F67241@BARILVM.BITNET (Dennis Rapaport) (06/19/89)
I tried installing the accelerator in an Amiga 2000 (4.3 board) with a hard disk and OMTI (IBM type) controller. The system hangs on the initial disk access, with a spurious "not a DOS disk" requester. The disk used here is a 30Mb drive; when the A2620 is installed in an otherwise identical 2000 with only a 20Mb drive and similar (but slightly different) OMTI controller it works perfectly. This is a question for the experts in subtle timing problems; what say you?
daveh@cbmvax.UUCP (Dave Haynie) (06/21/89)
in article <571F67241@BARILVM>, F67241@BARILVM.BITNET (Dennis Rapaport) says: > I tried installing the accelerator in an Amiga 2000 (4.3 board) with > a hard disk and OMTI (IBM type) controller. The system hangs on the > initial disk access, with a spurious "not a DOS disk" requester. The > disk used here is a 30Mb drive; when the A2620 is installed in an > otherwise identical 2000 with only a 20Mb drive and similar (but slightly > different) OMTI controller it works perfectly. This is a question for > the experts in subtle timing problems; what say you? If the drive controller was a DMA type device (which I suspect it isn't), I'd recommend you check the following: [1] Check to see if the expansion bus buffer at U605 is a Signetics part. If so, add a 3.3k ohm resistor between pins 20 and 11; the bus pullups don't pull the AS* quickly enough when used with a Signetics part (it doesn't hurt to make this change with any part, though it's a DMA issue mainly concerning CPU slot boards with local memory, like the A2620). [2] Make sure the bus request output of the controller card is correctly synchronized with the 7.16MHz clock on the bus. This is always a requirement, but could make more difference with an accelerator card of some kind. If the drive isn't DMA driven, it's basically set up looking like some memory mapped I/O. I don't have the slightest idea what such an IBM type device would look like specifically, but I suspect most of the work of making such a device act correctly in an Amiga bus is centered on the bus converter you're using. There are two very critical timings in a read cycle -- the time at which DTACK is asserted and the time at which data is asserted. All accelerator cards are basically 68000 emulators, and must follow the 68000 specification for these timings. However, that's no guarantee that an expansion card that doesn't follow the proper timing will fail with a 68000 -- most of the bus specs define a range of operation, and real live 68000s are prefectly valid follow best case as well as worst case timing. The A2620 is pretty much centered in the operating range. If the converter board is counting on the system generated DTACK, it's probably waiting too long to assert it's data on the bus. This sounds the most likely, as you mention that the two IBM controllers are slightly different, and the bus operation should have nothing at all to do with the hard drive you're using. Or the converter logic may be just plain marginal. Except for the first suggestion, there's relatively little you can do to make the thing work unless you're willing to dig into the timing with a 'scope or analyzer. Though complaining to the designers of the converter card may prompt them to attempt the same feat. -- Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy Be careful what you wish for -- you just might get it