Undumped Amiga chips

[Duke]

As you may or may not know, the Amiga had a microcontroller as keyboard processor. This chip, a 6500/1 from MOS Technologies, has 2kb of internal ROM. Has this been dumped by anyone? Later Amiga's used the 68HC05, which has a read protection, so it's probably very hard to get a dump.

There is also another (to my knowledge) not dumped ROM chip used in Amigas: The LC6554H LCD controller in CDTV's.

A dump of those chips is needed for the Amiga emulation in MESS (and would maybe benefit WinUAE too).

[Toni Wilen]

Emulating keyboard controller is imho quite pointless (emulation won't be any more accurate with or without it, only slower) except if you want complete emulation for some reason..

Not dumped ROMs:

Action Replay II/III ROMs need "real" dumping. First 4 ROM locations can't be dumped with "may the force be with you"-method.

[Duke]

I agress, it is kinda pointless for WinUAE as there will be no user visible changes, but MESS, like MAME, is striving for the most accurate emulation (and to document the hardware), which of course includes the keyboard controller.

[HymnsomniaK]

Well , I got an Amiga 500 with a CSG 6570-036 from 1985 and an Action Replay MKIII but is there any way to dump them ?

[jmmijo]

Originally Posted by HymnsomniaK

Well , I got an Amiga 500 with a CSG 6570-036 from 1985 and an Action Replay MKIII but is there any way to dump them ?

I thought the only way to do a complete/correct dump of these kinds of (EE)PROM's was to remove them from their sockets or desolder them as the case maybe, and use a ROM Burner to read the data and dump this to a file.

I've read/heard of rom protection that destroys/erases the rom if you attempt to remove it from the circuit or if the battery onboard dies

[ReadOnlyCat]

Thread, rise from your grave!

Originally Posted by Duke

As you may or may not know, the Amiga had a microcontroller as keyboard processor. This chip, a 6500/1 from MOS Technologies, has 2kb of internal ROM. Has this been dumped by anyone? Later Amiga's used the 68HC05, which has a read protection, so it's probably very hard to get a dump.
[...]
A dump of those chips is needed for the Amiga emulation in MESS (and would maybe benefit WinUAE too).

While tinkering with A500/A600 keyboard schematics and looking for official datasheets for the MOS 6570 keyboard controller I ended up on a post on the CBM-Hackers mailing list archive discussing the dump of its ROM.

The cbm-hackers guys have (in 2014, not exactly recent )

1) established that the 6570 is a 6502 + ROM
2) managed to dump its ROM:
http://cbm-hackers.2304266.n4.nabble...3p4659044.html

The approach (for a related chip, the MOS 6500/1) is described here:
http://e4aws.silverdr.com/hacks/6500_1/

A cursory search of "6570" on the EAB seemed to indicate that its ROM dump was only mentioned on your thread as a hypothesis so a necromancy seemed appropriate.

This is clearly very useful for accurate emulation and replacement of failing keyboards. MOS 6570 chips have become quite rare on eBay and cost above 40 USD so it's good that we now have everything that is needed to replace them should we need it.

Cheers

[Daedalus]

This is excellent stuff, and I wouldn't have known how to go about it... But wasn't it common knowledge all along that the keyboard controller was a 6502 core with some ROM and RAM included?

[Toni Wilen]

I am still quite sure it isn't needed for accurate emulation. Protocol and timing is very well known already.

[demolition]

It could be useful to have the 6500/1 ROM if you wanted to build a clone keyboard. Yes, you could implement a compatible functionality in a different uC, but it just wouldn't be the same.

[ReadOnlyCat]

Originally Posted by Daedalus

This is excellent stuff, and I wouldn't have known how to go about it... But wasn't it common knowledge all along that the keyboard controller was a 6502 core with some ROM and RAM included?

I have no doubt this was common knowledge but it was not to this kitten until I researched A500 keyboards a few days ago.

Originally Posted by Toni Wilen

I am still quite sure it isn't needed for accurate emulation. Protocol and timing is very well known already.

Who knows how many bugs lurk in that ROM?
Unless this is known, we cannot be sure that protocol virtualization is enough.

And even if there were none, the point of accurate emulation is to go beyond simple virtualization when corresponding information is available.

[Toni Wilen]

Protocol has to be correct because keyboard connects to CIA serial port (very simple, clock and data lines). CIA wouldn't either accept the data or there would be ghost key codes if protocol would do something weird.

Yes, it would be nice to have it completely emulated (and probably will be someday) but I am certain it makes no difference. Except most likely it will breaks some programs if emulation speed is too fast!

Main "problem" with keyboard is handshake pulse length requirement (too short delay: keyboard internal buffer fills and keyboard stops working). Bad keyboard routines have CPU delays. It would need even more complex (or stupid) workaround. Everyone would assume buggy emulation without automatic workaround..

[Daedalus]

Originally Posted by ReadOnlyCat

I have no doubt this was common knowledge but it was not to this kitten until I researched A500 keyboards a few days ago.

Ha, that wasn't really aimed at you, but at the work in the mailing list in 2014, which still feels quite recent in Amiga terms

It did get me thinking though, the chip is physically large enough to easily accommodate a PCB and suitable modern microcontroller to act as a compatible replacement. Now there's a project for a rainy day... *adds another entry to the never-ending list*

[demolition]

Originally Posted by Daedalus

It did get me thinking though, the chip is physically large enough to easily accommodate a PCB and suitable modern microcontroller to act as a compatible replacement. Now there's a project for a rainy day... *adds another entry to the never-ending list*

Sure, you could run the original ROM code in a 6502 emulator on an ATMega32 for example.
Here's one example: https://github.com/Klaus2m5/AVR_emulated_6502_4SBC

[Daedalus]

*adds yet another similar entry to the list*

[kipper2k]

Hi all,

I have made a replacement a500 KB controller board for those that need theirs replacing, just not got around to putting it out there yet...

http://www.kipper2k.com/a500controll...ontroller.html

[Daedalus]

*crosses two items off to-do list*

Progress

[jotd]

When fixing games, I'm interested that the game behaves like shit on fast machines.

The "keyboard stuck" issue because of bad handshake would be a plus (optionally of course, like blitwait/immediate blit, stuff like that)

[ReadOnlyCat]

Originally Posted by Toni Wilen

Protocol has to be correct because keyboard connects to CIA serial port (very simple, clock and data lines). CIA wouldn't either accept the data or there would be ghost key codes if protocol would do something weird.

You are right of course, the code was very likely tightly verified and its task is simple enough that the odds of there being bugs are super low.

Still, I love the idea that there may be bugs that could be exploited to use the controller in non documented ways. Most likely this is a pipe dream because sending data to the controller seems impossible looking at the schematics but we have seen so many impossible things happen on this machine that I cannot let that go (yet! ).

Originally Posted by Toni Wilen

Yes, it would be nice to have it completely emulated (and probably will be someday) but I am certain it makes no difference. Except most likely it will breaks some programs if emulation speed is too fast!

Main "problem" with keyboard is handshake pulse length requirement (too short delay: keyboard internal buffer fills and keyboard stops working). Bad keyboard routines have CPU delays. It would need even more complex (or stupid) workaround. Everyone would assume buggy emulation without automatic workaround..

Originally Posted by jotd

When fixing games, I'm interested that the game behaves like shit on fast machines.

The "keyboard stuck" issue because of bad handshake would be a plus (optionally of course, like blitwait/immediate blit, stuff like that)

This is a good point (for both) but my opinion is that it should probably not be the job of the emulator to fix buggy programs which already have issues on the actual hardware.
If users cannot be bothered to double check on the hardware then well, sorry for them. They can maintain a wiki page for all buggy games.

IMO fixing games is the job of WHDLoad, not UAE.
This said, I understand why Toni makes that choice.

Originally Posted by Daedalus

Ha, that wasn't really aimed at you, but at the work in the mailing list in 2014, which still feels quite recent in Amiga terms

Oh, I did not take it personally, I was just providing context for my ignorance.
Thanks for checking though!

Originally Posted by kipper2k

Hi all,
I have made a replacement a500 KB controller board for those that need theirs replacing, just not got around to putting it out there yet...

http://www.kipper2k.com/a500controll...ontroller.html

Oh, this is super nice. Thanks kipper2k!

You mention being able to plug into a 1200 but how would that work with only plugging in the 5 signals you mention? The keyboard MCU is going to drive the DATA signal by itself won't it? How can you ensure that your signals prevail?