VOGONS


First post, by dionb

User metadata
Rank l33t++
Rank
l33t++

A bit of a puzzle here:

I have two ostensibly identical MSI MS-6168 Ver:2 boards here (uATX, i440BX, Slot 1, onboard Voodoo3 and Creative/Ensoniq ES1373). Both are Packard Bell OEM boards with the same PB part number: 6770550100. Both had the identical 6.3V 2700uF cap next to the CPU burst and leaking, both replaced by a nice new Chemi-con with same specs. You'd think they'd support the same CPUs...

Board #1 (that I've had for about 5 months now) happily accepts any slot 1 CPU I feed it, having tested everything from a Celeron Covington 266 to a P3-1000EB on MSI MS-6905 Master slocket - running at 133MHz FSB.
Board #2 (that I've had for a few weeks, but only today got round to soldering) won't boot with a Coppermine of any description (Slot 1 native, P3 on slocket, Celeron on slocket). With default PB OEM AMI BIOS it hangs post on error code 27, with non-OEM Award BIOS it stops at C5.

All jumpers are in the same position, only things plugged into the board are a PSU, DIMM, the CPU and keyboard and VGA connector (and the ISA POST debug card). To 100% rule out external influences I've used exactly the same PSU, CPU and DIMM on both boards and even swapped the physical BIOS EEPROM chip. Still, whatever I do, board #1 keeps happily accepting Coppermines and board #2 doesn't boot with them. I've checked my soldering (in fact board #1 had a slightly loose solder, fixed that but it didn't change behaviour), I confirmed part# of every IC and component in the voltage regulator and it's identical. Double-checked all the other caps: both identical and equally good state.

There are a few minor differences, but seemingly unrelated to the CPU - different RAM chips for the Voodoo, date code differences of about 10-15 weeks on the ES1373 and the Conexant TV-out chip, and a few Q/A stickers in different locations - but apart from that they're the same. If anything, board #2 is optically in better state, #1 seems to have gotten rather battered here and there (and while soldering I melted part of the slot retention bracket...).

Oh, and back around 2000/1 when these things were pretty new and still in waranty I worked for Packard Bell's UK/IE support desk and don't recall any differences within the Ver:2 boards regarding CPU support. Of course as an OEM we never supported upgrades anyway, but the last systems with these boards were shipped with CuMine CPUs and if some batches hadn't accepted them I'd have expected the field technicians to have reported it.

Any idea where I could look for a cause or solution to this behaviour? I intend to push board #1 to its very limits (P3-1400S, probably with some custom cooling solution for it and the V3 onboard), and wanted a backup just in case. No CuMine is no backup...

Reply 1 of 4, by Roman555

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote on 2018-05-14, 21:39:

....
Any idea where I could look for a cause or solution to this behaviour? I intend to push board #1 to its very limits (P3-1400S, probably with some custom cooling solution for it and the V3 onboard), and wanted a backup just in case. No CuMine is no backup...

Hi,

Although it's an old topic I get curious.
MSI supposed to produce two hardware revisions of the mainboard. As it's said on https://de.msi.com/Motherboard/support/MS6168

For Intel(R) Coppermine(R) CPU support, please check first if your mainboard have a blue sticker with the letter "C" located in the I/O chip, if yes, that means your system could support Intel(R) Coppermine(R) CPU. (The Winbond83977 I/O chip is located near the bios chip)

I suppose the difference would be just PWM controller of VRM : SC1152 (does not support CuMine) vs SC1153 (supports CuMine).
But you wrote that both instances have the same components . So is it really SC1153 on both?

[ MS6168/PII-350/YMF754/98SE ]
[ 775i65G/E5500/9800Pro/Vortex2/ME ]

Reply 2 of 4, by dionb

User metadata
Rank l33t++
Rank
l33t++
Roman555 wrote on 2020-06-14, 16:58:
[...] […]
Show full quote

[...]

For Intel(R) Coppermine(R) CPU support, please check first if your mainboard have a blue sticker with the letter "C" located in the I/O chip, if yes, that means your system could support Intel(R) Coppermine(R) CPU. (The Winbond83977 I/O chip is located near the bios chip)

I suppose the difference would be just PWM controller of VRM : SC1152 (does not support CuMine) vs SC1153 (supports CuMine).
But you wrote that both instances have the same components . So is it really SC1153 on both?

Just double-checked and yes, it's SC1153 on both.

Reply 3 of 4, by Roman555

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote on 2020-06-14, 21:19:
Roman555 wrote on 2020-06-14, 16:58:
[...] […]
Show full quote

[...]

For Intel(R) Coppermine(R) CPU support, please check first if your mainboard have a blue sticker with the letter "C" located in the I/O chip, if yes, that means your system could support Intel(R) Coppermine(R) CPU. (The Winbond83977 I/O chip is located near the bios chip)

I suppose the difference would be just PWM controller of VRM : SC1152 (does not support CuMine) vs SC1153 (supports CuMine).
But you wrote that both instances have the same components . So is it really SC1153 on both?

Just double-checked and yes, it's SC1153 on both.

So it's not the case. Then it looks like something wrong with the Board #2 (loose contacts, cold joints, damaged components).
I suppose POST C5 and 27 might be caused by RAM-, bus-, northbridge- errors, instability of their power supplies (DC-DC voltage regulators).
I believe you know but it's worth to mention that 440BX runs AGP at 89MHz when FSB is 133MHz. IMHO it might cause damage of a northbridge and/or a graphics chip.
I've read about your Board #1 with graphic artefacts. You think that it might be faulty the video RAM, though it also might be faulty the graphics chip Voodoo 3.
But if it's really the video memory haven't you thought about upgrade it to 16MB? Because there's a version with 16MB video RAM onboard 😀
BTW, did you try to run Board #1 with PCI video ? Is it stable with FSB 133 and PCI video?

[ MS6168/PII-350/YMF754/98SE ]
[ 775i65G/E5500/9800Pro/Vortex2/ME ]

Reply 4 of 4, by dionb

User metadata
Rank l33t++
Rank
l33t++
Roman555 wrote on 2020-06-15, 14:46:
[...] So it's not the case. Then it looks like something wrong with the Board #2 (loose contacts, cold joints, damaged component […]
Show full quote

[...]
So it's not the case. Then it looks like something wrong with the Board #2 (loose contacts, cold joints, damaged components).
I suppose POST C5 and 27 might be caused RAM, bus, northbridge errors, instability of their power supplies (DC-DC voltage regulators).
I believe you know but it's worth to mention that 440BX runs AGP at 89MHz when FSB is 133MHz. IMHO it might cause damage of a northbridge and/or a graphics chip.

Not very likely. The AGP port is running out of spec, but the chip has its own clock and runs at regular 143MHz regardless. It supposedly overclocks happily to 166MHz, but I never did that.

I've read about your Board #1 with graphic artefacts. You think that it might be faulty the video RAM, though it also might be faulty the graphics chip Voodoo 3.
But if it's really the video memory haven't you thought about upgrade it to 16MB? Because there's a version with 16MB video RAM onboard 😀

Possible, but it shows all the signs of bad RAM: repeated patterns of artifacts, even in plain desktop mode, with severity depending on resolution in that case.

Here's what it looks like:
AUYA0SwGEDK7FdXbV9y67Dlh.jpg?f=user_large

BTW, did you try to run Board #1 with PCI video ?

Yes, it works fine with PCI video.