FWSI Analogic CD32 Diskdrive issue

[Retro1234]

So even if you put in a music CD you cant use the other buttons or bring up the NV-Ram menu?

Originally Posted by kaffer

Reviving this thread again, I got one of these expansion units this week and also stumbled into the fact that it nobbles gamepad buttons. I can say with certainty that it is a hardware-level issue as my SysTest disk accesses the gamepad hardware directly, and it fails to work on an Analogic-equipped CD32. I observe that the red and blue buttons make all buttons appear to be pressed, while all other buttons do nothing.

My gamepad logic is pretty close to lowlevel.library's. I tried sprinkling millisecond delays when clocking the gamepad's shift register and then accessing potgo (in case the expansion unit accidentally capacitively loads those lines, or something) but no change. It's a mystery. Could the unit be *deliberately* nobbling the gamepad? It was designed to be bolted to the back of CD32s for educational use, so.... maybe...?

[Keir]

From the main screen I can use the red and blue buttons to access the non-volatile store and the language menu. That makes sense though if the main screen is accessing the gamepad in legacy 2-button joystick mode, which works fine with the Analogic.

I haven't tried a music CD. If I try the Diggers/Oscar CD then I can't get past the selection screen (which requires use of shoulder buttons to select which game to boot).

[Retro1234]

Every time someone says this I always ask about the lowlevel.lib and never get a reply, I know you say its not but for conclusive I dont know Diggers - try music CD or the game legends that dont require lowlevel.lib.

[Keir]

Originally Posted by Retro1234

Every time someone says this I always ask about the lowlevel.lib and never get a reply, I know you say its not but for conclusive I dont know Diggers - try music CD or the game legends that dont require lowlevel.lib.

Again: My own test suite, downloadable via a link here https://github.com/keirf/Amiga-Stuff, does not use lowlevel.lib for gamepad access and does not work with the Analogic plugged in. That's definitive as far as I'm concerned.

And anything that uses only 2 buttons (the red and blue) may not be accessing the controller in gamepad mode anyway. In which case it will work with or without the Analogic just fine.

[Keir]

Originally Posted by kaffer

And anything that uses only 2 buttons (the red and blue) may not be accessing the controller in gamepad mode anyway. In which case it will work with or without the Analogic just fine.

In fact if red and blue individually work then it's certain that gamepad-access mode is not being used. Because in gamepad-access mode, red and blue cannot be distinguished: they have the same effect (all gamepad buttons appear to be pressed when either button is pressed).

[Retro1234]

I want to test your program whats its name I cant see a program of that description in your link?

I think I see it
- The CD player uses all buttons - hope someone can help you.

[Keir]

Originally Posted by Retro1234

I want to test your program whats its name I cant see a program of that description in your link?

Look under the "SysTest Binaries" heading in the README document at the bottom of the linked page.

[HelloBeautiful]

Just got two of these without knowing there's an issue like that. I guess a simple off-switch mod would be enough to make them usable.

[xArtx]

It could hold PotX high (or low whichever it is) to force the game controller into the legacy mode,
to ensure the games you’re likely to play from floppy all work with it.
Assuming it’s the Anagogic floppy expansion... the pics are broken in the link...
it would be interesting to know if another one does the same thing.

[Keir]

Originally Posted by xArtx

It could hold PotX high (or low whichever it is) to force the game controller into the legacy mode,
to ensure the games you’re likely to play from floppy all work with it.
Assuming it’s the Anagogic floppy expansion... the pics are broken in the link...
it would be interesting to know if another one does the same thing.

The POT lines aren't brought out to the expansion connector so it can't be anything that simple. I don't have any theories myself at the moment.

[Toni Wilen]

I can confirm this (I got FWSI drive few months ago).

It looks like CD32 pad code fails to detect the pad because red and blue buttons work fine in built-in CD player. (red = press selected button, blue = stop. Other buttons do nothing)

My guess is that firebutton (which is connected to CIA and part of CIAs are replaced with expansion) fails to work correctly in output mode (Pad's shift register won't advance) causing CD32 ROM code to not detect CD32 pad. This fault would also cause normal CD32 pad read routines to read all buttons pressed when red is kept pressed.

I'll measure pin output levels later.

[Toni Wilen]

Confirmed.

I connected scope to pin 6 of CD32, then checked what happens when writing to $bfe001 and $bfe201 fire button bits.

Writing to $bfe001 bits 6/7 or $bfe201 bits 6/7 does not seem to do anything at all if disk expansion is connected.

But writing to CIA mirror addresses, for example $bfe003 and $bfe203 does change pin logic state. For some strange reason FWSI only enables "cia replacements" when CIA address exactly matches documented CIA address. All mirror addresses still hit the original CIAs (inside Akiko).

(I used this FWSI feature to see how unconnected Akiko IO pins work, I originally thought it was going to be impossible until I noticed mirror addresses having different values compared to "official" addresses)

[Keir]

Originally Posted by Toni Wilen

I can confirm this (I got FWSI drive few months ago).

My guess is that firebutton (which is connected to CIA and part of CIAs are replaced with expansion) fails to work correctly in output mode (Pad's shift register won't advance) causing CD32 ROM code to not detect CD32 pad. This fault would also cause normal CD32 pad read routines to read all buttons pressed when red is kept pressed.

Awesome deduction, and also does fit with my observation that pressing red *or* blue buttons causes all buttons to appear pressed:

Blue pressed: Blue is connected to shift-register parallel input H and this is immediately visible on Q_H output when switched to shift mode. Hence it appears on gameport pin 9 and remains there as we fail to pulse pin 6 (due to FWSI messing it up) and thus do not clock the shift register.

Red pressed: Red is connected to shift-register parallel input G but also to the shift-register clock input. When red is pressed the clock input is held low. When gamepad routine starts it instead connects gameport pin 6 to the clock input, causing it to be pulled high. This rising edge at the clock input shifts G -> H, and hence red button state is visible at Q_H -> gameport pin 9. And remains there as we fail to pulse pin 6, as above.

I'm glad this mystery is solved and I will add FWSI detection via the CIA mirror locations to my gamepad test routine

It also means a hacked lowlevel.library would fix gamepad+fwsi for most games? (Most don't do pad direct access I believe?)

[AJCopland]

Is it possible to directly fix/hack the FWSI hardware to resolve this?

[Amigajay]

Would be amazing if this could be resolved using a hacked lowlevel.library, i've been holding off buying one because of the button issue, but having a floppy to save games, rgb to my collections will be awesome!

[Toni Wilen]

Originally Posted by kaffer

It also means a hacked lowlevel.library would fix gamepad+fwsi for most games? (Most don't do pad direct access I believe?)

Maybe, maybe not.

There are also some that read pad using custom code but are mostly system friendly with other things (cd access etc..)

Originally Posted by AJCopland

Is it possible to directly fix/hack the FWSI hardware to resolve this?

I don't think so. All the logic is inside programmable logic chips.

--

It makes no sense. They created CD32 expansion that can't work with CD32 pad!

There is no way they failed to notice it. Is there some technical reason? Can other floppy expansions do it?

Both fire button lines are connected to expansion connector (and Akiko chip). CIA (=Akiko) button pin is in input mode after reset, expansion device should be able to pull it down or up without problems. Perhaps logic chips they used didn't have IO pins that can handle situation where also Akiko is driving the pin? Does this require some extra hardware? (But on the other hand, if they would have "shadowed" all $bfe201 addreses, there would not have been any way to configure Akiko's fire button pin in output mode)

Oddly enough CD32 schematics shows Akiko firebutton pins being input only which of course can't be correct.

[Amigajay]

Originally Posted by Toni Wilen

It makes no sense. They created CD32 expansion that can't work with CD32 pad!

There is no way they failed to notice it. Is there some technical reason? Can other floppy expansions do it?

It makes sense because it was the Wallstreet Institute using them, so they had the cheap CD based computer and needed a way to save data, hence the floppy addon and clock, all they would have used was a mouse, no need for a joypad.

Perhaps they had a request to disable the buttons to stop workers playing games?!

[Keir]

Originally Posted by Amigajay

Would be amazing if this could be resolved using a hacked lowlevel.library, i've been holding off buying one because of the button issue, but having a floppy to save games, rgb to my collections will be awesome!

I'll be recapping my CD32 in the next week and I will drop in a modified ROM while I have it open and see how that goes.

[nujack]

Originally Posted by kaffer

I'll be recapping my CD32 in the next week and I will drop in a modified ROM while I have it open and see how that goes.

Great!
Would be nice to test it in combination with FMV-Card. With CD32fWSI and FMV-Modul you can only use self starting VideoCD's because Play button is not working.

[Keir]

Originally Posted by nujack

Great!
Would be nice to test it in combination with FMV-Card. With CD32fWSI and FMV-Modul you can only use self starting VideoCD's because Play button is not working.

Well I don't have an FMV module But you could try a modified ROM...