SuperMac PDQ, Spectrum 8, 24, Thunder ROMs

[Arbee]

Are there actual dumps of any of those cards? Would be worth a safety check to see if they're inverted or address-reversed (at least some RasterOps cards definitely do have the EPROM bits inverted, and those S-Records don't).

Also I thought I'd converted all the Mac line endings before committing those files, drat.

I need to go digging through that archive for more cards too.

 

[eharmon]

Are there actual dumps of any of those cards? Would be worth a safety check to see if they're inverted or address-reversed (at least some RasterOps cards definitely do have the EPROM bits inverted, and those S-Records don't).

Also I thought I'd converted all the Mac line endings before committing those files, drat.

I need to go digging through that archive for more cards too.

Yeah, I confirmed from other dumps none of these look inverted/reversed/etc for these cards. I might have missed one or two, but they look legit as-is.

 

[eharmon]

Are there actual dumps of any of those cards? Would be worth a safety check to see if they're inverted or address-reversed (at least some RasterOps cards definitely do have the EPROM bits inverted, and those S-Records don't).

Also I thought I'd converted all the Mac line endings before committing those files, drat.

I need to go digging through that archive for more cards too.

Ah and I forgot, the one with Mac line endings is ROM.24MxTV.S1.64K.

 

[David Cook]

@bigmessowires and @eharmon spotted this SuperMac Spectrum 8 Series II v2.0 ROM on eBay.



ROM dump attached. I'll be interested to hear what other cards it works on.

 

[eharmon]

@bigmessowires and @eharmon spotted this SuperMac Spectrum 8 Series II v2.0 ROM on eBay.

View attachment 68680

ROM dump attached. I'll be interested to hear what other cards it works on.

Nice! After unscrambling the ROM it identifies as a Display_Video_SuperMac_VGB and provides a driver for a Display_Video_SuperMac_SpectrumC.

SpectrumC is unique, but VGB is shared with the following devices:

• Spectrum SE/30
• Spectrum/8 Series III
• Spectrum/24 Series III

At minimum, the Series II and SE/30 both share the same HWId and Device ID, so it's probably worth a try. But I doubt the others are compatible. At least from the ROMs I have seen.

 

[jmacz]

Here's a SuperMac Spectrum 24 Series V ROM v3.1. I don't think I saw it uploaded on this thread yet (I saw 3.1 ROMs for Series IV and the Thunder 24 earlier).

The checksum seems valid.

I got this off a busted Spectrum 24 Series V I'm trying to fix still. Replaced two missing caps, two broken ferrite beads, two broken traces, and a broken Nubus connector. Mostly working now but I've got redraw issues where it's unable to erase every 8th vertical line or so (as the cursor passes over it) but it still seems to be able to blit an image onto the whole screen just fine. Probably something with one or more of the memory chips but too lazy to pull each one off right now.. maybe later.

 

[David Cook]

Here's a SuperMac Spectrum 24 Series V ROM v3.1. I don't think I saw it uploaded on this thread yet (I saw 3.1 ROMs for Series IV and the Thunder 24 earlier).

Awesome

 

[MacOSMonkey]

The Spectrum/24 Series III and Spectrum/8 Series III boards have unique ROMs, use different/newer components vs. the Spec C/SE30 boards (the SMT01 vs TMS) and the Spec C/2.0 ROM is not compatible.

The design of the Spec/8 Series II SE/30 board is similar to the Nubus version and the ROM is quite similar, but there was a different config process, different board IDs, and other special cases for the SE/30. I don't think the Spec/8 II 2.0 ROM is a unified one and the SE/30 version I recall is 1.0d1 (I think) -- not at home at the moment so can't check the boards. n.b. The board IDs for the Spec II Nubus and SE/30 boards should be different as far as I know.

If you are interested in QuickDraw-related bug trivia, Those boards both have the QuickDraw-related BFEXTU interaction bug. For example, if you use the monitors cdev to position the Spectrum 19" screen on the bottom edge of the SE/30 screen and shift it to the right -- say...halfway across the bottom, then put the menubar on the lower/Spectrum screen, then drag a folder window from the SE/30 screen region diagonally downward across the origin/menubar/upper left corner of the lower screen, QuickDraw will bus error the Mac (depending on OS version). It has to do with the fact that there is a bitfield extract that occurs at the origin that tries to access memory ahead of (and also after the frame buffer for lower-right accessess (one long per bit depth -- not available on those boards because, for example, baseAddr = base of video RAM). So, in this example, when the OS/QuickDraw tries to access memory ahead of the frame buffer (which also happens to be the start of video RAM), it bus errors. Earlier SuperMac boards did not have extra buffering protection, so there were certain (unlikely) corner cases where there could be a crash. Extra protection was added in later boards and Apple also eventually made changes to QuickDraw to correct this architectural issue -- certainly by System 7 and also System 6.0.x. I have commented on this issue elsewhere and I think there was also a tech note about it at some point. Anyway - if you are interested in retro design and want to create a retro video card (or do something that manipulates frame buffers), make sure you account for memory beyond the start or end of your frame buffer for maximum QuickDraw compatibility.

 

[jmacz]

@MacOSMonkey quick question.

SuperMac wanted to have a price and performance marketing spread across the 24-bit high-end video market -- for example:

"SLOWER/CHEAPER"
Spectrum/24 Series III -- 1989 -- no accelerator board (to replace Spectrum/24) (Codename: Easy)
Spectrum/24 Series III -- 1989 -- with accelerator (Codename: Easy Rider, "Rider" was the daughterboard)
Spectrum/24 PDQ -- 1990 -- (Codename: Shockwave)
Spectrum/24 PDQ+ -- 1991 -- (Codename: Poptart, a ROM gimped version of Thunder/24 without GWorld RAM or DSP board connectors)
Thunder/24 -- 1991 -- (Codename: Pop, full performance with GWorld acceleration RAM slots and DSP board connectors)
Spectrum/24 IV (aka Thunder/24)
Spectrum/24 V (but the same basic architecture)
etc.
"FASTER/MORE EXPENSIVE"

You mentioned the 24 PDQ+ was a ROM gimped version of the Thunder 24 and also missing the GWorld RAM slots. You also mentioned the Thunder 24 was related rebranded the 24 Series IV. But the 24 Series IV is also missing the GWorld RAM slots. So then looking at the 24 PDQ+ vs the 24 Series IV, they both don’t have the GWorld RAM slots so is the only difference between them the gimped ROM on the PDQ+? So basically the PDQ+, Thunder 24, and 24 Series IV when using the same latest ROM are the same card with same performance just that the PDQ+ and Series IV are missing the GWorld RAM slots?

What about the 24 Series IV vs the 24 Series V? Both the cards seem nearly identical, same chips if I remember correctly (don’t have the cards with me right now so can’t confirm), just a couple resistors added on the V. What’s the difference for these two?

As you know, I have a 24/III, 24/IV, 24/V, 24/PDQ, incoming 24 Original, there was no 24/II right? So to complete the collection I need a PDQ+ and a Thunder 24. But if they are that identical to the 24/IV that I already have, maybe not worth it given the Thunder 24 prices are out of control.

 

[MacOSMonkey]

- There was only the original Spectrum/24, then Spec/24 III and beyond.
- The Thunder/24, PDQ+, Spec/24 IV and Spec/24 V use all the same parts and the acceleration is SQD01-based.
- I don't know if the Spec/24 V ROM is backwards compatible (haven't tried it), but Spec/24 IV should be (just a rebranded Thunder/24 without GWorld slots).
- If you put a Thunder/24 ROM on a Spectrum/24 PDQ+ board, it will be a Thunder/24 board without GWorld support. All the board naming was mostly a marketing exercise using the same parts/products.

And yes, just ignore the gougers who are selling $100 boards at $500. Wait for something else to appear. And I also wouldn't recommend paying outrageous prices for VideoSpigot boards.

 

[jmacz]

The Thunder/24, PDQ+, Spec/24 IV and Spec/24 V use all the same parts and the acceleration is SQD01-based.

Sounds like my IV and my V are then identical. If there’s no perf difference, curious why they increased the series number. I think someone mentioned the V ROM works on the IV? I think I tried it and it works but can’t remember exactly.

 

[MacOSMonkey]

The V may have had more supported monitors or other hardware changes to consolidate manufacturing. And not sure if the DSP modules are the same or not. I know the Thunder/PDQ+ swap works, but I haven't tried the others. However, I think I have all of these boards.

 

[Bolle]

Thread is about ROM dumps, but I guess this fits in as well while we're at it... put my hands on one of my Thunder DSP boards, desoldered the GALs and extracted some fusemaps to add to the archive.

 

[eharmon]

@MacOSMonkey If I remember correctly, the Thunder II mainly adds (optionally) higher resolutions and drops GWorld. Is an 1152 with no GX board identical in performance to a Thunder/24? I've never seen a head-to-head, but it seems to also share the same chipset.

Which, I guess, confusingly, makes it a rebranded Spectrum/24 IV that adds DSP board connectors?

(I also find the Thunder II Light even more baffling...what is it...just the GX DSPs for a Thunder/24 and a new ROM??)

 

[MacOSMonkey]

@eharmon Yes - correct - more configs -- I was forgetting. Thanks. Eventually 1360x1024 and 1600x1200. The DSP board connections go all the way back to Thunder/24. The original Thunder/24 had a max of 1152x870. I don't know as much about later boards -- after '93/'94 -- but performance-wise, it's all going to be about the same without the DSP board (which also required external software support - Photoshop, etc.), barring any improvements to the original accelerator code. There might have been some enhancements/extensions in later code. But, in accelerating QD32, there was a point of diminishing returns. The acceleration is all SQD01-based. It's possible that the clock changed -- don't know/haven't looked. The GX boards are just DSP marketing branding, as I recall. I will look at the Spec/24 V board vs. Thunder/24 and try the ROM if it makes sense and see how it does.

 

[jmacz]

What is the latest ROM for an original (first series) SuperMac Spectrum 24?

 

[jmacz]

Attached the v1.0 ROM for the original SuperMac Spectrum 24 (series I, I guess?). Looks to be 16K and inverted. Checksum was valid after inverting it.

If anyone has a newer one, would appreciate a copy of it. Thanks.

 

[jmacz]

@David Cook would it be possible for you to get me a cleaner picture of the bodge wires on U43 and where those bodge wires go on your revision B1 PDQ+?

I believe on your board the U43 pin 9 bodge wire is going to the SMT-02 chip pin 39, and the U43 pin 13 bodge wire is going to Nubus pin C28? Can you confirm?

Also can you confirm that the board trace going from Nubus pin C28 to SMT-02 chip pin 39 has been cut/disconnected?

If the above is true, then that particular modification is making the traces on your board nearly identical to the Spectrum 24 series IV and V.

I also want to check if U43 pin 10/11/12 are connected and that U43 pin 11 and pin 13 are no longer connected. Again, that would match the Spectrum 24 series IV and V.

I am asking because I have a 24 PDQ+ revision A that is missing those bodge wires and thus is wired differently than yours or my Spectrum 24 IV/V. My revision A PDQ+ board has SMT-02 chip pin 39 connected to Nubus pin C28 (which is the /START signal). This PDQ+ board is fully functional at all color depths except has one single problem: software use of HiliteMode (ie. selection using the user preference color) has some garbage artifacts in the selection area. The issue is not on the video output path but on the write to VRAM path, proven by the fact that a screenshot also shows the garbage artifacts. It doesn't seem to be a VRAM issue because it only happens (anywhere on the screen) when HiliteMode is enabled and only in the hilited area. And the issue follows this PDQ+ card across multiple computers, on multiple versions of System 7, with extensions disabled so that suggests a hardware problem. Also occurs in ROM versions 1.262, 3.0, and 3.1 for the PDQ+.

I noticed your comments about the bodge wire and found it weird that my PDQ+ looks almost exactly identical to my Spectrum 24 IV/V yet this one section of the board is wired differently whereas yours matches my Spectrum 24 IV/V. I'm trying to rule out the modifications done on your B1 board as contributing to my issue so confirmations on my guesses above would help me.

Thanks!

 

[jmacz]

@David Cook would it be possible for you to get me a cleaner picture of the bodge wires on U43 and where those bodge wires go on your revision B1 PDQ+?

Nevermind. I was able to figure out what the modifications were and after making the modifications, the issue did not correct (although the rest of the board works fine with the modification). Going to leave the board in this modified state as it now matches my newer boards.

 

[thisisamigaspeaking]

How about a Radius Thunder IV, anyone know how that differs from a SuperMac Thunder II? Performance?

Is there any source for GWorld modules today? Apparently Finder could use it for something.

But, in accelerating QD32, there was a point of diminishing returns. The acceleration is all SQD01-based. It's possible that the clock changed -- don't know/haven't looked. The GX boards are just DSP marketing branding, as I recall.

Ah, that's what I came to this thread looking for. I had seen a few people state or imply the DSPs were also used for QuickDraw acceleration but that doesn't make much sense.