Farallon ETHERMAC LC NSC w/NuBus drivers in the SE/30 PDS?

[nickpunt]

Looks like 1996 according to JDW's Flickr photo. Dunno about the rest!

 

[Trash80toHP_Mini]

Pulled a pile of stuff out onto the dining room table to see what was up with that Address E (Slot $E) setting using the IIsi board @maceffects sent my way for the project. All is good with it using BMOW's ROM-inator II boot disk w/TattleTech handily on board. It's been a while since my IIsi boards have been usable (recap fraidycat still  [:I] ) so I relearned a couple of things.

While RBV/Vampire Video uses memory mapped to NuBus/PDS $E, the RBV system shows up at location Slot $00, which is the Mac board itself.

With no sense lines/display detected on the Video port at startup, memory for Vampire Video remains unbuffered and RBV disappears from the Slot $00 function listings.

With NIC set to $E when booting with display attached to either my SuperMac Card or to the RBV/Vampire Video setup it fails to make an appearance in the NuBus/PDS report.

So came up with more questions, no answers. Asante's $E setting appears to be a canard or might have been implemented for an unrealized third generation of 030 PDS machines equipped with VRAM and no Slot $E entanglements imposed by Vampire Video?

Alternately, I'm wondering if the drivers were loaded it might show up in the IIsi's Slot $00 motherboard function listings as a built-in Ethernet setup with or without RBV in the picture? That's unlikely as well as being well beyond the scope of what I've got lashed up for today's session.

Before heading off on that wild goose chase, I need to figure out the Address jumper mechanism of the Asante NIC et al. WAG would be that one of the involved lines for using $E heads to a pin reserved by Apple for adding a fourth interrupt to the 030 PDS for a third gen 030 PDS machine?

Anybody got a handle on the address setting  jumper setup in general for the 030 PDS?

_____________________________________________________________________________________________

Q605 compatible HDD booted just fine the first time, but I should have installed a SIMM before testing it. Got a 1MB free memory required dialog box when I tried to run TattleTech from stock RAM. After installing memory the drive appears to have died on me. I've got a few others handy, but they're all 7.1/incompatible. I remain curious about where things show up on the NuBus/PDS report for the LC PDS.

 

[Trash80toHP_Mini]

Well, looks like there must be something to setting the Asante MacCon Address Jumper setup to "E" after all given this result:



Couldn't repeat that result, came up with this screen thereafter  .  .  .



.  .  .  which looks to be the same as the pattern I get when installing a IIsi NuBus Adapter in the SE/30.

On a tangential note for a change: also got that same pattern running with the SE/30 Video ROM pulled, but the boot screen came right up up on the external display.



Display is out of whack at 19" on the LCD, but as you can see, SE/30 internal Video in Slot $E makes no appearance when the ROM is pulled.



With the Video ROM restored, the external display remained as the Startup Screen and the SE/30 Screen was fine as the aux screen.

ISTR seeing reports that the SE/30 tries to use internal video as the startup screen, even with the Video ROM pulled, but happily that doesn't seem to be the case! So the door appears to be cracked open a bit for developing a "relatively simple" internal only Grayscale Card for the SE/30. Just maybe?

But that's tangential information developed at the end of today's Asante MacCon/Slot $0E research. Been poking around at that GS Card notion in some loosely related threads.

 

[maceffects]

Was that testing a card with the jumper/switch or did you try to use the prototype card?   Looks interesting.  I’m certainly no expert but I think it sounds like there must be a way given the results. 

 

[Trash80toHP_Mini]

I was just testing that oddball "Address E" setting on the MacCon NIC for the IIsi. Need to read up on that Mac again, coming up empty on my guesses so far. I'm running down all possibilities for testing the adapter without major logical intervention. The mixed results were more interesting than promising today. Need to read over my posts after the brain comes online again in the AM. :blink:

 

[nickpunt]

That's interesting. As Gorgonops mentioned earlier in thread:

That stuff is *electrically* going to conflict with any other PDS device you try to put there.
 

Perhaps that's what you're seeing on the main monitor.

Not sure if I understood all the permutations you tried, but did you try pulling the video ROM *and* setting the MacCon to address E? Or would that be pointless? 

 

[PotatoFi]

I'm struggling to follow the conversation, which is my responsibility so I deserve nothing here... but if I'm reading this right:

1. The wire wrap is complete
   
2. You've installed a IIsi NuBus Video Adapter
   
3. It totally works, except there's a potential conflict with the SE/30 ROM and internal video no longer works
   

Am I reading this correctly?

 

[Trash80toHP_Mini]

I was exhausted when I posted. so I can't follow it either! :lol:

Wire wrap not finished, playing around with test cases for using it.

No NuBus Adapter involved, only mnentioned it because of the (identical?) screen pattern.

Things work strangely playing with everything BUT the wire wrap adapter/LC NIC

Not sure if I understood all the permutations you tried, but did you try pulling the video ROM *and* setting the MacCon to address E? Or would that be pointless? 

Good Question! Never tried that with the VidCard/External Display.

Machine state on first pass last night: Video ROM in place - MacCon set to address E. No VidCard/External Display

View attachment 28365

Just tried your suggestion with a new twist, now I'm sure I didn't try it last night!:

Machine state: Video ROM pulled, VidCard/Display in place!



MacCon shows up in place of internal Video in Slot $0E, so I've got promising results after all for testing the Farallon LC NIC and an immediate reason to get back to wrapping! Not expecting full function, but hopeful of seeing its DeclROM in the report above at $0E. []

 

[Trash80toHP_Mini]

Looks like SE/30 Video implementation follows Apple's PseudoSlot design conventions and the IIsi doesn't.

Anybody know offhand if IIci Internal Video shows up at $00 as a function of the System/Logic Board as in the case of the IIsi or if it follows the PseudoSlot model, having its own Slot ID assigned as in the SE/30? I'm guessing the former.

 

[maceffects]

So with the video ROM pulled it should yield favorable test results?  Would adding a custom ROM be of use?  Just trying to brainstorm. I have some machines I can use for testing too if we get to that point. 

 

[Gorgonops]

To reiterate again, it makes sense that you're not seeing a problem in Tattletech with the video ROM pulled out, but you *still* might very well have an electrical conflict from the video circuitry sitting on the bus if whatever you're sticking in there overlaps with its I/O space. (Remember, all I/O is memory mapped with the 68k CPUs, and ethernet cards also do have real RAM on them.) You could potentially damage your machine if this is the case.

(Scenario: Ethernet card has a RAM buffer on it that gets filled when an ethernet packet comes in. This RAM buffer overlaps with the 64k of VRAM in the SE/30, which may not be seen as a video buffer without the ROM pulled but is still sitting at its normal address. Packet comes in, driver goes to read it, and both the video RAM and the ethernet buffer RAM try to answer. Electrical sadness ensues.)

 

[maceffects]

That is an interesting situation, I think you may be right.  I am willing to sacrifice one of my boards if it will lead to learning something useful.  

 

[Trash80toHP_Mini]

To reiterate again, it makes sense that you're not seeing a problem in Tattletech with the video ROM pulled out, but you *still* might very well have an electrical conflict from the video circuitry sitting on the bus if whatever you're sticking in there overlaps with its I/O space. (Remember, all I/O is memory mapped with the 68k CPUs, and ethernet cards also do have real RAM on them.) You could potentially damage your machine if this is the case.

I'm actually expecting conflict to be the case, which is why I said I don't expect full function for the LC NIC even if it by some chance does show up in that slot space. I'll be happy if I can get its DeclROM to show up anywhere at all in that report.

Thanks for the warning, didn't realize such conflicts could cause damage to the circuitry? Wasn't planning on loading drivers for testing the MacCon card in that configuration, now I definitely won't.

 

[Trash80toHP_Mini]

Has anyone got a handle on how "Address Set E" is generated from the jumpers on the Asante NIC? Haven't buzzed the lines yet, but I have a sneaking suspicion this setup leads to GAL/PAL intervention? The Spectrum/8 appears to be hardwired to "Address Set 9." ISTR other cards being set up for only two "Address" Sets using the same two jumper setup.

Very curious, time to break out the continuity tester is seems, but any ideas/theories would be much appreciated.

 

[maceffects]

I know that @Bolle is developing a much better board, but I still hate leaving projects unfinished.  I'm going to try wire wrapping this myself, once I can find the right wire wrap capable 120 pin right angle card.  Or maybe I'll just make a PCB to try it.  I'll be using the approach @Trash80toHP_Mini discussed on page six.  Nothing ventured, nothing gained. 

 

[Trash80toHP_Mini]

We both hit a brick wall in the Technical Thread related to this project, but I can't find it ATM?

I can finish up my board and send it to you. That's the easy part, it's connecting the proper lines to the proper pins on one of my PLCC-20P-T sockets that's been holding things up. If someone can give me the schematic for that I can wrap it up quickly, so to speak.

Equations for and burning of the GAL for that socket as outlined in the TechThread are way above my pay grade. When I started here, theory was that we only needed to swap a couple of lines and run them through the inverter I have in hand for the socket that's already installed. Then we hit that GAL roadblock in the other thread. That'd be Hadrian's wall from my point of view, someone on the Roman side needs to take over, they were pretty good at engineering.

Bolle's new NIC development is fabulous, but probably not going to affect developments here. 10bT performance should be equal. If anything, the LC boards are more advanced in terms of being ASIC based, which of course makes them impossible to clone! The interesting thing though would be comparing the input load requirements of the those later generation technology ASICS as compared to the logic on Bolle's MacCon clone. If appreciably lower, the LC NIC may well make using it more reliable than MacCon/other 030 PDS NIC when pushing the limits of the SE/30 PDS expansion envelope?

There's also the notion on the previous page about adaptation of an LC NIC to the SE? Farallon didn't happen to have done a version of this NIC for the SE? If not, driver development would be the same thing. I'll tilt at windmills, but thaqt Hadrian thing is quite another thing entirely. :-/

edit, fabulous prototyping board! :approve:

 

[macdoogie]

Wow, you guys have some brass bolt-fastening hardware! However you may be approaching this from the wrong "direction". A lot of things to address here, since I'm coming in late.

1. Getting the device/card "on the bus" with no contention is one thing, but that's no guarantee that the drivers will load. The PDS slot for the LC is hardwired to the 00EXXXXX/FEXXXXXX address space and designed for machines that intended to only have/fit a single device in that space. There's a good chance that the drivers are hard coded to expect the card to be at that address space. You can probably successfully remap the card to respond to a different address as you are attempting here, but if the driver always looks for the card at $E, it'll never find it at it's new location. This stuff predates plug-and-play and the PDS slot is more of a pseudo slot as described in the Apple Reference guides and dev notes. Same thing with the IRQ, when that fires, if it's not mapped to where the driver expects it, it may not get serviced.

2. I highly suspect that NuBus drivers are not going to talk to the PDS version of the card. While NuBus is apparently not a true "bridged", isolated bus in that it shares address and data lines with the processor bus directly. There are the side state machines(control logic really) that the Nubus Driver likely uses to initiate transactions and such. I'm not a NuBus expert, but I came to Apple in the late PCI architecture days and well into the PCIE generation. I worked on the FireWire controller ASIC and did a lot of I/O integration so I have pretty comprehensive knowledge of processor and I/O architecture. I suspect the PseudoSlot scheme that the IIsi and SE/30 implement rely on the card's hardware emulating the effect of both the host and card NuBus state machine controllers, since those were likely just memory mapped devices in the NuBus Macs. The PDS card probably expects a different protocol.

3. To address some machine architecture comparisons earlier in the thread(where is was said that the MacClassic was an offshoot of the SE/30 and such), a more accurate lineage is this:

MacClassic = Cost Reduced Mac SE(not /30) with a QFP or PLCC(I forget which) 68000 chip and a slightly improved version of the BBU among other things like lighter weight and not being made in USA.

MacClassic II = Cost Reduced and Performance reduced "Sort Of" SE/30 only not at all really with only a 16-bit data path and no expandability other than memory and the FPU card.

Color Classic = Mostly related to the LCII architecturally. Again both have a 16-bit data path and I believe share ASICs. I worked with engineers who did these machines(and ASICs), but they were getting fuzzy on the details as the decades passed.

Color Classic II(FWIW) = Based on the LC-III architecture, with a full 32-bit data bus, albeit at 33Mhz instead of 25Mz, so actually a bit better than the LC-III. I think there was an LC-III+ that was 33Mhz? There definitely was a Performa 460 that was(The 450 was the 25Mhz model).

4. The biggest problem I see with any thoughts of porting these cards to the 68000 machine is that the drivers really won't know what to do with the hardware, probably would not even load(I suspect the machine Gestalt has some influence) and if they were forced to load in some way, they'd throw a trap the minute some 68020/68030 instruction was encountered since the drivers were probably compiled for post 68000 CPUs.

Not trying to discourage the effort here. Half of the reason to do this stuff is for educational purposes! Personally, I would have started by putting a logic analyzer on the bus and sniffing out the early accesses to the card space. You should be able to see either the OS or Tattle Tech "sniffing" out the declaration ROM to see what's out there. The thing to find out is whether the OS or drivers "poll" all of the virtual slots, or whether they go straight to, say $E, in the case of this card. Note that despite all this talk of Slots and PseudoSlots, all of this stuff, including the captive I/O on the logic board are just memory mapped into the CPUs address space. It's not too difficult to "relocate" the hardware, but if the OS or Drivers don't know where to look for it, or if the IRQs are hardcoded(in the software) then still won't work with stock drivers.

So far, my hacking experience with these machines involves making a Mac SE PDS-Slot ROM card(Anyone interested in this project? I can start a thread) as well as making interposer cards for both the SE PDS Slot and LC PDS slot with logic analyzer headers on them. I can make kits available for those as well if people are interested. Oh yeah, I've also made a small PCB replacement for the IIe Y-Cable and a replacement button board for the Color Classic with an RGB LED and a way to program the color mode with the panel buttons, while still allowing them to function as volume and brightness I plan on building a batch of these up and offering them for sale as well. Should I start threads for those? I've been keeping this stuff to myself for the last few years, simply because I've been super busy.

Anyway, I'm willing to help answer technical Q's and provide lab tips. Not sure I have a lot of bandwidth to do extra lab work at the moment, but maybe I can provide guidance where needed.