Control a 4usec switch in 1-wire

MacTuxLin · 2012-01-11 02:06

Hi All,

I need help in understanding PASM. I knew I'll have to touch on this sooner or later (was hoping the latter but), anyway, I need to control a 1-wire device that has a very tight timing requirement. Example:

Logic 1 = Low in 4.34 usec, then a High in 4.34 usec, then a Low in 4.34 usec, then a High in 25.98 usec

I need to set low at 4.34 usec but it seems I can only set either 4 usec or 5 usec? The device spec Min (4.1 usec) & Max (4.5 usec). Do I add 2 or 3 dummy instructions to achieve the 0.34 usec or do something in the _delay routine?

I've made slight changes but I think I have erred in the _newwrite sub-routine.

[SIZE=3][FONT=courier new]
[SIZE=2]cmd_newwrite            rdlong  value, t2               ' get the data byte
                        add     t2, #4                  
                        rdlong  bitCnt, t2 wz           ' get bit count    
          if_z          mov     bitCnt, #1              ' must be 1 to 32
                        max     bitCnt, #32
                        call    #_newwrite              ' write bits and exit
                        jmp     #endCommand


[/SIZE][/FONT][/SIZE][SIZE=2][FONT=courier new]'------------------------------------------------------------------------------[/FONT]
[FONT=courier new]' input:  value         data bits[/FONT]
[FONT=courier new]'         bitCount      number of bits[/FONT]
[FONT=courier new]' output: none[/FONT]
[FONT=courier new]'------------------------------------------------------------------------------[/FONT]

[FONT=courier new]_writeByte              mov     bitCnt, #8              ' write an 8-bit byte[/FONT]
[FONT=courier new]
[/FONT]
[/SIZE][SIZE=2][FONT=courier new]_newwrite                  andn    outa, dataMask          ' set data low for 4 usec[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        or      dira, dataMask[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        mov     t1, #4[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        call    #_delay[/FONT][/SIZE]
[SIZE=2][FONT=courier new]
[/FONT][/SIZE]
[SIZE=2]
[/SIZE][SIZE=2][FONT=courier new]                        ror     value, #1 wc            ' check next bit[/FONT][/SIZE]
[SIZE=2][FONT=courier new]          if_c          andn    dira, dataMask          ' if 1, set data to high Z[/FONT][/SIZE]
[SIZE=2][FONT=courier new]
[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        mov     t1, #4                 ' hold for 4 usec[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        call    #_delay[/FONT][/SIZE]
[SIZE=2]
[/SIZE][SIZE=2][FONT=courier new]                        andn    outa, dataMask          ' set data low for 4 usec[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        or      dira, dataMask[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        mov     t1, #4[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        call    #_delay[/FONT][/SIZE]

[SIZE=2][FONT=courier new]                        andn    dira, dataMask          ' set data to high again 25 usec[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        mov     t1, #25[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        call    #_delay[/FONT][/SIZE]
[SIZE=2][FONT=courier new]
[/FONT][/SIZE]
[SIZE=2][FONT=courier new]                        djnz    bitCnt, #_newwrite         ' repeat for all bits[/FONT][/SIZE]


[SIZE=2][FONT=courier new]_writeByte_ret[/FONT]
[FONT=courier new]_write_ret              ret[/FONT]
[FONT=courier new]


'------------------------------------------------------------------------------[/FONT]
[FONT=courier new]' input:  t1            number of usec to delay (must be multiple of 4)[/FONT]
[FONT=courier new]' output: none[/FONT]
[FONT=courier new]'------------------------------------------------------------------------------[/FONT]
[FONT=courier new]
_delay                 shr     t1, #2 wz               ' divide delay count by 4[/FONT]
[FONT=courier new]          if_z          mov     t1, #1                  ' ensure at least one delay[/FONT]
[FONT=courier new]                        mov     t2, delay4usec          ' get initial delay[/FONT]
[FONT=courier new]                        add     t2, cnt[/FONT]
[FONT=courier new]                        sub     t2, #41                 ' adjust for call overhead[/FONT]
[FONT=courier new]
[/FONT]
[FONT=courier new]:wait                   waitcnt t2, delay4usec          ' wait for 4 usec[/FONT]
[FONT=courier new]                        djnz    t1, #:wait              ' loop while delay count > 0[/FONT]
[FONT=courier new]_delay_ret              ret[/FONT]

[/SIZE]

I'm using the 1-wire object from OBEX

Thanks a lot!

MagIO2 · 2012-01-11 03:20

Why do you have a loop in _delay? You only need a loop in case you want to wait longer than ~51seconds.

You would simply create some delay-values for your different timings like

delay4_34us long 347
delay28_98us long 2078

At the beginning of your signal-sequence you would store cnt into another variable and after you set the signal to a level simply call the waitcnt with the right time:

mov waittime, cnt
add waittime, delay4_34us

andn outa, dataMask
or dira, dataMask
waitcnt waittime, delay4_34us ' this is the timing for the next wait
...

' this has to be at the very end of your PASM-dat section
waittime res 1

The call of a subroutine is not needed and the subroutine is wasting more memory than just having some constants for the timing.

MacTuxLin · 2012-01-11 05:25

Thank you MagIO2. I've changed the code, could you please take a look again?

_writeByte              mov     bitCnt, #8              ' write an 8-bit byte

_newwrite               mov     waittime, cnt
                        add     waittime, delay4_34us
                        andn    outa, dataMask          ' set data low for 4.34 usec
                        or      dira, dataMask
                        waitcnt waittime, delay4_34us


                        ror     value, #1 wc            ' check next bit
          if_c          mov     waittime, cnt
                        add     waittime, delay4_34us
                        andn    dira, dataMask          ' if 1, set data to high for 4.34 usec
                        waitcnt waittime, delay4_34us


                        mov     waittime, cnt
                        add     waittime, delay
                        andn    outa, dataMask          ' set data low again for 4.34 usec
                        or      dira, dataMask
                        waitcnt waittime, delay4_34us


                        mov     waittime, cnt
                        add     waittime, delay
                        andn    dira, dataMask          ' set data high 25.98 usec
                        waitcnt waittime, delay28_98us


                        djnz    bitCnt, #_newwrite      ' repeat for all bits
_writeByte_ret
_write_ret              ret




delay4_34us             long    347
delay28_98us            long    2078




bitCnt                  res     1                       ' bit counter
dataMask                res     1                       ' data pin mask
waittime                res     1

JonnyMac · 2012-01-11 06:34

Is the device a Dallas/Maxim 1-Wire device? If yes, it should conform to the 1-Wire specification which Cam's object, too. I've also written a PASM driver for 1-Wire which uses timing values I received directly from a Dallas/Maxim engineer. It's another approach that may help you with your project.

MacTuxLin · 2012-01-11 07:34

Thank you Jon. Unfortunately, its not a Dallas/Maxim chip but I'll use your code as reference as well. I'm reading the datasheet & I guess its never a better time than now to seriously learn PASM.

MagIO2 · 2012-01-11 08:34

That is not exactly what I meant. You should read the manual entry for waitcnt. Waitcnt already is doing the calculation of the next waittime. So, you don't have to read cnt again and add delay4_34 again. Just read cnt at the beginning and sum up on that initial waittime.

I'm pretty sure johns code will show you this way. (can't check it on the smartphone to make sure)

MacTuxLin · 2012-01-11 17:35

Thank you. (slap my head) I re-look my code again this morning & it looks bad. I'll continue with Jon's code. Thanks.

_writeByte              mov     bitCnt, #8              ' write an 8-bit byte

_newwrite               mov     waittime, cnt
                        andn    outa, dataMask          ' set data low for 4.34 usec
                        or      dira, dataMask
                        waitcnt waittime, delay_4_34us


                        ror     value, #1 wc            ' check next bit
          if_c          mov     waittime, cnt           ' if 1, set data to high for 4.34 usec
                        andn    dira, dataMask          
                        waitcnt waittime, delay_4_34us


                        mov     waittime, cnt
                        andn    outa, dataMask          ' set data low again for 4.34 usec
                        or      dira, dataMask
                        waitcnt waittime, delay_4_34us


                        mov     waittime, cnt
                        andn    dira, dataMask          ' set data high 25.98 usec
                        waitcnt waittime, delay_28_98us


                        djnz    bitCnt, #_newwrite      ' repeat for all bits
_writeByte_ret
_write_ret              ret

Control a 4usec switch in 1-wire

Comments