SuperMac PDQ, Spectrum 8, 24, Thunder ROMs

Phipli

Well-known member
Turns out I ordered a Pic 150k. Didn’t look too closely and it was $10 with next day delivery so didn’t think twice. Oh well, might be handy some other time. The cheapest TL886 variants seems to run about $85 😮
On eBay there are options like this :


There are likely cheaper options if you buy from AliExpress. Such as this, the first example I found (not researched) :

 

Arbee

Well-known member
@Melkhior As threatened, I've started uploading some of the ROPS stuff at https://github.com/rb6502/RasterOps/

Just a couple of simple TMS34061 based LC cards for starters. More to come, including the more mainstream stuff like the Horizon 24.

The 8LC ROM from that hard disk capture turns out to be a pretty early alpha. If anyone has a retail 8LC card and can provide the ROM dump for comparison/analysis (I'd prefer a real EPROM read to see if the bits are inverted or anything, but Paul Pratt's Slot ROM dumper will work if you don't have a TL866 or whatever) it would be greatly appreciated.
 

Arbee

Well-known member
@MacOSMonkey That would be cool. ROMs for LC PDS cards that aren't Ethernet seem to be pretty rare.

I've added the 24MxQ to the Github repo. More cards to come.
 

Arbee

Well-known member
The Github repo now has the sources for the MediaTime and the Horizon/PaintBoard Pro (which are compatible hardware, there's a #define that sets which logo bitmap and name it uses).
 

pl212

Well-known member
Wonder how easy it would be to extract the bitmaps from the c code? If I’m understanding correctly, you used to only see these flash on the screen at boot — it would great to have them as actual files…
 

eharmon

Well-known member
@MacOSMonkey That would be cool. ROMs for LC PDS cards that aren't Ethernet seem to be pretty rare.

I've added the 24MxQ to the Github repo. More cards to come.
Since if I understand correctly, the 24MxQ is a DAFB on-a-nubus, the 24MxQ source is pretty interesting. I wonder if the driver could be compiled into a Quadra onboard video driver instead. Not sure there’s much you could do with that though, since the DAFB wasn’t accelerated. Maybe convince later Quadra’s to support 2MB vram? I’ve never dug in to see if the vram memory can be configured to handle larger SIMMs, so maybe it’s impossible anyway.

It does reveal a toggle for the sync-on-green flag in PRAM though, which is pretty handy (and I’m pretty sure is basically what the Apple Basic Color Monitor control panel toggled to fix the green-tint on Quadras).
 

Arbee

Well-known member
@pl212 The bitmaps are byte arrays (usually at 8 bits per pixel) in the source, so it'd be pretty easy to convert them to PNG or something.

@eharmon djMEMC has the same VRAM control signals as standalone DAFB, and the DAFB driver dynamically checks the VRAM size on all machines, so 2 MB VRAM should be possible there. MEMCjr does actually reduce the pin count so no love for the Q605/LC475/LC575.
 

eharmon

Well-known member
@pl212 The bitmaps are byte arrays (usually at 8 bits per pixel) in the source, so it'd be pretty easy to convert them to PNG or something.

@eharmon djMEMC has the same VRAM control signals as standalone DAFB, and the DAFB driver dynamically checks the VRAM size on all machines, so 2 MB VRAM should be possible there. MEMCjr does actually reduce the pin count so no love for the Q605/LC475/LC575.
Yeah, it definitely looks like it's totally dynamic. At the risk of forking this thread: did you ever try messing with this @cy384?
 

Arbee

Well-known member
The ROM driver has the same dynamic checking. I had a bug in the Wombat emulation initially where I hadn't turned the VRAM down to 1 MB and it enabled the modes requiring 2 MB. Most of them worked fine except it didn't program the clock divider correctly for 824x632 at millions of colors. Not a terrible result for a probably-untested configuration.
 

Phipli

Well-known member
Maybe convince later Quadra’s to support 2MB vram?
The djMEMC chip should work with 2MB. I suspect you'd just need to replace the onboard chips with bigger chips, if the footprints are wired for it.

If they're not, it is a PCB issue more than a software one.

Edit... As others have said.
 

Phipli

Well-known member
Most of them worked fine except it didn't program the clock divider correctly for 824x632 at millions of colors. Not a terrible result for a probably-untested configuration.
I don't think the real machines physically can do 24bit regardless of how much VRAM they have. The function was apparently left out to simplify the design.
 

Arbee

Well-known member
The video DAC is the main difference. AC842 does 1/2/4/8/24 bit, and AC843 does 16 bit but not 24. That is auto-detected by the driver as well, so if you had a spare DAC from a battery bombed Q700 or 900, you could theoretically drop it into a Wombat (and extend the VRAM to 2 MB) and have 24 bit (the pinouts are the same).
 

eharmon

Well-known member

pl212

Well-known member
@pl212 The bitmaps are byte arrays (usually at 8 bits per pixel) in the source, so it'd be pretty easy to convert them to PNG or something.
My inspection of the c code so far has delivered the insight that "RasterOps blue/green" is defined as #00dfaa. Still working on the rest of it. :)


The logo in the 8LC board code, for example, is defined as a 31x37 array:


...however the values are encoded as a two-characterchar which I think represents a hex value. We can turn that into integer values (and either treat the image as black&white, or assume that they represent the luminance values of an image in RasterOps blue/green):

Regardless, 31x37 is not quite the right aspect ratio for what we're trying to produce:

31x37.png

Pulling the image to 600% of its original horizontal scale gets us closer:

stretched.png

...but at a certain point I wonder exactly how the drawing was done to get a horizontal logo out of a roughly square pixel array. Will try to look closer at the drawing code...
 

eharmon

Well-known member
Were the Thunder IIs unique cards or were they another rebrand from elsewhere in the line?

Relatedly, anyone know if 3.0 is the latest ROM for those? It appears that 3.1 for the Thunder/24 adds 1280x1024, at least from the ROM strings, which is nice for those stupid 5:4 ratio LCDs and it'd be nice to get that on the Thunder IIs.

Finally, I've only seen a 1360 ROM floating around so I'm also curious to compare it with the 1600. Looking at the pictures it seems likely the 1360 has sufficient memory and RAMDAC speed to run at 1600 with 24-bit....so it might just be ROM gimped?
 
Top