Did you try setting the 1 ws jumper on the Cirrus cards?

feipoa wrote:

Did you try setting the 1 ws jumper on the Cirrus cards?

yes they are setup for 1WS as 0WS never worked for me reliably 😵
Infact it is something that bugs me to this day: what system can use 0WS ? It's quite a bummer for the 5429 card as 0WS is indeed faster but stability is really bad....

From what I have noticed in general, but not always, is that when pushing my 386 further, e.g. with SXL2 at 50 or 55 MHz, and using the tighest RAM timings, that I need to use 1ws on the ISA graphics card. Otherwise, 0ws is fine with an AMD DX40 and fastest timing. Sometimes adding 16-bit wait states in the BIOS will help instead of using 1ws on the ISA card. The 16-bit BIOS wait states didnt seem to reduce benchmark results.

For me, though, the instability of 0ws on the graphics card manifests itself when loading Win311 graphics drivers.

Probably Cirrus uses irq12.
Did you check it?
The TSR really loads only if a mouse is connected.
The code of an ordinary bios is taken as a basis for the TSR. It does the same. It checks for the mouse presence before activating the functions necessary for a mouse to work.

I just tested with my WD90C30 vga - the TSR does the same thing as the CL cards -> fails to detect the mouse and locks up the kbc.
None of these cards have an IRQ jumper and I have used all of them successfully with systems that have ps/2 mouse support from the factory (a DFI 286/16 system with an ancient AMIBIOS , my IBM PS1/pro and a couple 486 mobos with working ps/2 ports).
If the IRQ was the issue wouldn't those systems also be affected ?

I run CHECKIT to check IRQ usage, in both CL and WDC cards it seems IRQ12 is free:

Can the controller be wired to a different IRQ for testing?
Also has anyone else tried CL/WDC cards or some that are not really basic vgas? Trying to see if this is a specific issue to my hardware or not

keropi wrote:

None of these cards have an IRQ jumper and I have used all of them successfully with systems that have ps/2 mouse support from the factory (a DFI 286/16 system with an ancient AMIBIOS , my IBM PS1/pro and a couple 486 mobos with working ps/2 ports).
If the IRQ was the issue wouldn't those systems also be affected ?

Yes, you are right.

keropi wrote:

Can the controller be wired to a different IRQ for testing?

Tell me which Irq you want to use, I will change and recompile the TSR.
But I very much doubt that the reason is this.

^ maybe IRQ11? or 10?
Do you have any ideas what could be the issue?

Do you have a link to CHECKIT? I assume it is freeware or shareware.

I did all my testing with a CL5434 without issue, except for the AMI board which wanted the TSR loaded first.

My PCBs have shipped and I will retest with this PCB if I can locate and desolder some SMD transistors of proper pinout.

I just load the TSR after boot ends, there is nothing really on that test system just himem for XMS. I am using a XT-IDE CF controller for quick tests, there is only DOS6.22 in it with a barebones autoexec/config setup and a bunch of test/bench programs and games to evaluate the system being tested.
Maybe it's an AMIBIOS thing - but that's what all my boards use unfortunately

So you aren't loading anything except for Himem? No sound card drivers, SCSI, etc? As with other things, I've noticed some drivers want to be loaded in particular orders, e.g. the DLC3 driver. Once I complete tested the most likely issues, I start testing the unlikely issues. What happens if you put the TSR into autoexec.bat and have nothing else in config.sys or autoexec.bat?

Yes I load nothing - no drivers of any sorts. The only things installed on the test system is a VGA and the Lo-tech CF card.
I haven't tried booting without autoexec/config, I will though but I doubt something will change :\

Yeah, I know it is doubtful. I also had little hope that changing the order of loading the Windows net setup in autoexec.bat would change anything, and it made all the difference.

Any reason why you want to be able to load the TSR and not have a mouse connected?

none, it is only useful if you want to test a bunch of controllers and not have to solder the ports to them in order to save time

I received the PCB's a few days ago and have assembled a unit. The transistors were desoldered from dead HDD PCB's. The resistors and capacitors are from whatever I had in my bin. Using 47 pF for the caps, 10K for mosue data/clock pull-up, and 2.05K for the transistor's base resistor. I will test it out later.

I forgot to point out that the PCB seems a bit thick for this approach, though I think it will work. I was just wondering if PCB's can be ordered thinner than what OSH Park has to offer?

sadly oshpark does not offer a thinner pcb ... it will work for testing though so just do it
it's not that bad - if the mod was working on my board with the faster VGAs I would have replaced the motherboard socket with round-pin machine milled female receptors and solder the kbc on a round-pin socket. Obviously this works when the kbc is not near the ISA slots...

I was unable to get the PCB working.

The PCB is too thick. When smashing the cable header between the KBC and the socket, one of my pins was not making contact. So I used an additional DIP-40 socket and cut away one of the supports such that the cable header can fit between without limiting pin contact. Put it back in the motherboard, still no working mouse. I discovered that the insulation on wire for pin 39 was getting smashed and making contact with the socket, such that pin39 on the KBC was shorting down to pin 39 on the motherboard's socket. I resoldered on a new wire and confirmed there wasn't a short. Still no mouse working. Its possible that one of the transistors died during the short... I don't know. I'll have to order some new transistors during my next digikey order and will be putting this on hold until then.

The other thought is, I am using 2.05K for the current limiting resistor into the transistor's base. You are using 4.7K. Did you determine that 4.7K was needed, or did you just use whatever you had?

no I did not think any of it, I just followed Rio444's schematic here and used the same parts: Re: Native PS/2 mouse implementation for 386/486 boards using the keyboard controller

Sorry, I should have been clearer. that last question was directed to Rio444.

semi-unrelated to the project but I tried both an AMIKEY-2 and VIA controllers on the mobo below and it could not boot with any installed (blank screen, no beeps)
It only worked with it's own controller (amikey-2 was modded , via I tested both modded and unmodded)

I have no idea what the mobo's kbc is but it seems the controllers are not always interchangeable.