Techknight™ - LC PDS NETWORK CARD IN A SE/30?

[uniserver]

who wants to pay $150 for a nic that works in the SE/30… Naaaaaaaaaaaaaaa! Boooooooooooooooooo!

who whats to pay $9.99 for a LC NIC and install it into the SE/30 Yaaaaaaaaaaaaaayyyyyyyyyy!

Techknight might have the ideas/knowledge to make this happen !

Postby techknight » 20 Dec 2013, 22:34
BTW, Off topic: I wonder if LC PDS ethernet cards which are still plentiful, would work in an SE/30. Technically the PDS is the same. Well not the same pinning, but same architecture.

Postby techknight » 21 Dec 2013, 00:49
Now my eyes went crosseyed, but it looks like the LC-PDS. The lower 96-pin portion of it can directly swap over to the SE/30. All the signals are the same, however the pinouts are ENTIRELY different so a remap PCB would be needed.

Only thing that i can see from a quick glance, is A28 to A30 arnt included in the lower PDS. Only the expanded PDS, whereas the SE/30 contains all address lines. Also the LC PDS contains a function code 3 (FC3) which the SE/30 doesnt have. But i dont think ethernet cards use this control pin. Also the LC only has /PDS.BG (bus grant), whereas the SE/30 has both a /PDS.BG and a /BG. No idea what impact this would have.

I believe that LC Ethernet cards, given a pinout-remap would directly swap over, They both map into Slotspace $E

Postby techknight » 21 Dec 2013, 16:07
Well the LC card wont go directly into the SE/30 PDS. the wiring is 110% different, However all the signals appear to be the same. So to make this work, I have to make an adapter PCB with the eurocard connectors on each end, and wiring adaptations.

Might be possible with this inexpensive adaptor card?

http://www.ebay.com/itm/SuperMac-PDS-Angle-Adapter-Vintage-Apple-Mac/151011026594?_trksid=p2047675.m1985&_trkparms=aid%3D444000%26algo%3DSOI.CURRENT%26ao%3D1%26asc%3D17905%26meid%3D3560927830306836012%26pid%3D100012%26prg%3D8262%26rk%3D5%26rkt%3D5%26sd%3D151117999850%26

 

[Trash80toHP_Mini]

I just checked an Asante Mac LCIII NIC against the SuperMac PDS Adapter and it's looking pretty good. I have a sneaking suspicion that there will be no problems at all from the driver side for these cards.

 

[techknight]

Shouldnt have an issue, as an 030PDS is 030PDS. Im sure using the correct connector, it would work on a Duo as well... at least HARDWARE wise.

Now software, that could be an issue if the LC ethernet driver recognizes the machine as not an LC, and fails to load. That would be my only worry which a software hacker would have to fix. But I dont think that will be an issue as im sure the cards DeclROM will fix everything anyway.

 

[techknight]

I did some more research in development notes, and designing cards/drivers etc etc...

I noticed that the LC uses Slot $E for its cards. Unfortunately, the SE/30 doesnt use slot E, as Slot E is taken by the on-board video.

So i dunno what im going to do about that. lol. Also the PDS address map is way different. The expansion address is the same though.

 

[techknight]

SE/30 Address map:

lc address map.png[/attachment]

Note the differences.

 

[techknight]

I guess the way around this is, Disable the on-board SE/30 video by removing its DeclROM, Use that IIsI video card hoping its not mapped in Slot $E. If its mapped in something other than Slot $E, then I can use that, and map Slot $E to the LC Ethernet.

 

[Trash80toHP_Mini]

Since you're adapting the card, you can hotwire the interrupt/SlotManager ID for the NIC to whatever interrupt line is available on whatever machine you're wedging it into, I would think.

It then becomes a question of the drivers. In the case of my Asante NIC, the driver will be looking for "a" NIC to be in whatever slot might be available on whatever Mac in which "a" NIC made by Asante may reside. I don't think the NIC needs to know jack about which interrupt is yanking its chain, only the driver needs to know.

Whatcha' think, bbraun? :?:

 

[Trash80toHP_Mini]

I found it on p.1 of the SuperIIsi™ thread

I finally got my IIsi delivered today to check this out.The 3 normal pseudo-slots for the IIsi are 9, A, and B. Pretty straightforward, and configurable card like the maccon ethernet can be configured for any of those 3 slots. The maccon ethernet can even be configured for slot E on the IIsi according to the documentation, which brings it to a total of possibly 4 usable pseudoslots.

But slots C and D are the trouble. Are these slots recognized by Slot Manager? They have no interrupt over the 030 PDS slot, but if you can get a card into those address ranges, will it be recognized?

I got one of these cards, and swapped A24 and A25, which swaps both address spaces: D and E, and 9 and A. Additionally I swapped /IRQ1 and /IRQ2 (9 and A's IRQs). The swap was performed by connecting pin C27 on the IIsi side to pin B27 on the passthrough side and vice versa, then severing the traces for those two lines on the PCB. Same thing for pin C4 and B4. I threw on one of these right angle dealies to get a more vertical arrangement. There may be more involved in swapping slots 9 and A, such as swapping /IPL0 and /IPL1 maybe.

[attachment=0]IMG_0121.JPG[/attachment]

I then configured the maccon ethernet for slot E and put it in the adapter, and the result is the above picture of it showing up in slot D.

So, it looks like the IIsi can have as many pseudoslots as you want, if you can:

1) Get the card showing up in the right address space.

2) Figure out interrupts.

But, the slot mgr will find it if you can deal with those two problems.

So, no more complaining about the IIsi ROM only allowing 3 pseudoslots.

I still haven't quite digested this info as yet, but I think it's tres applicable!

 

[techknight]

You can change the interrupt line from Slot E to whatever. Thats not the issue. The issue is the address decoder PALs on the Card itself, it wants to decode inside slotspace E, and that would have to be modified.

For example, if i modify the line so the card triggers Slot A, its going to read slot A's address, and it will throw a buss error. Why? because the card wont respond. its address decoders are setup for Slot E.

So In order to make it work, i would have to flip a couple of address lines around physically, so it appears in a different slot space than E, and obviously move the IRQ to the correct spot. Same thing bbraun did.

Now, the DeclROM I dont know if it is hardcoded for Slot E or not. I doubt it as the nubus slot manager will find it wherever it is, The whole purpose behind "plug and play".

According to the LC, I have C, D and E. But C and D are unsupported at least by the LCIII.

Now since I have no idea what slot space to throw it in, It might be better off for me just to make a wire-wrap PDS adapter card. even though i HATE wire-wrap boards. But oh well, it is what it is.

Edit: Also it appears that 99% of the LC ethernet cards, are 96-pin PDS, which means its a 24-bit card. Now the 120-pin slot is 32-bit. So, I assume even though mode32 is enabled, the card should still work in 24-bit mode? Of course the Modes are defined by the SIZE pins on the PDS. So that makes sense.

 

[Trash80toHP_Mini]

I still don't think the PALs on the card need to know Jack. The CPU and driver only need to know the Slot ID and Address Block in which the NIC resides. If the Slot manager can identify the NIC, it can run the NIC, the NIC itself is a relatively dumb I/O device. Only the DeclROM of the Card has any preconceived notions about what the card is supposed to be, what it's supposed to do and where it's supposed to be doing it . . .

. . . and that, can be changed. [}] ]'>

I'm not even certain the LC cards need to know what slot they are in, it's always the same and it's the slot that knows it's $E, wire the LC NIC to a different interrupt that's available on the /30 & si and the Slot Manager will find it and run it. The NIC doesn't run anything or know from where its strings are being pulled.

Buzz the connections on an SE/30/IIsi NIC with jumpers for changing the addresses. Dollars to Doughnuts they only A/B switch the card Slot Address input between whatever address lines it's wired to.

 

[techknight]

I hope the card's DeclROM is using indirect or relative addressing. Because if its using direct addressing for JSR/CALL routines then remapping away from slot E wont work unless its modified.

Reason I bring this up, becuase I remember back in the 6805 days that if i moved the ROM around to a different spot, It would break the operation on any ASM code that used direct-addressed jump to subroutines, or just jump blocks in general. As long as I used within-page calls/jumps I could move the ROM anywhere.

So basically any jump vectors that looked like this: JSR $E000 0000 would break if i moved the ROM to somewhere other than E. (of course DeclROM is above the slotspace).

 

[Trash80toHP_Mini]

The only thing the DeclROM does is tell the CPU/Slot Manager about the card its socket is soldered to, and sometimes supplies the driver to the system, it doesn't run anything.

 

[techknight]

It doesnt tell the mac its required slot-space?

 

[Trash80toHP_Mini]

Nope, the Mac KNOWS in which slot the DeclROM/Card resides by its reply when that interrupt/Slot ID is polled by the Slot Manager at startup.

Just because only $E is available on the LC Series, doesn't mean the card would know if it had been piped another interrupt, an interrupt is an interrupt, only the OS needs to keep which one is for what straight. A NuBus Card and its DeclROM don't care in which PseudoSlot Address they're dumped, they just need to respond when that particular chain is pulled. LC Cards are likely only set up for $E because that's all LCs have . . .

. . . until Herr Braun gets aholda' them. }

 

[Trash80toHP_Mini]

Actually, each NuBus Slot has its Slot assignment built in, I'm going to make a W.A.guess that it's on the same pin for each slot . . .

. . . just woke up, gotta make some coffee. |)

 

[uniserver]

offtopic

so i would assume the IIsi can do nubus as well because of nubus transceiver… on the LB?

Nubus network cards are so much cheaper and easier to find…

seems like you can get IIsi Nubus adapter cards for reasonable too.

just woke up too, slept like 10 hours... yesterday with the IIsi ram thing was an all nighter…

I just went to the coney island and got breakfast for the the wife/kids/myself.

 

[uniserver]

oops i was wrong. the IIsi nubus adaptor card is quite sophisticated….

[attachment=0]Screen Shot 2013-12-27 at 8.22.06 AM.png[/attachment]

DING DING DING!!!

Light bulb….

would this work in the SE/30 ? maybe heat gun off the fpu….

 

[Trash80toHP_Mini]

Good notion that, uni. Bunsen and I have been poking around that very hack for years.

Back on the original topic: I'll have some very interesting piccies to post later for this same hack in the IIsi! :approve:

< . . . resistance is futile . . . >

Anybody got a trashed SE/30 or just a MoBo for trade?

 

[uniserver]

Anybody got a trashed SE/30 or just a MoBo for trade?

make sure you specify ( working )

i'v got 5 dead se/30 mobo's

3 empty se/30 cases

I think when they are dead dead like the 5 i have… and the one that techknight has from japan…

the GLU I/C cooks… from the cap goo shorting just the right items.

thats is my hunch.

working but:

no sound = GOOD

no SCSI = GOOD

Gangster stripes = GOOD

garbled Video even if you can make out a some what flashing floppy disk icon = GOOD

Dead ADB ports = GOOD

working but NEEDS CAPS(leaking) = GOOD

battery explosions are almost always - bad

this screen = bad

[attachment=0]Screen Shot 2013-12-27 at 9.45.23 AM.png[/attachment]