Building an internal grayscale card for the SE/30

[Trash80toHP_Mini]

Is your existing Backplane Expansion Card adaptable to a 640x480/800x600 VGA converter built into it?

Might you add 600x800 VGA portrait output to the list? will 1024x768 in 1bit or 4bit grayscale fit within available VRAM?

 

[bdurbrow]

Hmm... I'm wondering if there's enough room on the recreated SE/30 board to remove the B&W circuitry (without the Video ROM the system doesn't even go looking for it) and put this in it's place instead...

 

[Bolle]

Lots of stuff to squeeze in. I'd say no

 

[Bolle]

Is your existing Backplane Expansion Card adaptable to a 640x480/800x600 VGA converter built into it?

It is, but I'd rather keep it DB15 so it will still be easy to wire it up like for like to other graphics cards.

Might you add 600x800 VGA portrait output to the list? will 1024x768 in 1bit or 4bit grayscale fit within available VRAM?

At some point I am limited by how many oscillators I can fit onto the board for all the different pixel clocks.

An option here would be to have a socket for "external monitor" and people would have to put a suitable oscillator in for their desired resolution.
I can tell from experience that even a simple socketed oscillator on an accelerator is going to cause problems to some so I'll most probably not include that as an option. If you want to tinker with weird resolutions warm up your soldering iron
It's going to be documented how to calculate and where to put the values in the ROM to set up your own resolutions.

will 1024x768 in 1bit or 4bit grayscale fit within available VRAM?

That would fit into the existing 512K VRAM.

 

[Trash80toHP_Mini]

Thanks for the in depth info! @Bolle

Using this card as the primary display with onboard disabled could prove interesting. Will it decouple the timings required for onboard video to A/B interaction etc. enough to raise the logic board's system clock?

You'd be boosting the PDS frequency as well which might be problematic for accelerators, but most (all?) work on the 20MHz clock in the IIsi so I'd think not? Network cards transition to IIsi without problem, while those in general use were built for the IIsi PDS to begin with.

Per Apple: some non-compliant/slow PDS cards for SE/30 might not make the 16MHz->20MHz transition when installed in the IIsi.

Overclocking the system would be worth exploring for increased memory access bandwidth.

 

[Trash80toHP_Mini]

. . . I'd rather keep it DB15 so it will still be easy to wire it up like for like to other graphics cards.

Thinking here is that it would be a modular setup enabling either internal VGA adapter board/HD15 or standard DE15 passthru module where either would bolt up to your existing I/O panel.

An option here would be to have a socket for "external monitor" and people would have to put a suitable oscillator in for their desired resolution.
I can tell from experience that even a simple socketed oscillator on an accelerator is going to cause problems to some so I'll most probably not include that as an option. If you want to tinker with weird resolutions warm up your soldering iron
It's going to be documented how to calculate and where to put the values in the ROM to set up your own resolutions.

Can't ask for more than that. Would the documentation on my programmable Spectrum24 be of use to you?

Check PM for another goodie that might be of use.

 

[Trash80toHP_Mini]

Aw, come on big guy. Just harvest the onboard video real estate, switch to a pair of 72pin SIMM slots and use the liberated square footage with tall bits on the component side and SMT chiplets on the solder side. Hook it up to $E and Bob's your uncle!

edit: add headers for a daughtercard in the vacated cubic from the six excised 30pin SIMMs and you'll have almost twice that much more acreage for the graft.

 

[Trash80toHP_Mini]

One more thought for the MoBo graft possibility. On the system board, you wouldn't need the PDS buffer setup, no?

 

[Bolle]

On the system board, you wouldn't need the PDS buffer setup, no?

That would definitely have to stay if you plan to use other cards in the PDS slot as well.

 

[Nixontheknight]

Good question, haven’t actually tried the card with anything but 7.1. The driver is sitting in the ROM, no extra system extension is needed.

Switching between internal and external display would require some sort of video signal switch which I don’t plan to implement. The card is going to be wired up in a way that as soon as you connect the CRT neck board wiring harness it will always output compact Mac resolution (or optionally 512*384 if you want that selectable by a jumper on the card)
With the internal harness disconnected it’s going to output 640*480 (or 800*600 - that should still fit into the 512k of VRAM as well)
The video signals are always present on both, the internal connector and the breakout header, at the same time. That’s why it has to be handled in a more restrictive way… neck board connected->always do compact Mac resolution so you won’t fry the analogboard by sending VGA or even 800*600 sync rates to it.

that's a good idea because most mac video cards don't support two displays at once, so adding this limitation seems right for a video card for Macs of this era

 

[Trash80toHP_Mini]

@Bolle I don't think internal video implementation on a system bus needs buffering? DCaD Spec requirement for buffering close to the base of a PDS Card is due to line driving limitations of the MLB at the PDS connector. Same would not necessarily be true if folding the GS card into the Logic Board itself, no?

@trag got any comment on the internal video buffering front? The only Video System buffering of which I am aware would be the protected mode buffering of the IIsi's first Meg of RAM when sense lines at the Video Connector are detected.

Just a matter of curiosity, question was asked, so I had to suggested a crazy workaround for the clearly obvious negative answer to fitting it onto the Logic Board as is. Will drop if too tangential, can't help it.

 

[chillin]

Just out of curiosity, is internal 24-bit grayscale @1024x768 possible? What are the drawbacks?

 

[Trash80toHP_Mini]

Nope and needn't be 24bit at all at any rate. The human eye can't distinguish between 8bit and 16bit grayscale IIRC, but certainly not 24bit if even possible.

Like Bolle said, 640x480 or 800x600 VGA timings/frequencies fed to the A/B would fry it. As I understand it, he's using a jumper on the board to set the internal GS neck board default to external output only mechanically to preclude such disastrous results. If swapping out the CRT for an 8" LCD 8bit color at 1024x768 becomes a possibility for a later version given the GS Card's lineage.

@GeekDot I see your little DeclROM corner, how's that going? @cheesestraws what have you been up to on this one? I'm sure the gang would love to here a bit from both of you, I sure would!

Awesome work one and all!

 

[chillin]

The human eye can't distinguish between 8bit and 16bit grayscale IIRC, but certainly not 24bit if even possible.

The Xceed MacroColor 30HR supported the Apple Portrait Monitor, which was grayscale, in 24-bit, at least according to gamba's Xceed spec chart.

There are a handful of photogravure presses that will duplicate B&W images from film burned to metal plate to paper without a halftone dot, meaning a smooth transition from white (no ink) to 100% black, exceeding 24-bits of grayscale.

idky I thought of custom photogravure presses when B&W film obviously has more than 24-bits if photographers have been known to scan B&W and infrared negatives and prints at up to 32-bits for editing before reducing bit depth for output.

 

[Trash80toHP_Mini]

B&W film emulsions are single bit, fixed silver or clear gelatin emulsion/base. Randomized matrices of silver making up the image would be analogous to a near infinitesimal halftone screen that's for all practical purposes continuous tone. That same mechanism would be what's going on in the photogravure process for intaglio printing as well. It's still a fine grained gelatin emulsion based etching process. At some point the gradations of what's called continuous tone are lost to human vision.

24bit Color vs Grayscale is a sticky wicket. RGB is 8bit per channel, no more. Scanning in 32bit is the same with 8bit Alpha Channel. So 8bits of RG&B add up to white. No bits of any are black and it may as well be eight bits of grayscale as I understand it?

30bit Deep Color at 10bits/Channel is something else entirely and well beyond the scope of anything in our world of 68K or PPC.

 

[chillin]

You'd scan your B&W film and prints with 1-bit depth? I can't really agree with your characterization re: randomized matrices of silver, poetic though, nor the insinuation that your analogy is somehow more than an analogy, and literal, that ultimately the universe is in halftone with dots so small our eyes just can't see it.

 

[Scott Squires]

He's not saying that the universe is half-tone, just b&w film. (Is it actually? I don't know.) If that were the case, would you scan your film at 1-bit? Depends on the resolution of your scanner, of course.

 

[chillin]

The image on film made by an SLR camera is not... dots. It captures the image from light (which has no stepped gradations or dot) using optics, and duplicates at whatever the film density is, but I don't think we can just hand wave away what is going on at literally the molecular level on film with 1-bit depth and say it is halftone, even if technically is similar, in theory, at that molecular level. This ignores the scale of the vast chasm between macro world and Interspace. I'm not clear on resolution, but I know with color, 24-bits is sufficient to be known as "true color." I recall when I finally got a 8-bit graphics card for my Mac II c.1991, having lived with 1-bit monitor for 2 years and seeing 256 shades of gray on it felt somehow miraculous. But since then, I have seen more than 256 shades of gray, and I think many can tell the difference between 8-bit and 16-bit grayscale. What I mean is, whatever 8-bit grayscale is, it can't be "true grayscale," if such a thing existed.

 

[Scott Squires]

24-bit color is only 8-bit grayscale. Most people have not seen a monitor that can display more than 8-bit grayscale. My monitor is capable of 10-bits per channel, but I have never used that feature because only very specialized software actually understands how to leverage it (not to mention requiring a professional GPU; not sure if that is still the case). Anything beyond 256 grays is technically tinted, not gray, but for colors very close to gray it may not be noticeable.

Edit: Actually, I guess a lot of people have access to 10-bit panels now in the form of HDR. If you're using an HDR display then you have a 10-bit image. With Dolby Vision, potentially 12-bit. Most media besides movies and games won't leverage that color space though.

 

[Trash80toHP_Mini]

We're getting pretty far afield here. We're quibbling over jargon in the computer imaging, photographic and printmaking fields all at the same time! No worries and certainly no offense intended.

The silver bits fixed in a developed B&W negative are literally the remaining granular chunks (not dots) of photosensitive silver halide suspended randomly enough to appear continuous tone to the human eye. It's still Black OR White in fine enough a grain matrix that it's referred to as continuous tone, and is for all practical purposes.

This is easily seen if you've developed a series of high contrast ortho process film 8x10s for posterizing B&W photscreen prints in levels of gray and black on white substrate. That film's grain is big and coarse specifically for that purpose or use in photolithography.

As for the 'verse, it's most definitely a MATRIX done up in ever diminutive bits and clearly continuous tone as every single bit is whirling about in constant motion.

Shiny!