I assume you have tried with different cable... but have you tried using a driver other than the stock MS one?

Yes, I have swapped cables, and no, I don't know of any other driver for the Roland MPU-401 other than the one provided by Microsoft. However, as the issue also seems to appear in DOS, I don't think its a driver issue.

I tried the following debug commands :
- i 331
BF
- o 331 97
- i 330
FE
-q

That should allow Exclusive to Host, but it doesn't make any difference.

well, if the same happens to DOS then it's a hardware problem I guess... maybe you can check with a multimeter the continuity of the DB25 cable and visually inspect the unit inside for any fault... can't think of anything else atm 😐

btw there are at least 2 more drivers, the patched one from GSPanel and the Cakewalk one, both are multi-clients that allow more than one program to access the mpu.

keropi wrote:

well, if the same happens to DOS then it's a hardware problem I guess... maybe you can check with a multimeter the continuity of the DB25 cable and visually inspect the unit inside for any fault... can't think of anything else atm 😐

btw there are at least 2 more drivers, the patched one from GSPanel and the Cakewalk one, both are multi-clients that allow more than one program to access the mpu.

The port is not completely dead at least.

If I connect the MIDI OUT of my SB Live! in my Windows 98 machine to the MIDI In of the MPU-401 and then connect the MIDI OUT of the MPU-401 to the MIDI IN of the MT-32, the (non-sysex) data will pass to the MT-32. This is how the LAPC-I's synthesizer source is connected to its MPU-401 interface, and how you would send data to the card from an external source.

PeterLI has several MPU-401 boxes he got for his ongoing project. Try borrowing one from him to rule out the port failure?

I made sure in each of my systems that IRQ2 was not in use, disabling AHCI in my Windows 98 machine.

I used a DOS program called MIDI Data Filer, which can poll data as well as use the MPU interrupt. When receiving data, it did NOT work with IRQ2, as I expected, but it did work CORRECTLY with the polling mode. I tried it on both an MT-32 and FB-01, sent the dumped data back and did not see any errors on their LCD screens.

So the problem with my MPU-401 is definitely related to the IRQ, and since I have no conflicts, I suspect that either the transistor inside the MPU-401 on the IRQ line is not working that goes from the LSI QFP chip to the connector.

Otherwise it could be the '04 on the interface card, but I think that less likely since it also drives the reset line. Most likely the DTC124 transistor in the MPU-401 needs replacement.

With considerable help from my friend Cloudschatze, I determined that the problem was the 74LS04 on my MIF-IPC-A. I removed the chip and replaced it and now things are working. The MIDI Data Filer program works with the IRQ mode (set it to IRQ9) and the polling mode after I modified the card. NewRisingSun sent me an MPU-401 IRQ test program, and where before it stated no MPU IRQ received, it now says MPU IRQ on IRQ9.

I tried It Came from the Desert, which didn't work previously with the MT-32, and now it does. This game does rely on MPU-401 IRQs.

great news then! glad you got it sorted 😀

My work will not win Best Soldering of the Year. However, it just goes to show that even with a card and a package that came to me in great shape, there may still some subtle flaw that will impair the functionality without causing total failure.

I hope that PeterLI and the other guy making the MIF-IPC-A clone include IRQ functionality in their testing.