• 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.

SuperMac Spectrum/24 Project

MacOSMonkey

Well-known member
Hmm...didn't see Phipli's post, but yes, they were on a Q4ish cycle. I think Spectrum/24 Series III was towards the end of 1989, with a possible ROM update in 1992.
 
Last edited:

Trash80toHP_Mini

NIGHT STALKER
@MacOSMonkey There's s hole in that lineup. No Series II references. Did they go straight from 1.0 to Series III because it sounded spiffier or was there an unreleased code named product? Might that reflect the lineup branching out to three pricing levels with the Specrum/24 being top of the line?

If so, I've got five high resolution pics of fourteen in the timeline project.
 

MacOSMonkey

Well-known member
Here is what I think I remember, but it hurts my neurons: :p

Spectrum/24 Rev A -> Spectrum/24 Rev B (maybe a VRAM change? hmm...something like that) -> Spectrum/24 Series III

...but I think the "III" was to prevent confusion with the Spectrum/8 Series II (Spectrum/8 "C") and establish an across-the-board product line at the III level.
 

MacOSMonkey

Well-known member
I can make the Spectrum/8 Series II work on an old 2007 Dell monitor at 800x600 with some kind of variant SVGA timing with a 40Mhz oscillator in the extra socket. But, I will have to put it on the scope to try to get the correct timing. It shows the classic problem of a green-polluted black level and it won't sync to the Apple monitor. So, the timing is slightly off. The problem with green is that the signals are typically sync-on-green, so if the sync (green) is in the wrong place, it ends up screwing up the black level measurement. With a scope, I should be able to see exactly what signal the card is generating. I am sure my calculations are correct, so something else is happening at the board level (which is not too surprising).

Also, in the process of trying to make the board work, I remembered that custom configs load at secondary init (desktop loading). And, on the Spectrum/8 Series III, the board expects certain oscillators to be in certain spots. So, if you put a 40Mhz oscillator in the extra socket, it will incorrectly detect it as a 57.28Mhz oscillator (Apple 16" 832x624).

It is also interesting to see the horizontal divider at work when tweaking values. For example, when the divider is 8, any change to a blanking value alters the visible screen by 8 pixels (expected outcome).
 

MacOSMonkey

Well-known member
Well...unfortunately...it looks like I need to call Scotty to fix the deflector dish on my Multi-Scan 15". It seems to have no video in any mode - probably failed caps or bad solder after over 30 years. So, I will have to find another test source for my 800x600 SVGA experiments. I guess I am also Scotty, so maybe I will call myself. :D
 

Phipli

Well-known member
With the Spectrum/24 SIII installed I can't see the custom option in SuperVideo 2.7.4 or 2.7.5. I held down option while clicking on the resolution tab. I have a 30.24MHz clock installed and recognised in the spare socket.

I don't believe custom timings are possible from the software unless it is extremely hidden. Please prove me wrong :)

IMG_20220918_182118.jpg
 

Bolle

Well-known member
it was the most chip-heavy board because it used a bunch of Xilinx Virtex-3 parts to do the video acceleration - all the X parts.

I don't think Virtex FPGAs did exist back then... so they're probably Xilinx 3000 series FPGAs. Just noticed the Spigot uses one of those "X" labeled SuperMac chips with an XC1736 config PROM sitting right next to it. So indeed a bog standard Xilinx part with a SuperMac label. XC3020 or XC3030 in case of the Spigot judging from the PLCC68 footprint. The PDQ probably has at least XC3042s then.
 

MacOSMonkey

Well-known member
Yeah - IIRC, the original PDQ was Virtex-III-based. The 3x00s appeared in '88 and were cutting edge, at the time. And yes, the IDs were reprinted. On the Spectrum/24 III's, SuperMac sanded off the GAL IDs on the daughter card ICs as an anti-reversing practice. I will try to remember which part it was - I'll think about it. It has been a long time.
 

MacOSMonkey

Well-known member
Phipli - If you have the v1.1 installed, there is no custom support (and no lock menubar, as above). You would need a 1.0 ROM to do custom, but it's really not worth going backwards. The 1.1 version had significant changes that addressed many problems.
 

Phipli

Well-known member
Phipli - If you have the v1.1 installed, there is no custom support (and no lock menubar, as above). You would need a 1.0 ROM to do custom, but it's really not worth going backwards. The 1.1 version had significant changes that addressed many problems.
I think I have v1.0 on this card. I can check in TattleTech, but if it is a UV chip I doubt someone would have got the label back on this neatly.
IMG_20220918_224733_edit_200065088942387.jpg
 

MacOSMonkey

Well-known member
I thought v1.0 on Spec/24 III supported custom. Try the option key. But, if it's not mentioned in the manual, then maybe not. You might also need to use an older version of SuperVideo -- try 2.1 or 2.49.

But anyway, just use the v1.1 ROM. It's significantly better and supports more video modes (and the timing is more accurate). v1.1 also makes the Spec/24 III a great board for A/UX. Earlier SuperMac products had various A/UX-related issues -- depth changes, video reset, losing the PRAM config, etc. Maybe other things also...not sure. Interestingly, A/UX calls VideoReset at launch, which the MacOS does not. Also, for low-level hackers, if you want to detect the presence of A/UX, I think there is a low-memory global that sets the $200 hex bit at memory address 0x0b22 (maybe). If someone wants to check and confirm, that might be helpful - you would have to compare in Mac vs. A/UX modes. I don't know if it's documented anywhere...it might be. And, I'm not 100% sure about the address, but it's what I remember at the moment.
 

MacOSMonkey

Well-known member
Attached is the Spectrum/24 Series III v1.0 ROM from my board, along with its device type, in case anyone is interested in reading their ROM and doesn't want to peel the label. You could try the same device ID as a starting point.

n.b. At least 1 other person should verify their v1.0 ROM vs. this file, then we can be sure there are no errors and that it constitutes a master version. I would recommend not using this ROM until it has been verified by someone else. Also, I don't recommend using it anyway. :D

On my board, my accelerator card was having what may have been a platform-specific issue and I don't think it was related to the ROM integrity. Maybe @Phipli can help verify the ROM.

I found a 27C512 and double-cloned (upper/lower 32K halves) v1.1 into it and it works fine. v1.1 solved my accelerator issue. I also reflowed my accelerator card. Some of the soldering was not very good. My board is now working great on an upgraded Mac II(x) in 24-bit and 32-bit modes/24-bit color.

My accelerator daughter card is Rev C and it has 2 reworks on it. You can find these under the paper backing. I think these reworks were rolled into a later revision of the accelerator daughtercard.
Spec24III-RevC-AccelRework.jpg
 

Attachments

  • Spec24III_v1.0_NMC27C256B.zip
    14.7 KB · Views: 5

MacOSMonkey

Well-known member
@Bolle - I meant to agree about Virtex-III above (they were later, as you pointed out) - should say "I thought the original PDQ,,," (but can't edit) And the more I think of it, the more the 3042 might ring a bell. The 3042 was the most functional part that could fit in an 84-pin package in 1988-89. But, the 3020 and 3030 were also 84-pin package devices.

The Xilinx code was manually-routed to run at 40Mhz to squeak out every ounce of performance. IIRC, the best the Xilinx auto-router could do at the time was around 20Mhz. The QuickDraw acceleration had 2nd-gen improvements and fixes, including additional Xilinx-related extensions vs. the 1st-gen version in the Spectrum/24 Series III (and ColorCard/24 III).

Also, I think I am remembering that the Spectrum/24 PDQ board worked under A/UX 2.0 with the v1.27 ROM. However, v1.1 of Spectrum/24 Series III from 1991 may offer better A/UX compatibility.
 
Last edited:

Aeroform

Well-known member
Super interesting thread. I have a somewhat related hypothesis on these cards. I just acquired a Supermac ColorCard SE/30 (courtesy of fellow 68kmla user @ktkm). After having vacuumed the web for details (not that I managed to find that much), I'm hoping this card may be identical hardware wise to the Spectrum/8 series II SE/30.

They only image I've managed to find is from an old ad here (https://books.google.se/books?id=vD...I#v=onepage&q=supermac spectrum se/30&f=false), unfortunately only about half the card is shown. But, from what I can see they look identical. Also the backside om my ColorCard pcb says "Spectrum".

My hope is that they simply limited some functionality in the rom and dropped that in the same card they already had, would save a lot of hardware development work I gather.

Anyhow, if anyone happens to have a SE/30 version of the Spectrum/8 II (or the normal one), it would be super interesting to see some pics to compare the h/w. My hope would be that replacing the rom with the spectrum variant perhaps would unlock 1024*768 @ 8 bits on my card (limited to 16 colors on the color card variant).

IMG_2262.jpeg

IMG_2266.jpeg
 

MacOSMonkey

Well-known member
That board (Spectrum/8 Series II SE/30) has 1Mb of RAM and should do 1024x768 @ 8-bit Color. The 30.24Mhz oscillator is for the Apple 13" @ 640x480, the 64Mhz oscillator is for 19" 1024x768 @ 60Hz and the 57.28Mhz oscillator is for 15" Portrait monitor 640x870. I think it also probably supported MultiSync -- I think Apple Mode, and probably NTSC if you add a 14.31818Mhz oscillator to the socket. If it is limited to the 16 colors, it is because it is a ColorCard SE/30 that was only designed to support 13" monitors. IIRC, the Spectrum codename was "Astro" and the ColorCard version was "Puggsley."

I am just trying to remember whether or not the ColorCard SE/30 could do 8-bit on the Portrait. Sorry - I see that yours is a ColorCard and it has 8 Zip RAMs, so the only difference was the ROM. If you swap it, it should work.
 
Last edited:

Aeroform

Well-known member
That board has 1Mb of RAM and should do 1024x768 @ 8-bit Color. The 30.24Mhz oscillator is for the Apple 13" @ 640x480, the 64Mhz oscillator is for 19" 1024x768 @ 60Hz and the 57.28Mhz oscillator is for 16" 832x624 @ 75Hz.

Thanks, that strenghtens my belief these are indeed the same card, with just the rom being different.

According to the info I could find, the Specteum variant came first retailing for $1895 supporting up to 1024*768 at 256 colors. They later followed up with the budget oriented ”ColorCard”, retailing for $995, supporting 640*480 at 256 colors up to 1024*768 @ 16 colors.

Would be REALLY interesting to replace with a Spectrum rom on my card to see wether that unlocks the higher res/color combo. If anyone has a dump or could help in making one, let me know (I’d be happy to cover any potential costs in getting it done 🙂)
 

MacOSMonkey

Well-known member
@Aeroform - The answer is yes - I think it will work. It was just a ROM gimp. However, both of those ROMs had compatibility and other issues. I don't think they were ever fixed. Hopefully, you can find someone who can read in the Spectrum version of the ROM.
 
Last edited:

Phipli

Well-known member
n.b. At least 1 other person should verify their v1.0 ROM vs. this file, then we can be sure there are no errors and that it constitutes a master version. I would recommend not using this ROM until it has been verified by someone else. Also, I don't recommend using it anyway. :D

On my board, my accelerator card was having what may have been a platform-specific issue and I don't think it was related to the ROM integrity. Maybe @Phipli can help verify the ROM.
I ripped the ROM from mine and ran diff on it - my ROM is identical to yours.

Mine is labelled "1003061-0001B SPEC/24 SER III Ver. 1.0 © 1989" etc.

I'll rip the PDQ and Thunder IV too and upload them shortly.
 
Top