Sega Katana Devkit Issues

Are any jumpers set on the HDD ?
(back of the HDD next to the SCSI connector, and/or on the motherboard of the HDD itself)

 

Last I checked, Termination Power from Drive was set on J6, and no address jumpers on J2.

 

That's in line with a typical setup.
Active SCSI terminators connected to the SCSI ports at the back of the devkit I suppose ?

 

Yup.

 

Interestingly, setting the HDD ID to #2 does this:

 

Other things I would try before checking for capacitors/... :
- try with another SCSI cable (shorter ? mine are max 2m I think. We've had people with bad scsi cables)
- try with another drive
- check the connectors & cables inside the devkit
- try another reflash - did you respect the 30 seconds wait time with everything shut down after the reflash, as indicated in the manual ?
- check the jumpers - I don't think there are jumpers on the GD-M board, but on my mainboard (with the back connector on my left side, from left to right the jumpers are: D- DUUD DUU (D=down, U = UP, - = nothing) cfr here: http://devkits.handheldmuseum.com/Sega_Katana/SegaDC_KatanaMBa_1.jpg - not sure what these jumpers are for, remark... found no documents about them / didn't investigate

 

Will have to get a hold of a drive + cables. I'm leaning away from the thought that it's the drive as they work fine directly connected to the adapter.

I'll try reflashing, and yeah I have respected the 30s wait and reboot of Katana -> PC.

Jumpers are the same (and were not touched).

 

Got another drive + long SCSI 50pin cable.

The same symptoms are showing themselves. I'm going to try flashing the bootrom again but I have run out of things to test. Anyone have any ideas or is the GD-M basically busted? I don't get how the hell this happened =\.

I wish I knew how to monitor what's going on in the system. Examining the ROM there should be debug output somewhere but you probably need special equipment to access it.

I'm wondering if the SCSI ID settings for the GD-M were in flash ram and got cleared when I ran the above code? Though I think it broke before resetting the flash ram, the days are merging together.

Tangent: Was playing with the samples, some good info on various techniques there. Loving modifier volumes.

https://imgur.com/a/1fa7TNm

 

So I tried tackling the problem another way. Cross Products had given us two debug ports so I was probing around trying to find the TX line. It's pin 8 (middle bottom). Transmits at 38400 baud. Attached is the msgs I got.

Hmm.... Initer on 0, should be on 7 =\.

I got the Japanese SDK as well and they pack in some extras including a fwhistory file (firmware history). In one update it says:

So might be able to reset it in this debug menu. Currently have a hacked up Arduino and holding the wire to read the pin, going to get a 10pin ribbon cable, should help with the corruption issue.

 

some old scsi drive Termination will not work,
try use SCSI cable with Termination adapter

 

Good news, got a SCSI2SD card and made more progress. The debug log now says:

However the GD-M then stops appearing as a device when booting, and the HDD/CD LEDs are blinking a pattern. Doesn't happen if I unplug it.

Maybe the card is configured wrong?

 

Did some more configuring and..... IT WORKS!!!! YAAAAY!

I had to turn off the built in termination as it was screwing up the bus (since we terminate with physical terms at the back).

Brought the devkit back to working order. Now I just need a mount for this thing. I'm still surprised the HDD died, but I think HI_Ricky is right, the HDD's termination had fried or something. I guess the Adaptec card is better at overcoming this than the Cross Products hardware hence why it works ok in the pc.

 

Congratulations !
You mentioned you also had problems with the GD-M disappearing with the SCSI2SD, I suppose it was because of the SCSI termination activated on the SCSI2SD then / and then it worked with SCSI2SD termination off ?
Or did you do something else too ?

 

Most likely me shutting off the termination on the card. It would only disappear if I removed the scsi terminators on the back, and come back if one existed. Once I shut off termination on the card it worked as intended.