Can you take a good picture of your board ?
Hmm yes many old ISA cards only work well at the designed 8Mhz (max of about 8.5Mhz), many newer ISA cards can work up to about 10Mhz or more.
Since you overclocked cpu by 125%, and if the ISA bus clock is derived off the cpu crystal thru dividers/multipliers and not a separate crystal/clock then you also multiplied the ISA bus clock by 125%.
At turbo/full speed would assume 8Mhz ISA clock now at 10Mhz which does not work and at the 6Mhz setting would be 7.5Mhz ....which should work.

Horun wrote on 2022-11-27, 21:14:

Can you take a good picture of your board ? Hmm yes many old ISA cards only work well at the designed 8Mhz (max of about 8.5Mhz […]
Show full quote

Can you take a good picture of your board ?
Hmm yes many old ISA cards only work well at the designed 8Mhz (max of about 8.5Mhz), many newer ISA cards can work up to about 10Mhz or more.
Since you overclocked cpu by 125%, and if the ISA bus clock is derived off the cpu crystal thru dividers/multipliers and not a separate crystal/clock then you also multiplied the ISA bus clock by 125%.
At turbo/full speed would assume 8Mhz ISA clock now at 10Mhz which does not work and at the 6Mhz setting would be 7.5Mhz ....which should work.

it is the same board from the faulty floppy controller post. (I am using an addin floppy controller temporarily while I test stuff, hopefully I can get it fixed by the time I put it all back together, and I checked: the floppy controller is fautly at either 12.5 or 6mhz.)

The board's ISA clock is the same as the CPU's (25/2), no dividers. It is a pretty old-style clone board I think. The MFM controller was fine at 10mhz.

Do you know of any MFM controllers that do work on a 12mhz bus?

here is a link to the previous post with some pictures of the board in it: Troubleshooting faulty integrated floppy controller on AT clone motherboard (if you need a closeup of something else let me know)

Ok, do you know all the jumper settings for your board ? Saw the pictures but hard to tell which crystal you replaced. One rectangular near floppy stuff is Y4 24Mhz in the picture. The others y1, y2, y3 look stock but hard to tell...
So which did you replace? Also curious why you want to overclock the board if the onboard floppy does not work (is it related to the overclock ??).
The pictures you posted are still not good enough for me (or probably anyone) to see the board like we need to see it
(take a look at these: 80286 BIOS image collection ) those are good pics of 286 boards that we can use to figure out where an issue could be...
I suggest you go back to 10Mhz cpu speed and get everything running first... then we can talk about overclocking. just my opinion....

Horun wrote on 2022-11-27, 21:14:

Hmm yes many old ISA cards only work well at the designed 8Mhz (max of about 8.5Mhz), many newer ISA cards can work up to about 10Mhz or more.

Hi there! Sorry for interrupting you, I just want to say that there's sort of a paradox or exception here:
ISA was defined as a 8,33 MHz bus, that's completely right. The "Gang of Nine" did this is in 1988 or so, when EISA saw light of day.
However.. Before that, when ISA still was "AT-Bus", high-speed 16-Bit systems already existed.
Some ran on 10 MHz and beyond. Makers of old memory boards did advertise such high bus speeds in their manuals and ads, even.
So it might be, that an AT-Bus card from the mid-80s (-before the bus speed was capped-) can handle 10 or 12MHz, but not a similar ISA card from the late 80s or 90s.

Jo22 wrote on 2022-11-28, 07:09:

ISA was defined as a 8,33 MHz bus, that's completely right. The "Gang of Nine" did this is in 1988 or so, when EISA saw light of day.

not really, IBM did that when they released the 8MHz AT, the 1988 spec was simply a formalization of an already well established convention

maxtherabbit wrote on 2022-11-28, 15:25:

Jo22 wrote on 2022-11-28, 07:09:

ISA was defined as a 8,33 MHz bus, that's completely right. The "Gang of Nine" did this is in 1988 or so, when EISA saw light of day.

not really, IBM did that when they released the 8MHz AT, the 1988 spec was simply a formalization of an already well established convention

But if so, was IBM still relevant or an authority at the time? It's PS/2 line was widely ignored, for example.
The Gang of Nine was a collective or consortium of independent companies in the clone/third-party business at the time.
A bit similar to VESA, which specified the VBE graphics modes, maybe.
Their recommendation/specifications might have been playing a greater role.

Edit: Or maybe, the 8MHz AT was simply the latest model, hence the IBM specification was mentioning 8 MHz? 🤷‍♂️
If so, clone makers couldn't know that 8 MHz marked the end of the line.

Edit: Sorry for the many corrections. I'm typing on a mobile device. 🙁

Edit: Personally, I'm afraid that VGA was IBM's last masterpiece that was accepted by the PC industry.
After Compaq released the first AT/386, IBM was out of business. Hardware wise, I mean.
PC DOS sold well and OS/2 Warp was sort of a success, before Windows 95 blew it out of the water. 🙁

Edit: That being said, it's good to always question history.
Often, it turns out that the media or mass hasn't a true/complete perception of it.
For example, let's take online services. Old online services or BBSes from the 1980s are mostly forgotten.
The media makes it look as if the internet started it all in the 90s.

Edit: Never mind. I was just thinking out loud. 😅

Horun wrote on 2022-11-28, 03:31:

Ok, do you know all the jumper settings for your board ? Saw the pictures but hard to tell which crystal you replaced. One recta […]
Show full quote

Ok, do you know all the jumper settings for your board ? Saw the pictures but hard to tell which crystal you replaced. One rectangular near floppy stuff is Y4 24Mhz in the picture. The others y1, y2, y3 look stock but hard to tell...
So which did you replace? Also curious why you want to overclock the board if the onboard floppy does not work (is it related to the overclock ??).
The pictures you posted are still not good enough for me (or probably anyone) to see the board like we need to see it
(take a look at these: 80286 BIOS image collection ) those are good pics of 286 boards that we can use to figure out where an issue could be...
I suggest you go back to 10Mhz cpu speed and get everything running first... then we can talk about overclocking. just my opinion....

Yes, I have checked all of the jumpers and none relate to bus divider (I attached a picture of the manual page with jumper info). I don't think the FDC is affected by the overclock since it has the same behaviour at 6mhz setting as it does at the default 10mhz and overclocked 12.5mhz modes. I mostly want to overclock the board to get a bit more performance in stuff like flightsim and maybe just a bit of "because I can"

Here are some better images of the board, although since they were over 5MB I had to upload them with imgbox. Sorry for not getting better images sooner, my phone's camera is really bad so I had to wait and borrow a decent camera.
https://imgbox.com/7DMWHrDl
https://imgbox.com/UmZ5sELN
https://imgbox.com/Fae8wBiu
https://imgbox.com/6FStTU1V
https://imgbox.com/5whNn3EM (mfm controller board)
https://imgbox.com/aoZljmuN (manual page)

Also, to note: the images may look like the crystal is shorting the nearby capacitor legs, but it isn't; there is a piece of tape between them and they are separated by a small air gap.

Thanks for the better pictures !
Going back a bit OT: Agree with Max. Sure some adapters could run on a faster ISA bus than standard like some VLB Video cards could run faster than the "standard" 33Mhz though usually you have jumper some wait-states in (for VLB both on the board and the adapter). If you think about the late XT era and later 286 era there is a reason the majority latter 286 boards had 12 (or 16Mhz cpu's ) all easy to divide the cpu clock to get 8Mhz for ISA. The 10Mhz variants were sort of bastard board requiring a complex divider+multiplier circuitry to get a 8Mhz ISA bus IF they did it at all. Same goes forward to 386 and 486 boards and why a 33Mhz clock became so important (easy to divide out to 8Mhz) until they figured out to Async the cpu and ISA bus speeds....
Not saying anything you all already know.. just having a conversation 😀

Horun wrote on 2022-11-29, 03:09:

Thanks for the better pictures !
Going back a bit OT: Agree with Max. Sure some adapters could run on a faster ISA bus than standard like some VLB Video cards could run faster than the "standard" 33Mhz though usually you have jumper some wait-states in (for VLB both on the board and the adapter). If you think about the late XT era and later 286 era there is a reason the majority latter 286 boards had 12 (or 16Mhz cpu's ) all easy to divide the cpu clock to get 8Mhz for ISA. The 10Mhz variants were sort of bastard board requiring a complex divider+multiplier circuitry to get a 8Mhz ISA bus IF they did it at all. Same goes forward to 386 and 486 boards and why a 33Mhz clock became so important (easy to divide out to 8Mhz) until they figured out to Async the cpu and ISA bus speeds....
Not saying anything you all already know.. just having a conversation 😀

So it sounds like most boards of the era that ran faster than 8mhz had a way of dividing the CPU clock down to 8mhz and I am just very unlucky to have a board that does not seem to have any form of divider?

Most of the cards the system came with seem suited to handle up to 12mhz: the parallel/serial card says it is designed for 12mhz systems, the AboveBoard AT has a 12mhz option in the config options, EGA board clearly works (or else I assume I wouldn't get much video output). Even some of the cards I added myself later, namely the CT3600 and the smsc elite 16 didn't seem to mind. It is only the MFM controller that (seems to) not like 12mhz.

I have thought of 2 possible reasons for the problems and maybe you would be able to say if they sound likely.
1. Variable capacitor vc1 has been changed from its stock state, I have heard that vc1 is used for "stability" but not much beyond that, could it be generating a noisy or bad clock signal?
2. I at one point (stupidly) tried to clean the very bad battery leakage using vinegar, since nothing else seemed to work and it did clean most of the leakage but I later found that vinegar and PCBs aren't exactly friends. I cleaned the board with isopropyl alcohol after that and everything seemed fine but I have since noticed a white-ish tint forming around chip legs, could that be from the vinegar and somehow causing just enough of a short somewhere to make the clock (maybe also the FDC) unstable in a way the MFM controller doesn't like? And if so, how would I clean this? Because I have tried rubbing alcohol and acetone without much success.

I think the MFM card just can't keep up and your motherboard is perfectly fine

maxtherabbit wrote on 2022-11-29, 04:49:

I think the MFM card just can't keep up and your motherboard is perfectly fine

That is what I have been suspecting. Are you aware of any decent MFM controllers that support 12.5mhz systems? Or if I just happen to come across some is there any decent ways of determining whether or not it may support 12.5mhz?

another_808286 wrote on 2022-11-29, 05:24:

maxtherabbit wrote on 2022-11-29, 04:49:

I think the MFM card just can't keep up and your motherboard is perfectly fine

That is what I have been suspecting. Are you aware of any decent MFM controllers that support 12.5mhz systems? Or if I just happen to come across some is there any decent ways of determining whether or not it may support 12.5mhz?

Nope, I do not know of any MFM cards that overclock.

IDE Multi-io = yes
Modern XTIDE + CF CARD = yes

Good luck

Could the problem be related to bus noise or harmonics, maybe?

I mean, generally speaking, some problems with data lines on old or large PCBs are due to electrical characteristics.

A too long trace or wire does not carry a high frequency signal very well.
Capacities increase, reflections occur, etc.

Same goes for traces too thin or too close to each others (cross talk)..

Anyway, maybe an oscillograph can help finding out what happens?

If noise is an issue, it maybe helps to add some extra capacitors accross the power/gnd pins of the ICs.

In most cases a 100 nF or 1000 nF (1 µF) capacitor greatly reduces noise.
Or rather, reduces power fluctuations caused by the ICs themselves.

Digital ICs use switching transistors. If they think, they will cause fluctuations in voltage, which is similar to harmonics caused by square wave oscillators (TTL oscillators).

A little capacitor, say tantalum/ceramic, in the 10 to 1000 nF range, added to each individual IC, will absorb these fluctuations, making the whole power rail more smooth.
In addition, it has a low-pass effect, as well.

It's also possible to attach multiple capacitors in parallel (say, 10 nF+100nF+1000nF).

Variable capacitors (variable condensers) or variable capacity diodes are used to "drag" an oscillator down to a preferred frequency. Or the crystal, more precisely.

The capacity of this part will slow down the oscillations a bit. By minus 3 or 4 KHz at best.
Variable capacitors of about 30 pF are used for this (variable; they can provide less capacity ofc). They're often mistaken for trim pots.

The problem is, if the capacity becomes too high.
It's not just stopping having an effect, but can make the whole oscillation coming to an halt.
The oscillator ceases to function.

Edit: Correction. In the absolutely worst case, the oscillator goes totally bonkers
and runs at frequencies that have nothing to do with the crystal frequency anymore.

The opposite device in this application is a tiny coil, by the way.
Or an "inductor" if we're fancy. 😉
If added, it will increase frequency by a few KHz.

Jo22 wrote on 2022-11-29, 08:22:

Could the problem be related to bus noise or harmonics, maybe? If noise is an issue, it maybe helps to add some extra capacitors […]
Show full quote

Could the problem be related to bus noise or harmonics, maybe?
If noise is an issue, it maybe helps to add some extra capacitors accross the power/gnd pins of the ICs.

In most cases a 100 nF or 1000 nF (1 µF) capacitor greatly reduces noise.
Or rather, reduces power fluctuations caused by the ICs themselves.

Yes back in the late 70's/early 80's it was common to have a 0.1 ceramic parallel an Electrolytic (like a 10uF or 25uf) due to the caps at that time were not very good at filtering HF noise off the power at the IC.
Tantalums are a bit better at HF than standard Electrolytics of the 80's and early 90's iirc.

another_808286 wrote on 2022-11-29, 05:24:

That is what I have been suspecting. Are you aware of any decent MFM controllers that support 12.5mhz systems? Or if I just happen to come across some is there any decent ways of determining whether or not it may support 12.5mhz?

I do not know of any either. If a video card do know some ISA variants that work well on overclocked bus.
Just relized your board is in TH99//theretroweb.com https://theretroweb.com/motherboards/s/itt-co … ional-series-40
can you take a picture of the manual page that describes the front panel connectors (think they are J18, 19, 20) ??

Horun wrote on 2022-11-30, 02:12:

I do not know of any either. If a video card do know some ISA variants that work well on overclocked bus. Just relized your boar […]
Show full quote

another_808286 wrote on 2022-11-29, 05:24:

That is what I have been suspecting. Are you aware of any decent MFM controllers that support 12.5mhz systems? Or if I just happen to come across some is there any decent ways of determining whether or not it may support 12.5mhz?

I do not know of any either. If a video card do know some ISA variants that work well on overclocked bus.
Just relized your board is in TH99//theretroweb.com https://theretroweb.com/motherboards/s/itt-co … ional-series-40
can you take a picture of the manual page that describes the front panel connectors (think they are J18, 19, 20) ??

There is no page in the manual with the pinouts but it is simple enough that I figured it out by looking at it. I attached a diagram I made.

It looks like the page for the board is kinda empty. Would I be able to provide any info to help fill it? I have the original DOS disk that the system came with and I have made various backups of it, it contains some software such as one that sets the clock speed 6/10mhz for the board. Would these be helpful? (and of course, I will scan and upload the full manual when I get a chance)