It's odd to see Cx code so soon in the boot process - I assume it's AWARD BIOS? I haven't done much with these lately so my memory of it might be hazy. Did you try to reset the mobo via reset switch after it has been powered? Could be another case of bad power good signal.

Deunan wrote on 2024-05-25, 17:03:

It's odd to see Cx code so soon in the boot process - I assume it's AWARD BIOS? I haven't done much with these lately so my memory of it might be hazy. Did you try to reset the mobo via reset switch after it has been powered? Could be another case of bad power good signal.

Yes, reset does not help. Forgot to mention, PSU is also known good component. 😀

Well, the only thing that comes to mind when I hear that siren is the CPU temp sensor tripping. I looked at some photos on the net and I see this mobo does indeed have such sensor. Is it working properly on yours? I would assume it's just a thermistor, so you can check the resistance and even if it changes if you hold the end of the sensor between your fingers.

Deunan wrote on 2024-05-26, 16:29:

Well, the only thing that comes to mind when I hear that siren is the CPU temp sensor tripping. I looked at some photos on the net and I see this mobo does indeed have such sensor. Is it working properly on yours? I would assume it's just a thermistor, so you can check the resistance and even if it changes if you hold the end of the sensor between your fingers.

Seems to be in working order - 6,6kOhm at room temp, went down to ~5kOhm when i held it between my fingers for some 10 seconds or so.

One looping long "beeeep" is common POST alarm code for RAM problems, but in your case there's a complete "concert" after. I've never heard such a strange sounds from PC-speaker on failed booting motherboard.

I wonder where the thermistor connects. W83977TF doesnt have any analog inputs, there is another chip under BIOS, U2 - can you read its markings?

IT WORKS!

I do not know, why it started to behave differently today, when i started to "play" with it once again.
Immediately, today, i heard that the ambulance alarm is gone. I only got single long beeps. So i continued playing because it was at least some progress after about half a year i've tried to make it work...

And what i found eventually, seems like, that it is pretty picky with memory. At one point i tried 2x64MB ram sticks, and it booted! Then i tried different slots and ram i had and found one 133MHz 256MB stick that works.

It seems to be really cool board! It's a 440BX board that supports 133MHz FSB! The fastest CPU i have to try is S370 Pentium III 866MHz. It works at correct speed. (using slotket)
But there seems to be some kind of instability issue with anything that isn't "era appropriate" Pentium 2 or Celeron. Celeron 300A (even at 450MHz) and Pentium II 450MHz are rock solid. All kinds of Pentium III CPUs so far have had some marginal stability issues. (I have SLOT1 677MHz Pentium 3 and the maximum S370 i have is 866MHz tried so far. I have some more.)

I'm running the rom.by patched variant of BIOS.

As i also have some old Win98 install currently on it, it may be some software issue, though. Have to reinstall OS at some point to confirm.

Thank you for thinking along! And if you have any experience with seemingly unstable operation with this board and Pentium 3 kinds of CPUs, I'm all ears to learn. 😀

Congrats. So, I was right by guessing memory issues. Do you have any explanation about the "concert" part of the sound error codes? Did you changed PSU or some other hardware components in your last successful test?

analog_programmer wrote on 2024-06-02, 09:00:

Congrats. So, I was right by guessing memory issues. Do you have any explanation about the "concert" part of the sound error codes? Did you changed PSU or some other hardware components in your last successful test?

I think i do, but it doesn't help.

On some random events, i got it into this "alarm state". Once was when i tried to set my Celeron 300A to run at 600MHz (133MHz FSB). Another time it was different AGP videocard i tried. Once i think it started to react like it when i tried different RAM sticks.
But you know what i discovered after a while? My ATX extension cable started to become warm! I used the cheapest aliexpress extension cable that i also modified and took molex connections from it, so i have a single 20/24pin connector to connect/disconnect and my whole bench is supplied from it. (The devices on molex connectors don't much matter as i have nothing physcally moving there - ssd and gotek are active)
I did notice that bios reported a little uncomfortably low 12V and 5V lines, but when i checked (from molex connector) with multimeter, everything was okay. When i now discovered warm wires, i checked voltages from the motherboard's atx connector and "lo and behold!", the voltage was very measurably lower! The bios did not lie!

So here's a PSA! The cheapest ATX extension cables from china are using below-standard wire widths! 18AWG should be standard so make sure your extension cable is at least that.

GigAHerZ wrote on 2024-06-03, 18:01:

The cheapest ATX extension cables from china are using below-standard wire widths! 18AWG should be standard so make sure your extension cable is at least that.

That reminds me that I bought a pack of breadboard connection wires and for some reason my contraption didn't work properly. Turns out the circuit pulled some 700mA of current and the wires, which I also used for power delivery, didn't have enough copper in them - so the voltage sagged quite a bit. Same with wires with alligator clips - those were so bad that I ended up throwing them away about 15 minutes after I unpacked them. I got me some much more expensive ones from local parts seller and these are great. And the funny thing is good chances are both were made in China.

Anyway, out of curiosity, did you inspect the wires? There should be markings - if so, does it say 18AWG on them? If you are going to trash this extension then cut the wire in half and make some photos.

Currently testing stability on this machine with Pentium III 866MHz @ 100x6.5 (650MHz) on slotket with looping 3DMark2001 together with Radeon 9250 with 128bit memory bus. So far so good. (I have a 512MB PC133 CL2 RAM stick installed. It's dual-ranked, so the board only addresses 256MB, but that's fine. I have currently no other 133MHz CL2 sticks of RAM right now...)

But no matter how loose i set all kinds of memory timings and other such stuff, i can not make it work at 133MHz FSB stably. It might be caused by the out-of-spec AGP, though. At one point i have to test this machine with PCI graphics card to see, if the board itself is stable at 133MHz FSB.

On another note, there seems to be some kind of issue with Internet Explorer 5.5. When i use it for some 5-15min, then at one point the machine just hangs completely. (I'm using protoweb and oldnet services to get any content on this old machine) It hasn't hanged on anything else (yet!), but with IE, it seems to be a pattern...

Deunan wrote on 2024-06-03, 18:59:

GigAHerZ wrote on 2024-06-03, 18:01:

The cheapest ATX extension cables from china are using below-standard wire widths! 18AWG should be standard so make sure your extension cable is at least that.

Anyway, out of curiosity, did you inspect the wires? There should be markings - if so, does it say 18AWG on them? If you are going to trash this extension then cut the wire in half and make some photos.

The wires themselves claim to be 18AWG. Full line: "1007 VW-1 80°C 300V 18AWG YOUCHUANG - LF". But i have some power cables with markings of 20AWG that feel stronger and thicker inside than this extension cable's wires.

Fake wire is a problem with cheap cables in general. Most of it isn't even copper or CCA. It's some weird alloy that has strangely high resistance (and inconsistent - each conductor in the cable has different resistance while being the same length), is weakly attracted to a magnet, and is the color of old oxidized copper.

So, i've been playing around with this board and seems like there is some kind of issue with it.
I've tried numerous CPUs, RAMs and graphics cards. (I most like Radeon 9250 with 128bit ram - i have 3 of them.)
And there seems to be some kind of issue with any pentium 3 CPU i try with it.

Pentium 2 (3.5x100) and Celeron 300A (4.5x66 & 4.5x100 OC) work perfectly.
One Pentium III i have (5x133) locks up even at 5x100 and other Pentium III on slotket (6.5x133) does the same also even at 6.5x100.
Everything works well until i do something graphically demanding. I can cause the lockup easily with 3DMark2001. Sometimes it takes a bit of time until it locks up, but then it seems to degrade for that day and usually i even can't get passed the first car demo on second+ run.

I'm using Windows 98SE and i've re-installed it countless times. I've also used both SSD with SATA-IDE adapter as well as CF card for storage to have at least 2 different storage mediums tested - no behavior change.
With windows, I've tried all kinds of different orders of installations after windows install. On top of windows, i install 440bx chipset drivers, graphics drivers, soundcard (ymf724) and NIC drivers and all main windows updates. For windows updates i've used windows 98 SP3 (only the "main updates" parts) as well as restored windows updates through browser. The behavior does not change.

I have an oscilloscope, i believe pretty decent one. But I'm very new in this. I tried to measure CPU voltage and 3.3V from graphics card (from an unpopulated capacitor pad) and everywhere i read anything, i see noise/ripple with peak-to-peak of 250-400mV. It seems crazy, especially when considering that my Pentium 3 wants 1.65V. The voltage itself with multimeter seems to be on correct level.
Because oscilloscope is pretty new to me, please instruct me, what to measure and what settings should i have on my oscilloscope to do the measurement. Like AC/DC coupling, the 10x/1x on probe, etc. I don't believe the ripple is that large - the CPU wouldn't be able to work AT ALL with such big noise on its power supply.

So any suggestions and instructions are welcome! It is a beautiful board, i would like to get it fully working.

NB! Not using any ATX extensions for now. All voltages are within normal range measured from the motherboard. I use the latest bios from retroweb, that is patched with rom.by for better support of CPUs and whatnot. For oscilloscope, i used serial port's shield as ground, as it was closest bare and shiny metal. (it should be good ground, right?)

EDIT: Just to confirm.

Celeron 300A @ 450MHz.
128GB SSD over IDE-to-SATA adapter
Fresh windows 98 install.
All drivers + Main updates from Service Pack 3.
DX 8.1 + 3DMark2001

And is rock-solid. I made it to loop 3DMark2001 low-detail cars demo (as that should stress more the CPU, motherboard and RAM instead of being bottlenecked by GPU) and it just continues to loop with no issues.
May there be some kind of incompatibility with P3 type CPUs somewhere/somehow? Not even sure is it incompatibility with the board or with windows or something else... Maybe some of you have encountered similar issues and can suggest some ideas?

GigAHerZ wrote on 2024-06-16, 15:21:

EDIT: Just to confirm. […]
Show full quote

EDIT: Just to confirm.

Celeron 300A @ 450MHz.
128GB SSD over IDE-to-SATA adapter
Fresh windows 98 install.
All drivers + Main updates from Service Pack 3.
DX 8.1 + 3DMark2001
And is rock-solid. I made it to loop 3DMark2001 low-detail cars demo (as that should stress more the CPU, motherboard and RAM instead of being bottlenecked by GPU) and it just continues to loop with no issues.
May there be some kind of incompatibility with P3 type CPUs somewhere/somehow? Not even sure is it incompatibility with the board or with windows or something else... Maybe some of you have encountered similar issues and can suggest some ideas?

I m not familiar with that particular Chaintech board but according to the info on the web it is an early BX and originally designed for Pentium II (66Mhz). Most if not all the parts that you tried and that failed are in principle out of specification. Anything of a higher order that you can make work is therefore a bonus and in my opinion there is not necessarily anything ' wrong' with that board because it is flakey when run out of spec. I have here a later BX Chaintech board (s370 - 6BJM) with a VRM that was designed for Coppermine and it runs everything that is thrown at it without problems.

PARKE wrote on 2024-06-17, 14:44:

GigAHerZ wrote on 2024-06-16, 15:21:

EDIT: Just to confirm. […]
Show full quote

EDIT: Just to confirm.

Celeron 300A @ 450MHz.
128GB SSD over IDE-to-SATA adapter
Fresh windows 98 install.
All drivers + Main updates from Service Pack 3.
DX 8.1 + 3DMark2001
And is rock-solid. I made it to loop 3DMark2001 low-detail cars demo (as that should stress more the CPU, motherboard and RAM instead of being bottlenecked by GPU) and it just continues to loop with no issues.
May there be some kind of incompatibility with P3 type CPUs somewhere/somehow? Not even sure is it incompatibility with the board or with windows or something else... Maybe some of you have encountered similar issues and can suggest some ideas?

I m not familiar with that particular Chaintech board but according to the info on the web it is an early BX and originally designed for Pentium II (66Mhz). Most if not all the parts that you tried and that failed are in principle out of specification. Anything of a higher order that you can make work is therefore a bonus and in my opinion there is not necessarily anything ' wrong' with that board because it is flakey when run out of spec. I have here a later BX Chaintech board (s370 - 6BJM) with a VRM that was designed for Coppermine and it runs everything that is thrown at it without problems.

It has 66/100/Auto (MHz) jumper physically on board so i would expect it to be stable at 100MHz FSB. I understand that at 133MHz, the AGP is also running out of spec. But why does it have issues with architecture itself? Can it be somehow fixed/improved?

The earlier boards had VRM model 8.1 or 8.2 that could handle currents of 14.2 amps or 16.2 amps respectively. In 1999 Intel published guidelines for VRM model 8.4 that was designed to accomodate Coppermines and could handle 22.8 amps. It seems not unlikely that running Coppermines on earlier boards with VRM 8.1/8.2 may result in instabilities like you describe. Running cpu's at 100 Mhz should be ok as long as it are for example Mendocino Celerons or Deschutes P2's.

It is behaving really weirdly. It seems as if it degrades over running time.
I just tested some more slot 1 CPUs i have and have never tried anywhere yet. One of them was Coppermine 7.5x100. At rated speed, it did full games benchmark in 3DMark2001 and i got the score, but then when i continued with looping low detail car demo, eventually it hanged. And after that, it hanged without being able to go through the demo even once.
Then i tried 4.5x100 Katmai and it hangs too.
From earlier, Coppermine 667 hangs sometimes even at 5x100MHz.

It's almost as if there's some kind of thermal expansion issue or something that let's the P3 CPUs work for ~10min and then it starts to hang on 3D workloads... I do have decent airflow - open bench with extra 120mm fan right on top of the board creating additional airflow both to the motherboard and its chipset and a bit to the graphics card.

Is there some decent ways to maybe replace few components to slightly "uplift" the VRM easily? I do have decent soldering station and my thumbs are not exactly coming out from my palm, so i might be able to do some works like that. 😀

UPDATE:
I searched out my "highest multiplier 66MHz S370 Celeron" and found it. It is a Coppermine-128 11.5x66MHz.
The board didn't want to even boot with that.
Then i searched out the next best thing - Coppermine-128 10x66MHz.
It booted but hanged right away. But the screen came on and it thought it is a 1GHz Celeron. 😁 I had forced the FSB to be 66MHz from jumper so it was truly miss-detected.

I became curious and started to suspect the BIOS i have on it. It is the latest from retroweb with supposed rom.by patches. In that bios, there are some signs of patching, so there is something.
But then i decided to take the latest non-patched bios are patch them myself. Both with 4.51 and 6-alpha9, the patching worked and they look nothing like the one from retroweb. Not only that, the one from retroweb seems to have a bunch of additional CPU microcodes that are not available on rom.by website. Not even as a separate microcode file they also provide for you to manually inject.
In addition, this bios from retroweb also reports "unknown flash rom" or something similar...
Really weird.

So i started with the bios patched with 4.51 patcher. It's currently in my machine and the 766MHz Celeron (Coppermine-128) is happily running the 3DMark2001 in an endless loop.
So i can at least confirm that some sort of coppermine on some sort of slotket works. (All the coppermines from earlier descriptions were native SLOT1 CPUs, unless mentioned otherwise)
I did retry the 7.5x100MHz Coppermine and it still did hang, unfortunately. But maybe that truly is because of VRM.
Gonna try the 5x100MHz Coppermine and 4.5x100MHz Katmai soon as they should consume considerably less current. - EDIT: They hanged. I found 5.5x100MHz Katmai variant, too, an it hanged as well.
EDIT: S370 Mendocino 433MHz on slotket was stable. Also some CPUs, that refused to boot at all, now also booted, like one Celeron 333 that i have.

Also the bios patched with patcher 6-alpha9 is still on my hdd waiting for testing. We'll see...

Nice to see some progress. Maybe it is a combination of curent draw and bios. When you get the higher end cpu's to run you may consider some extra cooling on the VRM. Coppermine Celerons draw the same current as P3 as far as I know. Out of curiosity: that 766Mhz Celeron that you have is that with sSpec SL52X or SL5EA ?

PARKE wrote on 2024-06-19, 13:40:

that 766Mhz Celeron that you have is that with sSpec SL52X or SL5EA ?

Mine is SL4P6.