• Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.

Compact Mac Video Timings Needed for SuperVideo Input Parameters

Ah! But I might as well buy a 24mhz crystal just for the fun of it  .  .  .  unless you've got a better suggestion? [}:)] You haven't said if it might be fed to the GS neck board thru the A/B gatekeeper?

As I've mentioned before, using a newer card that actually *does* have a programmable frequency generator for the pixel clock would totally eliminate these problems. It's actually kind of pathetic that the Supermac card only has a single crystal on it. I mean, I guess I'm kind of confused again; can you even change resolutions without changing crystals?
Good question? I don't think I've ever tried to lower the resolution, why would I as a dyed in the wool pixel junkie? In a quick check, neither resolution switching nor QuickDraw acceleration are mentioned in the manual, so I'd say not offhand. It was designed to drive SuperMac 16" and 19" 1024x768 displays which were almost certainly fixed at the 64mhz standard clock frequency. The card was bundled with a 30.24mhz crystal to drive two of Apple's fixed frequency displays. The card dates from 1989 and was geared toward panning around virtual desktop resolutions up to 4096x1536! Slap a 24mhz crystal on it to support MultiSync monitors at 640x480 or a 14.31818mhz crystal for NTSC RGB at 640x480. That last sounds very good indeed now that I've found it in the docs, I was afraid it might not support a low enough clock to output the Compact Mac's 512x342.

Dunno, the card always sounded pretty awesome to me given that I was mired in B&W, albeit in the 1182xwhatever range at that time. Pathetic is in the eye of the beholder. In 1989 I think this was something to behold. By the second rev of the Spectrum/24 series, not so much. The market had moved beyond big virtual desktops to multiple big displays for CAD and DTP had begun to require QuickDraw acceleration. It'll do for this.  [;)]

______________________________________________________________________________________________________

I'm getting waaaay ahead of myself here, but is there currently a RAMDAC type part available that's suitable to a GS/30 card design? In the HDMI output NuBus VidCard thread, the VRAM side of things sounded practical enough, but the back end was a formulaic/done deal kinda thing for digital output from an FPGA. How would the analog side break down in terms of available, off the shelf components for pushing just a single, variable intensity analog feed to the GS neck board? Sounds like a strange beast?

 
You haven't said if it might be fed to the GS neck board thru the A/B gatekeeper?


I kind of thought it went without saying/repeating that if you actually want to feed *anything* from your SuperMac card into the compact monitor you'll need something like a Micron neck board (or requisite hacks to the original analog board) *and* the necessary hacks to the horizontal drive outputs.

If you have a sufficiently flexible multisync monitor lying around you can *prototype* the mode on that, but ultimately you're nowhere without doing some actual soldering.

How would the analog side break down in terms of available, off the shelf components for pushing just a single, variable intensity analog feed to the GS neck board.


Find the data sheet for a CRT similar to the one in a compact Mac and figure out what the drive voltage is for the intensity pin leading to the electron gun, and from there look for a high speed analog op-amp that can convert the 0-0.7v range that's standard for VGA-era analog monitor inputs into what the tube wants. Presumably that is what the Micron board has on it, unless said op-amp is on the card itself.

 
.  .  .  ultimately you're nowhere without doing some actual soldering.
If I can get the resolution up and running on the Radius display, VidCard or GS neck board will be making a long trip, likely the former. If I can't, your Linux/PC/VGA suggestion comes into play, but not here in the Bermuda Triangle. Man's got to know his limitations. :wink:

Op-amp sounds great, but hadn't thought that far ahead, those limitations being what they are. Analog I understand to a limited extent, it's the digital to analog conversion I was curious about. In the HDMI NuBus Card thread it sounds like the frame buffer section is straightforward enough. I'm slogging through the DCaDftMF series for this, frame buffer would be one black box I can see well enough in the Toby documentation. RAMDAC is another story entirely, docs assume an understanding of such, else why would someone even own the book? Documentation is for RGB channels, so an available single channel GS part has me wondering. I see it as something that might be documented for a different, but loosely related application in the Pi'verse?

Pet project for AI playtime will be tying relevant sections of SE/30 logic board and A/B schematics with the Xceed card and neck board inputs and outputs together in one tidy package. ISTR seeing the card to neck board/neck board schematics to throw into the mix. Probably no real need for such a diagram, but fun for me to try to visualize what's going on in what I think will make for a cool graphic.

 
Last edited by a moderator:
There isn't anything magical about a ramdac for a grayscale system, it's still just a high speed digital to analog converter. Only difference is if you *only* want to support grayscale you only need one output channel instead of three. I'm sure almost any off the shelf three channel part could be convinced to do the job. Or if we're back to talking about FPGA video cards you could make your own single channel output out of eight GPIO pins and a resistor ladder.

Not to criticize your methods or anything, but if the goal is to ever make any grand idea actually come true you might want to try concentrating on a single piece at a time, preferably ones that are close to achievable with your current tool and knowledge set.

 
LOL! None taken, that's what I've been doing. Bolle has already built the Neck Board and a simplified, internal only harness. If I can get the SuperMac baby to roll over, it's up to him to get it creeping along toward that grand idea. [;)]



He suggested hacking Toby just to see what happens when I inquired about Mavericks/Goose on the Xceed card. I suggested taking a whack at getting my stone age Spectrum/24 card to output the real deal instead. If that's a fail, it sounds like he's all over your VGA card suggestion. GS/30 isn't my idea/project, it's my code name for something that's been simmering around here for a very long time. I'm just trying to crank the burner up to reach the heat/temp plateau. No way I could ever get farther than that, if even that. Reaching heat of vaporization is obviously up to others, continued discussion about what's needed for that in terms of hardware here in this thread is fun for the sake of learning.

Looks like I'll be getting my crystal from eBay. Available cans I've found have been SMT, so this type is looking better and better. Connector conversion should be easy enough.

s-l1600.jpg.cc313a30c3c29dfdeb190ab37b287df0.jpg


I need to order caps for two or three of my IIsi boards and figured getting some crystals at the same time would be the way to go. Soldering iron will be applied to revive a test bed for the LC NIC wire wrap adapter board project. I like to do play at these things in spurts as information on them comes together. ATM the SE/Radius TPD is on the bench for hooking output up to Comp.HD instead of G on the 2150 as I did on the previous run at it.

 
Last edited by a moderator:
14.31818mhz crystal for NTSC RGB at 640x480. That last sounds very good indeed now that I've found it in the docs, I was afraid it might not support a low enough clock to output the Compact Mac's 512x342.
FWIW, if you actually had the NTSC crystal (the SuperMac manual implies it was included with the card originally, but I'm guessing if it ever had a box with the alternate crystals described in the manual it's been long consigned to the dustbin of history) playing with the Xfree modeline calculator implies you *might* be able to put together a functional 512x342 mode using the 14.31818mhz frequency. You have to lump about a third off the hsync back porch pixel count, which I *think* would translate to needing to fairly radically adjust the hsize pot on the monitor, if it has one.

 
Nope, unpacking/contents of package lists card (with stock SuperMac 16/19 inch display crystal on board) and 32.4mhz crystal for the two pathetic Apple Monitors it supports at 640x480.

View attachment 27930

Two other modes are supported with optional crystals. It's the wide open support for configuring custom displays that has always intrigued me about this card. the virtual desktop would have been great for panning around in a huge CAD drawing and 24bit at 1024x768 is nothing to sneeze at, but it's still an un-accelerated video card. Not that that really matters, Quickdraw acceleration's benefits are terribly overblown. It's fabulous if you click the scroll bars or use Page Up or Page Down on the few Apple KBDs on which they were available, but I've never seen anyone work that way. As soon as you press down on the space bar to use the hand tool move the page around manually, you've given up all benefit of QuickDraw Acceleration. It doesn't work the way people do, go figure.

 
Last edited by a moderator:
Two other modes are supported with optional crystals.
Apparently so, I guess I missed that the NTSC crystal wasn't in bold. Honestly it's kind of a head-scratcher to me why the manual/software preset specifies a 24 mhz crystal for that "Multisync" 640x480 configuration, when you'd think it'd make more sense to specify the 25.whatever dot clock that's standard for VGA? Who knows.

(Just for laughs I looked up the programming manual for the ATI Mach64 family, a typical mid-to-late-90's consumer video card, and skimmed through its clock generation section. If I understood it correctly the clock generation PLLs in those cards, depending on the version, is capable of generating fairly arbitrary clocks anywhere from sub-NTSC up to a couple hundred Mhz with resolutions within a few tens of Khz of ideal. Funny how much difference a few years makes.)

 
Last edited by a moderator:
Back
Top