VOGONS


MSI Slotket woes

Topic actions

First post, by neonblurb

User metadata
Rank Newbie
Rank
Newbie

I recently built a Pentium III Slot 1 machine which went through a number of upgrades, after which I found the system was running really poorly and would not stay up for longer than a few minutes at a time. Through a gradual process of elimination, I managed to figure out that the issue was to do with the Slotket adapter I was using, an MS6905 Ver 2. Swapping it for an Asus S370-DL completely fixed all the issues I was having, but I'd like to see if anyone in the community has seen something similar to my experience so I can try to figure out what went wrong with the MSI slotket and learn fromshare my experience with this kind of issue.

My basic setup:

  • Asus P3V4X motherboard running latest official 1006.004 BIOS (not the oezmen modified one)
  • Pentium III 1000 EB (Coppermine) SL52R running through slotket
  • 1 GB PC133 RAM (2x 512MB)
  • CF to IDE adapter
  • GeForce FX 5200 AGP

After mixing and matching components, I found that the system was much more stable after swapping the RAM for a set of 3x 128MB PC100 sticks I used with a prior 440BX build whether I ran these at 100 or 133, but I was still having severe instability. I was at least able to get XP to install this way and get the system to boot after a couple of attempts, but shortly afterwards it would hang.

The glitches I was seeing were mostly:

  • File access related glitches - the system would fail to read files from known good media (e.g. Windows XP install CD which had worked in the same drive in the past)
  • Errors extracting known good ZIP or RAR archives - the program would claim a specific file in the archive was corrupt, always the same one, although this archive was known to be good
  • Hanging after a certain amount of file access attempts - the system would still be running, the mouse cursor would move, etc. but absolutely no disk access would happen often leading to minor file system corruption after a reboot

I did try switching the CPU as well for a second SL52R I had, which gave exactly the same issues, so I think it's very unlikely it was just a bad CPU in this csae. Switching to a standard Slot 1 PIII 550MHz (Katmai SL3FJ) after this gave me a completely stable system which was what finally confirmed to me that it was the slotket. Other things I tried in addition to switching the RAM: I tried a number of different AGP video cards including S3, 3dfx and later GeForce models, I also ran with no other cards installed in the system and also tried using a number of different CF cards from different manufacturers, types and sizes. I even tried swapping my 235W power supply for a newer 500w Thermaltake unit. Another oddity was that the system would be basically stable under DOS and would never hang prior to Windows starting up in the boot sequence -the hangs would only happen under Windows XP. Memtest ran through perfectly well on the 1 GB of RAM, and the system was stable throughout the long testing cycle.

In terms of the slotket itself, I'm fairly confident the jumpers were set correctly for my CPU (everything on Auto basically) although the only instructions I could find were for the 2.3 revision, while I have a 2.0. Everything looked fine in BIOS: CPU was being detected fine, I was getting a reasonable thermal diode reading, I was able to set the frequency, and it was picking up and running at the correct voltage (1.75V). I did try overvolting slightly to 1.8V to see if it would improve stability at all, but saw no change. I also flipped pretty much every setting in the BIOS to the most conservative value but it made no difference, and I was having the same issues with an earlier BIOS revision. I suspect the particular version of the slotket I was using was just not stable with higher clocked/later Coppermine models running at 133 MHz, although I wasn't able to find anything to that effect online.

My questions to anyone who might have experienced something similar are:

  • What might be causing the underlying instability?
  • Why would it have improved after swapping out the RAM?

I'm just hoping to learn from this and maybe pass on my experience in case anyone else is having or has had a similar issue.

Thanks!

Reply 1 of 9, by cloppy007

User metadata
Rank Newbie
Rank
Newbie

I'd run memtest86+ as a first stability test and then some prime95 (some hours), if you can handle the temps.

Reply 2 of 9, by neonblurb

User metadata
Rank Newbie
Rank
Newbie

Thanks, Memtest ran through fine with the MSI slotket but I can't run Prime95 because the system is unstable under Windows - it's really odd.

Reply 3 of 9, by rasz_pl

User metadata
Rank l33t
Rank
l33t
neonblurb wrote on 2024-11-22, 16:51:

[*]Asus P3V4X

I have boardview for this board 😀

neonblurb wrote on 2024-11-22, 16:51:

I found that the system was much more stable after swapping the RAM for a set of 3x 128MB PC100 sticks

ASUS hates using 3.3V directly from ATX connector.
ASUS loves generating VTT 1.5V from 3V rail.
Ram is supplied from 3V. VTT 1.5 V is derived from 3V thru Q7 RFD3055S mosfet operating in its linear region. Nearby CE7 14 make it smooth.
My theory: you might have insufficient 3V, maybe caps got old? ASUS slotket is of higher quality and can mitigate somewhat worse VTT.
3V is generated by circuity between SLOT1 and DIMM slots. CE18 19 22 23 30 31 32 are responsible for making it smooth and steady.

Scientific approach - put a scope on VTT and 3V rails and look for spikes/noise.
Practical approach - start desoldering caps I listed and throwing them on an LCR meter to check ESR.
Shotgun approach - start replacing caps I listed with new ones.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 4 of 9, by myne

User metadata
Rank Oldbie
Rank
Oldbie

If it's during file copy, wouldn't it more indicate something on the surerio side?

To test: 66fsb/ram (33pci) and see if it does the same thing.
If yes, everything else is underclocked, so it is most likely that side.

If no, you might be right about the cause of the instability.
That said, pre p4, almost everything was derived from the PSUs 5v. New psus are weak on the 5v circuit because everything is taken from 12v now.
Even if you have a 1000w psu it might not be enough.

I built:
Convert old ASUS ASC boardviews to KICAD PCB!
Re: A comprehensive guide to install and play MechWarrior 2 on new versions on Windows.
Dos+Windows 3.11+tcp+vbe_svga auto-install iso template
Script to backup Win9x\ME drivers from a working install
Re: The thing no one asked for: KICAD 440bx reference schematic

Reply 5 of 9, by rasz_pl

User metadata
Rank l33t
Rank
l33t

P3 wont take more than 20-30W, rest of the board another 30W? GPU will take the most as its 3V is also sourced from 5V, but GPU will not be polluting 3V rail as much as DIMMs.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 6 of 9, by PARKE

User metadata
Rank Oldbie
Rank
Oldbie
neonblurb wrote on 2024-11-22, 16:51:

In terms of the slotket itself, I'm fairly confident the jumpers were set correctly for my CPU (everything on Auto basically) although the only instructions I could find were for the 2.3 revision, while I have a 2.0. Everything looked fine in BIOS: CPU was being detected fine, I was getting a reasonable thermal diode reading, I was able to set the frequency, and it was picking up and running at the correct voltage (1.75V). I did try overvolting slightly to 1.8V to see if it would improve stability at all, but saw no change. I also flipped pretty much every setting in the BIOS to the most conservative value but it made no difference, and I was having the same issues with an earlier BIOS revision. I suspect the particular version of the slotket I was using was just not stable with higher clocked/later Coppermine models running at 133 MHz, although I wasn't able to find anything to that effect online.

There is one topic here on Vogons that disscussed instability of revision 2.0 at fsb 133 although with different symptoms: MS-6905 slotket V2.3 works fine, V2 only at 100MHz
The release of the MSI Master, which was a new generation meant to support Coppermines, was a bit fuzzy to call it so. The last/final revision of the earlier generation (MSI 6905 rev 1.1 / pcb B) was called back because of functional problems with Coppermines and replaced with the Master revision 2. Then followed three revisions in rather quick succession of which revision 2.1 was built on a different pcb. It would seem that those three revisions would not have been necessary if revision 2 had been sound but that is just my impression.

The attachment stamp.jpg is no longer available

Reply 7 of 9, by neonblurb

User metadata
Rank Newbie
Rank
Newbie

Hi guys, thanks a lot for your replies on this and sorry I didn't come back for a while, got very busy with work but now finally have some time off over Christmas 😀

The last reply was really helpful, I think the fact that there were a number of different revisions, people were having issues with running some of the pre 2.3 ones and I was trying to use a later/faster Coppermine convinced me to look out for a replacement Slotket, and I ended up grabbing an Asus S370-DL - it was a bit expensive but worth it as this completely solved the problem, and my machine now runs perfectly with the above configuration.

Reply 8 of 9, by Grem Five

User metadata
Rank Member
Rank
Member

I have the same motherboard and I use one of these no problems with the same CPU model.

The attachment Shuttle.jpg is no longer available

Reply 9 of 9, by myne

User metadata
Rank Oldbie
Rank
Oldbie

Someone has to grind one of the better ones down for science.

But it looks like the main difference is a shitload more terminating resistors

I built:
Convert old ASUS ASC boardviews to KICAD PCB!
Re: A comprehensive guide to install and play MechWarrior 2 on new versions on Windows.
Dos+Windows 3.11+tcp+vbe_svga auto-install iso template
Script to backup Win9x\ME drivers from a working install
Re: The thing no one asked for: KICAD 440bx reference schematic