VESA LB Motherboars does not come to live

[gnallkoff]

Hi Folks
I try to bring a M-Board back to live from which I assume it is a VL-M-Board: HOT-419DZ (REV 3)

The Problem is that way, that the board don't pass the Power On Self Test. I did insert an omni POST Card and it tells me Step 04 Postcode passed stop at 05 : "Soft reset / power on determined. Going to enable ROM i.e. disable shadow Ram if any.

but after removing D-Ram Simm from bank 0 and bank 1 the postcode stops at step #13
that encourage me to add a VL-Video Card and an ISA HDD/Fd Controller. Bank #0 was fitted with one Simm module only

I did power up again and the Post Code stops now at step # 4F
Now I changed the Simm and the post code was stopped at step # D0
changing the Simm again the POST Code now always stops at step #13.

does some Body have a clue what ther could be happend?
is that possible the the Bios does not fit the Main Board ?

How can I determine the Size of a Simm Module from the inscription of the Chips ?

Best Regards Joe

 

[deathshadow]

I know that board quite well, sold several hundred of them back in '93 to '94ish.

Well, first off being a 486 class rig you cannot boot it without at least two 72 pin SIMM's due to how memory is mapped. (be thankful it's not 30 pinners or you'd need all four filled!). If at all possible make sure they match.

That revision of the board lacks the transistors for voltage selection, so I'm not sure it will run a 3.3v cpu (like that AMD) properly. See how you don't even HAVE a JP64? Compare to this model:
http://www.amoretro.de/wp-content/uploads/shuttle_hot-419_dz_r2_486_vlb_motherboard_vesa_local.jpg

That GREATLY limits what you can put in there for a CPU. You can't set the AMD 3.45v or the Cyrix 3.3, or the intel DX/4 3v settings. You may in fact be frying that poor AMD486DX-40. I also think you are jumpered for 33mhz, not 40, though it's hard to make out the silkscreening to know which way is which.

Pretty sure that board can't support that CPU, and even if it does the jumper settings are all wrong. Track down a Intel DX2/66 or real DX/50, get the jumper settings right, and you'll be golden.

 

[Old Thrashbarg]

Well, first off being a 486 class rig you cannot boot it without at least two 72 pin SIMM's due to how memory is mapped. (be thankful it's not 30 pinners or you'd need all four filled!). If at all possible make sure they match.

486 boards work just fine with single 72-pin SIMMs. 32-bit SIMM, 32-bit data bus. (Note that 30-pin SIMMs are 8-bit, hence the need for four of them.) Some boards did support interleaving, where you'd get a (very) slight boost from using a matched pair, but it wasn't really a common feature and the boards would still work just fine with a single stick of RAM.

That AMD CPU is a 5V chip, not 3.3v. And that board does have explicit support for the AMD chips, not that it really matters since the AM486DX is interchangeable with the Intel 486DX...

The seemingly random changes in POST codes, and the fact that it's now stalled at 13 (which, IIRC, is chipset initialization), makes me wonder if maybe one of the Opti QFP chips has some legs broken loose from the board. I've encountered that sort of ailment on several 486 boards, and it can cause all sorts of hard-to-track havoc.

 

[deathshadow]

486 boards work just fine with single 72-pin SIMMs. 32-bit SIMM, 32-bit data bus.

Goes over to 5x86-133 (clocked at 50x3) and pulls out 192 of it's 256 megs, does not POST. Swaps out remaining 64 EDO for a standard 8, does not POST... Entirely consistent with what I remember from 20 years ago. You SURE on that? Cause... uhm... no. I don't remember being able to go to singles until DIMM's came along. Despite being 32 bit DATA path, isn't there some funky addressing issue since there are only 24 address lines on the BUS unless you go EDO or have sticks and boards new enough (which being VLB it isn't) to meet the JEDEC standard? ... and of course don't forget the "extra address strobe" headache... actually, hah. Yeah, that's why they aren't working here. Yup, put the 8 in bank 1 it recognizes as 4, but won't even POST in bank 0. That would be the magic of RCAS lines 1 and 3 in action. Put two in, boom it's 16.

Wait.. my 486/33 will boot with a single 4 in it... doesn't like this 8... and of course that one doesn't even support EDO. I REALLY don't miss SIMM's a whole lot.

That AMD CPU is a 5V chip, not 3.3v.

Never even heard of an AMD 486 class that was 5 volt. Really odd duck that one if so.

And that board does have explicit support for the AMD chips

I would REALLY question that with the large amount of circuitry not even present on the board -- but then I never heard of an AMD that was 5v or "directly" DX compatible; though to be fair I've never actually seen an AMD DX/40 -- by the time that came along weren't DX-2's all the rage? If it was a 386, sure... but a 486/40 from AMD? That's NOT a DX2? (and yes, I've SEEN DX/2-40's and SX/2-40's from AMD... which were some real herpaderp)...

and the fact that it's now stalled at 13 (which, IIRC, is chipset initialization)

Chipset init AND the start of the memory test. Usually this stage means the chipset HAS started, if it hangs here there's a problem accessing the RAM... like there not being any, chips in the wrong slots, or the memory map not corresponding to what it's expecting from either the SENSE pins or the Presence Detect lines.

 

[Uniballer]

My experience was that, in general, you could POST a 486 (socket 2, 3) board from the VLB era with a single 72-pin SIMM. Not a Pentium bus CPU, though (socket 5, 7).

I personally have AMD 486DX2/80 chips in both 5V and 3.3V. I think they quickly switched to 3.3V for all of them. And you're right that early VLB motherboards didn't have the required voltage regulator stuff to handle the 3.3V chips.

The problem could be as simple as the single SIMM being in the wrong socket (some boards didn't care, some did), or the socket being dirty.

 

[Old Thrashbarg]

You SURE on that? Cause... uhm... no. I don't remember being able to go to singles until DIMM's came along.

Yes, I am absolutely sure of this. You don't have to take my word for it, but you'll get the same answer from any of the various computer hardware books, or the manual for any 486 motherboard with 72-pin SIMM slots. That was one of the big advantages and driving factors of the switch to 72-pin SIMMs-- you only needed one stick of RAM, rather than four. It made life a lot easier... almost enough to make up for what a PITA VLB was.

Incidentally, here is the spec sheet / configuration manual for the Shuttle board. I don't know if OP has already found this, but maybe it'll help.

Never even heard of an AMD 486 class that was 5 volt. Really odd duck that one if so.

If you call every AMD 486 up until 1995 an 'odd duck'. The early AMD chips were direct clones of the Intel 486 (i.e., functionally identical), and that includes all their SX and DX chips, as well as a fair number of DX2 chips. It was only after legal issues with Intel* that they split off and started using their own designs, known as the "Enhanced Am486". That's when they switched to using 3.3/3.45V. Here is a listing of the various AMD chips.

*fun fact: AMD's cloning of Intel chips was part of the reason Intel came up with the name 'Pentium'. Intel wanted a name they could trademark, which they couldn't do with '586'.

 

[Scali]

Yes, I am absolutely sure of this. You don't have to take my word for it, but you'll get the same answer from any of the various computer hardware books, or the manual for any 486 motherboard with 72-pin SIMM slots. That was one of the big advantages and driving factors of the switch to 72-pin SIMMs-- you only needed one stick of RAM, rather than four. It made life a lot easier... almost enough to make up for what a PITA VLB was.

I can attest to that. One of my 486es has 4 30-pin slots and 2 72-pin slots. I am running the machine with a single 72-pin SIMM of 16 MB.

The early AMD chips were direct clones of the Intel 486 (i.e., functionally identical)

Yes... or?
I had an early Am486DX2-66, which burnt out after a while, so I replaced it with an Intel 486DX2-66.
So I could do a nice apples-to-apples comparison, with everything the same but the CPU.
The CPUs seemed to perform EXACTLY the same in all software I ran on it. Benchmarks would yield the EXACT same numbers, so I think the AMD is a cycle-exact copy of the Intel design.
There is only one exception: I could never get OS/2 Warp to install on the AMD. I thought it was a BIOS compatibility issue. It would install, and run in safe mode, but I could never boot it properly. It would always give some weird exception.
Once I had replaced the CPU with the Intel, OS/2 Warp worked just fine. So it must have been in the AMD CPU.
I have no idea what OS/2 Warp could possibly do that made a difference... There's no CPUID on these early 486es, so I don't think there's a way to tell them apart with software.
It could be some bug in the AMD implementation of pmode, but then it'd be odd that it only surfaced in OS/2 Warp, and not in all the other 32-bit software I used, including 32-bit versions of Windows (I ran Windows 95 on it at least, perhaps also NT3.51).

 

[deathshadow]

I can attest to that. One of my 486es has 4 30-pin slots and 2 72-pin slots. I am running the machine with a single 72-pin SIMM of 16 MB.

... and yet my PCI 486 motherboard will not, at least not with a 5x86-133... though I just tested again and if I put one of the 128's in the wrong slot (1), it comes up as 64m just like how the 8 comes up as a 4... but won't recognize in the proper slot (0)

... BUT, my 486/33 will run with a single 4, or with the 8 in the wrong slot as 4, or a pair of 8's as 16's, but won't see a single 8?

Though, 2's, 8's, 32's and 128's all have the extra pair (1 + 3) of RCAS lines active, so maybe it's a board implementation issue? I know I've had perfectly good 8's refuse to recognize before without a pair. Don't have any 16's to test with (laughably in 72 pin I have a slew of 8's and a quartet of 128's.. and that's it)

Laugh is I build HUNDREDS of machines on that board and ones similar, and we NEVER shipped one with just a single stick, never even tried it because it didn't work back then. It was SOP to go with two sticks ALWAYS. Admittedly that was over two decades ago, but still...

 

[Caluser2000]

I can attest to that. One of my 486es has 4 30-pin slots and 2 72-pin slots. I am running the machine with a single 72-pin SIMM of 16 MB.

I've got a board sitting next me of similar configuration that I use for testing cards. It's an ISA/VLB board as well. The OPs has 4 72pin simm slots.

Later boards may be set up differently. Only way to confirm it I guess to look at the specific mobo memory requirement list or just try it like d_s has just done.

 

[Old Thrashbarg]

There is only one exception: I could never get OS/2 Warp to install on the AMD.

I guess I should have said that they were supposed to be functionally identical clones. It's not as though AMD could just run an Intel chip through a photocopier and print out their own... they had to reverse engineer it, so it wouldn't be at all surprising if some obscure bug made it into the AMD version, or it's equally plausible that some undocumented behavior in the Intel version (which IBM made use of, either deliberately or unintentionally) didn't get translated into the AMD implementation simply because AMD didn't know about it.

 

[deathshadow]

Incidentally, here is the spec sheet / configuration manual for the Shuttle board. I don't know if OP has already found this, but maybe it'll help.

I'd point out that said spec sheet references jumpers that don't even EXIST on the OP's board. Check his pic.

 

[Caluser2000]

RE OS/2 and AMD

It may have something specific to that particular mobo,cpu, video card combo as well. Who knows. The fact it ran in "safe mode" indicates running in plain old 16 colour vga was fine -ie forced to plain vga from the ALT-F1 selection menu.

I've got a Warp v3 box here but no AMD486DX2/66 chips to test it with unfortunately.

 

[Scali]

RE OS/2 and AMD

It may have something specific to that particular mobo,cpu, video card combo as well. Who knows. The fact it ran in "safe mode" indicates running in plain old 16 colour vga was fine -ie forced to plain vga from the ALT-F1 selection menu.

Well, as I say, I just swapped the AMD CPU with an Intel. So mobo and videocard were the same in both cases.

 

[Caluser2000]

The exception error code may've given some clue to what was wrong.

I'm not doughting you had an problem at all. I know from experience it can be very picky. It'd be nice to see if swapping to an AMD cpu, or any non Intel clone chip(changing voltage setting if need be) for that matter, on my box has a similar issue. You also mentioned the AMD burnt out eventually. Was that a common issue with those?

I've also got a hdd with NT3.51 server on it I can try out, also my old OS/2 v4 FP14 drive.

Here's pic of my current Warp v3 setup: