Managed to find a set of generic setup programs and a speed utility from here:
https://www.minuszerodegrees.net/software/

The speed utility refused to work because the keyboard controller must be from Award, whereas mine is AMI. But this got me to highly suspect that this issue is caused by the replacement keyboard controller. Unfortunately the original one is completely dead, but it probably had the programming needed to handle the clock selection.. whereas the AMI replacement I added is missing on that, so I'm getting this behavior.

If this is the case then this board will be pretty much stuck in hi speed mode all the time..

I was watching a necroware video recently about fixing a turbo issue on a 486 board and the keyboard controller acts as an input to the turbo system on some mainboards based on keyboard shortcuts hard coded into the controller IC. If the keyboard controller was replaced and it uses this pin for a different purpose, the turbo mode would act according to the logic on that pin.

If you probe the keyboard controller for any output pin which changes state at the same time the system goes from 12.5mhz to 25mhz during boot and you find one, lifting this pin will likely allow the turbo button header to work again.

The strange thing is though that you would expect the turbo light to be on even if you use the keyboard shortcut

Ok, I finally got to get back to this one. I did probe all keyboard controller pins, and I couldn't find anything that particularly corresponds to the turbo behavior I'm seeing. One of the keyboard controller pins was going high during the memory count, then going low as soon as counting is done. That's the A20 gate, and it's working fine. The turbo up behavior happens when Starting MS DOS shows up, so it's not correlated.

One thing caught my attention though is how the clock signal looks like at the CPU.. See the attachment. As if both 12MHz AND 25MHz are being fed at the same time.. even the scope is a little confused.. some times it reports it as 25MHz and sometimes 12MHz every second or so.. I don't think this is normal, no?

Can you trace where the turbo header routes to? Which chip is controlling the turbo behaviour that takes this as an input?

NeoG_ wrote on 2025-12-18, 01:39:

Can you trace where the turbo header routes to? Which chip is controlling the turbo behaviour that takes this as an input?

I was only partially successful in my traces.
I traced the turbo switch to a NAND chip that fed into a flip-flop that fed back to the Turbo LED. So that solves why the turbo LED is working regardless of the actual chipset's decisions.

Now I expect that the output from the flip flop should also feed into the chipset to tell it about the turbo switch's state.. but so far I couldn't find the trace. Either it is severed by corrosion and I missed fixing it, or the output of this flip flop isn't used by the chipset (but that doesn't make sense.. as the whole point of relying on the flip flop is that it's in sync with the CPU's clock so turbo switch changes do not take effect at bad times).

The flip flop in question is SN74LS74AN. Here is its data sheet:
https://www.ti.com/lit/gpn/sn54s74

1D is ultimately connected to the turbo switch. It goes to 4.6V when the turbo switch is shorted and 0V when open. 1PRE and 1CLR are both always high at 5V. 1CLK is also connected to clock, and has the same signal shape I can see at the CPU (a mix of both 25 and 12.5MHz).
1Q goes ultimately back to the turbo LED.. and the turbo LED is lighting on/off correctly.

I'm trying to find now if I can trace 1Q or the inverted 1Q to anything else other than the turbo LED. But I have a feeling that I won't find anything else, and that would be the issue. That would be a bummer though because it will be tough to find the responsible broken trace.

according to the little settings document on TRW

JP1 - "Hispeed" setting to chipset 101; pins 3-2 closed = ON (CLK*2 selected) open = CKLx2 /4. If left open CPU will go half as fast.
JP19 - TURBO LED
JP2 - TURBO; 3-2 = ON. TURBO is desabled on power on.

this would make me think the turbo switch should be connected to JP2.

my 2cents.

weedeewee wrote on 2025-12-21, 21:53:

according to the little settings document on TRW […]
Show full quote

according to the little settings document on TRW

JP1 - "Hispeed" setting to chipset 101; pins 3-2 closed = ON (CLK*2 selected) open = CKLx2 /4. If left open CPU will go half as fast.
JP19 - TURBO LED
JP2 - TURBO; 3-2 = ON. TURBO is desabled on power on.

this would make me think the turbo switch should be connected to JP2.

my 2cents.

While JP2 does change the speed, I don't think it's the standard turbo switch. My claim is because JP2 doesn't result in Turbo LED being affected, whereas JP1 does. Another consideration is that other motherboards using the same chipset name have a jumper called Dynamic Memory Wait State [0 or 1] in addition to TBSW. I think this might be the same as JP2 on my board. JP2 doesn't cause a clock change at all on the CPU.
Finally, JP2 is changing the benchmarks from 12MHz to 16MHz. The turbo switch usually halves the speed rather than overclock the CPU.

Scouring other boards with the same chipset, I noticed some of them have jumpers to allow keyboard control of turbo instead of hardware. I topk this one:
https://theretroweb.com/motherboards/s/chaint … -elt-286b-120gm

Tried it BIOS, and the BIOS exposes options to boot with fast or slow CPU and to change the wait state. Alas, neither of those options seem to be effective on my board. Furthermore, when I do Ctrl Alt + and Ctrl Alt -, I get a beep that reflects the choice of turbo (high pitch vs low pitch), yet no actual speed change is occuring.

This leads me to believe that my board is hardwiring the chipset to not allow keyboard control of turbo.. which is ok.. but I still want to fix the hardware switch.

Thanks for the suggestion anyways.