Log..
I spent the day converting the SID Speech code to xa. I then hacked at it to operate on an Oric with AY and then compiled and ran it.
No sound. Oh what a surprise!
But i now realise one of the reasons. It uses the C64's CIA chips hardware Timer facility to control delays between operations together with an IRQ routine.
However whilst the IRQ routine was corrected the Oric or rather the VIA does not contain a timer of the sort required for the Speech code.
It requires a timer that allows the software to select when it should start or stop. This consists of a write in Bit 0 to its control register.
In fact the speech sid uses 4 CIA registers to play with the timer..
CIA2_TimerBLo and CIA2_TimerBHi are stored with $0064 (100 decimal) only. This sets the time period to 100 clock cycles.
CIA2_ICR is the Chips Interrupt Control Register and will reflect different things if Read or Written. Most of the time it is being read in order to detect the timeout but in one place it is being written to with $82 and that value will set Bits 7 and 1 which tells the CIA to generate an NMI interrupt on the timeout of Timer B.
However i think (tho not totally sure) that CIA2 interrupt line is not connected to the 6510 processor. This is why it is being read in the code.
Finally CIA2_CtrlRegB is CIA2's Control Register for Timer B.
In the code it is more often written to than read.
Infact the only time it reads Bit 0 which holds the current state of Timer B (started/stopped).
When it is written it is done with $90 or $91.
Set TOD Alarm, Set flag for 'Force load Timer B' and depending on the "1" start or stop Timer B.
Now how does this equate to us getting it to work on the Oric?
Well i guess we could use VIA Timer 2 to countdown 100 cycles because to do it in software would most likely upset the timing of the speech.
More later..
