yesterday well packed in mint conition in it´s original tray, no legs bent.

I couldn´t wait to fit it in so first I restored the lost soldering pad with a thin wire (see blue arrow in the picture), then carefully adjusted the new chip, soldered the pins at the corners and all remaining pins afterwards.
After some cleaning and drying this was the final result:

PT-2003_4.JPG

Everything is working perfectly now, all FSBs including 66 MHz are running stable. Another old SS7 board saved.

Very impressive!