JOY - a new Falcon030 demo from New Beat

[jd]

If anyone else is having problems running JOY, please get in touch with me. Thanks.

For a few days on the main board, on which this demo does not work, I tests CT60e, unfortunately the same, it confirms so much that it is certainly not a problem with the power supply or capacitors. Unfortunately, even after 80 minutes, this demo does not work at random. Why it work so before replacing the power supply I do not know, maybe it was a random case that it came to an end.

This is just my suggestion and my idea, I tested this demo on three other mainboards, where the only change is a small patch installed from the website of R. Czuby. It's exactly about the soldering resistor, which is well shown in the picture below.

SDMA.jpg

On each such motherboard works perfectly.
This change on these three mainboard were made by my friend.
The only idea is to install such a patch also on this main board, I do not have another idea at the moment.

Did this resistor patch work on a non CT60e Falcon?

[jd]

Ah, that's interesting. That happens to a friend of mine as well. His problem was that his Blitter does not do well while the machine is fairly cool. Anything using the Blitter extensively will crash the machine when cool. If he starts the machine and lets it heat up for some 30 minutes, the demo works. The demo uses the Blitter pretty heavily. Dunno if you have the same problem though, but so far I haven't heard about anyone else having problems that have not been hardware related.

Exactly what happens to mine, can run all other demos from cold then about approx half hour 'warmup' and only then will Joy run flawlessly. My falcon is from new and has had the mandatory clock patch at the time (piggy backed buffer chip type). I haven't yet done the full motherboard recap and runs from ATX PSU, pretty much always has as original blew early in life (5.01V and 11.59V when running the demos) , no other additions apart from powerup2 (disabled for demos).

While its 'warming up' if i have NVDI running Joy can sometimes play through fine (as blitter hardware replaced by software?), have to wait a bit longer if i run with a bare bones nothing loaded, always running RGB for this. Also passes the DMA2DSP test (green pattern always)

Very odd but maybe it does need a full recap + newer super clock patch (phantom) ??

I don't think this is clock patch related, since it used to work fine and i never installed it. Now it does not anymore, idon't know why.
One thing i noticed, using the diag cart : If i do the quick blit test with the blitter disabled, it passes. If i turn the blitter on, i got a red screen with the following message :
EA Address error Access address : 01FFA0 Program Counter : FA3F08
EA Address error Access address : 01FF7A Program Counter : FA3F08

Must be the same on all Falcons as blitter off (default) passes blit test, blitter on exactly above whether its cold or 'warmed up'

Can anyone else confirm if their falcon passes test with blitter turned on in diag cart : D (diag menu) > E (modify sys) > C (sys controls) > E (enable blit). Then run Blit test again (G).

[mikro]

Must be the same on all Falcons as blitter off (default) passes blit test, blitter on exactly above whether its cold or 'warmed up'

Can anyone else confirm if their falcon passes test with blitter turned on in diag cart : D (diag menu) > E (modify sys) > C (sys controls) > E (enable blit). Then run Blit test again (G).

I wasn't exactly thrilled to reset the NVRAM but what one doesn't do for a fellow Atarian. ;-) On my Falcon with CT60e (booted in 030 mode, of course) I see the two messages although with a different addresses (01FFA8 & 01FF82, PC is the same as in your report).

It's quite possible that the test is disabled for this very reason, it doesn't work. ;-)

I am able to run Joy without issues and I have just some very basic clock patch installed.

[LaceySnr]

Seeing this in the recent posts made me have another go at running it on my machine. Pretty sure the last time was over a year ago, and it never got all the way through without crashing. Yesterday it got most of the way through then stopped. Today I powered on, left the machine for maybe 10-15 minutes, and I've just got all the way through no problem. Running it again now to see what happens.

I have moved in the meantime (hemispheres no less) but my home office is warmer than my old one (at least in Winter). My Falcon's not been recapped or had a clock patch (did do the NVRAM) so quite impressed it can get through ok. Recap for the PSU to come soon just for the sake of sanity.

Still, I've watched the demo a few times on YouTube before, but it's great to see it on hardware. As fresh as it was when you released it!

[jd]

Seeing this in the recent posts made me have another go at running it on my machine. Pretty sure the last time was over a year ago, and it never got all the way through without crashing. Yesterday it got most of the way through then stopped. Today I powered on, left the machine for maybe 10-15 minutes, and I've just got all the way through no problem. Running it again now to see what happens.

I have moved in the meantime (hemispheres no less) but my home office is warmer than my old one (at least in Winter). My Falcon's not been recapped or had a clock patch (did do the NVRAM) so quite impressed it can get through ok. Recap for the PSU to come soon just for the sake of sanity.

Still, I've watched the demo a few times on YouTube before, but it's great to see it on hardware. As fresh as it was when you released it!

I have exactly this problem with mine, if i run NVDI it speeds up the 'warmup' time and is more stable. I have the bird running off an old ATX PSU (to be replaced with a nice modern Amiga PSU this winter) need to recap motherboard also. It has the original clock patch on it (piggy backing chip) so might replace with a more modern clock patch too depending on how brave i feel.

Let us know how you get on with it and whether just the PSU recap fixes it bit i would suspect it defo needs a clock patch and maybe recap motherboard (the big 4700uf at the minimum).

Good luck.

[Sturm]

Mhhh i'm confused now, joy 1.1 works flawlessly in 060 mode, but both versions crash in 030 mode

[dhedberg]

When running version 1.1 in 060 mode the Blitter is not used at all, but in 030 mode it's used pretty extensively (both in version 1.0 and 1.1).
It seems like quite a few Falcons have issues when hammering the Blitter. It's also a chip that is not commonly used in Falcon software, which could explain why people never experienced issues with their machines until they found Joy!

[dlfrsilver]

Mhhh i'm confused now, joy 1.1 works flawlessly in 060 mode, but both versions crash in 030 mode

it works in 030 mode on mine.

[jd]

Seeing this in the recent posts made me have another go at running it on my machine. Pretty sure the last time was over a year ago, and it never got all the way through without crashing. Yesterday it got most of the way through then stopped. Today I powered on, left the machine for maybe 10-15 minutes, and I've just got all the way through no problem. Running it again now to see what happens.

I have moved in the meantime (hemispheres no less) but my home office is warmer than my old one (at least in Winter). My Falcon's not been recapped or had a clock patch (did do the NVRAM) so quite impressed it can get through ok. Recap for the PSU to come soon just for the sake of sanity.

Still, I've watched the demo a few times on YouTube before, but it's great to see it on hardware. As fresh as it was when you released it!

I have exactly this problem with mine, if i run NVDI it speeds up the 'warmup' time and is more stable. I have the bird running off an old ATX PSU (to be replaced with a nice modern Amiga PSU this winter) need to recap motherboard also. It has the original clock patch on it (piggy backing chip) so might replace with a more modern clock patch too depending on how brave i feel.

Let us know how you get on with it and whether just the PSU recap fixes it bit i would suspect it defo needs a clock patch and maybe recap motherboard (the big 4700uf at the minimum).

Good luck.

Just an update on my Falcon, FINALLY got round to recapping the whole machine and the new external medical grade PSU (dare i say an Amiga one), which gives a rock solid 4.93v no matter what the Falcon is doing (floppy access, DSP decoding, AudioTracker, Aniplayer etc.) changed NVRAM and still have the same issues with this demo.

I already had a clock patch on it (original atari piggy back one) but after trying the demo with a 8MHz blitter (done using FALCONSX.ACC) the demo actually ran and didn't crash at the usual early place (just after bee disappears and the cube flies around the screen) so must be timing/clock related.

Enter new clock patch . . . phantom one to be exact . . . and the results . . . . .

It works, brilliantly!

Reliably this demo now flies (Time demo also used to crash) and adds the extra of knowing my Falcon was always a bit broke, now every demo I throw at it works, still trying to see if i can break it though, PowerUp2 connected with clock patch all good, BadMood now even more reliable.

Moral of the story (as mentioned in previous posts) - Your Falcon might 'work' but in regards to clock patch or to not clock patch if you run this demo on an otherwise working falcon and it crashes you NEED a clock patch or in my case a BETTER clock patch.

[viking272]

Nice update jd, thanks for that.

[mrbombermillzy]

@jd Theres nothing like a properly sorted F030!

[stormy]

Hey jd, did you make your own phantom clock patch? Did you break the sdma trace and run your own cable etc?

[jd]

Hey jd, did you make your own phantom clock patch? Did you break the sdma trace and run your own cable etc?

If I'm guessing handles correctly I bought one off you (? @Steve from Exxos forum ?) at the last Cyber Legend event in the UK, the PCB has Steve at Solid State Society on it and its website has Stormy as email contact.

Yep cut trace by SDMA, ran trace there and soldered 150pf cap from there to Ajax GND as per phantom patch instructions :

2023-10-01_17h37_23.png

What i did notice is there are two additional caps on the PCB which is in addition to the following (from the excellent website - https://mikrosk.github.io/clockpatch/ which imo should be mandatory reading for all Falcon owners) :

2023-10-01_18h02_26.png

Are they just for power filtering?

Seems the SDMA bit was the key, I know Exxos has done a hell of alot of analysis on this but one thing he commented was that the SDMA clock drifts its voltage level over time (the long track must get affected with heat via its route) which would explain what i was seeing originally (Joy would sometimes run ok if I left the Falcon on for an hour or so) so possibly this 'drift' was actually helping my Falcon. Anyways replace current patch with phantom patch, bypass the PCB track (and cut) with the SDMA wire + plus cap to remove any additional interference and it works superbly. Spent another few hours today trying to break it and its been 'perfect'.

So many thanks for the PCB if i indeed have the right person

[stormy]

Yes that's me So glad it worked out for you. The extra ceramic caps on the PCB are just decoupling caps for the IC's. When I spoke to Ctirad the original author, he told me to add them for extra stability. They provide a bypass for transient currents.

[dhedberg]

Reliably this demo now flies (Time demo also used to crash) and adds the extra of knowing my Falcon was always a bit broke, now every demo I throw at it works, still trying to see if i can break it though, PowerUp2 connected with clock patch all good, BadMood now even more reliable.

Thanks for the update! Glad you got it all sorted out!
So it seems that the 'frequent use of the Blitter crashing the machine when cold' problem is actually due to Atari's poor routing of the PCB tracks on the Falcon motherboard. Good to know!

[jd]

Reliably this demo now flies (Time demo also used to crash) and adds the extra of knowing my Falcon was always a bit broke, now every demo I throw at it works, still trying to see if i can break it though, PowerUp2 connected with clock patch all good, BadMood now even more reliable.

Thanks for the update! Glad you got it all sorted out!
So it seems that the 'frequent use of the Blitter crashing the machine when cold' problem is actually due to Atari's poor routing of the PCB tracks on the Falcon motherboard. Good to know!

Yet some falcons with no clock patch from factory condition can run it fine! I think each Falcon will be different so for me a combo of poor routing and better clock stabilisation.

Forgot to say BRILLIANT DEMO!!

When's the next one coming out. . . . .

[LaceySnr]

99% sure mine doesn't have a clock patch - it's been a long time since I opened it up though and I didn't know about them at the time. Going to be recapping the PSU soon so will check, but the need to cut a trace is a little off putting!

[dhedberg]

Forgot to say BRILLIANT DEMO!!
When's the next one coming out. . . . .

Thanks! It makes me very happy to hear that you like it!
Next one has been in the works for quite some time. High-priority rl issues are causing some delays...

[jd]

After speaking to @Stormy at the weekends Cyberlegends event he mentioned that the Phantom clock patch (that i purchased from him the year before ) is on PCBWay if anyone is interested in using the same one i used as he has none left himself :

https://www.pcbway.com/project/sharepro ... 1f878.html

All details/info on that page also.

[dlfrsilver]

OK, sounds like it's the MP2 player (which uses the DMA) that is the problem. Would you mind trying to run the It's that time of the year again demo on your Falcon030 (needs to be run in 030 mode)? It uses the same MP2 player and with pretty much the same configuration as Joy. Also, do you have 4MB or 14MB ST-RAM?

This is another demo that i also need to install on my Falcon 030

Glad you like it. Perhaps you'd like MORE as well?

Hi Daniel, each of them work great

I'm also posting for a falcon 030 user, who has a purple screen and then nothing. Does it sound like something you know or heard about the joy demo reported by other users ?

[dhedberg]

I'm also posting for a falcon 030 user, who has a purple screen and then nothing. Does it sound like something you know or heard about the joy demo reported by other users ?

Yes, I think I remember somebody else having that issue. That's when the music kicks in. So, it freezing at that moment is an indication of either SDMA issues or possibly corrupt MP2 data. If his Falcon has 4MB of RAM the MP2 is streamed from disk which adds another dimension to the problem.

"It's that time of the year again" uses the same MP2 player and streams from disk on 4MB machines. Does it freeze as well?

[dlfrsilver]

I'm also posting for a falcon 030 user, who has a purple screen and then nothing. Does it sound like something you know or heard about the joy demo reported by other users ?

Yes, I think I remember somebody else having that issue. That's when the music kicks in. So, it freezing at that moment is an indication of either SDMA issues or possibly corrupt MP2 data. If his Falcon has 4MB of RAM the MP2 is streamed from disk which adds another dimension to the problem.

"It's that time of the year again" uses the same MP2 player and streams from disk on 4MB machines. Does it freeze as well?

He has 14mb of ram on his falcon, the PSU has been serviced with fresh new capacitors. So the only thing that remains is the clock patch that he doesn't have on his machine (he confirmed it).