Falcon clock buffer patch V2 progress

General discussions or ideas about hardware.
User avatar
kohli79
Posts: 63
Joined: Fri Feb 05, 2021 8:23 pm
Location: Saxony Anhalt, Germany

Re: Falcon clock buffer patch V2 progress

Post by kohli79 »

exxos wrote: Thu Jan 27, 2022 9:24 pm Looking at the files for my buffer patch V3 (not yet produced) , it seems for some reason I added another delay to the "DELAY ON" thing. As it was about two years ago I cannot really remember why :lol: Running a quick simulation it seems I increased the delay by 10ns.

There was a post in the German forum were reporting that the delay on was causing problems. Maybe @kohli79 was involved and can remember ? I cannot remember if this was on a stock machine or not ?

I might have increased the delay further to mimic the gated delay on the Phantom. But by all accounts, the delay is probably not really needed And could be creating problems.

At this point I am unsure if to leave that on the board, the delay and the extra delay :shrug:
My falcon board that you had last year had these problems with delay on. All before the 100K resistor was installed. I then resoldered it to delay off and sent it to you. I did a bit of testing on another Falcon. With your Colckpatch (delay on) but without 100K resistance almost the same problems as mine. Then 100K resistor soldered in, sound problems gone, DSP test program all green.
But after 2-3 hours under load/stress, reading errors occurred. After soldering to delay off it was gone and everything was stable!

I had already written something here, penultimate post.
https://www.exxosforum.co.uk/forum/viewt ... 2&start=40

From the German Atari Home Forum. And yes, the Falcons were original without accelerators etc.
https://forum.atari-home.de/index.php/t ... 28.40.html
Life is too short not to have an Amiga/Atari. 8-)
User avatar
exxos
Site Admin
Site Admin
Posts: 23493
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Falcon clock buffer patch V2 progress

Post by exxos »

kohli79 wrote: Fri Jan 28, 2022 12:11 pm After soldering to delay off it was gone and everything was stable!
Thanks for the update! Yeah I am really wondering if the delay is going to cause more problems rather than solving them.
Steve wrote: Thu Jan 27, 2022 11:45 pm Hmm I think if it doesn't add any extra cost to the production but you get three choices: no delay, delay, extra delay. That would be something useful.
It wouldn't add cost, just makes the board a fraction larger because of the extra parts.If I was assembling them, then I would go for a more minimal setup, but as I am looking for a production build then it does not matter so much. I guess I will leave the delays in there for now in the hope that some people will test delay on and off and report back. Then if the delay on is causing problems ,I will just leave it off in future batches.
https://www.exxosforum.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxosforum.co.uk/atari/store2/ - All my hardware mods for sale - Please help support by making a purchase.
viewtopic.php?f=17&t=1585 Have you done the Mandatory Fixes ?
Just because a lot of people agree on something, doesn't make it a fact. ~exxos ~
People should find solutions to problems, not find problems with solutions.
Steve
Posts: 2570
Joined: Fri Sep 15, 2017 11:49 am

Re: Falcon clock buffer patch V2 progress

Post by Steve »

Yeah having the options is best, I know in my testing that the delay allowed my bus speeder to work. Without delay it didn't.
User avatar
exxos
Site Admin
Site Admin
Posts: 23493
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Falcon clock buffer patch V2 progress

Post by exxos »

Steve wrote: Fri Jan 28, 2022 1:19 pm Yeah having the options is best, I know in my testing that the delay allowed my bus speeder to work. Without delay it didn't.
In that case, the delay is causing a stock machine to malfunction, but not bus speeders.

Now I have woken up a bit, I think the original Phantom patch had multiple gates all in series for the delay. But on my current V2 boards, it only actually had 2 gate delays because I only have one IC on there. so I think I "emulated" the additional gated delay and artificially added 10ns extra to bring it in line with the original Phantom delay..

But again if the Phantom works with just 2 gate delays anyway, the extra delay probably be pointless, but like you say, at least it would give people the option to try it. Better to have and not need it kinda thing..

I was toying with going with smaller SMT parts as I am using 1206 sized parts. But I went for that in part because of their higher power rating and easier to assemble by hand. I was thinking of going down to 0805 to make the board smaller, but if someone shorts the board out then I would rather have a higher wattage resistor in place...
https://www.exxosforum.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxosforum.co.uk/atari/store2/ - All my hardware mods for sale - Please help support by making a purchase.
viewtopic.php?f=17&t=1585 Have you done the Mandatory Fixes ?
Just because a lot of people agree on something, doesn't make it a fact. ~exxos ~
People should find solutions to problems, not find problems with solutions.
User avatar
kohli79
Posts: 63
Joined: Fri Feb 05, 2021 8:23 pm
Location: Saxony Anhalt, Germany

Re: Falcon clock buffer patch V2 progress

Post by kohli79 »

Maybe interesting, the Falcon I used for testing later had a Magnum Fast Ram card installed.
So far there have been no issues related to the Clockpatch/100K resistor. Everything is stable!

https://www.stcarchiv.de/stc1996/03/mag ... -ram-karte
Life is too short not to have an Amiga/Atari. 8-)
User avatar
exxos
Site Admin
Site Admin
Posts: 23493
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Falcon clock buffer patch V2 progress

Post by exxos »

This is what the V3 boards will basically look like.

1.jpg
1.jpg (10.78 KiB) Viewed 3287 times
https://www.exxosforum.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxosforum.co.uk/atari/store2/ - All my hardware mods for sale - Please help support by making a purchase.
viewtopic.php?f=17&t=1585 Have you done the Mandatory Fixes ?
Just because a lot of people agree on something, doesn't make it a fact. ~exxos ~
People should find solutions to problems, not find problems with solutions.
Rustynutt
Posts: 224
Joined: Fri Sep 29, 2017 8:24 am
Location: USA

Re: Falcon clock buffer patch V2 progress

Post by Rustynutt »

A bit of a late follow up, want to present my findings.
Recently testing several buffers. One coming with the Nemesis, couple of "Atari" versions, Centuriontech and the last Exxos buffer with DMA delay ability.

Focusing only on the Exxos buffer, machine stability and SCSI data integrity.
None of the others managed to operate under all test conditions successfully.

Guess I could check audio but use digital for that anyway and never noticed an issue. The DMA test gets green screen. Couldn't hear clicks at my age anyway, not knowing if it's a bone or not.

The COMBEL clock speed change is arrived simply by swapping out the main oscillator.
Tried to explain how "simple" this is to do on the Falcon, without need for a Nemesis or Phantom, but failed miserably explaining it :)

In short

"Stock"
Falcon at 32 MHz COMBEL clock, Exxos patch with delay enabled.
No stability issues, SCSI integrity intact. Verified by transferring 1GB of disk data IDE to SCSI CF, HD and using Kobolds Correct and Treecheck to verify.

Falcon at 46 MHz COMBEL clock, Exxos patch with delay enabled.
No stability issues, SCSI integrity intact. Verified by transferring 1GB of data IDE to SCSI CF, HD and using Kobolds Correct and Treecheck to verify.

Falcon at 55 MHz COMBEL clock, Exxos patch with delay enabled.
No stability issues, SCSI integrity intact. Verified by transferring 1GB of data IDE to SCSI CF, HD and using Kobolds Correct and Treecheck to verify.

Not tested with the CT, don't see where there would be any difference in performance there.

Now the other beast, the Afterburner, no matter what type of buffer is used, the only way I've managed reliable data transfer is to reduce BUS speed to ~12MHz. With the Exxos patch, the above mentioned test conditions are successful there. This has been an existing issue since the board was available in 1996.

This new version of the buffer, with additional delay (+10ns) would be of interest. Less expensive than an oscilloscope and degree anyway :)

Played with some mods for the Afterburner only, using a ripple counter to step down the DMA clock input to 1/4 the COMBEL clock, 1/2 BUS (COMBEL input 32/2=16/2=8 on DMA) (placement near the DMA as in the Phantom gate mod) with limited success. There is about the same delay there at the proposed patch, but in this instance, think ran into sync issues between COMBEL and DMA data transfers by changing the clock speed with the ripple counter (totally clock patch unrelated here). If this is confusing, just ignore it. Only present it for academic purpose.

Didn't install the resistor in a previous Falcon repair thread, and seeing how well everything is working, best to leave well enough alone.

So very happy camper (caravanner?) now :)
User avatar
exxos
Site Admin
Site Admin
Posts: 23493
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Falcon clock buffer patch V2 progress

Post by exxos »

Thanks for the update, did you try with the delay disabled ?

I would also recommend doing the resistor fix, I've not done it on my falcon and it works fine. But as there is a known issue, doing the fix sooner rather than later can't hurt.

I've got the new boards made and came a couple days ago. But need to test first. But my RSI in my hands is acting up bad , so need to take a break from many things now for a couple weeks while I recover.
https://www.exxosforum.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxosforum.co.uk/atari/store2/ - All my hardware mods for sale - Please help support by making a purchase.
viewtopic.php?f=17&t=1585 Have you done the Mandatory Fixes ?
Just because a lot of people agree on something, doesn't make it a fact. ~exxos ~
People should find solutions to problems, not find problems with solutions.
Rustynutt
Posts: 224
Joined: Fri Sep 29, 2017 8:24 am
Location: USA

Re: Falcon clock buffer patch V2 progress

Post by Rustynutt »

exxos wrote: Sun Feb 20, 2022 12:28 am Thanks for the update, did you try with the delay disabled ?

I would also recommend doing the resistor fix, I've not done it on my falcon and it works fine. But as there is a known issue, doing the fix sooner rather than later can't hurt.

I've got the new boards made and came a couple days ago. But need to test first. But my RSI in my hands is acting up bad , so need to take a break from many things now for a couple weeks while I recover.
With delay off anything over about 38MHz caused problems.
Enabled, sky's the limit on the (1) stock Falcon. 55MHz is usable, but CPU does get very hot.
Think 50MHz is prolly the optional to be used. IMHO.
Definitely improves the DMA clock issues at higher clock speeds.

Yes for the resistor, before suffering a major setback :) After 25 years of dangling in the air, the DMA clock input pin broke off!
Had to break out the drill press to bore a hole to access a piece to connect a wire too. I'll not be touching that area again until actually
attempting to replace the DMA IC :oops:

Don't recall any difference with my setup with or without the resistor, but that's just seat of the pants.
Read the post when you repaired and added the resistor mod to a Falcon improving the clock balance between HI and LO.
Sure there's an electronic term for this :)

Can't read the REV on this board, it's one of the later versions if not last before CLAB bought them up.
It's been the Afterburner Falcon since 1996 or so.

Purchased 2 more main boards that came direct out of Atari Sunnyvale never sold.
One, no Rev No. is a Peritel board, B52A04000100.
It was tagged "bad solder joint".
Just put a new ROM in and it booted right up.
The only issue I could find was the DSP wasn't functional. Ran the SRAM test, and indicates one is bad. Have the IC, not got to it yet.

Would like to test the new buffer boards when you get to them. Have neuropathy from the neck down, know about pain and not being able to do things requiring dexterity, or without pain :(
User avatar
exxos
Site Admin
Site Admin
Posts: 23493
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Falcon clock buffer patch V2 progress

Post by exxos »

Rustynutt wrote: Sun Feb 20, 2022 12:58 am With delay off anything over about 38MHz caused problems.
So a further delay may help but hard to tell currently.

Rustynutt wrote: Sun Feb 20, 2022 12:58 am Would like to test the new buffer boards when you get to them.
I have the boards but want to test in my own falcon before I send out to beta testers ( can mark you down for one if you want).
Rustynutt wrote: Sun Feb 20, 2022 12:58 am Have neuropathy from the neck down, know about pain and not being able to do things requiring dexterity, or without pain :(
:( yep I'm typing on my tablet and its hard going. Every key press is like a sledge hammer across the back of my hands. Fine control I have problems with. It's why I don't solder much anymore as well.
https://www.exxosforum.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxosforum.co.uk/atari/store2/ - All my hardware mods for sale - Please help support by making a purchase.
viewtopic.php?f=17&t=1585 Have you done the Mandatory Fixes ?
Just because a lot of people agree on something, doesn't make it a fact. ~exxos ~
People should find solutions to problems, not find problems with solutions.
Locked

Return to “HARDWARE DISCUSSIONS”