WarpSE: 25 MHz 68HC000-based accelerator for Mac SE

[JDW]

Here's what I see on the CRT when FLASHING and when the Power-ON artifacts appear:

Stock SE Motherboard:


SE Reloaded Motherboard:


My testing of 0.6e showed the same old problems:

1. Artifacts (see above) at first power-on, but I can usually get a normal screen if I power off for a few seconds and then power-on again.

2. Mouse movement is bad when using my SE Reloaded board, but no issue with the stock motherboard. It's also clicking on its own again! Nope, it can't be the SE Reloaded board being bad because I don't get that in stock 8MHz mode or with the 16MHz 68000 Levco SpeedCard installed.

On the upside though, Tetris Audio is as good as stock. Benchmarks in Speedometer 3.23 & SCSI Director Pro 4.0 are identical to firmware 0.6b.

I made a video to show you all the details, but it's been rendering in Compressor for more than 1 hour and still has a ways to go. I am dead tired and need to go home. I will uploaded the video tomorrow morning.

 

[Kai Robinson]

I wonder if it's a difference with the 74 seire logic in the Reloaded. Iirc we went with ACT and AHCT instead of LS chips.

 

[JTRetro]

@JDW -I was getting checkerboard screens sometimes at startup when I was using the last firmware update from Zane. So far with the latest update that I've applied-.6b- I haven't had that problem, but then again I have only used me SE twice so far with this firmware. @Zane Kaminski -do you want me to upgrade to the latest firmware, .6e?

Also, I finally found my Macintosh SE with the Superdrive. I'm at work now, but when I get home I'm going to do some testing with that machine as well.

 

[Zane Kaminski]

@JTRetro @ppuskari @JDW Thanks for all your hard work testing!! Attached is the latest firmware, 0.6f!

0.6e fixed some problems but also introduced some other ones. I just now finished 0.6f which fixes multiple timing issues. I've tried this on two prototypes and three MC68HC000 CPUs and it's been working for me. Let me know how it goes!

 

[Zane Kaminski]

Here's what I see on the CRT when FLASHING and when the Power-ON artifacts appear:

Stock SE Motherboard:
View attachment 18145

SE Reloaded Motherboard:
View attachment 18144

My testing of 0.6e showed the same old problems:

1. Artifacts (see above) at first power-on, but I can usually get a normal screen if I power off for a few seconds and then power-on again.

2. Mouse movement is bad when using my SE Reloaded board, but no issue with the stock motherboard. It's also clicking on its own again! Nope, it can't be the SE Reloaded board being bad because I don't get that in stock 8MHz mode or with the 16MHz 68000 Levco SpeedCard installed.

On the upside though, Tetris Audio is as good as stock. Benchmarks in Speedometer 3.23 & SCSI Director Pro 4.0 are identical to firmware 0.6b.

I made a video to show you all the details, but it's been rendering in Compressor for more than 1 hour and still has a ways to go. I am dead tired and need to go home. I will uploaded the video tomorrow morning.

When the Mac boots up, the RAM contents and thus the screen contents are undefined. Until the 68k runs code to clear the screen, it'll look like that. What you're seeing is the uninitialized contents of RAM. If you power on and let the Mac draw something, then power off and on quickly while holding reset, you will see sort of a faded ghost of what was there. Such is the nature of DRAM. While flashing, this occurs because the USB power holds the Mac in reset, so neither the fast nor 7.8336 MHz CPU can clear the screen. When the WarpSE fails to boot, same thing. It never gets to the screen clear routine.

Anyway, 0.6e had some issues created in the course of fixing other problems. I suspect 0.6f will work better but let me know if it still does not boot reliably.

One more thing--does your Reloaded board have a 15.6672 MHz crystal? Or is it 16 MHz? I'm trying to get to the bottom of the phantom clicking and mouse stutters.

 

[Kai Robinson]

Pretty sure I sent a brand new 15.6672MHz one to @JDW

 

[Zane Kaminski]

Pretty sure I sent a brand new 15.6672MHz one to @JDW

Thanks for the info! I know some people just use a 16 MHz crystal, which I am skeptical of

 

[ppuskari]

Booting with 7.1 and firmware F seems to be okay now on my machine with the dual external zip 100 scsi drives and no internal HD or internal termination. 4megs ram. SCSI drive validations with Apple HD SC Setup 7.3.5 seems to be just fine so banging on the SCSI buss seems to work.

Localtalk (SCC) ports are still not working with AppleTalk shares or printers, etc. Can you make a build with SCC slowdowns? Will test for sound in a little bit. Address Errors on shutdown or restart seem to be fixed in "f"

UPDATE: Nope Address Error showed back up again on shutdown after it sat running for about 45 minutes.

 

[JDW]

@ppuskari
I've not tested LocalTalk, but I have used my home-made Serial Loopback Plugs (that link to my video shows you how to make them) in combination with Snooper 2.0 to check the Serial Ports, and Snooper gave my motherboard (with WarpSE installed and enabled) a clean bill of health from its Serial Ports test.

@Zane Kaminski
I've downloaded 0.6f, but testing will need to wait until my lunch break.

@Kai Robinson
Those who wish to see exactly how my SE Reloaded board was built and with what components were used need only check my SE Reloaded Playlist here:

Finding a particular part is easy. Just access my BOM here. It tells you in which of my SE Reloaded videos a given part was installed.

More specifically, U10D, the 15.6672MHz XTAL is mentioned at 46:48 in my Part II video.

---

Here's the video I made last night (3 videos I merged into 1), which I still feel is relevant and worth watching, despite the new 0.6f firmware:

If my testing of 0.6f later today leads us nowhere, it may be best for me to ship my SE Reloaded motherboard to Zane for analysis. (At my expense, of course.) And I can include my Levco SpeedCard too, because it needs to be determined why it has no mouse issues on the SE Reloaded motherboard, yet WarpSE does. Both use a 68000 CPU, although the SpeedCard only runs at 16MHz, versus 25MHz for the WarpSE.

 

[Zane Kaminski]

Thanks for all your hard work testing!!


Okay so there seem to be four remaining issues/symptoms:

1. Boot failure -- sometimes the system displays bars on boot. There has been several causes for this. 0.6a, b, c, d, and e all had their own independent issues causing this. Maybe it's fixed now?
   
   
2. LocalTalk/SCC issues -- will have to investigate this. @ppuskari LocalTalk worked in 0.6a for you but has been broken since 0.6b? The 0.6b changes were strictly limited to slowdown and the init sequence. Actually slowdown has been enabled for all I/O devices since 0.6a and earlier but in 0.6b the slowdown period was extended from something like 35 microseconds to 200 microseconds. Maybe this is too long? On the initial build on the card when I sent it, there was no slowdown except for sound, and that actually worked better.
   
   
3. Address error on shutdown -- @ppuskari When did this start? Was it in 0.6a or 0.6b?
   
   
4. Mouse jumpiness -- tricky issue... will have to think hard about this

@JDW I'd feel uncomfortable borrowing your Reloaded board unless absolutely necessary so let's try exhausting all the other options first. And of course GW would pay for shipping both ways!

 

[JTRetro]

So I tried to fire up my SE Superdrive......and nothing. Long story short, it looks like the power supply in it is in dire need of a re-cap. Just another of the 138 or so projects around here that I need to get to at some point in time......

Anyways, I pulled SE Superdrive board out of that machine and put it in my 800k machine (the one that I have been using) for testing purposes for the moment. Specifically, I wanted to test both the internal and external floppy ports for 1.44Mb support.

The first test I did I went all out: I used a PC-formatted 1.44Mb Floppy to see the machine would both be ale to read it and access it; it was able to do both.

At the same time, I attached my BMOW Floppy Emu drive to the external floppy port and tested it with a 1.44Mb image on the drive. Again, I had no problems with access:

With this test, I once again used my original 20Mb Harddrive with O.S. 7.1 and 4Mb of RAM. Additionally, I had several devices on the SCSI chain: an Apple CD300 drive, a 100Mb Iomega drive, and an external SCSI2SD drive terminating the chain.

 

[Zane Kaminski]

So I tried to fire up my SE Superdrive......and nothing. Long story short, it looks like the power supply in it is in dire need of a re-cap. Just another of the 138 or so projects around here that I need to get to at some point in time......

Aha!! So maybe the bars issue you were seeing at powerup was from your PSU. If that's the case then I think I all of the power-on issues have been accounted for.

What version are you on currently?

 

[JDW]

@Zane Kaminski

My first test of 0.6f shows no artifacts at all at first power on. That's a VERY positive sign! But I need to disconnect for a while and then try again to be sure. I will do that tonight after work.

Sadly, the mouse problems remain.

I filmed my short testing session for you here:

 

[Zane Kaminski]

@Zane Kaminski

My first test of 0.6f shows no artifacts at all at first power on. That's a VERY positive sign! But I need to disconnect for a while and then try again to be sure. I will do that tonight after work.

Sadly, the mouse problems remain.

I filmed my short testing session for you here:

Thanks for your help!! And there's no longer any need to power up and down a bunch of times or wait between power cycles to check for the failure to POST. There were various issues causing that in previous versions and I think they have all been rectified if indeed @JTRetro's issues were due to his power supply. Also no need to probe reset anymore since I think I have ruled out any issues with the initialization sequence. Also the audio and floppy fixes seem stable as well, so no need to test those extensively either although we may have to back off on the slowdown somehow to fix LocalTalk. So maybe audio and floppy will need to be re-tested later.

Soon I am going to determine some more specific next steps for fixing the remaining issues, which I think are just the mouse issues, address error on shutdown, and the LocalTalk issues

 

[JDW]

Tested 0.6f again after work:

1. No artifacts at Power-ON. I think you eradicated that problem entirely, Zane. Great job!

2. SCSI Director Pro 4.0 scores are exactly the same as I showed in this post where I was testing 0.6b.

3. Speedometer 3.23 scores are only slightly higher (within margin of error) than what I reported here with 0.6b.

4. Tetris Audio sounds great! Just like the stock SE to my ears.

5. I spent 2 hours testing, working in the Finder, launching Utilities, playing various games, booting from my FloppyEMU in HD20 mode. No crashing or freezing at all. BTW, "Mac Man" operates faster than normal with WarpSE, but the game is still very playable. Overall, WarpSE works quite nicely.

6. Mouse continues to be a problem. I'm good at Crystal Quest, but I could only achieve a score of 420,000 due to the mouse glitches shooting me all the way across the screen or suddenly slamming me into obstacles. So while you might be able to live with it if you only click around the Finder, game play when using the mouse is definitely out. So is precision painting in graphics apps. Why this only affects my SE Reloaded board is a mystery. But like I said, if I press INTERRUPT at power-on to disable WarpSE, all is well with the mouse. If there are specific spots on the motherboard or WarpSE you need me to test with my oscilloscope, just let me know where and I'll be happy to do that.

 

[Zane Kaminski]

@JDW Thanks for your help testing!!

I think the mouse issues may be related to the additional 16 fixed wait state sound slowdown function. Looking back at the thread, I think the mouse issues were highly correlated to whether the version you were testing had extra sound slowdown. I will have version 0.6g ready soon which removes sound-specific slowdown but still of course goes into PDS-synchronized slowdown after accessing the sound buffer. The slowdown still lasts for 200 microseconds. That was the main change I implemented to improve the quality the rest of the way compared to when the interval was 35ish nanoseconds. I am not sure if the sound quality will be affected by the elimination of the "fixed wait state" feature, so can you take a recording from the current version 0.6f as well as the future version 0.6g to compare? Then if there has been no regression in sound quality we can just take the "fixed wait state" sound slowdown out permanently. Hopefully that will fix the mouse, although if it does, I can't say I understand why.

 

[Kai Robinson]

@JDW are you using the original ADB chip or the new PIC16F84 one?

Tested 0.6f again after work:

1. No artifacts at Power-ON. I think you eradicated that problem entirely, Zane. Great job!

2. SCSI Director Pro 4.0 scores are exactly the same as I showed in this post where I was testing 0.6b.

3. Speedometer 3.23 scores are only slightly higher (within margin of error) than what I reported here with 0.6b.

4. Tetris Audio sounds great! Just like the stock SE to my ears.

5. I spent 2 hours testing, working in the Finder, launching Utilities, playing various games, booting from my FloppyEMU in HD20 mode. No crashing or freezing at all. BTW, "Mac Man" operates faster than normal with WarpSE, but the game is still very playable. Overall, WarpSE works quite nicely.

6. Mouse continues to be a problem. I'm good at Crystal Quest, but I could only achieve a score of 420,000 due to the mouse glitches shooting me all the way across the screen or suddenly slamming me into obstacles. So while you might be able to live with it if you only click around the Finder, game play when using the mouse is definitely out. So is precision painting in graphics apps. Why this only affects my SE Reloaded board is a mystery. But like I said, if I press INTERRUPT at power-on to disable WarpSE, all is well with the mouse. If there are specific spots on the motherboard or WarpSE you need me to test with my oscilloscope, just let me know where and I'll be happy to do that.

@JDW are you using the original ADB chip or the PIC16F84 remake? It might be worth swapping it out for one if using the original as i've seen plenty of marginal OG ADB controllers fail in this way.

 

[JDW]

@JDW are you using the original ADB chip or the new PIC16F84 one?

@JDW are you using the original ADB chip or the PIC16F84 remake? It might be worth swapping it out for one if using the original as i've seen plenty of marginal OG ADB controllers fail in this way.

Original ADB chip from the stock SE motherboard you shipped me, Kai.

But like I’ve said, the mouse works perfectly fine with no accelerators installed. The mouse also works perfectly fine with the 16 MHz 68000 Levco Speedcard installed.

 

[ppuskari]

Thanks for all your hard work testing!!


Okay so there seem to be four remaining issues/symptoms:

1. Boot failure -- sometimes the system displays bars on boot. There has been several causes for this. 0.6a, b, c, d, and e all had their own independent issues causing this. Maybe it's fixed now?
   
   
2. LocalTalk/SCC issues -- will have to investigate this. @ppuskari LocalTalk worked in 0.6a for you but has been broken since 0.6b? The 0.6b changes were strictly limited to slowdown and the init sequence. Actually slowdown has been enabled for all I/O devices since 0.6a and earlier but in 0.6b the slowdown period was extended from something like 35 microseconds to 200 microseconds. Maybe this is too long? On the initial build on the card when I sent it, there was no slowdown except for sound, and that actually worked better.
   
   
3. Address error on shutdown -- @ppuskari When did this start? Was it in 0.6a or 0.6b?
   
   
4. Mouse jumpiness -- tricky issue... will have to think hard about this

@JDW I'd feel uncomfortable borrowing your Reloaded board unless absolutely necessary so let's try exhausting all the other options first. And of course GW would pay for shipping both ways!

I saw the address error in OS 7.1 which is what I am running for all the test boots so far, I saw it sporadically on fastscsi as well as on the couple of times I got D to boot as well as f and e. I have NOT tried A yet I made it down to B where it sorta works then tried your other newer firmware. I may go back to Fastscsi for a bit just to see the accelerator working with Localtalk again, however. I know JDW gets a pass with TechTool testing but I can definitely say it doesn't work. IF I boot with the accelerator OFF via the programmer key, then Localtalk works just fine off the same boot media.

 

[ppuskari]

Thanks for all your hard work testing!!


Okay so there seem to be four remaining issues/symptoms:

1. Boot failure -- sometimes the system displays bars on boot. There has been several causes for this. 0.6a, b, c, d, and e all had their own independent issues causing this. Maybe it's fixed now?
   
   
2. LocalTalk/SCC issues -- will have to investigate this. @ppuskari LocalTalk worked in 0.6a for you but has been broken since 0.6b? The 0.6b changes were strictly limited to slowdown and the init sequence. Actually slowdown has been enabled for all I/O devices since 0.6a and earlier but in 0.6b the slowdown period was extended from something like 35 microseconds to 200 microseconds. Maybe this is too long? On the initial build on the card when I sent it, there was no slowdown except for sound, and that actually worked better.
   
   
3. Address error on shutdown -- @ppuskari When did this start? Was it in 0.6a or 0.6b?
   
   
4. Mouse jumpiness -- tricky issue... will have to think hard about this

@JDW I'd feel uncomfortable borrowing your Reloaded board unless absolutely necessary so let's try exhausting all the other options first. And of course GW would pay for shipping both ways!

I will test with 6a in a little bit and see if Appletalk is broken or working. I can go back but I believe the last time I saw AppleTalk working was with "Fastscsi" build. so far I have tried all the builds except for 6a because I didn't get that far with thinking b was the stable one working down from D-B then E and F for my testing.

For the address error issues I will go back and test with fast scsi and try the others again to rule things in an out including no accelleration.