VOGONS


Reply 20 of 44, by yawetaG

User metadata
Rank Oldbie
Rank
Oldbie

If you have a multimeter, you could check for trace continuity between slots. Just measure the resistance (Ohms) and put one probe at one end of the trace and the other at the other end - if resistance is 0, there's no connection...

I guess there are also diagnostics cards that could be used to test this.

Reply 21 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie
yawetaG wrote:

A motherboard that has 16 bit ISA slots that are wired like 8 bit slots...

I don't follow. You're referencing a board that has 16 bit ISA slots with no traces to the 16 bit extension slot? Do you have some examples of this on a 486? Fake cache is one thing, but fake ISA slots? I'll admit that it is a rather unique looking motherboard.

appiah4 - What motherboard is this? What is the BIOS string when you boot?

You could configure the SB not use the high DMA channel and see if that makes a difference.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 22 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

The motherboard is a UMC OEM motherboard that appears to be made by Biostar. I posted a photo of the motherboard before. Here it is again, for this second page of the thread:

gallery_60983_11505_111867.jpg

The board has no markings and the closest board I can find to it is the MB457 V1.1 as found on amoretro.de. It is not 100% identical, however, my board lacks several components found on the MB457 such as capacitors and resistors. They don't appear to have been removed, they were just never soldered on.

Here is the boot screen and BIOS string:

med_gallery_60983_11505_114176.jpg

med_gallery_60983_11505_14698.jpg

med_gallery_60983_11505_95477.jpg

My attempts at trying to locate a bios update have failed.

As you can see the board's model number is MB-1433UTI. Those model leads me to several MB-1433 models produced by Biostar at one time or another but references to this particular board are nonexistant.

Here's a recount of my attempts today:

I have relocated the sound card to ISA 1, 3 and 4 and they all have the same issues. The IO card and VGA work in all 16-bit ISA slots.

I have tried disabling the LPT on the Multi IO card and it has had absolutely no effect.

I have found the issue to be that 16-bit digital audio crashes the system. Every game appears to run fine with just OPL audio. I can run DOOM with Sound Blaster music and no SFX and it will run just fine. SFX hard locks the system upon loading, usually right at the splash screen.

Running the CPU at 40MHz appears to make things worse, as at 33MHz I can run DIAGNOSE.EXE and 16-bit testing actually runs fine. If I run DIAGNOSE.EXE at 40MHz, 16-bit testing causes crashes and or screen blacking. This is leading me to believe something's not right with the ISA bus speed, as 40MHz makes it worse. There are no ISA BUS dividers on the board that I can identify and the BIOS does not present any options either.

When I run CTCU it reports no Legacy or PnP cards on my system that I can tell, there is nothing to choose, and the only IRQs selectable are 1 and 6.

I have tried to run the card with SET BLASTER=A220 I5 D1 H1 P330 T6 and SET BLASTER=A220 I5 D1 P330 T6, and the issue persists, although in this instance Doom appears to load further before locking up for whatever reason.

Kind of stuck here, and considering just getting rid of this board and doing a Pentium build instead 🙁

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 23 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie
appiah4 wrote:

The board has no markings and the closest board I can find to it is the MB457 V1.1 as found on amoretro.de. It is not 100% identical, however, my board lacks several components found on the MB457 such as capacitors and resistors. They don't appear to have been removed, they were just never soldered on.

That mb457 board is using an Award BIOS, and you have an American Megatrends. Amoretro even mentions a lot of fancy settings in their BIOS for the mb457.

The manufacturer ID string has been cleared out of your BIOS (set to 0000).

Part of me speculates that this motherboard is a clone of a clone, perhaps? A bit of a knock-off.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 24 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++
CkRtech wrote:

That mb457 board is using an Award BIOS, and you have an American Megatrends. Amoretro even mentions a lot of fancy settings in their BIOS for the mb457.

The manufacturer ID string has been cleared out of your BIOS (set to 0000).

Part of me speculates that this motherboard is a clone of a clone, perhaps? A bit of a knock-off.

That is my feeling as well but does being a cheap knock off explain not working with any motherboards?

Part of me wants to say Fuck It and just flash the mb457 bios on mine with uniflash and trash the motherboard if it messes up completely.. It currently is useless as a 486 without a sound card anyway..

If the bios chip is even flashable at all and if I can actually find that award bios. If I cant resolve this I will shelf thşs build dump the mobo and just do a Pentium DOS build..

EDIT: Regardless of whatever I do with the I/O card (enable/disable COM/LPT ports) the IRQs for these devices are ALWAYS assigned to the respective COM and LPT ports as reported by Norton System Info 8. That leads me to believe that this board's PnP is basically garbage, and when the system tries to access the card through IRQ5 or IRQ7 it just fucks up, for whatever reason. Running out of ideas really. I will try setting the card up as IRQ2 and/or DMA 3/5 tonight to see if that helps with anything.

Last edited by appiah4 on 2017-10-31, 06:37. Edited 1 time in total.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 26 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++
yawetaG wrote:

Dumb idea: not quite production-ready prototype board or sales sample meant for display only? Did the board have any traces of previous usage?

Yes, it obviously came from some terminal or oem machine, it was certainly used. I'm assuming at this point that it's not a 'complete' MB457, but actually a cheap, no name version meant to run simpler terminals - maybe ATMs or SOP computers or whatever. The ISA bus simply can not handle a sound card, any sound card. At least for digital sound. So sad. I will try IRQ2 and DM3 tonight, and if that fails, I will give up and switch to a Pentium build. I've done all I can, I feel.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 27 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie
appiah4 wrote:

When I run CTCU it reports no Legacy or PnP cards on my system that I can tell, there is nothing to choose, and the only IRQs selectable are 1 and 6.

I must have missed this earlier. IRQs 1 and 6? It doesn't see your SB16? There should be various configurations to choose from so you can set your resources as well as what you want enabled/disabled. Some configurations will have a high DMA, some won't, some won't have adlib/joystick/etc. This is how you basically setup CTCM.cfg so the card knows what resources to use.

Regardless of whatever I do with the I/O card (enable/disable COM/LPT ports) the IRQs for these devices are ALWAYS assigned to the respective COM and LPT ports as reported by Norton System Info 8.

I haven't used Norton System Info since the 90s. I realized I actually had it on my main retro rig (Pentium 200 MMX), so I ran it.

The IRQ screen is more or less worthless. It just lists typical items that use the associated IRQ number (i.e. IRQ 7 = LPT1) and then reports any additional information it can get in the right column. For instance, my mouse is on COM1/IRQ4, so CTMOUSE is reported. IRQ5 says LPT2 and BIOS - and my AWE64 uses IRQ5. You can't really use System Info for IRQ assistance.

CTCU does report conflicts to me for IRQ, DMA, etc. I run multiple sound cards in the same box, so that feature has been helpful whenever I need to shuffle resources.

That leads me to believe that this board's PnP is basically garbage, and when the system tries to access the card through IRQ5 or IRQ7 it just fucks up, for whatever reason. Running out of ideas really. I will try setting the card up as IRQ2 and/or DMA 3/5 tonight to see if that helps with anything.

It is technically possible that your board doesn't have PnP. That's fine - it just means you can't control resources (like reserve IRQs) via the BIOS. But this does bring me back to CTCU. If you're running DOS 6.22 or prior (rather than Windows 95+) and using a PnP card (or a partial PnP card like your SB16), you should be able (and need) to use CTCU to configure your card.

Kind of stuck here, and considering just getting rid of this board and doing a Pentium build instead

Aww. Don't bail on a 486 just because of a squirrely mobo. Go after one that gives you options - a socket and VLB should you want it at some point.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 28 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Thanks ckRtech I will post more screenshots and videos of the issue tonight along with videos of available bios optipna..

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 29 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

I think I may be onto something.

This SB PnP Manual on Creative's site has an interesting point in Troubleshooting (Page C-4): http://files2.europe.creative.com/manualdn/Ma … 781/english.pdf

Problem System hangs or reboots during the 16-bit digital sound test, but it works fine during the 8-bit test. […]
Show full quote

Problem System hangs or reboots during the 16-bit digital sound test, but it works fine during the 8-bit test.

Cause Your system's motherboard cannot handle 16-bit DMA transfer properly. On some machines, the DMA controller on the motherboard does not function properly during High DMA transfers. High DMA transfers on such machines might corrupt the data in main memory and cause the system to hang or encounter a parity error.

Solution Use the 8-bit DMA channel to transfer 16-bit data. Run the Plug and Play configuration utility and select a configuration that uses Low DMA channel only. 16-bit audio data will then be transferred through the Low DMA channel.

When you set your High DMA channel to Low DMA, you will lose the full-duplex operation, which requires two seperate DMA channels.

By the "Plug and Plan configuration utility" I believe it means DIAGNOSE? Or maybe CTCU? Because I already tried manually setting the BLASTER variable H to H1, rebooted, the card was configured by CTCM/CTCU and once I loaded DIAGNOSE it found High DMA as H1, but Doom crashed anyway.. If it's CTCU then I'm hosed as the card does not show up on CTCU.. I will post photos of the screen tonight. Alternatively, this could mean, if it's potentially a memory access issue, things could work fine with another EDO or FPM memory stick?

One final possibility: I've always tried these with Expanded Memory. Now, reading that same manual, on Page 18 there is this, when troubleshooting CTCU:

Question My system hangs or reboots whenever I load CTCM

Answer The memory area of your PnP BIOS machine is probably mapped by EMM386 using the HIGHSCAN option and thus, can get corrupted easily. When it does, CTCM will not work properly.

Granted, this is not the issue for me - but could it be something related? PnP info is somewhere in the RAM where it gets corrupted by something else? Only bringing this up because there are a ton of RAM shadowing options in the BIOS I have no idea about, and it seems to reserve 384K for SOMETHING as it's always absent from total memory..

Grasping at straws at this point..

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 30 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie

CTCU is the PnP config utility. Diagnose is for testing or "diagnosing" your sound settings. The same utility can be used on jumper-based, non-PnP cards.

If it's CTCU then I'm hosed as the card does not show up on CTCU

Are the menu options greyed out? There are other PnP config utilities out there if you want to try them...but CTCU not finding the card is a potential sign of bad times. I am interested in seeing any resource conflicts it reports.

Regarding EMS - You could add NOEMS to your EMM386 line. You could also boot and bypass config and autoexec completely and then manually run CTCM. I have my doubts about this making a difference.

I totally understand the problem-solving mentality - solve the problem and reap the benefits as a reward - but how much money do you have in this that you don't want to go after a more "normal" motherboard? The board is pretty much a dead end in terms of processors and VLB should you eventually want to change things up. The only advantage it would seem to have over a more standard board is form factor.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 31 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

At this point I just want to resolve the issue and move on, even if it is 'this board is junk, throw it away and wash your hands clean' but my OCD mind can't let go until I know what the hell is wrong.

It's not that I can't get a better (say a VLB) 486 but I just found this little motherboard interesting and fun to work with, if I can get it to function I'll be happy. Otherwise, I may invest in a real 486 mainboard for a DX4-100 I already own or just go for a P5 system. Those options are on the side. At the moment, I just need to know.

31UUbCcfzBL._SX300_.jpg

Here is some info dump on the issue.

My CONFIG.SYS:
gallery_60983_11505_21010.jpg

My AUTOEXEC.BAT:
gallery_60983_11505_297891.jpg

Video of SB16 DIAGNOSE crashing at 16-bit testing with High DMA5: https://youtu.be/uG1QbXp48BM

Video of SB16 DOOM running with just music but crashing with SFX using High DMA1: https://youtu.be/x_nQfr7BbNE

Video of CTCU weirdness followed by more Doom crashing: https://youtu.be/pKQJ7bQb10k

As a final resort, I tried to boot the PC with all CTCM/CTCU startup lines REM'ed. I loaded Doom, and it actually worked! For about 2 seconds, after which I got this error message:

gallery_60983_11505_32625.jpg

So I tried running Doom a few more times..

Video of Doom giving me all kinds of weird errors before hard resetting when CTCU/CTCM is not loaded: https://youtu.be/_EdZmsws2E4

This last phenomenon gives me the hope that it may actually be RAM related, I will try with other sticks of RAM tomorrow night.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 32 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Final update. No, this does not have a happy ending.

I tried to launch Doom tonight with a different stick of RAM, and got a strange EMM386 error:

EMM386 has detected error #12 in an application
at memory address 1901:1FAB. To minimize the chance
of data loss, EMM386 has halted your computer.
For more information, see the README.TXT file.

To restart your computer, press ENTER.

So I thought to myself, how about I run the PC without EMM386? Well, the game actually launched and ran for a few seconds, then crashed.

So I went into the BIOS and set the loosest memory timings I could. I also disabled the External Cache option, as there is no cache on this motherboard.

To my surprise Doom launched, and I actually played about 5 minutes of it with SFX and Music. I then quit. I was ecstatic, I had solved it! It had to be the stupid external cache option!

I launched Doom again and the PC rebooted.

I set the FSB down to 33MHz thinking it could have to do with ISA divider not working right, and no luck again.

At this point, I'm done. I've had enough. Unless anyone has any further suggestions, this motherboard will end up in the dumpster this weekend and I will restart this build with a P54C or P55C CPU.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 33 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie

I mean I suppose you could increase your stacks to assist with the EMM386 issue - but I think your problems are a lot greater than that.

Although I don't have the history of things you have attempted memorized, I feel that if you just handed this off (like you lived next door) to me and said "have fun" - I would try to reestablish baseline. I would reset BIOS defaults, walk the BIOS settings for possible oddities, get the I/O card down to only giving the system hard drive & only hook up the system drive (no CD-ROM), pull the sound card, bypass your start-up files, and just run some stuff like memtest and some benchmarks (Doom timedemo without sound). Let stuff loop, etc.

I would then put the card back in, still bypass startup, execute CTCM to bring up the sound card, and run the Doom timedemo.

Regarding the (possibly custom) BIOS due to this being designed for a single task, I imagine there could be things in there that are a bit quirky - certain resources reserved for custom software or hardware. I am really reaching out into imagination-land with this one, but I assume specialty stuff (like a terminal) could do something funky with a particular setting in a custom BIOS and then require certain software (say power management related. I dunno) to utilize said resources. It could also be expecting a specific expansion card that is missing, and also work under the assumption that the OS used would have the proper drivers loaded or know how to talk to "x hardware" just fine. These are just possible assumptions.

Btw - Sorry if I missed it elsewhere, but do you have a photo of the RAM you are using in the system? It isn't EDO, is it?

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 34 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Hey thanks for sticking with me and trying to help me out here. Aside from trying a seperate IO controller (which I don't have) I have pretty much tried everything you named, and I sue can't try the card without an IO controller as I need at least a floppy controller for that.

I will try a larger STACKS but I doubt it's the issue.

I think there is something very quirky about this board's memory access or the way ISA slots are wired up or the BIOS is configured so that the ISA Bus doesn't play ball with DMA, but memory access from the ISA bus seems to be the issue here and I can't work around it, not with the settings available in BIOS.

Here are the RAM I tried so far, in chronological order:

gallery_60983_11505_389236.jpg

These are the specs (AFAIK) from top to bottom:

TMS444000J-70 chips, 16 modules, double sided, 1Mx64 configuration, 8MB Non-parity EDO: Detected as only 4MB (Combined ram most likely incompatible with board)
VG2618165BJ-6 chips, 4 modules, double sided, 1Mx64 configuration, 8MB Non-parity EDO: Did not work
HM5117405S6 chips, 8 modules, single sided, 4Mx32 configuration, 16MB Non-parity EDO: Did not work
HM5117400S6 chips, 9 modules, single sided, 4Mx36 configuration, 16MB Parity FPM: Did not work

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 35 of 44, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie

I'm gonna say that for that motherboard (and a majority of 486s..but my majority take is an assumption) you can't use EDO RAM. So that pretty much knocks out those top three modules that you mentioned. It is generally a good idea to just stick with FPM for 486s because I believe boards that support EDO do so for convenience alone with no performance gain.

Your last option - the HM5117400S6 is probably the best option. I don't know about your system requirements of parity vs non-parity as well as module sizes supported. Your best bet would probably be to find a 486 motherboard manual online for a board that uses that same UM8498F/UM8496F chipset and see what it says.

Is your battery good? It doesn't always say "CMOS System options not set" every time you turn it on/boot does it?

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 36 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++
CkRtech wrote:

I'm gonna say that for that motherboard (and a majority of 486s..but my majority take is an assumption) you can't use EDO RAM. So that pretty much knocks out those top three modules that you mentioned. It is generally a good idea to just stick with FPM for 486s because I believe boards that support EDO do so for convenience alone with no performance gain.

Your last option - the HM5117400S6 is probably the best option. I don't know about your system requirements of parity vs non-parity as well as module sizes supported. Your best bet would probably be to find a 486 motherboard manual online for a board that uses that same UM8498F/UM8496F chipset and see what it says.

Is your battery good? It doesn't always say "CMOS System options not set" every time you turn it on/boot does it?

Battery is absolutely fine.

The parity FPM may also be causing issues, and unfortunately I have no FPM memory to test.

There are very few memory settings in the BIOS, and one of the very few is Fast Page Mode: Enabled (or something similar). I had this Enabled while using EDO RAM, which may have caused issues? I tried the FPM memory yesterday (possibly with this setting Disabled) but that didn't go far IIRC.

I will do further testing tonight with the parity FPM sticks plugged back in. I'll try them with the loosest timings (3 Wait State etc) as well as disabling all memory related bells and whistles.. Let's see if that works. In the meantime, let me hunt down some 72-pin FPM and some 30-pin SIMM memory.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 37 of 44, by keropi

User metadata
Rank l33t++
Rank
l33t++

does this motherboard have any cache on it? I can't see any is it embedded in the chipset or something?

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 38 of 44, by appiah4

User metadata
Rank l33t++
Rank
l33t++
keropi wrote:

does this motherboard have any cache on it? I can't see any is it embedded in the chipset or something?

No external cache just 8K internal cache on the CPU:

med_gallery_60983_11505_199949.jpg

No idea why it has an external cache enable option. Its a strange motherboard.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 39 of 44, by keropi

User metadata
Rank l33t++
Rank
l33t++

OK....
I know it's not an answer you are looking for but this motherboard does not really worth the effort to use it - even if it was not a clone/cheap one (it certainly is a budget mobo don't know if it really is a clone) the fact that it has no L2 cache at all is a red flag... is there something special about it?

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website