Sync with ADTPro grub; tightens timing more

Signed-off-by: David Schmidt <1110325+david-schmidt@users.noreply.github.com>
This commit is contained in:
David Schmidt 2023-10-04 17:01:46 -04:00
parent a101949f8a
commit 2ad0e7432e

View File

@ -1,6 +1,8 @@
;
; Bitbang RS-232 (using a pushbutton input in our case)
; Tiger Learning Computer Joystick serial bitbanger (using a pushbutton input in our case)
;
; Custom serial-to-joystick cable is required. RS-232 settings are 9600 bps,
; no parity, 8 data bits, 1 stop bit; 1ms pacing seems to be required.
;
; CPU cycle counting must be done to meet bit-time requirements on the wire.
@ -10,7 +12,7 @@
; microseconds long.
;
.org $300 ; For now
.org $300
; Some system constants
WAIT = $FCA8
@ -42,7 +44,7 @@ done: rts
pb0_recv:
; State is currently unknown
lda #$09 ; We'll be watching for 8 bits plus one stop bit
lda #$09 ; We'll be watching for 8 bits plus one stop bit
sta bits
clc
@ -58,23 +60,20 @@ poll_for_0:
lda PB0
bmi poll_for_0 ; if negative, branch to poll_for_0
; State just became 0 (start bit)
; State just became 0 (start bit) sometime in the last 4 or so clock cycles
; Wait 1.5 bit times (104.2 + 52.1 = 156.3us at 9600 baud) to get into the middle of the first bit
; Wait 1.5 bit times (104.2 + 52.1 = 156.3us at 9600 bps) to get into the middle of the first bit
; Approximately 152.8 ($99) CPU cycles
; When falling through to here, the above branch was not taken - consuming 2 cycles to get here
ldx #$14 ; 2 loop count
: nop ; 2 \
dex ; 2 |-- 7 * loop count - 1
bne :- ; 3 / final exit of the loop adds 2, branch not taken
; $8F cycles to get here
beq :+ ; 3 burn
: beq :+ ; 3 baby
: nop ; 2 burn
ldx #$1D ; 2 loop count
: dex ; 2 \ = 5 * loop count - 1
bne :- ; 3 / final exit of the loop only adds 2, branch not taken
; $94 cycles to get here
bit $00 ; 3 don't care about results
; $97 cycles to get here; final 2 will be consumed by clc below
pull_byte:
; We now have one bit time (104.2us at 9600 baud) to process this bit
; Approximately 106.6 ($6B) CPU cycles
; We now have one bit time (104.2us at 9600 bps) to process this bit
; Approximately 101.8 ($66) CPU cycles
clc ; 2
lda PB0 ; 4
bmi :+ ; 2 if positive, 3 if negative
@ -88,17 +87,14 @@ push_bit: ; We now have a bit in the carry
ror ; 2
sta ring ; 4
; $1D cycles to get here (since center of bit time)
; We are now done with processing that bit; we need to cool our heels for the rest ($6B - $1D = $4E) of the
; We are now done with processing that bit; we need to cool our heels for the rest ($66 - $1D = $49) of the
; bit time in order to get into the middle of the next bit
ldx #$0A ; 2 loop count
: nop ; 2 \
dex ; 2 |-- 7 * loop count - 1
bne :- ; 3 / final exit of the loop adds 2, branch not taken
; $47 cycles to get here, burn $07 more (includes our jump back to top of loop)
nop ; 2
nop ; 2
jmp pull_byte ; 3 Loop around for another bit - we burned $4E cycles
; $6A
ldx #$0E ; 2 loop count
: dex ; 2 \ = 5 * loop count - 1
bne :- ; 3 / final exit of the loop only adds 2, branch not taken
; $47 cycles to get here
jmp pull_byte ; 3 Loop around for another bit - we actually burn $4A cycles
; $67
byte_complete:
; Carry now holds stop bit (clear/0 indicates framing error, because we end with set/1)
lda ring ; Exit with the assembled byte in A