Falcon + CUBASE AUDIO + SCSI - Experiences?

[whomper]

Found the issue; didn't connect to the EXT pad as I have the same trace cut as you had in your image.

Now that I did, Falcon powered up DFB1X in 16MHz mode.

Cubase Audio crashed the first time loading the song.
Second time loaded but playback was bad as seen in the video.

[exxos]

Now that I did, Falcon powered up DFB1X in 16MHz mode.

Cubase Audio crashed the first time loading the song.
Second time loaded but playback was bad as seen in the video.

You have OPTION jumper still connected ?

That should all really work like stock Falcon now

[whomper]

Yup, OPTION jumper in place.

[Badwolf]

Cubase Audio crashed the first time loading the song.

13 bombs isn't an normal error -- that's suggsetive of bigger issues going on. Are you using TT-RAM? That'll probably be slower than ST-RAM in 16MHz mode so best avoided.

Second time loaded but playback was bad as seen in the video.

Oh wow. That's extreme. The LED blinking off for *human scale* time is really weird. I think that's still the bus mastery signal on DFB1X. That suggests the board is relinquishing control for an extended period of time. Truly bizzare.

I'm really leaning towards Cubase doing something really finely tuned which the timing changes (to bus arb as much as anything else), are throwing off.

@whomper the original title of the thread involces SCSI. Is that still a factor? Does the problem occur with IDE?

BW

[whomper]

I also have that 100K resistor on DMA path, maybe that's the issue?

[whomper]

@whomper the original title of the thread involces SCSI. Is that still a factor? Does the problem occur with IDE?

Cubase Audio only works with SCSI drives as they have some sort of SCSI optimized code as part of Cubase's Direct to Disk behavior.

[whomper]

13 bombs isn't an normal error -- that's suggsetive of bigger issues going on. Are you using TT-RAM? That'll probably be slower than ST-RAM in 16MHz mode so best avoided.

I think it is 14 bombs and if it's the CPU, then its Format Error, whatever that means here.... perhaps it's something to do with the SCSI reading of the song as it happened upon song load, so a SDMA bus malfunction.

[exxos]

I also have that 100K resistor on DMA path, maybe that's the issue?

No that would be fine. You could try replacing the wires with like 33R to see if it helps.. but I fear not..

I'm really leaning towards Cubase doing something really finely tuned which the timing changes (to bus arb as much as anything else), are throwing off.

It does really make me wonder as well.. Though there is ACEtracker in the mix for some people as well.

But don't forget there are problems even without the DFB1X plugged in.. The SDMA clock has to be delayed. So I don't even think this is really DFB1X or clock patch related as such.

Cubase Audio only works with SCSI drives as they have some sort of SCSI optimized code as part of Cubase's Direct to Disk behavior.

Maybe that by itself is a clue..

[exxos]

I think it is 14 bombs and if it's the CPU, then its Format Error, whatever that means here.... perhaps it's something to do with the SCSI reading of the song as it happened upon song load, so a SDMA bus malfunction.

It would suggest why delaying the SDMA clock is needed. Possible now the DFB1X timing is different, its aggravated more what ever the problem is.

I think we should leave the DFB1X out of the equation for now.. Go back to the V3 setup, and keep things simple to work out why the SDMA even needs that delay in the first place..

Can you run ACE TRACKER without problem ? That uses the DSP I believe. It that works fine, then we are circling back to SCSI related issues.

[Badwolf]

13 bombs isn't an normal error -- that's suggsetive of bigger issues going on. Are you using TT-RAM? That'll probably be slower than ST-RAM in 16MHz mode so best avoided.

I think it is 14 bombs.

Oh yes. Again, not a normal one and suggests stack pointer corruption. Possibly TT-RAM related in 16MHz mode.

Cubase Audio only works with SCSI drives as they have some sort of SCSI optimized code as part of Cubase's Direct to Disk behavior.

I'd suggest that's the smoking gun, then. Some special sauce driver that ekes the very limit out of a DMA channel that now has its timing slightly changed? It feels like a strong candidate.

I suspect the only way to fix that would be for the DFB1X to some how be able to mirror the bus arbitration cycle exactly. That may be possible for a verilog whizz, but even then the GALs may have something to say about things.

It does raise the question: does this Cubase SCSI trick work with CT60?

BW