LSS10999 wrote on 2023-12-19, 06:31:

I think I might have caused damage on the Z97M board by actually connecting the LPCPD# wire without thinking this time. On newer versions of the LPC bridge LPCPD# is not needed anymore and is connected to +3.3V via a pull-up resistor on the PCB. The reason was that on some boards the LPCPD# pin might not have been actually wired to the TPM header. It may be possible that the board's PCH might not like that pin being externally pulled up, as the boardview suggest TPM header's LPCPD# is wired to the PCH, yet it does not appear to have been wired to the SuperIO...

I would doubt a pullup would damage anything, even though it's relatively strong (4.7K) that's still a tiny current flow.

LSS10999 wrote on 2023-12-19, 06:31:

I don't have any usable 25Q64 (8MByte) SPI flash at the moment... I think BIOS expects a flash ROM of matching size and would not accept larger ones such as 25Q128 (16MByte) which I do have.

It's an interesting question, I didn't try myself yet if it's possible to replace flashrom by larger size. Maybe some MB manufactuers would use different sizes on the same chipset but not sure. What may work - duplicate the current bios image to fill bigger flashrom by 2 copies (one by one), one of them could catch on. My gigabyte mobos has dualbios so it gives me backup if something messed up...

And AFAIK any common spi flashrom programmer doesn't run at full speed of SCK,.e.g. 100MHz that flahrom can run.

duplicating a rom to fit it into an spi eeprom is silly.
parallel eproms need it due to the fact that the higher address lines can be in an unknown state which causes the need for duplication
another option for the parallel eeproms would be to fix the higher address lines to a fixed state .
Neither of these options are valid or necessary with a serial eeprom

PCH maps SPI flashrom at top of CPU address space and I'm not sure if it has some size autodetection according to device ID command or if it's hardwired/strapped some way. If the size is autodetected, then eg. 8MB flash wil be mapped at 0xFF800000-0xFFFFFFFF while 16MB flash will be at 0xFF000000-0xFFFFFFFF. And when the 8MB image will be flashed only into the 1st half of 16MB flashrom then BIOS code will be wrongly mapped at 0xFF000000-0xFF7FFFFF while 2nd half will be empty and CPU will have nothing to execute at reset vector. So in such case the duplication of 8MB image to both halves of 16MB flashrom would ensure proper execution for various cases. But as I told I'm not sure about it. But at least duplicating the image will not hurt anything while it _may help so why not try it...

RayeR wrote on 2023-12-20, 05:51:

PCH maps SPI flashrom at top of CPU address space and I'm not sure if it has some size autodetection according to device ID command or if it's hardwired/strapped some way. If the size is autodetected, then eg. 8MB flash wil be mapped at 0xFF800000-0xFFFFFFFF while 16MB flash will be at 0xFF000000-0xFFFFFFFF. And when the 8MB image will be flashed only into the 1st half of 16MB flashrom then BIOS code will be wrongly mapped at 0xFF000000-0xFF7FFFFF while 2nd half will be empty and CPU will have nothing to execute at reset vector. So in such case the duplication of 8MB image to both halves of 16MB flashrom would ensure proper execution for various cases. But as I told I'm not sure about it. But at least duplicating the image will not hurt anything while it _may help so why not try it...

Thanks for the explanation. I've read such suggestions before, about duplicating the ROM contents when using larger ones.

I recall there were reported cases of success, though I've never tried myself. If the PCH indeed manages SPI flash the way you said, then it's very likely duplicating the ROM contents will work.

Just that I've no idea what would happen to the ROM if I use its built-in Instant Flash.

Quick question, how hard it is to solder these adapters ? Or do PCBWay or other are able to fully populate them on order ?
It would be interesting to see a picture of both sides of the adapter, I did not find one.

anton23 wrote on 2023-12-20, 09:47:

Quick question, how hard it is to solder these adapters ? Or do PCBWay or other are able to fully populate them on order ?
It would be interesting to see a picture of both sides of the adapter, I did not find one.

As of v0.2/v0.3 there are no parts on the other side of the PCB, though for v0.2 (v0.1 also) you'll need an additional resistor between two adjacent ISA slot pins to fix an issue.

You'll want to start with F85226 first. If you're good at drag soldering it's not the hardest component to solder, just make sure it's positioned and oriented correctly.

As for other SMD parts, resistor arrays can be difficult due to them being relatively small, but SMD resistors, caps, fuses and LEDs should be easy. Should note that all five LEDs have the same orientation. The + markers on the LED footprints for negative ones are misleading.

Most through-hole parts shouldn't be difficult, either. You'll want to start with the caps and headers then the ATX and ISA slot connectors. That leaves only L7905 which is the tallest and also the most difficult component to solder.

I think a higher solder temperature may be needed when doing L7905, because its heat spreader will divert the heat from the soldering iron (as well as the tin) which could interfere with the soldering process.

IMO it's not difficult to solder in overall but it's not easy to ensure everything works. You'll want a good magnifier to be able to carefully inspect and verify everything's soldered correctly, and without bridges (especially F85226 and resistor arrays).

PS: While the BOM specified 1K for resistors connecting to the LED, in my case it's a little too bright when operating. Perhaps increasing these resistors a bit could mitigate the problem.

My PCBs including LPC2ISA just arrived today from JLCPCB so I started with component assembly. 1st I have to try SMPS inverter if it works properly...

Assembly completted. I only tested the DC-DC SMPS yet and it seems to work...

My 1st test in PC was not successfull. I attached it to my Gigabyte GA-P67A-D3-B3 with LDRQ1# mod and put ES1869F in the ISA slot. I run SAPPHISA.EXE utility without any args and it detected the Fintek bridge and set default I/O windows. Then I run Unisound 0.80 and it detected ES1869F with IO 220h, IRQ5, DMA1 but Doom and other programs don't play any sound (neither adlib) and ESSCFG and SBpro test couldn't find the soundcard at 220h. Also HWINFO utility detected the ES1869F. If I don't run SAPPHISA.EXE before, then neither Unisound nor HWINFO sees the soundcard. So it is partially working (I guess at least data and address bus) but I don't know what's wrong. I have to doublecheck schematics and soldering. I tested ES1869F in my temporary Tillamook setup in native ISA slot and it works fine.

RayeR wrote on 2024-01-02, 04:45:

My 1st test in PC was not successfull. I attached it to my Gigabyte GA-P67A-D3-B3 with LDRQ1# mod and put ES1869F in the ISA slot.

You have another soundcard to try? Maybe sapphisa doesn't open up enough ports for the ES1869 or something

I think it don't use anything special than usual 22x/330/38x ports for SBPro, MIDI, OPL...

RayeR wrote on 2024-01-02, 12:26:

I think it don't use anything special than usual 22x/330/38x ports for SBPro, MIDI, OPL...

Does UNISOUND mention any other port than the usual Sound Blaster ones?

Some non-Creative cards do use more ports, such as YMF718 which additionally used port 370h for control registers, though you could move it to an accessible one (like how SETYMF moved it to A28h).

I'm not at that PC now... Mumak wrote me that HWINFO use PnP detection mechanism when supported, if I remember well the ES1869 was listed in ISA/PNP section so it was detected by PnP not direct access.

UPDATE: I just inspected PCB soldering under stereoscope magnifier and didn't find anything suspicious. Also I checked ISA bus and fintek bridge pinouts and wiring-should be ok. I use config with default IO address, disabled ROM decoding and disabled kbd/mouse.

I got a bit further...
1st I got an idea to test with my POST card. So I setup single IO window for 80h:

C:\DOS\DRIVERS>sapphisa.exe 80 0f
Found Intel LPC Controller.
Found Fintek F85226FG LPC-ISA Bridge
Enabling Range 0 : Base 80, Mask f, LPC f0081
Ports : 80-8F

and send some data there:
smb /dbg
CMD:> outportb(80, aa)
AAh -> port[0080h]
CMD:> outportb(80, bb)
BBh -> port[0080h]

and yeah, the display light up, so it seems at least address and data bus works.

Then I looked again at Unisound output and it shows some extra port CTR:660 that was not included before:
C:\DOS\DRIVERS>unisound.exe
Universal ISA PnP Sound Card Driver for DOS v0.80a. (c) JazeFox 2019-2022
PnP card found: [ESS1869] ESS ES1869 Plug and Play AudioDrive
Loading Unisound default settings...
ADD:220 OPL:388 MPU:330 IRQ:5 DMA:1/0 CTR:660 JOY:201
Initialization done.
ESS Mixer [VOL:85 WAV:80 FM:80 LIN:0 CD:0 MIC:0 3D:0]

So I included it to IO windows (a bit confused how base and mask actually works for a range but should be OK for now):
C:\DOS\DRIVERS>sapphisa.exe 200 fc 300 70 388 1c 660 3f
Found Intel LPC Controller.
Found Fintek F85226FG LPC-ISA Bridge
Enabling Range 0 : Base 200, Mask fc, LPC fc0201
Ports : 200-2FF

Enabling Range 1 : Base 300, Mask 70, LPC 700301
Ports : 300-303,310-313,320-323,330-333,340-343,350-353,360-363,370-373

Enabling Range 2 : Base 388, Mask 1c, LPC 1c0389
Ports : 380-39F

Enabling Range 3 : Base 660, Mask 3f, LPC 3f0661
Ports : 640-67F

Then important change happened - ESSCFG found and initialized the soundcard 😀
But now I'm facing probably some IRQ problem. I tried both options 5 and 7 but none works. SBPRO Test programs cannot find IRQ and Doom SFX is silent. But now I got at least the FM and MIDI music. Also Boost demo by Doomsday (uses MIDAS) no sound. But MPXplay plays MP3 file fine. I think MPXplay may not use IRQ/DMA? It shows snd as SBPro.

For this test I disabled LPT port to save IRQ 5/7 but according to HWinfo it seems some other PCI devices is occupying it and I cannot get rid of them. The SETUP don't have common PCI&PNP section that allowed to reserve some IRQ for legacy, no way 🙁

RayeR wrote on 2024-01-03, 06:36:

I got a bit further... 1st I got an idea to test with my POST card. So I setup single IO window for 80h: […]
Show full quote

I got a bit further...
1st I got an idea to test with my POST card. So I setup single IO window for 80h:

C:\DOS\DRIVERS>sapphisa.exe 80 0f
Found Intel LPC Controller.
Found Fintek F85226FG LPC-ISA Bridge
Enabling Range 0 : Base 80, Mask f, LPC f0081
Ports : 80-8F

and send some data there:
smb /dbg
CMD:> outportb(80, aa)
AAh -> port[0080h]
CMD:> outportb(80, bb)
BBh -> port[0080h]

and yeah, the display light up, so it seems at least address and data bus works.
hw-setup.jpg

Then I looked again at Unisound output and it shows some extra port CTR:660 that was not included before:
C:\DOS\DRIVERS>unisound.exe
Universal ISA PnP Sound Card Driver for DOS v0.80a. (c) JazeFox 2019-2022
PnP card found: [ESS1869] ESS ES1869 Plug and Play AudioDrive
Loading Unisound default settings...
ADD:220 OPL:388 MPU:330 IRQ:5 DMA:1/0 CTR:660 JOY:201
Initialization done.
ESS Mixer [VOL:85 WAV:80 FM:80 LIN:0 CD:0 MIC:0 3D:0]

So I included it to IO windows (a bit confused how base and mask actually works for a range but should be OK for now):
C:\DOS\DRIVERS>sapphisa.exe 200 fc 300 70 388 1c 660 3f
Found Intel LPC Controller.
Found Fintek F85226FG LPC-ISA Bridge
Enabling Range 0 : Base 200, Mask fc, LPC fc0201
Ports : 200-2FF

Enabling Range 1 : Base 300, Mask 70, LPC 700301
Ports : 300-303,310-313,320-323,330-333,340-343,350-353,360-363,370-373

Enabling Range 2 : Base 388, Mask 1c, LPC 1c0389
Ports : 380-39F

Enabling Range 3 : Base 660, Mask 3f, LPC 3f0661
Ports : 640-67F

Then important change happened - ESSCFG found and initialized the soundcard 😀
But now I'm facing probably some IRQ problem. I tried both options 5 and 7 but none works. SBPRO Test programs cannot find IRQ and Doom SFX is silent. But now I got at least the FM and MIDI music. Also Boost demo by Doomsday (uses MIDAS) no sound. But MPXplay plays MP3 file fine. I think MPXplay may not use IRQ/DMA? It shows snd as SBPro.

For this test I disabled LPT port to save IRQ 5/7 but according to HWinfo it seems some other PCI devices is occupying it and I cannot get rid of them. The SETUP don't have common PCI&PNP section that allowed to reserve some IRQ for legacy, no way 🙁

hwi-irq.jpg
hwi-dma.jpg
hwi-peri.jpg

Hmmm... system devices taking up IRQ 5 and 7? From what I searched... 1C12 is a root port and 1C26 is a USB controller.

You're unlikely to find options to reserve IRQ for legacy on such new boards... Maybe you can just leave printer port on either 5 and 7 and put your card there. Back then LPT ports appear to permit its IRQ shared by sound cards, since both had a default IRQ of 7. Letting the printer port taking up that IRQ might at least block PCI/PCIe devices from taking it.

By the way... from your test result, looks like the A00 address range is only needed during PnP setup and most likely not needed anymore once configured, and it can then be assigned for a different purpose until you need to configure it again.

Good point. When I reenabled LPT1 with IRQ7, it kicked PCI device off. Then I configured ESSCFG with IRQ7 and modified BLASTER env. I hope it could be shared as I wouldn't use any LPT communication during gaming. SBPro test still cannot find the IRQ but Doom catches on and I got SFX! 😀 But Boost demo couldn't find it and still silent. I don't know if possible to force it to IRQ7, probably no cmdline args, MIDAS should have some autodetection. It could also affect other programs using MIDAS. Need more testing...

I don't know what port A00 was for but yes, not needed at least for ESS. I think ESSCFG tries communicate at 660h. Maybe only needed for init. I could try to remove this window after ESSCFG if it still keeps working...

RayeR wrote on 2024-01-03, 12:53:

Good point. When I reenabled LPT1 with IRQ7, it kicked PCI device off. Then I configured ESSCFG with IRQ7 and modified BLASTER env. I hope it could be shared as I wouldn't use any LPT communication during gaming. SBPro test still cannot find the IRQ but Doom catches on and I got SFX! 😀 But Boost demo couldn't find it and still silent. I don't know if possible to force it to IRQ7, probably no cmdline args, MIDAS should have some autodetection. It could also affect other programs using MIDAS. Need more testing...

If Boost demo expects IRQ5 you may consider assigning LPT1 to 278h/IRQ5 to see if it makes a difference.

Also, make sure you configured BLASTER environment variable correctly. Note that UNISOUND also uses BLASTER to configure the sound card, and it accepts some new parameters for configuring extra features. You can refer to UNISOUND's manual for details.

Sound card detection mechanisms may or may not work for a variety of reasons, but if the game detects the sound card and plays fine then it's okay. In my current systems SETD (from DIGPAK) never succeeded in detecting my Sound Blaster PnP cards (including a Sound Blaster 16 PnP) but audio worked fine when I manually set the driver up.

RayeR wrote on 2024-01-03, 12:53:

I don't know what port A00 was for but yes, not needed at least for ESS. I think ESSCFG tries communicate at 660h. Maybe only needed for init. I could try to remove this window after ESSCFG if it still keeps working...

I was referring to A00-AFF range. ISA PnP uses A79 as WRITE_DATA so it's kind of needed when you're initializing the ISA PnP card.

Looks like once you're done with ISA PnP you can safely reuse that range for a different purpose, like control register as well as using it for something like WSS.

Is there a way to make this less wider? I for one, much prefer to use a longer board (most cases can hold cards as long as 330mm+), and lose less PCI slots.
Also, i bought an X99M killer for testing, but I cannot determine via the Boardview where the LDRQ is. Can someone help me out on this one?
Thanks,

nakos1212 wrote on 2024-01-03, 13:47:

Is there a way to make this less wider? I for one, much prefer to use a longer board (most cases can hold cards as long as 330mm+), and lose less PCI slots.

Unless your case is tall enough that allows 3 or more PCI holes past ATX specs, I don't think it's feasible to use it with a full ATX board. Most cases have just the amount of PCI holes an ATX motherboard would have, or one more to allow a 2-slot PCIe video card on the bottommost PCIe slot if available.

nakos1212 wrote on 2024-01-03, 13:47:

Also, i bought an X99M killer for testing, but I cannot determine via the Boardview where the LDRQ is. Can someone help me out on this one?
Thanks,

I don't own your board's boardview so I'm not sure. There's a good chance LDRQ0# is wired to SuperIO which make it a bit difficult to reuse.

You may consider looking for LDRQ1#. Either it's used for special purposes (as it can be used as a GPIO), or it's left unused. For X99 chipset, LDRQ1# is AA35 (refer to X99 chipset's datasheet for more info).

There's a caveat, however, that make sure there's nothing else on the LPC bus other than SuperIO. You can check for the presence of other devices on the LPC bus by looking at LAD0-LAD3 nets.

I got myself an ASUS X99M-WS whose LDRQ1# was apparently being used for CPU_OV (so theoretically no need to mod) but the adapter could not be used there because of the onboard TPU chips (overclocking-related) which were also on the LPC bus and were interfering with the adapter.