TFMSX rev 2 features & problems

[terriblefire]

Sorry but i cannot help people flash CPLDs. If i did i'd never do anything else. Which route you go depends on what jtag programmers you have.

The .jed attached to my previous will swap slots 2 and 3 from what you have currently. Its probably worth sending your board back to chucky if the ram tests are failing anyways.

EDIT: These may help

[arkadiusz.makarenko]

Since my knockoffs programmer stopped working after Window 11 Upgrade I switched to Raspberry Pi for programming cpld.

Arm can be programmed using usb A - usb A cable.

[terriblefire]

Since my knockoffs programmer stopped working after Window 11 Upgrade I switched to Raspberry Pi for programming cpld.

Arm can be programmed using usb A - usb A cable.

@arkadiusz.makarenko is there a tool for capturing USB HID keycodes on a PC?

[arkadiusz.makarenko]

Since my knockoffs programmer stopped working after Window 11 Upgrade I switched to Raspberry Pi for programming cpld.

Arm can be programmed using usb A - usb A cable.

@arkadiusz.makarenko is there a tool for capturing USB HID keycodes on a PC?

Never had to do it, as I just captured it from debugger on stm32.
But there must be one.

[arkadiusz.makarenko]

I will look at keyboard firmware when I am back in Scotland.

[NYYRIKKI]

Yeah, sorry that I'm not that much into HW. Buying Raspberry Pi or ordering custom programmer cable for programming the cpld seems a bit overkill ATM as I likely need this device only once and it is not even sure this solves all my problems as the chip may indeed be dead. I have only USB Blaster cable, but it does not seem to work in this case. Thank you anyway for making the update available. I will try to agree maintenance with Chuck at some point.

Direct USB A - USB A cable is anyway something I can likely even find or build from the stuff I have around already.

BTW here is the test program I used to test the keyboard.

Code: Select all

10 CLS
20 FOR I=0 TO 10:D(I)=255:NEXTI
30 LOCATE 0,0
40 FOR I=0 TO 10
50 D(I)=D(I)ANDPEEK(&HFBE5+I)
60 PRINTI,RIGHT$("0000000"+BIN$(D(I)),8)
70 NEXT I
80 GOTO 30

... If the keyboard works fine, you should be able to turn all the bits to "0" by pressing all the different keys on keyboard. (STOP pauses the program, so skip or press two times)

One thing that would be good to test as well is that CTRL+SHIFT+GRAPH+CODE keys work at a same time. This combo is considered kind MSX version of "CTRL+ALT+DEL". One way to test is to execute POKE &HFBB0,1 on BASIC and then pressing the keys... You should see some weird behavior and extra "Ok" messages printed to screen. I've had some problems in past with emulators. If it does not work, then it is likely good idea to add something like:

Code: Select all

{KEY_F8,	KB_ROW_6,	0xE8},

Maybe if you post your constants, we will find the few missing codes... As I said, I expect only AltGr to be (maybe) a bit problematic. Rest of the codes I expect to be same on both layouts, just different prints on keys. Maybe Chuck can help here as well? He sure has same kind of keyboard than I have.

[terriblefire]

From memory it wasnt that the keys werent mapped. it was that different USB keyboards have different codes for the "same" keys so they get missed.

Also we discovered that the layout was set on a per rom basis.

[NYYRIKKI]

From memory it wasnt that the keys werent mapped. it was that different USB keyboards have different codes for the "same" keys so they get missed.

Also we discovered that the layout was set on a per rom basis.

It may be... but nothing on the code that was posted earlier here pointed to this direction... The problem was clearly caused by trying to allocate the same key code to multiple different keys.

[NYYRIKKI]

Also we discovered that the layout was set on a per rom basis.

Yes, but here you talk about logical layout that works on top of physical layout. On MSX the physical layout is same, no matter if you have JCUKEN-keyboard or QWERTY-keyboard. Ok, ok... There are differences where ie. some special keys like SELECT are physically placed as well, but you can safely ignore these things. If you run my test program, it does not even care about the logical layout. If you look hard enough, I'm sure you will find some MSX ROMs where you get problems with row 9 & 10 as they don't exist in all machines, but those you have practically mapped out ok already (except for ",")

By just looking the table content, I can see that you can't pass my test even with the keyboard you have.

[terriblefire]

By just looking the table content, I can see that you can't pass my test even with the keyboard you have.

This is useful. So we can use this when @arkadiusz.makarenko gets back from his vacation