Dual / Multi LC PDS slot

[nickpunt]

Anyone ever try doubling up on LC PDS cards? Would they just work if stacked?
 

There have been multi-LC PDS solutions in the past, specifically MicroQue's TopHat which supported 4 LC PDS cards in a weird thick LC basement case. See my gallery here of images of that. What's interesting is in the TopHat brochure, it states (emphasis mine):

Many PDS cards work simultaneously with others. However, the LC and Performa were designed to use only one PDS card, so problems sometimes occur when you use more than one card at a time. To prevent conflicts and to make card selection easy, MicroQue provides a convenient card switching system on the back panel, so you can turn each card on and off individually. For current PDS card compatibility chart, please contact your dealer, or MicroQue.

This strongly suggests that it's possible to use multiple LC PDS cards, so long as you can physically connect them. Something I've wondered is if I could stack another LC PDS card atop my IIe card, simply by plugging in a female-female EuroDIN adapter on the solder through holes from the IIe card, then plugging another card into that. For demonstration purposes using a right angle f-m eurodin: 

View attachment 5947

I guess @Trash80toHP_Mini you're probably the one to ask, since you were in that thread mentioned in my gallery

 

[Trash80toHP_Mini]

The problem is that the LC PDS is Slot $E and all cards for it are hardwired to work in that slot only. Sounds like the TopHat employed an A-B-C-D switching setup for using them independently, one connected to the $E interrupt line at any time with the others floating offline on what looks like a multiple slot bus, which it pretty much is, but not really?

Look at it as if it were a Centronics-50 A-B-C-D  Rotary Switchbox with four devices set to the same SCSI ID on each of four cables. Any one of the devices is selected by the switch connecting all fifty lines at once from any of those four cables to one connected the Mac's SCSI port. I think it's probably easier to imagine the device connections splayed out in four different directions from the Mac in that topology.

The TopHat has all 96 PDS lines but one set up on the PCB in a linear bus fashion. It only needs to switch that single $E interrupt line around from any given card's slot to another to achieve the same result.

Funny you should mention the IIe Card, that's probably the single exception to the rule. It's a single board computer that just happens to be located in the LC PDS. In your other thread I found out the LC and IIe are entirely different computers, taking turns running in the same box. The IIe Card ought not to notice a piggy-backed, dormant LC expansion card when it's running? The LC probably won't notice a dormant IIe computer acting as a riser for it's LC expansion card when it's running?

Dunno, trying it out in a 630 series Mac with the headroom to do it like in your pic is the way I'd do it. I'd probably hack an LC NIC for the test, they're expendable. [}]

edit: Missed your Mystic/Video Card/IIe Card thread. I hope the above makes sense, it's really late here and I'm a bit doper than usual. With a printed back panel you won't need to worry about connector offsets. Not up on CC lore, Mystics got CS? I'd still try killing a NIC off first off. []

 

[Trash80toHP_Mini]

Got to get some sleep, but was curious. Copy you quoted, pic of the independently acting switches and markings on the TopHat PCB connector pins all suggest that some cards can work well enough with others. I wonder if that chart is available anywhere? I also wonder about CS Slot Location and what that will screw up? That wrinkle arrived well past the LC's era.

 

[NJRoadfan]

MicroMac had a multi-LC PDS adapter too. You can likely share interrupts similar to how two serial ports piggybacked on IRQ 3 or 4 on PCs. NICs generally don't use interrupts either. Looking at the pinout, there are additional SLOTIRQ D and C pins on the 32-bit LCIII version of the slot but none of these upgrades appear to have taken advantage of this.

Overall its a fairly "dumb" slot since its supposed to be a mostly direct pinout from the CPU, similar to PC's ISA or Amiga's Zorro (although both of those also have DMA controllers).

 

[Trash80toHP_Mini]

Yeah, last night I was thinking the video card probably doesn't use an interrupt and that a "third" card might be added into the mix. Interesting that a NIC wouldn't need an interrupt, that might be the explanation why the NIC Daughtercard for the Futura II/SX VidCard combo works in one NuBus Slot on its single (unnecessary to begin with?) interrupt? *****

Yes, those interrupt lines are implemented in the slot connector. But if you check the table of signal functions in the DevNote for any given LC or LC III equipped Macintosh you'll find the parenthetical: Not supported in the (insert model designation here) exception. Thanks for the PC serial port info, never hear about that one.

***** If that's the case, yet another tangential topic deserving its own thread pops into mind. It may be possible to wire up a "Parasitic LC NIC" to any NuBus VidCard?  TOBY is a horribly limited VidCard, but its ASIC and memory expansion sockets offer convenient wedge points for testing such a hack. Adding a "Slot E LC NIC" to the Slot A/B/C limited IIcx would be a neat trick.

Now I'm wondering if an LC NIC would require space in its host's Slot $E memory map? If not, a "Parasitic LC NIC" hack for the SE/30 might be greatly simplified as there might be no conflict with its Slot $E Video subsystem? The same might also be true in the case of the Slot $E Vampire Video subsystems of the IIci and IIsi?

Coffee is kicking in, so morning muse mode is fading to black. Where a NIC fits into I/O and memory map is the critical question here. My IIfx PDS and SE/30 sockets are eagerly awaiting some answers.  :blink:

@Bolle little help here? @Gorgonops whatcha' think? I'm lined right up in your shoot down sights again here. []

edit: one last bit of insanity just surfaced. @trag might a Parisitic LC NIC be added to the daughtercard slot on your Futura II/SX? The NICs for those things are 23K Unobtanium.

 

[NJRoadfan]

I am not too familiar with NuBus, but if it uses memory mapped I/O, its likely the slot spaces are hard wired on the logic board (ID0 to 3 pins) eliminating the need for address decoders like ISA has (similar to how the Apple II's slots are mapped). PDSes aren't so restrictive, its likely you can add an address line decoder for additional "slots" besides $E.

Also, some ethernet chips may require an interrupt request line. Others like the cs8900a can be run in polling mode, which has a performance hit.

 

[Trash80toHP_Mini]

Yep, the four Slot Identifier bits are independently wired at each NuBus Slot. Apple played a big part in establishing the IEEE 1196-1987 NuBus "Standard" all the while setting the Mac II up with that sub-standard implementation. :

I'm wondering specifically about @maceffects dragon's hoard of LC NICs here. IIRC they're Farallon or some other make that has a NuBus Card/Driver equivalent? I'm thinking that the NuBus driver, if not the LC NIC driver might be workable in the SE/30. Just asked Bolle about the interrupt/memory map thing in PM, but he's probably asleep.

Depending on his or someone else's answers, there's a very slight possibility of using the M->F Breadboard Jumper Slot Riser Hack (somebody raise their hand for pops here :approve: Post a link too, that one is one very elegant solution for this project) to wire one of his NICs up to an RCPII/IIsi VidCard's passthru slot. That would allow testing in the SE/30 bench setup with and without the Video ROM mucking things up in Slot $E.

 

[maceffects]

Yeah, I have tons of those cards, and they are Farallon and share the same driver as the NuBus ones, not sure about the SE/30 version, but I think they are slightly different.  It would be nice to find more uses for these cards. 

 

[Trash80toHP_Mini]

I was looking for the spec sheet on the MX98803AFC controller on an LC NIC I just found in one of the three PC-Compatibility Boxes I'm repacking. Only worthwhile hit to far is an Applefritter Thread.

Do you or does anyone else have the Farallon SE/30 card or a pic of it handy to confirm presence of jumpers? My MacCon IIsi NICs have a pair of jumpers that are probably for Interrupt/Slot ID settings as do the RCPII/IIsi VidCards. As I loathe network setup, I've yet to test one of the IIsi NICs, so I haven't the foggiest. I'm now thinking this has been another wild goose chase unless we can come up a card that runs in the "polling mode"  @NJRoadfan just mentioned.

We're probably right back to hoodwinking the LC NIC so that it starts working from one of the three available interrupts of SE/30 and IIsi, but at least we've got a simple approach for a test setup by using the M/F breadboard jumper methodology out of this mess. :mellow:

 

[Trash80toHP_Mini]

Looks like it might be time for a reboot thread. Found a couple of others, trag's comment has one good use for your metric crapload of NICs. [}]





Closely related project developed into LC NIC in SE/30 tangent:





I'll see if I can dredge up the more of the topics, as this notion has come up A LOT over the years. Has anyone got links other topics handy? Panning the flow of threads for unobtanium nuggets might be in order.

 

[trag]

edit: one last bit of insanity just surfaced. @trag might a Parisitic LC NIC be added to the daughtercard slot on your Futura II/SX? The NICs for those things are 23K Unobtanium.

I doubt it would be an easy adaptation.   It all depends on what kind of GLUE is used to hook the E-Machines network card into the NuBus Futura card and how different that is from the LC slot GLUE.

 

[Trash80toHP_Mini]

I've got one of each flavor if you'd like to try. That connector makes for a sweet wedge point. [}]

 

[trag]

I've got one of each flavor if you'd like to try. That connector makes for a sweet wedge point. [}]

I would like to, but I'm not going to any time soon.  Struggling just to keep up with the laundry at this point.

 

[Trash80toHP_Mini]

LOL! I hear you there. For me it's the dishes ATM, ick is beginning to make the transition to YUCK!!!!!! Gotta keep fixing stuff up in the apartment and painting on the terrace while the sun shines, then its on to downsizing and organizing what I'm keeping when it goes down. I've been forgetting to eat.

__________________________________________________________

Just to get it straight in my head:

Fairly solid plan developing ATM is to bolt an LC NIC and RCPII/IIsi up together on a piece of plexi. I only have the one SE/30 board so I'll be testing in one of my several IIsi boards if I can ever get around to recapping them. : Running off the Radius Vidcard only and with no sense pins detected at startup, Vampire Video at $E disappears from TattleTech/SlotInfo reports. Gotta hijack the $E interrupt from the logic board from somewhere convenient, but that shouldn't be a problem. Everything else will be connected using short M->F breadboard jumpers per Slot Conversion Diagram.

View attachment 12609

LC PDS conversion is greatly simplified as compared to the LCIII PDS. Ratline diagram was for a converter, only the bold lines could be done straight thru in copper so I was shifting to horizontal/vertical methodology and then got side tracked by something else  .  .  .  imagine that! : I'll do another one pared down to LC sometime showing the connectors of the two cards aligned correctly. First I've gotta look at the board and schematic to purloin $E. [}]

Gotta get my hands on a Farallon LC NIC to see if the NuBus drivers will work in that kluge, but first I need to get some jumper wires in from China to build that rat's nest. Faster and easier than wire wrap that way, but I might need to go the far more reliable route using header strips in the RCPII/IIsi's passthru. Gonna have to be scheduled for the oppressive heat and humidity of AC summer here or maybe come next monsoon season.

 

[NJRoadfan]

I'm assuming the need to tap the $E IRQ is because the driver doesn't support anything else? The IIsi PDS does have other IRQ lines available (that is if they are electronically compatible with what the /SLOTIRQ.E pin in a LC outputs).

 

[Trash80toHP_Mini]

HRMMM: The diagram above is the only one I recall developing thus far with just LCIII and IIsi. Here's a simple diagram with Address/Data grayed out but with the IIci Cache slot as a distraction in the middle:

View attachment 12055

The LC PDS is by definition Slot E and any card for that slot is hardwired to SLOTIRQ.E as it's the only connection available in the LC.*****

The NuBus drivers will support a NIC any of the of the six slots in the Macintosh II, IIx and IIfx from Slot $9 thru Slot $E.

So we're stuck with testing the LC NIC in Slot $E to see if the NuBus drivers will work at all for the LC NIC in that location in one or the other of the 030 PDS twins. After that would be playing ID games until we can fool any LC card about where it is as opposed to where it thinks it ought to be and drive it at that location. In the IIsi and SE/30 target machines that would be Slots $9, $A and $B which just happens to be Macintosh IIcx NuBus Slot availability. Apple threw a wrench at the works by using a $E PseudoSlot for SE/30 internal video, they managed to get it wedged firmly into the works of the infernal Vampire Video of the IIci and then doubled down, using a pipe extension on that wrench to well and truly bork the single, adapted NuBus Slot limited IIsi.

Be that as it may, booting the SE/30 with Video (Decl)ROM pulled and starting the IIsi up with nothing attached to the Vampire Video Connector are the two paths to testing Smaug's hoard of LC NICs in the target machines just to see if they might work in the simplest of implementations on an ad hoc basis. May work, may not, dunno, techknight posted some possible roadblocks to the approach some time ago, but it's best to just try the damn thing out!

*****The LCIII PDS was expanded to fill memory address lines out to the full 32-bits. While /SLOTIRQ.C and /SLOTIRQ.D lines were added to the rows 35-40 slot extension, those are tagged as "not supported in the Macintosh xxxx" in the signal descriptions of the DevNotes of any Mac with any type of LC Slot I've seen, though I'd love to hear differently.

Personally, I think that when the bat rastards at Apple marketing found out developers were building multi-slot expansion chassis for the LOW COST bottom of the barrel LC line they had engineering put a stop to any possible use of /SLOTIRQ.C and /SLOTIRQ.D so nobody, but nobody could build a fully functional three slot FrankinTosh. Me, I'm wondering if that was done at the trace or ASIC level by the hardware engineers or at the ROM level by the firmware wizards. The former we can't do anything about, but for the latter we have sorcerers with mad skillz in that area hereabouts. I'm not anywhere even close to either, I just play around at torquing the Legos in the Block Diagrams all outta shape.  [}]

 

[maceffects]

@Trash80toHP_Mini if you want me to send you a free new LC PDS card with driver disk, send me a PM. 

 

[Trash80toHP_Mini]

Thanks! You just got me to reread my last post. On second thought, if the marketing vermin had the hardware guys hobble /SLOTIRQ.C and /SLOTIRQ.D at the trace level, that we can patch wire! But I seriously doubt it would be that easy, I'm guessing a ROM patch would be required.

Iit would be great if some of you would buzz those connections on various LC Slot Macs to see if the lines for those interrupts are there just to see. My LCIII is out on loan and I expect those to be wired up anyway, My Q630 board is AWOL and wire wrapping the test kluge is probably time better spent than buzzing my LC475 board anyway. :mellow: The Q630 series would be the main line of inquiry anyway, it's the only form factor with head room to spare for a stack of LC PDS Cards.

 

[Trash80toHP_Mini]

View attachment 26733

View attachment RosettaCache-01-000.PDF