NSF Sync w/ FamiSlayer (Page 1 of 3)

Enclaimer: I'm an expert. (I'm an expert at making up words as well.)

Another thing that could be used for syncing is the DIN sync output of a drum machine or TB-303.

I was going to suggest reading the joypad more often than the regular every VBlank, but you're already doing that. Good. However, I think (from reading the source code) that your joypad check routine is buggy in how it handles multiple, For example, I think if you say, hold up permanently pressed, and then press start, that the start button press will be registered as an up press.

There's a way to solve this to create a trigger mask sing XOR then AND. Pseudo code, in a more Z80-ish syntax:

call READPAD         ; Let's sy this function returns the currently pressed keys in A
ld   B,A             ; Save current mask for later (copy to B)
ld   A,[OLDJOYPAD]   ; Get previous joypad state (load to A)
xor  A,B             ; XOR with current state to detect any changed buttons (target: A)
and  A,B             ; AND with current state to only store currently pressed (ie newly pressed) buttons
ld   [JOYTRIGGER],A  ; Save trigger mask
ld   A,B             ; Load B into A
ld   [OLDJOYPAD],A   ; Save current state as old state for the next comparison

Now check all bits in JOYTRIGGER.

Another idea: You could make the program have a master mode as well. In this mode, the joypad is only checked once every frame (or more exactly, once every player tick, if a song should happen to use a different sync source than VSync (NMI). (The sync source must be internal in this mode.) This would read out the controller like normal, but it would also have the side effect of ending out 8 clock pulses, one per bit that is being read. If you connect the NES controller clock line to the Gameboy's clock line, (and ground-ground of course) these 8 bits will be read by the Ganeboy as an incoming byte on the serial port. In LSDj's MIDI mode, this will be enough to synch the song.

Great work!

That's my plan exactly ;P

40106 generator was my hope, glad to hear that's a possibility. I wonder what the duty cycle is like, or if it even cares how wide it is.

Thanks everyone for checking it out.

@uxe Your FamiiDI project is awesome! When I saw your post, it totally inspired me to get the ENIO board. It just came in the mail today, cant wait to mass around with it.  Moving the controls to controller 2 to work with FamiiDi should be easy enough.

@nitro2k01: Great idea on the master option.  I see what you are saying about checking against the old button state code.  But I think that if they press start while pressing up, then it will register the New state as Up & Start and will then pass the check because it is a different button state I think. I believe in that case then both the Start and Up code would both get triggered when ANDing the button state against each button.  I will def try out your algorithm, thanks for the heads up!

@TylerBarnes Try it with your NSF's.  Only the Rewind, Fast Forward and Restart options were developed specifically for FamiTracker NSF driver.  But the sync and change of song number should still work I think.

@DSC Thanks man!  I know I still owe you a beatbox!

Yogi, when / if you do end up building an interface based on that code we finally settled on in the FamiiDI thread, all you'd need to add to generate Nanoloop clock (12 pulses per quarter note) from MIDI clock (24 ppqn) is:

if (MIDI.getType() == Clock)
  DDRB ^= bitMask[0]; // Toggle 'Start Button' Bit

Edit: Oh, and get rid of the 'MIDI.getType() < B11110000' condition! (I added that in to filter out System messages because I was only interested in Note / Pitch Bend messages)

Last edited by uXe (Jan 7, 2014 12:26 am)

Quick update, was able to get the Commodore 64 to clock the NES using the user port on the C64.  Will post the code after I get a chance to clean it up a bit.

So far I am loving how this combo sounds. Think this might end up being my new live setup.

Sorry if this is a dumb question but what program are you running on the commodore?