• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

MagicalBus™ - PDS reimagined in 68000/68020/68030 compatible form.

Trash80toHP_Mini

NIGHT STALKER
Been playing around with reverse engineering boards here and there and something twisted up in my head this morning.

Inspiration would be the Radius Magic Bus setup for the SE with FPD and TPD cards:

LCIII-MagicalBus-02.JPG

Pie hole in the front is too constraining for reverse engineered boards headed into the Compact Mac form factor, so here goes:

MagicalBus-11.JPG

Expansion chimney of SE/30 is unavailable for the SE so this is the PCB real estate required for implementing this connector:

MagicalBus-Connector-00.JPG
View here is from front of machine, width is limited in the rear cutout for SCSI and FDD cables:
_ far left section unused as it fits on board but full length card is a few MM too wide for opening
_ plenty of contacts to divvy up any which way for matching up like signals of all three target Processor Direct Slots.
____ even more contacts available for PDS specific signals and a few spares for what-not
_ move SCSI and FDD DIP connectors to baseline card w/ prototyping area?
____ pads for bus transceivers would be provided on "stock" card with prototyping area like ISA ProtoBoards of yore
____ saves real estate on logic board and opens up a world of opportunities! }:-D
_ a pair of SCSI connectors go between controller and internal HDD connector
____ acts as internal cable with termination/switch on third connector up in lieu of HDD
____ offers ability to plug SCSI/SD/NIC implementations into new "SCSI riser" of sorts.

"Stock" baseline riser card required for SE board replacement, but that's no biggie for any such retro hack of Macintosh boards.

Front edge position leaves room on the logic board of SE for full SD card housed in printed external FDD D-Sub connector "SD slot." Traces for SD lead to the "stock" riser card's connections. With board's PCI-X connector mounted close to front edge of chimney, plenty of cubic remains for any kind of stock backplane breakout board I've seen to date.

SE/30 chassis implementation provides width necessary for full universal 64-bit PCI Card Connector and even more shenanigans.

Looking for feedback on how to best approach this. Suffice it to say that no stock Compact Mac boards are involved in this craziness. The only "original" Compact board would be the SE as it's in dire need of a PDS passthru for a VidCard when something nasty like a slot hog of an accelerator plugs up PDS and cuts off access to its single interrupt. 😬
 
Last edited:

Trash80toHP_Mini

NIGHT STALKER
Misleading topic title? Slots aren't compatible, PDS signal locations for each are identical where in common.

Signal positions for Address/Data/ would be set in stone, however many are implemented per CPU in system. Control Signals, where universally present, would be in common on the pinout, however the systems might make use of them. Machine specific signals and Apple's infamous "Reserved Pins" would be mapped to specific contacts, of which there are plenty. I think there might be enough contacts to provide alternate ground wherever such might be wanted?

While implementation of the new connector in any of the three specific CPU target platforms would share this skeletal pinout where common, this is in no way a PDS type conversion mechanism.

This polyglot PDS merely arranges the basics of any given PDS onto a common framework for development on those platforms.

I think that might clear it up a bit . . . or further muddy the waters? :rolleyes:
 

ymk

Well-known member
Interesting idea.

Were I prototyping a PDS card, I'd go for a PDS to (multiple) IDC40 breakout.

The signals should be fine for reasonable runs of ATA/66 cable.

The card could be easily probed, flipped over, etc, without disturbing the host machine.

What I wouldn't do is introduce a 'T' into the SCSI bus. The reflections could cause termination issues.
 

Trash80toHP_Mini

NIGHT STALKER
What do you mean by a "T" in the SCSI bus? The way I'm seeing it would be the equivalent of a multiple connector internal cable or an external chain configuration that really amounts to exactly the same thing. Each device on an IDC cable is a "T" connector in either configuration.

The two connectors on the "stock riser card" would be for SCSI to Whatever devices to plug directly into as if it were a short IDC cable setup.

For your IDC 40 setup you'd need a baseline card with cable drivers. I'd probably go with a standard PCI connector on the baseline board and use a single PCI slot cabled riser for flipping it up and down at times. Normally it would be plugged into an outrigger slot through the SE/SE/30 chassis' side opening.

But I do like your Triple-40 approach to 120pin PDS. ATA/66 has interleaved ground lines, no?

edit: you're looking at prototyping in general, The MagicalBus connector and "stock board" implementation is intended to break expansion outside the dungeon of the SE's solid roof. Use in the SE/30 chassis would be a flexibility/cost saving matter. I've been mentioning using the PCI connector with the edge card setup on 10cm boards for quite some time now. ;)
 
Last edited:

ymk

Well-known member
But I do like your Triple-40 approach to 120pin PDS. ATA/66 has interleaved ground lines, no?

It does, and I'd expect it to carry the bus signals of 68K Macs without additional buffers/drivers.

Are you thinking capacitance might be an issue?

What do you mean by a "T" in the SCSI bus? The way I'm seeing it would be the equivalent of a multiple connector internal cable or an external chain configuration that really amounts to exactly the same thing. Each device on an IDC cable is a "T" connector in either configuration.

If your riser card replaces the internal ribbon cable, then it would not be a T.

I think it's difficult to improve upon the simplicity of either the internal or external connectors.
 

Trash80toHP_Mini

NIGHT STALKER
It does, and I'd expect it to carry the bus signals of 68K Macs without additional buffers/drivers.
I can't imagine that it wouldn't require them, just got up, so muddled ATM. Every PDS card spec I recall offhand requires buffering/line driving as close to the interconnect as possible. PDS spec is limited to driving only a couple of inputs. Driving them across a cable connection would seem to be wildly out of spec, no?

Interesting case would be KILLY Klipped and Socketed interface applications. Haven't looked at those implementations, but now curious. @Bolle would know that answer offhand.

While 030 PDS adapters have no buffering across their trace runs, it's implemented on every in spec card that's connected to them. So your Triple40 system would almost certainly require line driving on the interface card. Implementing at the end of the cable on every prototype card makes no sense to me, but . . . 😴

Are you thinking capacitance might be an issue?
Not specifically at the time, but every interconnect introduces that stuff. Not an electron pusher, so my understanding is very limited.

If your riser card replaces the internal ribbon cable, then it would not be a T.
That would be the case as it would be a "SCSI slot" as seen in SCA disk arrays. But temptation at use them outside that spec at the user end can't be assumed.

I think it's difficult to improve upon the simplicity of either the internal or external connectors.
I think your are spot on in that observation, thank you. KISS principle correction applied. ;)
 
Last edited:

ymk

Well-known member
I can't imagine that it wouldn't require them, just got up, so muddled ATM. Every PDS card spec I recall offhand requires buffering/line driving as close to the interconnect as possible. PDS spec is limited to driving only a couple of inputs. Driving them across a cable connection would seem to be wildly out of spec, no?

I meant I wouldn't expect the cable to require additional buffering, beyond what the PDS card spec does.
 

Trash80toHP_Mini

NIGHT STALKER
Likely not, but putting the drivers on the cable interface card in lieu implementing such each rev. of the ProtoCard would seem to be in order/closer to spec? What Machine/PDS are you working on?
 
Top