Issue between Rapidus and 16K cartridges

[Joey Z]

So I'm working on a 130XE for someone, for way too long now actually, and I have installed a rapidus and a U1MB into the machine. It took some additional troubleshooting to make it work, but the last problem I can't seem to get rid of is that some cartridges don't load.

 

Basically, rapidus freezes while loading the FPGA core. The loading bar never fills.

 

I've got it narrowed down to 16K cartridges, and I've actually narrowed down that it has nothing to do with the cartridge itself, but that RD4 is pulled high (and thus RAM at $8000 disabled). Has anyone with rapidus been able to consistently get 16K cartridges to load? This machine has no issue with 8K carts at all, just 16K. The worst part, is that I SWEAR I had 16K carts working 2 times exactly (no more, no less) but it wasn't because I fixed anything, or did anything, it was just random chance, apparently.

 

I've removed the U1MB and gone back to the normal MMU and OS ROM, no change. I even reflashed the CPLD and loaded the latest FPGA core, also no change. I've tried different CPUs, no change.

 

I've found other mention of issues with carts, but just from one person: http://atariage.com/forums/topic/257225-rapidus-questions/?p=3597056

 

he says DK didn't work, that's one of the ones I tried here, and it's a 16K cart. Earlier, he said defender didn't work, another 16K cart.

 

Also, inserting a cart when the system is on (after the rapidus core has loaded) and resetting lets the carts load fine. So it's something specifically during the FPGA core load. I don't know, maybe the rapidus designers forgot that there might not be RAM at $8000? sounds doubtful, but not impossible.

Edited April 4, 2017 by Joey Z

[flashjazzcat]

Did you check the firmware sources at bitbucket to see if the core loader uses $8/9xxx as a buffer?

 

https://bitbucket.org/laoo/ptb/overview

[pasiu]

Unfortunately I've never checked Rapidus with 8/16k cartridges, so it seems we missed it during sw development.

I think the bios uses some memory locations at $8xxx, so I will try to work on this to correct it.

[Joey Z]

Did you check the firmware sources at bitbucket to see if the core loader uses $8/9xxx as a buffer?

 

https://bitbucket.org/laoo/ptb/overview

 

I hadn't realized this was available to be viewed, I figured it would have been closed source.

 

Unfortunately I've never checked Rapidus with 8/16k cartridges, so it seems we missed it during sw development.

I think the bios uses some memory locations at $8xxx, so I will try to work on this to correct it.

 

And if this is true, then that's great news, because it's not my problem anymore (because it's fixable in software, and I am just installing the hardware).

[+JAC!]

Just as confirmation: I installed a Rapidus last weekend and came across the same issue. Original 8k cartridges and also Atarimax 1MB work fine. All 16k cartridges stop the FPGA boot process.

[Joey Z]

Just as confirmation: I installed a Rapidus last weekend and came across the same issue. Original 8k cartridges and also Atarimax 1MB work fine. All 16k cartridges stop the FPGA boot process.

yes, I think we're pretty confident this is the issue. Some labels from NewDevice6502.asm:

Scratchpad   = $8000
JumpTableLo  = $8100
JumpTableHi  = $8180
GenCode      = $8200 

So it's definitely using the cartridge region's RAM, which won't work with a cartridge that fills that space. The good news is that it's probably not too bad to fix, just move around the memory map a little bit.

[laoo]

Some time passed, but I've managed to move around the memory map a little bit and prepared a flasher for core loader that does not use memory above address $3000. Pasiu checked that it loads the core correctly but could not check it with any suitable cartridge. Could someone give it a try?

Here is the file: FLND6502.XEX

Edited September 11, 2017 by laoo

[Tigerduck]

Some time passed, but I've managed to move around the memory map a little bit and prepared a flasher for core loader that does not use memory above address $3000. Pasiu checked that it loads the core correctly but could not check it with any suitable cartridge. Could someone give it a try?

Here is the file: FLND6502.XEX

 

Tested with Joust, Hero, E.T. and Moon Patrol (They should have all 16K). Works fine now.

Just one remark, it is not possible to activate the RAPIDUS menu with "RESET and Invers" while Moon Patrol is running.

Maybe a "magic thing" in the programming routines of Moon Patrol.

Thanks a lot for fixing that, laoo!

Edited September 13, 2017 by Tigerduck

[drac030]

Just one remark, it is not possible to activate the RAPIDUS menu with "RESET and Invers" while Moon Patrol is running.

Maybe a "magic thing" in the programming routines of Moon Patrol.

It may be that Moon Patrol's cartridge is a diagnostic cartridge. In such a case entering the menu will indeed not be possible.

[+JAC!]

Yes, the "Allen Miller Bit" is set in Moon Patrol ($BFFD=$80), like in some other early games. It's a diagnostic cartridge.

[FULS]

Yes, the "Allen Miller Bit" is set in Moon Patrol ($BFFD=$80), like in some other early games. It's a diagnostic cartridge.

Thank you for that link,JAC!

Very interesting. David Crane starts talking about it at the 14:40 min. mark.