myne wrote on 2025-05-26, 10:52:

Why does yours look so much prettier?

I know, I was wondering the same thing! Seeing the PCI connections so direct, makes me think we can do full length matched PCI traces, and let the ISA section be a mess/some % tolerance for lengths. Since ISA and PCI are generally low frequency buses, I'm not sure lengths will even matter much though. There's far more room to play with on ISA side, so it makes sense to route PCI first then ISA. Although I did some trace length calculations and I'm not even sure it would matter on PCI either? At 30ns clock period (33MHz PCI), even 6 inches of trace length difference only adds ~1ns of skew, which is usually within PCI's timing budget. So a 1 inch trace vs 7 inch trace on PCI would likely be fine. (IT8888 only does 33MHz PCI)

Basically I think we don't really need to worry about lengths and we can make it look as good as far as we are willing to put the time in. (I only bring matched lengths up because almost all pci cards I have, have length matched traces but I don't think it matters much for us, maybe it does for PCI graphics/sounds cards I guess.)

rasteri wrote on 2025-05-26, 15:44:

When I lay out a board I try and get as much routing done on the component side as possible, then leave the bottom side for a gr […]
Show full quote

When I lay out a board I try and get as much routing done on the component side as possible, then leave the bottom side for a ground plane. So the first thing I do for all PCI signals is get them onto the top side.
I'm guessing freerouter prefers to leave signals on the layer they start on, which for the PCI edge connector is sometimes the bottom side.
Also just in general hand-routed boards always look prettier than autorouted ones 😀

If it's any consolation, my ISA section is a complete mess :

I was going to say the ISA pins are what mess up the routing from what I was experiencing. The SA/SD pin placement on the IT8888 are wack imo. Idk why they split them up like they did. The chip layout essentially treats the ISA data bus as a low byte + optional high byte setup rather than one unified 16 bit bus which combined with the layout of control signals makes the ISA stuff a right mess from the start.

Yeah Freerouter is terrible for high pin counts, I'm sure it's good for low pin count but with the amount of connections there are, it doesn't seem to be able to handle it at all. I was surprised myne got a functional routing from it at all, it would just loop for far too long on my end. It seems to have zero logic and just basically throws traces willy nilly. I wonder if making buses on schematic side would influence it's trace routing but I doubt it.

dartfrog wrote on 2025-05-26, 18:07:

I was going to say the ISA pins are what mess up the routing from what I was experiencing. The SA/SD pin placement on the IT8888 are wack imo. Idk why they split them up like they did. The chip layout essentially treats the ISA data bus as a low byte + optional high byte setup rather than one unified 16 bit bus which combined with the layout of control signals makes the ISA stuff a right mess from the start.

Actually - I've had a look at your original project and I think you have the numberings on your ISA footprint wrong.

If I replace it with the footprint from the disappointment kicad project then it seems to make a bit more sense, layout-wise.

Autorouter will probably have an easier time of it too.

(Of course I've only routed address/data so it'll definitely get messier once control lines are added).

Did you try 45* like I did?

rasteri wrote on 2025-05-26, 18:41:

Actually - I've had a look at your original project and I think you have the numberings on your ISA footprint wrong. […]
Show full quote

dartfrog wrote on 2025-05-26, 18:07:

I was going to say the ISA pins are what mess up the routing from what I was experiencing. The SA/SD pin placement on the IT8888 are wack imo. Idk why they split them up like they did. The chip layout essentially treats the ISA data bus as a low byte + optional high byte setup rather than one unified 16 bit bus which combined with the layout of control signals makes the ISA stuff a right mess from the start.

Actually - I've had a look at your original project and I think you have the numberings on your ISA footprint wrong.

If I replace it with the footprint from the disappointment kicad project then it seems to make a bit more sense, layout-wise.

Autorouter will probably have an easier time of it too.

(Of course I've only routed address/data so it'll definitely get messier once control lines are added).

Hmm I copied directly from the datasheet. I double checked and it looks right, I couldn't find anything wrong.

I think it was something else as I just tried freerouting again and didn't loop forever like it was and it isn't routing traces into nothingness / dead-ending them. It did want me to install a new version of Java so maybe it had something to do with that. Although there's a lot to be desired like it placing 90 degree / acute angled traces, vias/traces at the bottom of the edge connector which would get routed/beveled off, etc, maybe there's setting i don't know about and possibly using keepouts can fix the traces below the edge connector problem.

FWIW I was always going to hand route the board I just wanted to see how well freerouting worked.

Try the settings I posted

dartfrog wrote on 2025-05-27, 07:41:

Hmm I copied directly from the datasheet. I double checked and it looks right, I couldn't find anything wrong.

I mean the ISA port pinout rather than the IT8888.

I've attached a (terrible quality) pic showing what I mean - your pinout is on the left, the pinout from wikipedia is on the right.

Unless I've gone mad (entirely possible).

myne wrote on 2025-05-27, 01:25:

Did you try 45* like I did?

I did try 45, and yeah it's probably how I'd do it for a full-PCI-card sized board, but I wanted to make my version as small as possible and diagonal requires more space.

Oh. Shit. The footprint is probably my fault.
The joys of not having the same numbers between footprints.
It looks like my footprint is 1-49, 50-98 A,B
But dartfrogs was ABCD.
And yes, it also looks backwards.

A question about the most recent version of dISAppointment on GitHub (v0.4). Has that version been tested functional? Haven't checked the files in detail so I'm not sure about all the changes compared to v0.3 which I could not get working back then.

On the other hand, I've been experimenting with some other ASUS boards that I could find boardviews for using my previous dISAppointment v0.2 builds but little progress have been made. I've tried P8B75-M and Z97M-PLUS, though I'm leaning on using the former as it has both PS/2 keyboard and mouse ports, plus native PCI. On both boards I've soldered wires connecting their LDRQ1# to the pin 20 of the TPM slot (which was NC according to the boardview), so I could easily connect the adapter and the motherboard as they have matching pinout.

Results so far:
- The F85226AF bridge itself can be detected and its registers can be read fine without any abnormal FFs.
- UNISOUND can't detect any supported sound card that I tried plugging into it at the moment. Haven't narrowed down the reason yet, though it's possible the ISA-side connectivity might be a bit loose.
- When configuring the address mapping using LPCISA (part of my lpcexp tools) on both ASUS boards the bit setting LDRQ1# as GPIO23 were set. I did try unsetting that bit in hope to restore its native functionality, but not sure if that could lead to side effects... On my working setup (ASRock X99M Killer/3.1) the bit doesn't appear to be set, so I wonder if it's something specific to ASUS.

Very quick autoroute with fixed isa slot

LSS10999 wrote on 2025-05-27, 10:05:

A question about the most recent version of dISAppointment on GitHub (v0.4). Has that version been tested functional? Haven't ch […]
Show full quote

A question about the most recent version of dISAppointment on GitHub (v0.4). Has that version been tested functional? Haven't checked the files in detail so I'm not sure about all the changes compared to v0.3 which I could not get working back then.

On the other hand, I've been experimenting with some other ASUS boards that I could find boardviews for using my previous dISAppointment v0.2 builds but little progress have been made. I've tried P8B75-M and Z97M-PLUS, though I'm leaning on using the former as it has both PS/2 keyboard and mouse ports, plus native PCI. On both boards I've soldered wires connecting their LDRQ1# to the pin 20 of the TPM slot (which was NC according to the boardview), so I could easily connect the adapter and the motherboard as they have matching pinout.

Results so far:
- The F85226AF bridge itself can be detected and its registers can be read fine without any abnormal FFs.
- UNISOUND can't detect any supported sound card that I tried plugging into it at the moment. Haven't narrowed down the reason yet, though it's possible the ISA-side connectivity might be a bit loose.
- When configuring the address mapping using LPCISA (part of my lpcexp tools) on both ASUS boards the bit setting LDRQ1# as GPIO23 were set. I did try unsetting that bit in hope to restore its native functionality, but not sure if that could lead to side effects... On my working setup (ASRock X99M Killer/3.1) the bit doesn't appear to be set, so I wonder if it's something specific to ASUS.

Which exact asus board and revision?
I'll see if I have the kicad for it

Nevermind. They're too new. Found boardviews though.

For the Z97 you are REMOVING the SR7 resistor, right?
The way registers work, is that they can be set by hardware - in this case it's pulled up to 3v. So I don't think software can actually change it.

LSS10999 wrote on 2025-05-27, 10:05:

A question about the most recent version of dISAppointment on GitHub (v0.4). Has that version been tested functional?

No I haven't tested it yet, I've been meaning to order one the next time I do a pcbway order but I haven't done one in ages.

Will be soon though.

myne wrote on 2025-05-27, 10:22:

For the Z97 you are REMOVING the SR7 resistor, right?
The way registers work, is that they can be set by hardware - in this case it's pulled up to 3v. So I don't think software can actually change it.

No there's no resistor to remove. Normally the pull-up resistor for LDRQ1# (SR71 in case of Z97M-PLUS) is never populated so I can simply solder a wire to bring it out.

However, those ASUS boards' BIOS configures that pin as GPIO (GPIO23) for some unknown reasons...

PS: SR7 is GPIO50 which I don't think it's of any relevance for LPC stuffs. It's an unmultiplexed GPIO pin.

Gpio is a safe setting, I assume.

I'm take a quick look at the bios.
Latest version I assume?

LSS10999 wrote on 2025-05-27, 11:03:

However, those ASUS boards' BIOS configures that pin as GPIO (GPIO23) for some unknown reasons...

Whoops, I have the same board (Z97M-Plus) waiting for LDRQ1 soldering. And two Z87 Gryphon (has LDRQ1 exposed) and Maximus VI Z87 (LDRQ1 not exported but LDRQ0 can be hijacked from Nuvoton, coz no LPT and seems no reason to use old style DMA for something else).

myne wrote on 2025-05-27, 11:35:

Gpio is a safe setting, I assume.

I'm take a quick look at the bios.
Latest version I assume?

Yeah the boards are on the latest versions. I did not see any GPIO related options in BIOS though there may be hidden menus...

Even without functional LDRQ it should be possible to play some adlib FM synth just via IO ports if properly passed to the soundcard so I would start there and then trying something with DMA...

You can try some nonPNP soundcard like old SB16 or PicoGUS...

LSS10999 wrote on 2025-05-27, 11:45:

myne wrote on 2025-05-27, 11:35:

Gpio is a safe setting, I assume.

I'm take a quick look at the bios.
Latest version I assume?

Yeah the boards are on the latest versions. I did not see any GPIO related options in BIOS though there may be hidden menus...

I had a quick look in amibcp but didn't see anything that jumped out. It has to be in there though. Registers don't just set themselves.
I'm... Better than a noob, but no where near a pro with it though

>Registers don't just set themselves
Yes, but it's a bit needle in a haystack, don't expect it would be in hidden setup menu but anywhere else in the code...

RayeR wrote on 2025-05-27, 14:03:

Even without functional LDRQ it should be possible to play some adlib FM synth just via IO ports if properly passed to the soundcard so I would start there and then trying something with DMA...

You can try some nonPNP soundcard like old SB16 or PicoGUS...

PicoGUS is a PnP-ish card by themselves, other init port though 500-ish IIRC, not classic. I have no luck with it on 0.2 adapter on z87m pro4 asrock's. After setting up ports window and successful pgusinit, no sound or music in duke3d's setup. Not tested sb or mpu mode though.