jmacz journey

[jmacz]

I had picked up a set of 3 NEC 15" AccuSync 52V LCD monitors a while back to get native 1024x768 resolution from my set of SuperMac Spectrum 24 cards. The LCD panels are native 1024x768 and support both 60Hz and 75Hz. But once hooked up, I was getting washed out colors on the panels. All three were exactly the same. I had posted about this before and gotten suggestions to use the Gamma control panel to help. It did help but still wasn't great. Then I found a hidden service menu within the LCD on screen controls which let me dial down the gain. That also helped a lot but still not great. So I've had a sub optimal result for a while now.

Recently in @bigmessowires's thread about his syncinator beta program, there was a conversation about washed out colors and @bigmessowires mentioned how this could be caused by the sync on green signal, well, sync on red/green/blue (all channels). If the monitor doesn't support these sync signals, it could cause it to fail to correctly determine its color range, or these signals could raise the floor on the black signal causing the washed out look.

I went to test this out by putting the SuperMac Spectrum cards RGB signals into a scope. Sure enough, the sync signals were there, on all three channels (red, green, blue). My NEC LCDs don't seem to support this hence the washed out colors.

I figured it was the RAMDAC that would be adding these signals to the red/green/blue channels so took a look at the datasheet. All the RAMDACs on my SuperMac Spectrum cards (24/III, 24/IV, 24/V, 24/PDQ) have RAMDACs that allow this signal to be optionally added. But you can also ground the SYNC pin to disable this. The SuperMac cards have this pin enabled. I tried lifting this SYNC pin and grounding it. It worked. The colors were no longer washed out and looked normal. Fantastic! But now, how to make this configurable.

I then decided to take a look at the ROMs for these cards to see whether the monitor configurations (stored in ROM and shown in the control panel) had a hidden flag for enabling/disabling sync on the RGB channels (ie. send a signal to the SYNC pin on the RAMDAC). I could make out a lot of the configurations (resolution, oscillator requirement, etc) but there's a lot of bits in each record. I was about to start looking at some of the code in the ROM when I noticed on the board that the sync signal seems to be going both to the SYNC pin on the RAMDAC as well as to the DB15 pins. This probably means I can't selectively turn off just the sync to the RGB channels as turning that signal off means it would disable the other sync pins on the DB15 cable.

So I decided to punt on the ROM and just hack something on the card itself.



Crude. But it's configurable now via the jumper. Pin 1+2 will enable sync on the RGB channels for my monitors that need that. And pin 2+3 disables it for my monitors that don't support sync on the RGB channels and want separate sync.

cc: @trickydee

 

[jmacz]

Made the same modification above to the following cards, and it's working great with my LCDs that do not support sync on red/green/blue:

• SuperMac Spectrum 24 PDQ (BT458 series RAMDAC)
• SuperMac Spectrum 24 Series III (BT/ADV 473 series RAMDAC)
• SuperMac Spectrum 24 Series IV (BT/ADV 473 series RAMDAC)
• SuperMac Spectrum 24 Series V (BT/ADV 473 series RAMDAC)

I have had so many issues with these SuperMac Spectrum cards primarily due to age and that I purchased them in not-so-perfect state so every success is a big win. As of now, the PDQ and the Series IV are fully working.

The Series III still has artifact issues which I'm slowly debugging when I have time. Most recent hunch is the delay chip on the accelerator daughterboard. The chip on the daughterboard clearly has issues and it looks like there was a factory bodge wired installed to work around it. But still testing out whether my problem is in that area.

The Series V is a card I recently acquired with massive damage. Broken nubus connector, two broken ferrite beads, two sheared off tantalum capacitors, cracked PCB on one edge that broke a trace, and another trace that was cut by some PCB gouging. The slot bracket also is bent. Looks like this card endured some physical trauma. All of it those issues have been fixed now but I'm getting vertical stripes and some other strangeness which suggests something wrong with the video memory or the data lines coming from the nubus connector. Fun stuff.

 

[croissantking]

I'm happy with how it turned out. I need to look into the memory gotchas on the SE/30. I put 16MB in 4 sticks into it and I'm running Mode 32. But need to understand what the differences are with mode 32 and replacing the ROM with a 32bit clean one.

For the purposes of running with more than 8MB RAM, having a 32bit clean ROM is more of a convenience than a necessity. The most annoying thing with Mode32 is the long wait at startup while the RAM is tested, this is eliminated with a custom ROM i.e. the variants based on Rob Braun's work inlcuding the Rominator II. Note that a stock IIsi or IIfx ROM will still have the startup check.

There are some other gains to fitting a custom ROM, such as a bootable RAM disk and HD20 support.

 

[LaPorta]

When I start my IIfx with 128 MB, I just take a walk and do something else for a bit!

 

[croissantking]

When I start my IIfx with 128 MB, I just take a walk and do something else for a bit!

For sure, some will not be bothered by this. It's totally down to the individual. I run 68MB in my IIci and SE/30s and the delay was slightly inconvenient - but I have custom ROMs in all of them now.

 

[LaPorta]

Just a funny counterpoint is all: I manually disabled the RAM check in my PT Pro for this very reason.

 

[olePigeon]

@jmacz Were you able to test out Compact Virtual? I'm curious if it solved your RAM issue.

 

[jmacz]

For the purposes of running with more than 8MB RAM, having a 32bit clean ROM is more of a convenience than a necessity.

Great, thanks for confirming!

When I start my IIfx with 128 MB, I just take a walk and do something else for a bit!

Haha, same!

@jmacz Were you able to test out Compact Virtual? I'm curious if it solved your RAM issue.

No, I haven't had a chance yet. It's running 4MB on the accelerator board right now and it's working great and stable. I could not get 8MB (in 2MB sticks) or 16MB (in 4MB sticks) to work without crashing randomly. I have to get some time to open it back up, try 16MB again and try Compact Virtual.

 

[jmacz]

Recapped an Apple 17" Multiple Scan Monitor this past weekend. A few weird random distortion issues are gone. There's still an issue where the screen size settings sometimes get lost though. The monitor seems to have an eeprom (at least based on the maintenance guide) and not sure if there's an issue there. But the picture is nice and crisp.

While trying to get all my floppy drives to read all disks consistently with each other, found a problem on my SE/30. It reads all types of floppies just fine. But whenever I try to write, the system crashes. Same thing happens on an external floppy drive. So it's not the drive. Interesting thing is that my SE also seems to have an issue writing floppies. It reads them all just fine but on writes, it quickly errors out. Strange thing is it also happens with both the internal and an external floppy.

 

[jmacz]

I think you need Compact Virtual for that (3.0.2 version available on the Garden.) It'll convert any RAM in excess of 4MBs into a RAM disk, then use the RAM disk for virtual memory, thereby cleverly bypassing the 4MB RAM ceiling.

Only problem is that I don't know if the commercial version is the same as the version that shipped with the myriad of accelerators.

Tried to test this but I could not make much progress.

The only driver I have found online for the AE Transwarp SE is a Warp 030 v1.5 driver. That's the one I'm using on my SE right now.

There were some links to a AE TransWarp Compact with a "2,1" which I assume is a v2.1 driver but the link is broken in most places. Macintosh Repository seems to have a file but upon downloading it, it seems busted and extracts to an empty folder. I've tried multiple different versions of Stuffit and all the same.

Reason I'm mentioning this is I tried installing both 3.0 and 3.02 versions Compact Virtual. Neither recognize my accelerator. The README for v3.0 says it supports v1.04 of the Transwarp driver which I don't have. The README for v3.02 says it supports the v2.2+ of the Transwarp driver which I also don't have.

So I think I'm going to let this go for the moment. I don't have much use for a large ram disk with my SE. And it looks like there "might" be one case out there of someone who had AE modify their board to fully support more than 4MB of memory but the default card only officially supported 4MB which I think I have. I just get random memory access errors with anything more than 4MB.

 

[olePigeon]

@jmacz This took some internet sleuthing, but user CowgodA2R was gracious enough to make Flux image copies of his TransWarp Compact disks. I've added them to my website for easy access because it would not show up in a regular search via Archive.org. All the other archives I found online from Garden to elsewhere were incomplete. They simply included a Control Panel, but not the full installer with support Inits and everything.

TransWarp Compact 2.1

 

[jmacz]

Awesome!! Thanks for that @olePigeon! I will give this a try.

 

[jmacz]

Unfortunately still not working.

The TransWarp Compact 2.1 driver disk image itself is good. I executed the included Installer and it installed an extension (TW Init SE) and a cdev (AE Cache). The AE Cache cdev seems to be part of the "2.1" package but its own version is "2.0". The TW Init SE extension does not seem to be able to detect the accelerator. The extension on boot has an X through it and after booting, TattleTech shows the system as a 40MHz 68020 with no FPU. The AE Cache cdev also complains that no accelerator is installed. And Compact Virtual also doesn't recognize the presence of the accelerator with just the TW Init SE and AE Cache installed.

Removing just the TW Init SE extension and using the TransWarp 030 v1.5 control panel results in detection of the accelerator. TattleTech shows the system properly as a 40MHz 68030 with FPU. I left the AE Cache control panel installed as well and that cdev now recognizes the accelerator and mirrors the configuration shown in the TransWarp 030 v1.5 control panel. Any change I make in one of those two cdevs automatically reflects in the other. So at this point tried Compact Virtual again (with both cdevs running) but it still doesn't recognize the presence of the accelerator.

At this point, the TW Init SE which seems to be the driver, doesn't work for my card. The AE Cache seems to just be a configuration cdev which does complain if no accelerator is present (and complains if I am trying to use the TW Init SE extension). The TransWarp 030 v1.5 cdev I have been using seems to be a combination driver/configuration cdev and it does recognize my card. And with that loaded, the AE Cache cdev also works. But in no case does Compact Virtual work.

The accelerator I have is an Applied Engineering TW1340 Mac SE II Accelerator. I believe the TW in TW1340 is for TransWarp. The only driver that has worked for me is the TransWarp 030 v1.5. Sounds like the TransWarp SE Compact is a different accelerator or a different revision as the TW Init SE driver does not work for me. Management/configuration interfaces seem to be the same as AE Cache works with my TransWarp 030 v1.5 driver and it's UI mirrors the UI included in the TransWarp 030 v1.5 cdev.

 

[olePigeon]

Darn. Was worth a shot.

 

[jmacz]

Yeah, was hoping it would work but it's ok. I've got a fast 030 based SE limited to 4MB which is fine, I'm running 6.0.8 on it. I've got my SE/30 if I need more.

 

[jmacz]

I have been debugging a SuperMac Spectrum 24 Series V. Making some progress and documenting along the way here:

SuperMac Spectrum 24/V Display Problems

Another year, another SuperMac Spectrum 24 card to debug. :oops: Yes, I am 4/4 on obtaining cards with issues and batting 2 for 4 so far in fixing them. I fixed a dead 24/IV... fixed a slightly malfunctioning 24/PDQ... still continue to have some problems with a 24/III. The latest is a SuperMac...

68kmla.org

Not sure why I keep subjecting myself to punishment, but saw a SuperMac Spectrum 24 board that’s not working up on eBay and couldn’t resist. I believe this is the original unaccelerated 24bit nubus card. It is completely dead with no video signal. It’s on its way to me and will be a future project. LOL cc: @MacOSMonkey

It will be my 5th Spectrum 24 with the others being:

• Spectrum 24 PDQ - fixed and working.
• Spectrum 24 Series III - still getting artifacts from time to time at 24bit.
• Spectrum 24 Series IV - fixed and working.
• Spectrum 24 Series V - some fixes but still in progress.

More debugging pain incoming.

 

[samurai-j]

I also have a TW1340.
I set it up quite a while ago,
I don't remember how I did it,
Compact Virtual recognizes the TW1340 and the 16mb memory.

 

[jmacz]

Your Warp 030 Control looks slightly different than the version I am using if I remember correctly. Could you possibly share the version number you have and attach a stuffed/binhex’d copy of it? Also your Compact Virtual looks to be version 3.0. I have only had up to version 2.1. Could you possibly share that as well? Thanks!

 

[samurai-j]

The system should be v7.1.
If it is v7.0.1, there is a problem.

There is a warp030 v1.5 in the control panel,
The only difference is that mine is the Japanese version.

Put TransWarp compact in the INIT folder,
The coprocessor is recognised,
Remove it, only 68030 MPUs are recognised.

I think Compact Virtual needs v3.0.2.

I believe the above is available from Macintosh Garden.

These are my environments.
I hope it works.

 

[jmacz]

Actually I was wrong, I did try Compact Virtual 3.0 and it didn’t work. I mentioned it earlier, I just forgot.

But, I did not try 7.1. I have been using 6.0.8. Will have to try again.