VOGONS


Can't get into CMOS setup on old 486.

Topic actions

Reply 20 of 68, by BitWrangler

User metadata
Rank Oldbie
Rank
Oldbie

Tried the old magic thumb treatment yet? put it on firm surface like antistatic foam on the bench, mash the BIOS chip into the socket with your thumb, mash the KBC chip in, mash anything else in sockets, put thumb in middle of each SIMM, mash and wiggle. Put thumb on CPU, mash and work the lever back and forth a few times, and lock it down while pushing, exercise the VGA card up and down in the slot a bit, clear CMOS again for good luck, stare at it threateningly for two minutes, then power it on again just DARING it to have a problem now.

2017: Basement full of ancient PC stuff, starting to go through it. 2021: Still starting, heh, many setbacks. So what's this BitWrangler guy's deal ??? >>> Taming the pile, specs to target?

Reply 21 of 68, by Keatah

User metadata
Rank Member
Rank
Member

I believe the CPU is a DX2/50. I was told the board was a pull from a Gateway system. It’s likely an early version as GW didn’t sell a lot of the DX2/50 models. I mean they did, but didn't, because the time on market was short. The 66 and VLB rapidly took over and became the standard IIRC.

The CPU is in a non-ZIF socket so I didn’t pull it yet. The board has a DX2 label in the corner. And there is 50MHZ oscillator present. There is an adhesive heatsink which I don’t want to remove. So no reading the S-spec. But it’s likely made in 1992 judging by the datecodes on all other motherboard chips. And if it’s that early it likely has a Write Through cache.

I bet the CPU was either installed by the motherboard maker, Micronics, or Gateway 2000. Well it had to be one or the other. I mean it has a low-profile heatsink that has that pedestrian "OEM look" about it. Anything user installed or aftermarket would look more substantial and bold.

I might be able to find another CPU in my parts bin. It will have to be the same rating because there are no settings to try something else different. Unless it’s like a PNY PowerStacker or Kingston TurboChip.

Last edited by Keatah on 2021-06-09, 08:45. Edited 1 time in total.

Reply 22 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Yes. That's exactly what I believe should happen here. The board currently is simply omitting the last line of text. The one reading "Strike the F1 key to continue, F2 to run the setup utility." Reboots instead. And from what I gather the only other way is to use <ctrl> <alt> <esc> from MS-DOS or other OS.

At first I tried without any cards except a basic videocard. Then I tried the Multi-IO without any floppy connected. Then I figured it's best to actually plug in a floppy drive. Same response all three ways.

I have a TL866II+ and dumped the BIOS chip. PCEM shows the same screen. My BIOS version is a revision earlier at G21 instead of G22. But I don't think that matters here.

Reply 23 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Yikes I don't want to scare the board into running away. I can certainly try reseating all the socketed chips and SIMMs. Maybe even use a spit of Deoxit D5 here and there. I will try that later today and see what happens. Thanks for all the ongoing suggestions!

Reply 24 of 68, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

Wait, this is a 486 mobo and it has 2 BIOS ROMs? First time I see one like that.
Anyway, I wanted to suggest to de-populate cache sockets, don't bother rejumpering, just test with no chips. Could be the BIOS is trying to verify cache size, or test it, but it fails and results in reboot. BIOS should complain about cache not being usable but continue with boot sequence and let you enter settings to disable it.

Same can happen with bad RAM so try a different set of sticks, and try putting them in other bank as well. If nothing helps it might be a mobo issue (broken connection somewhere, bad chipset, you name it). A POST card might help but then again a rapid reset can easily wipe the last code from display and you won't really know where it gets to before reboot.

Reply 25 of 68, by BitWrangler

User metadata
Rank Oldbie
Rank
Oldbie

Yah, machines that have been kicking around a bit might have has the tasty 15ns tag "robbed out" to borrow an expression from archaeologists and the 25ns store left there, leaving the motherboard a bit confused.

2017: Basement full of ancient PC stuff, starting to go through it. 2021: Still starting, heh, many setbacks. So what's this BitWrangler guy's deal ??? >>> Taming the pile, specs to target?

Reply 26 of 68, by Am386DX-40

User metadata
Rank Newbie
Rank
Newbie
Keatah wrote on 2021-06-09, 08:21:

Yes. That's exactly what I believe should happen here. The board currently is simply omitting the last line of text. The one reading "Strike the F1 key to continue, F2 to run the setup utility." Reboots instead. And from what I gather the only other way is to use <ctrl> <alt> <esc> from MS-DOS or other OS.

At first I tried without any cards except a basic videocard. Then I tried the Multi-IO without any floppy connected. Then I figured it's best to actually plug in a floppy drive. Same response all three ways.

I have a TL866II+ and dumped the BIOS chip. PCEM shows the same screen. My BIOS version is a revision earlier at G21 instead of G22. But I don't think that matters here.

Do you have a spare chip to burn the newer bios and try that with your motherboard? That may make it work.

Reply 27 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Reseating socketed chips didn't affect anything.

Yes this board has 2 EPROMS for the BIOS. Each marked HI or LO. I don't have any 27C256 EPROMS on hand (packed away) but the originals read & verify fine in my TL866II+ 'prommer. Can't really make new BIOS chips till like later next week.

I've tried enough RAM combinations to feel confident it isn't the RAM SIMMs themselves. And some of them have been tested a long time ago in a SIMM tester.

I can get the board to display the following..

"Memory parity interrupt at F000:80AC."
"Type (S)hut off NMI, (R)eboot, other keys to continue_"

..if I purposely mis-configure the RAM via the DIP switches to say it has a different layout or amount. Something like saying it has 4 4MB x 9 in bank0 vs 16 1MB x 9 spread across bank0 bank1 bank2 bank3. Banks 0 & 1 are onboard. Banks 2 & 3 are on a proprietary memory expansion card.

After the board displays the above quoted text it stops hard. Won't respond to the keyboard. It also does not make it far enough to even test & count the memory. Intuition tells me the DRAM subsystem is working ok because I can make it fail, or not.

Reply 28 of 68, by BitWrangler

User metadata
Rank Oldbie
Rank
Oldbie

Have you tested with SIMMS only or has the proprietary expansion been in all the time?

2017: Basement full of ancient PC stuff, starting to go through it. 2021: Still starting, heh, many setbacks. So what's this BitWrangler guy's deal ??? >>> Taming the pile, specs to target?

Reply 29 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Like a good little technician I'm testing with the onboard SIMMs only. One problem at a time. One problem.

Bank0 and Bank1 are on the motherboard. Bank2 and Bank3 are on the optional proprietary memory card. Each bank has 4 SIMM slots. I never got far enough to play with the expansion card. Trying to keep everything as basic & simple as possible. I just know it's gotta be something obvious I'm missing. What with the thousands of configurations, spotty technical documentation available to consumers, and pre-dotcom-era websites that are long gone, something.. Something is bound to be incorrect or have gotten lost over the years.

Anyhow. The proprietary expansion card (Micronics M810) doesn't seem that proprietary because it's just some 74LSxxx buffers or transceivers connecting a seemingly "bigger-than-ISA" slot to 8 additional off-motherboard SIMM sockets. No PALs or GALs. Easily copied today I suppose.

I have tested the following arrangements:
16MB RAM = Bank0 with 4-SIMMs [4MB x 9] @ 60ns.
4MB RAM = Bank0 with 4-SIMMs [1MB x 3] @ 70ns.
8MB RAM = Bank0 with 4-SIMMs [1MB x 3] @ 70ns + Bank1 with 4-SIMMs [1MB x 3] @ 70ns.
..and got the exact same results.

This further increases confidence in the DRAM memory subsystem.

Last edited by Keatah on 2021-06-09, 23:31. Edited 1 time in total.

Reply 30 of 68, by BitWrangler

User metadata
Rank Oldbie
Rank
Oldbie

Oh you got motherboard banks too, do they disable? Just 'coz it's meant to be a thing that if a board isn't managing to run long enough to give you sensible errors it's the first 64kilobytes of RAM that should be suspect.

Edit: derp, nevermind the SIMM banks are on the mobo. Thought you meant soldered down discrete banks.

2017: Basement full of ancient PC stuff, starting to go through it. 2021: Still starting, heh, many setbacks. So what's this BitWrangler guy's deal ??? >>> Taming the pile, specs to target?

Reply 31 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Now that's interesting. Something is beginning coalesce in my tiny head.

But.. No.. I don't believe they can be disabled. If they can I don't know how, and the organization chart starts with bank0, then moves on to include 1, 2, 3 as memory size increases. There's no documented configuration that uses the offboard expansion card banks 2 and 3 only. No configuration that doesn't use bank0.

The M810 is almost a full-length card, with half of it being devoid of traces or parts. Big enough to extend to the back where it can be screwed to the chassis. Like an ISA card. Like an ISA card with its chirality messed up! This whole board is messed up!

Reply 33 of 68, by Keatah

User metadata
Rank Member
Rank
Member

Unfortunately I do not. I have it packed away, and it would be a while before I go tearing through stuff to get at it. I hadn't expected to be working on an ISA computer so soon. And well here we are!

This mobo has 20ns cache chips on it. The numbers end in 20. And that would be a match for the 50MHz clock crystal sitting nearby.

On occasion I'll note a stripped board. Cache swapped with slower ratings or missing entirely. It seems so cheap and petty to do that in this day and age as SRAM is cheap. But. Then. Bottom barrel scrapers.

I did find some 27C256 EPROMS. But I would need the G2-22 BIOS files if I was to try the later version. The original G2-21 copied into PCEM works like I'd expect. Burning it back to different EPROMS didn't make any difference.

Reply 35 of 68, by Keatah

User metadata
Rank Member
Rank
Member

There's no lengthy BIOS ID string like in a modern PC.
After power up, the video BIOS banner is shown.
Then the regular BIOS banner.
And after it counts the memory it will reset and start over.

Attachments

Reply 36 of 68, by jakethompson1

User metadata
Rank Oldbie
Rank
Oldbie

Unfortunately I think you're going to need the POST card. If there's any way to figure out the last code before it resets at least you could dig through the BIOS ROM and try to figure out what it's doing...

Another shot in the dark. Do you have a PC speaker connected? I notice one of the things it tries to do is beep before displaying the F1/F2 message. Presumably it tries to beep for the parity error too.

Reply 37 of 68, by Keatah

User metadata
Rank Member
Rank
Member

In a couple of weeks I can get to my POST card.

I hooked up a speaker and there were no beeps of any kind. Total silence. Including when I made it make the mis-configured (via incorrect DIP switch setting) RAM parity error.

I merged the HI + LO dumps with romwak and looked through the resulting single file. Other than seeing a lot of text related to things that would be in the setup menu I only saw the usual ASCII-HEX gibberish. Interpreting that and disassembling it is likely beyond my current understanding.

Reply 38 of 68, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

BIOS has (or should have) internal checksum. It will usually stop and beep if that is incorrect. I was merely surprised that there are 2 chips, that is usually only something you see on 286 and some very early 386SX mobos that don't really support ROM shadowing in RAM. Therefore you need two 8-bit chips for the 16-bit bus, or else the code fetching would be at half bus speed (plus any waitstates the ROM might need).

Then again your BIOS version 0.10 doesn't exactly inspire confidence, could be the chipset is that braindead and can't do shadowing, or someone figured it's better not to enable it.

Reply 39 of 68, by Keatah

User metadata
Rank Member
Rank
Member

This board and BIOS should do shadowing. This is also a board that has many design cues taken from the 386 era. It's not refined with a later chipset, or VLB, or PCI, or support for multiple CPU types & voltages. It supports the out of style Weitek 4167. The BIOS options are not extensive. Basically 2 screens. Once screen for the disks and time. The other for cache and shadowing. And it uses the older style 30-pin SIMMs. I believe there is only 1 or 2 upgrade paths for it. If that.

The 486 is an evolution of the 386. And is much much closer to it than it is to the Pentium. And the board I'm trying to get rolling is much the same way. The Pentium 60 & 66 represented a whole new architecture.