SE/30 the final frontier: Fast/Wide SCSI 2 for the 030 PDS

[Trash80toHP_Mini]

It's always annoyed me that there was no equivalent of the JackHammer for the 030 PDS  .  .  .  so this morning's coffee musings:

We currently have:

SCSI2SD that's capable of supporting Fast Narrow SCSI 2 that's bottlenecked at the incredibly slow SCSI bus of the SE/30

SCSI-NuBus Test Card design example and DeclROM/code documented in GttMFH2e, that's capable of booting from a SCSI device.

Crazy notion here was to:

1 - Prune the NuBus Test Card back to the NuBus transeiver chipset on the Mac side, converting it to DeclROM/interface MUXless PDS speak.

2 - Prune SCSI2SD board back to where it makes the SCSI interface connection.

3 - Develop CPLD or summat to interlink the two chopped back interfaces at the Fast Narrow SCSI 2 capability level of current SCSI2SD

Insane version:

Tweak the DeclROM code of the SCSI-NuBus Test Card to support a Fast Wide controller chipset interfaced to the likes of my 2.5" Savvio SCA Server Drives.

Irrational version:

030 conversion and DeclROM hack of the SCSI-NuBus Test Card and hardware interface to SATA SSD

Way out of my skill sets, but images of some mix-n-match possibilities popped up and just had to float the notions. :blink:

 

[trag]

You would need drivers for the specific SCSI chip used.   The SCSI chip is going to have registers used for various purposes (configuration/communications) that must be addresses, and a driver tells the Mac how to address them.

Also, I don't think you can get the on-board SCSI and an expansion SCSI to work at the same time without SCSI Manager 4.3 or greater.    Or, more accurately, they'll both work, but they'll appear to the Mac as a single logical SCSI bus.   So no duplicate SCSI IDs on either/both busses.

It's been years since I looked into this and I did post about it in a thread somewhere (that's probably lost, grrrr) but there are structures in Device Manager or something like that called Shims or something.  Maybe in one of the threads that Henr3y participated in?   Or there was that guy who dropped in for about 10 minutes who seemed to know a lot about it, but didn't seem to have much patience for actually helping us at a detailed level.  

Basically, to really make it spiffy, you'd need to dig down to that level, replace the apple managers and device handlers with your own custom units or get SCSI Manager 4.3 to load from the expansion card's ROM, or use a ROMinator and get Doug (or someone) to figure out how to add SCSI Manager to the firmware that's being run at boot time.

The software effort should be done some time as it's a hacking milestone that we need for other stuff.   Specifically, the device manager/shim(?) rewrite would be needed to add non-SCSI peripherals that are not disquised as SCSI peripherals.

 

[Trash80toHP_Mini]

HRMMM? I'm not talking about using a SCSI chip. I'm talking about doing exactly that same kind of masquerading as SCSI in the ToolBox routines used to get IDE up and running as "SCSI" in the first IDE Mac generation: Quadra 630 & PB190. No controllers or bus bottlenecks involved on the hardware side.

The IDE implementation on the PowerBook 190 logic board would be my choice as a reference design for getting something going. The 190/5300/2300/1400 all use a bridged 030 I/O bus structure functionally identical to the SE/30/IIsi PDS from where I'm sitting.

I've got an extra 1400 board or two if you wanna poke around in hardware from that direction? Sure would be nice to get a SATA interface wedged into an 030 PDS in the same manner. h34r:

 

[Trash80toHP_Mini]

Check that:

Crazy notion: No SCSI controller in between most of the SCSI2SD and just the DeclROM section of the Test Card Design Example on the PDS.

Insane version: yep you got me, was pretty much talking about an UltraSCSI Controller setup, you make it sound far more than non-trivial.

Irrational version: yep, that'd be the one I've been mulling over, see above. [] All the Duos implement most of the 1400's bridged 030 I/O bus' function on the docking connector more or less. Plus one: DuoDock II/AAUI and minus one: T-REX/PCMCIA and probably some other stuff I'm overlooking.

I'm wondering at what model/processor level the Duo's would have become compatible with the vaporware PCMCIA UltraDock? 68040 almost certainly, as the Blackbirds had PCMCIA. I wonder about 68030 Duos and T-REX and especially about these two ASICS:

From 1400 DevNote:

Bus Bridge
The PBX IC acts as a bridge between the processor bus and the I/O bus, converting
signals on one bus to the equivalent signals on the other bus. The bridge functions are
performed by two converters. One accepts requests from the processor bus and presents
them to the I/O bus in a manner consistent with a 68030 microprocessor. The other
converter accepts requests from the I/O bus and provides access to the RAM and ROM
on the processor bus.
The bus bridge in the PBX IC runs asynchronously so that the processor bus and the I/O
bus can operate at different rates. The processor bus operates at a clock rate of
33.33 MHz, and the I/O bus operates at 25.175 M

Baboon Custom IC
The Baboon custom IC provides the interface to the expansion bay. The IC performs four
functions:
-  controls the expansion bay
-  controls the IDE interfaces, both internal and in the expansion bay
-  buffers the floppy disk signals to the expansion bay
-  decodes addresses for the PC card slots and the IDE controller
The Baboon IC controls the power to the expansion bay and the signals that allow the
user to insert a device into the expansion bay while the computer is operating. Those
signals are fully described in “Expansion Bay” beginning on page 30.
The Baboon IC controls the interface for both the internal IDE hard disk drive and a
possible second IDE drive in the expansion bay. For information about the internal IDE
drive see the section “Internal IDE Hard Disk Drive” beginning on page 16. For
information about the IDE drive signals in the expansion bay, see “Signals on the Floppy
Disk Connector” beginning on page 34, particularly Table 4-3 on page 35.
The Baboon IC also handles the signals to a floppy disk drive installed in the expansion
bay. For more information, see “Signals on the Floppy Disk Connector,” particularly
Table 4-3 on page 35.
The address decode portion of the Baboon IC provides address decoding for the IDE
controller portion of the IC. It also provides the chip select decode for the TREX custom
IC and address decoding for the two PC card slots.

So the Baboon IDE controller runs on an I/O bus and instruction set necked down the 68030 level. The drivers may well be running in a 68040 and up instruction set on the CPUs, but Baboon has no clue about that. It only talks 68030.

I'm half asleep and wondering about a few things:

Didn't the Apple II gang develop an IDE controller card?

Is there an open source example of a SATA to PATA bridge?

Might it be most direct to go straight to the SATA interface from the 68030 PDS using some programmable logic device for the translation and the DeclROM code from the NuBus Test Card design example to make the CPLD/whatever look like the SCSI controller in the design example to the OS? :blink:

zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz