Project30 - IIsi NuBus adapter in SE/30 trials - take 3

[Trash80toHP_Mini]

Worked out the grand scheme of things in two previous incarnations of this guerilla campaign. So the goal is in sight and it's been time to do the work.

Took one big step back from the mock-up to begin a serious project: Reworking a riser/right angle adapter card for the IIsi that I knew wasn't going to work as was. I tried it on a lark way back when, got nada, but figured it wasn't a true PDS Passthru as it had been specifically designed to adapt a SuperMac Video Card to the IIsi PDS form factor, not as a general purpose PDS Angle Adapter/Passthru kind of deal. Turned out to be right, lots of control signals missing, not hard to patch, but I got bored. :

Was looking at the assemblage and wishing this or that weren't in the way and it hit me: who needs a friggin' chassis for a test setup anyway! HEH!





Bolted the clear inner shell from a test rig for another backburnered hack from waaaay back when up to the Chassis/AB/PSU of the SE/30. Marked out the four drill points for mounting it to the AB/PSU and then bolted it right up without that dang chass!. It was plenty stable with it's snaggle-toothed edge tucked up inside the nicely notched rearedge of the front bezel, but figured I'd still manage to knock the thing over right on top of the MoBo. Fabbed a quickie bottom plate that inserts into the slot at the bottom of the back of the Front Bezel, duct taped the three sides of an empty box together and went to town!

Was ready to plug the NuBus Adapter in when I remembered that blasted FPU, so I fired up the hot air rework station and promptly set the smoke alarm above my head to wailing (note to self: find a new spot to do hot air rework) and beat a hasty retreat, smoking card in hand into the bathroom to turn on the exhaust fan. Curses, foiled! Dropped the smoking card (fresh from its anti-static bag) to grab the one I've been pawing over for months without an anti-static care in the world and chopped that FPU's legs off with a handy knife! h34r:

It was ugly, figured it might have a short, but I popped it in and fired up the SE/30 anyway to a great death knell. Cleaned up the mess a bit and got a startup chime and this for video output.

View attachment 11408

Installed and hooked up the Radius Color Pivot II/IIsi to the KDS/Radius LCD I've been using for testing, checked this or that, made sure the magic smoke was still where it belonged and now I get repeatable results. Thought maybe I'd get video on the LCD despite mobo video borkage, no joy there. Hunky-dory without the IIsi NuBus Adapter, startup bong and the same memory interference pattern on the monitor with the NuBus adapter plugged directly into the logic board or into the RCPII/IIsi with its full on PDS Passthru implementation. Nice thing that!

Here's another shot of what I get:



One of the times I removed the NuBus Adapter to reboot without it, the pattern had persisted in video memory, being wiped away with the successful reboot.

Now if just ONE of my FOUR friggin' IIsi boards worked to verify function of this particular NuBus adapter I'd say I had some interesting results to show, but this way it's only a sneaking suspicion:

NuBus in SE/30 will work but it will require a Slot E video memory conflict fix along the lines of the active component PowerCache adapters.

That's my WAG and I'm sticking to it! :blink:

 

[Trash80toHP_Mini]

OUCH! forgot to right-size the pics, sorry:



View attachment 11415





View attachment 11418

 

[Trash80toHP_Mini]

RuhRoh! :wacko:

View attachment 11433

The Good, the Bad and the Fugly© :rambo:

Not even close to figuring out which is/was which, but it got very interesting overnight. Now it's time to figure out which what will be. So in that light:

Who's seriously interested in having one, possibly two NuBus Cards installed within their SE/30?

Who might be interested in getting involved in an open source development project for the above?

I've been at the Fugly stage of this hack on and off, very literally, for ten or more years and that's about to change. The above pic shows HackLabBTNC™ in what is very likely the most organized hardware state achieved in any of my hacks to date.

That's about to change, I'll probably be spending the time earmarked for the project in developing an exoskeletal, clear plexi case for my SE/30 before continuing the project. As it stands, I can't really run it continuously because its stock cooling system (overburdened is about as PC as description as you'll ever get out of me) is well and truly borked in its present state. It won't be nearly a elegant as the PlexiProtoMac, but it will have to do.

At any rate, it'll bolt up to the front bezel and it will have a slide out logic board drawer. Not to the back, it'll slide out the side just far enough to cycle expansion cards in and out while protecting the CRT assembly from my grubby little paws. A pleasant side effect will be making it safe for others to do so without case disassembly or risk of shock to any but the most ham-fisted possible operator.

I'll be keeping it an IKEA correct design (given the remotest possibility of calling that peculiar manufacturing/flat packing/shipping/user assembly hostile model correct) for others to help in refining for laser cutting and distribution.

Enough for now, it's time to get ready for work. If I return to notices of interest it will be a welcome change. I've got  Delta Change test results to post as well as taking a first stab at putting together my research materials list.

Is anyone out there interested?

p.s. forgot to mention that the auxiliary display is a totally unnecessary, very optional feature/complication.   

.

 

[Trash80toHP_Mini]

This is the original screen artifact pattern. It's repeatable. I added the passive adapter prototype into the mix and got a new result that is repeatable:







Agent of change the change:





Don't worry, it'll be a level higher soon, I've got a replacement RCPII/IIsi readied for the wrong way connector installation upgrade.

Reasons for thinking it's a memory conflict: How the Macintosh II NuBus Works

 

[Trash80toHP_Mini]

Here's why I'm almost positive a memory conflict is to blame:

DCaDftMF3e: pp. 316-317

Once the PDS adapter card has been installed in the Macintosh IIsi, the expansion
connector is physically and electrically identical to the one found on the Macintosh SE/30.

However, there are two major design differences between the Macintosh SE/30 and the
Macintosh IIsi. These differences may prevent PDS cards developed for one MC68030
computer from working in the other.

The clock speeds of the Macintosh SE/30 and the
Macintosh IIsi are different, so that a PDS expansion card developed for the Macintosh
SE/30 must be able to run at the Macintosh IIsi clock speed of 20 MHz.

Here's the important part:

The other difference between the Macintosh IIsi and the Macintosh SE/30 is the way that RAM
accesses are structured. This difference has serious implications for MC68030 PDS master
cards. The Macintosh SE/30 responds to RAM accesses with /DSACKO and /DSACKl. The
Macintosh IIsi responds only with /STERM. Because of this difference, expansion cards,
especially PDS master cards, may not be interchangeable.

If ever there were a PDS Bus Master card, I think it would be a NuBus Adapter. This might be confirmation of my gut response to this puzzle:

NuBus in SE/30 will work but it will require a Slot E video memory conflict fix along the lines of the active component PowerCache adapters.

That's my WAG and I'm sticking to it! :blink:

Dunno, might be an entire bus lockup and the video artifacts are a shot of the machine state at the memory fault?

The Byte article about designing a NuBus card for the Mac II describes a utility the authors developed for playing "move the artifact around on screen" by inducing specific address conflicts with the frame buffer (Toby card) of the Mac II. Of course the conflicts didn't hang the system between boot chimes and  .  .  .  well  .  .  .  that'd be the question!

Anyone got a notion about this? ANY thought or question about it would be greatly appreciated.

 

[Trash80toHP_Mini]

View attachment 11450

General Purpose 68030 PDS Prototyping Board anyone?

PDS broken out to 68030 socket thru-holes interface to Rpi etc.

GAL Pad & socket interfaces for building SE/30/IIsi PowerCache Adapters

Prototyping area with .05" spacing

Need to add the rest of the IIsi board as snap-off for other two mounting points.

Kitchen sink?

Whatcha think?

 

[joethezombie]

really cool stuff trash, but way past my feeble hacking abilities!  I found mention of the ExpanSE/30 Nubus expansion in an online Computerworld scan, that I thought you'd get a kick out of:



That'd be sweet.  An HD20 styled box under the SE/30 which supports four Nubus cards.  WOW!

 

[Trash80toHP_Mini]

What makes you think this isn't way over my head? This is a learn as you go process for me, that's why I keep begging for help!

Thanks ever so much for your support in posting that tidbit and especially for your encouragement. It gets really old just talking to myself in these threads. Reminding me of the Expanse/30 came at just the right time. I'm burned out on this project for a while and that'll keep it in post processing in the back of my head as I move back to something simple like reducing the height of the 1400 stack by a couple of units.

Next rev of the 68030PDS/GPPB (General Purpose Prototyping Board) will be in its own hacks topic. Open sourcing a four layer development board for the 68030 PDS is long past due as far as I'm concerned. We'll see what happens. Not burned out on that one at all, it sits smack between the NuBus and PowerCache Adapter projects. It's meant to wind up a low volume production board for SE/30 and IIsi PowerCache adapter kits.

I'd love to see somebody develop something useful for that teensy little 68030 PDS on its ass. Something like an Orange pi/Firefox under ubuntu browsing co-processor card for the SE/30 would be killer! That would be a direct, open source Linux descendant of my PC Compatibility Cards.

 

[Trash80toHP_Mini]

 

[Trash80toHP_Mini]

Last shot of the hardware for the time being, very different point of view. I'll be going on hiatus for a while, need to do some other stuff to wind it out.

 

[Trash80toHP_Mini]

OK, now we'll try bribery to get somebody to post in one of my hacks threads. We're going to play "Where is Waldo?"

Spent yesterday morning buttoning the chassis free testbed up in its cozier, safer enclosure and began feasibility testing for the end game of the NuBus hack at the same time. I've been clearing the decks for another run at the 1400 eradication campaign next WED/THU. Gotta get one together and get it to the USPS office fairly soon and I'll mail out the prize(s) at the same time, if and when.

You've got to spot it.

You've got to find the name and link to info on the inspiration for it.

You've got to figure out the reason for it. ***

Big whoop for a prize, just a VGA adapter and whatever other trade trinket (maybe a lot more worthwhile, depending) I decide to throw into the package. Maybe some bragging rights or street cred to be gained? Dunno, but let the games begin!



*** There are actually two reasons, one I've posted about and another that was posted by someone else who wants to do something similar. Bonus points for getting both reasons at the same time.

p.s. I'll throw a VGA adapter at the first two to get answers to any of the three listed questions right.

p.p.s. If I've PM'd info on any of this to you, you're obviously disqualified.

 

[IlikeTech]

A proto Mac style enclosure? Complete with a logic "board"

 

[Trash80toHP_Mini]

LOL, nope, it'll be a standard SE/30 case, some clear plexi sections aside. Upgrade though, I just can't leave this hack "stock."

 

[Trash80toHP_Mini]

Now that joethezombie and Bolle are demolishing the very foundations of the ProtoCache1 windmill, I'm back to banging my head on this even older project. The quest for NuBus in SE/30 has been going on for some five years or so in its current iteration. Started researching it about thirteen years ago!

I've got a few new lines of attack on the list, most important of which is finally getting my 50MHz PowerCache into an adapter so that it will be running the show. [] ]'>

I've got the chassis-free test setup reconfigured. Still needs some work to lock the front bezel down and fix HDD and a Zip drive into place. Cooling be damned, I'll just set up a fan to blow on it for now!

1 - test 32bit Clean ROM SIMM assortment

2 - set the RCPII/IIsi and Radius LCD up as the boot screen

3 - retest in configurations above

4 - start running TattleTech reports for each configuration.

Quantification of results in all the different config. states ought to help.

5 - install something more capable than SimpleText so I can READ the friggin' reports!

6 - test with Video ROM pulled!

Maybe I'll get some interesting error reports from TattleTech with the Video subsystem at PseudoSlot $E disabled. h34r:

 

[Trash80toHP_Mini]

IIsi DevNote p.73

RAM access from a PDS expansion card
The memory cycle for a Macintosh IIsi processor-direct expansion
card operating as bus master is substantially different from that of the
Macintosh SE/30 computer. It has been changed to support burst
transfers using the /STERM signal generated by the MDU rather than
the /DSACK signal generated by the general logic unit chip. If bus
master cards look only for /DSACK, they will not work. Figures 6-8
through 6-9 show the timing for both random and burst writes to
RAM and reads from RAM.

p.54

Processor-direct expansion interface
A 68030 Direct Slot adaptor kit, available from an authorized Apple
dealer, allows a customer to install an existing Macintosh SE/30
processor-direct slot expansion card in the Macintosh IIsi computer
and have it work exactly as it does in the Macintosh SE/30 computer,
provided that the expansion card is able to operate at the Macintosh
IIsi clock speed of 20 MHz. If it cannot, the expansion card developer
must revise the card to be clock speed independent before it can be
used in a Macintosh IIsi computer. Also, any new or revised cards
should be designed to use 32-bit rather than 24-bit addressing.

p.12

Address space
The address space is decoded by the Memory Decode Unit (MDU). On
power-up, ROM is mapped by the MDU to physical location $0000
0000. This mapping enables the starting address retrieved by the
68030 on reset to be stored in ROM. After the first access to the true
ROM address space ($4000 0000 through $4FFF FFFF), the normal
memory map is imposed by the MDU. The only change from one
map to the other is that in the power-up map, ROM is selected for
addresses $0000 0000 through 3FFF FFFF, whereas the normal map
selects RAM for that address space, as shown in Figure 2-1.

 

[IlikeTech]

I know this is an old post, but I think this is a very interesting subject and something worth working on.  How is this going @Trash80toHP_Mini?

 

[Trash80toHP_Mini]

Not sure, thanks for the interest. t's getting a lot more thought these days. I just carpet bombed joethezombie and Bolle with a litany of crazy, mixed up notions about it in PM, let's see if I can do a better job here. They're based on a tidbit found on a page omidomo's recent content contributions:

View attachment 21711

The next to last paragraph gives some hope that there might be a way around the SE/30's borked bus mastering setup. Hoping that "It runs all your hardware  .  .  .  bus masters and all software." might include a fix for SE/30 PDS and its memory addressing issues.

I got the signals involved mixed up when I was trying to explain the conceptualization that morning. It goes something like this:

Pretty sure I've mentioned it, I've long wondered if my 50MHz PowerCache P33 running in its little IIci Cache Slot niche might address the IIsi adapter within its sphere of influence. Hoping that the NuBus ChipSet just showing up on the bus will kick its employment into action by the Slot Manager. The much missed bbraun floated that notion

The way the NuBus chipset in the DuoDock remains active on the Duo's docking connection(an 030 PDS) when its Declaration ROM is pulled to disable the Dock's shutdown of the Duo's LCD was the topic in that exchange.

The way I'm seeing this would be hanging an independent 030 PDS passthru off the Cache Connector in the block diagram of the IIci. The "Cache Slot" is the initial example of what was first described in DCaDftMF2e's Part III: Application-Specific Expansion Interfaces.

So the Cache Slot on any 030 PDS accelerator adapter falls outside Apple's definition of a PDS. I'm hoping that isolating the passthru slot on Bolle's DiiMO adapter clone from the SE/30's borked 030 PDS and wiring its control lines directly to the full IIci pinout might allow PowerCache to address the NuBus chipset on the IIsi adapter or the memory on the logic board directly in the way they need to be addressed. It can't be doing both at the same time. The universal PowerCache is extremely flexible, with any luck it will make use of the SE/30's "IIx ROM" to run NuBus in SE/30.

Some gate logic might be necessary, not clue yet, but that's the next avenue for exploration.

Now if someone conversant with ToolBox routines and Inside Macintosh would like to help out, we also have the Expanse30 approach to the problem at hand.

“If we knew what it was we were doing, it would not be called research, would it?”

- Albert Einstein

 

[Trash80toHP_Mini]

SE/30 PDS/NUBUS Adapter Project?

SE/30: IIsi NuBus Card might work in Artmix adapter??????

Mac II PowerCache Adapter - divining its unique secrets . . . maybe.

Project30 - IIsi NuBus adapter in SE/30 trials - take 2

Project30

< /project notes >

 

[nickpunt]

The ExpanSE/30 went by eBay back in Nov 2013, I saved the pics. Maybe they're helpful: