cannot find the reason for such a strange behavior

[Alberto]

if you can find the troublesome code i'd like to have a look at it with the view to seeing why it fails

Door_Lock var PortB.4
DmyFlg var bit
Report var byte[8]


Report[2] = !Door_Lock + 48 ' (didn't work! can't remember if it fails with True or false)

The work around was

DmyFlg = Door_Lock
DmyFlg = !DmyFlg
Report[2] = DmyFlg + 48

Sorry, not exactly as I mentioned in my previous post! I should say "bit into byte"

Alberto

[richard]

The work around was

DmyFlg = Door_Lock
DmyFlg = !DmyFlg
Report[2] = DmyFlg + 48

I'M not sure you solution is correct

I did a few experiments as follows and conclude that,

if you invert a bit var (or bit alias) wether logical or bitwise only bit 0 in the result is valid , I you go on to use the result mathematically then all the other bits need to be masked off

for a correct calculation



D_Lck !D_Lck ~D_Lck D_Flg !D_Flg ~D_Flg R+!L R+~L R+!F R+~F
WRONG
0 65535 65535 0 65535 65535 47 47 47 47
CORRECT
0 255 65535 0 65535 65535 49 49 49 49
WRONG
1 65280 65534 1 65280 65534 48 46 48 46
CORRECT
1 0 65534 1 65280 65534 48 48 48 48
WRONG
0 65535 65535 0 65535 65535 47 47 47 47
CORRECT
0 255 65535 0 65535 65535 49 49 49 49

   TRISC = %11101111        ' set PORTC I/O
 

 D_Lck var Portc.4
 D_Flg var bit
 Report var byte[4]
 
 
    DEFINE DEBUG_REG PORTA
    DEFINE DEBUG_BIT 0      ;  if not used for pwr  
    DEFINE DEBUG_BAUD 9600
    DEFINE DEBUG_MODE 0     
    pause 2000
    Debug "Start",13 ,10      
 
 Debug 13 ,10 ,"D_Lck",9 ,"!D_Lck",9, "~D_Lck",9, "D_Flg",9, "!D_Flg",9, "~D_Flg" ,9, "R+!L" ,9, "R+~L" ,9, "R+!F" ,9, "R+~F"
main:
 PORTC.4=!PORTC.4
 D_Flg=D_Lck 
 Debug 13 ,10 ,"WRONG"
 Report[0] = !D_Lck + 48 
 Report[1] = ~D_Lck + 48 
 Report[2] = !D_Flg + 48 
 Report[3] = ~D_Flg + 48 
 Debug 13 ,10 ,#D_Lck,9 , #!D_Lck,9, #~D_Lck ,9, #  D_Flg ,9,  #!D_Flg, 9,# ~ D_Flg ,9,# Report[0] ,9,# Report[1] ,9,# Report[2] ,9,# Report[3]
 Debug 13 ,10 ,"CORRECT"
 Report[0] = (!D_Lck&1) + 48 
 Report[1] = (~D_Lck&1) + 48 
 Report[2] = (!D_Flg&1) + 48 
 Report[3] = (~D_Flg&1) + 48 
 Debug 13 ,10 ,#D_Lck,9 , #!D_Lck,9, #~D_Lck ,9, #  D_Flg ,9,  #!D_Flg, 9,# ~ D_Flg ,9,# Report[0] ,9,# Report[1] ,9,# Report[2] ,9,# Report[3]
  
 
 
 
  pause 1000
 
  goto main

[Alberto]

Today I have finaly found the time to implement the corrections to my code!

I have corrected the number of bytes int the array Pulse from 75 to 80. Next I have changed the following two lines of code:

DMY = W0//256
Pulse[75] = (255 - DMY)

with the simple form suggested by Richard

Pulse[75] = ~W0

Compiled (22Kb of code) and updated the mcu. Power up and checked the packet transmitted via esp NOOOO! CRC is still txed as 253 all the time!

I did change the code again setting Pulse[75] = 123 (costant value) and re-checked. Well this time CRC was txed as 123! This proved that there is no byte corruption in the code, but the problem should be in setting the byte variable Pulse[75].

So I did change the setting of Pulse[75] = 123 to Pulse[75] = W0.lowbyte. Re-checked and it worked! CRC was txed correctly!

Hence I added : Pulse[75] = ~ Pulse[75] (inverting bits) and CRC was still txed as 253!!!!

I have removed Pulse[75] = ~ Pulse[75] and kept the CRC as the low byte of the packet sum that works!

But as I stated in this thread title: "I cannot find the reason for such a strange behavior"

Alberto

[richard]

try as I might I cannot produce any code where
Pulse[75] = ~W0
gives an incorrect result , the problem is elsewhere
I believe you are just masking the real problem/s
my challenge to you is :-
provide some fully compliable code that demonstrates the problem , I would bet you cannot


Pulse[75] = ~W0 converts to a 1 line macro when compiled

NOT?WB  _w0, _Pulse + 00075h

where can that go wrong ?