Post by thread:Serial Blues Continued

Hello everyone- thanks for tolerating and providing help with the ongoing saga of my serial problems. In summary, I am having difficulty getting a PIC serin and serout program to work together. They both work if I run them independently, by commenting out the jump to the other routine. In the latest attempt to solve the serial blues I've encountered I added an additional handshake line to provide a PIC output to tell the BS2 when it is ready for serin data. I did use the BSF/BCF on portb,5 to send this command. I think that the BCF/BSF commands are messing with my port states. When I run this program my BS2 timesout on both the serin and serout. I did find a problem in that once I set the _INSHAKE line high it doesn't go low. I added a BCF instruction at all the exit points of the serin routine and the _INSHAKE line still stays high all the time. If I reset the PIC it has a 50/50 chance of being high or low all the time. I am going crazy about now. I don't know why this part of the code is so difficult for me. I found a reference in a book I have by Mike Predko that talks about the problem using BSF and BCF that a read takes place and if the line is pulled high via another device, it will be set high after the read takes place. I believe that exactly what one of my problems may be. I am not sure what the simpliest way is to solve this problem because I use the BSF/BCF instruction throughout the PIC serout routine which is timing sensitive. Am I to understand using the MOVLW AND MOVWF PORT,B or the AND (other) command will not inadvertently alter the pin states? Should I rewrite all my routines to eliminate BSF/BCF instructions? What is the most efficient way to alter a single output line? For example if I want to set portb,5 high what instructions can do this efficiently without altering the other pin states? Maybe the AND command? Could someone provide an example? By the way I am not sure what the term "bit banging" means. I think it means bit manipulation using instructions such as RRF,RLF, AND, OR etc. and not MOV or SWAP etc. Am I correct? Thanks

At 07:01 PM 18/06/98 EDT, you wrote: {Quote hidden}>Hello everyone- thanks for tolerating and providing help with the ongoing saga >of my serial problems. > >In summary, I am having difficulty getting a PIC serin and serout program to >work together. They both work if I run them independently, by commenting out >the jump to the other routine. > >In the latest attempt to solve the serial blues I've encountered I added an >additional handshake line to provide a PIC output to tell the BS2 when it is >ready for serin data. I did use the BSF/BCF on portb,5 to send this command. I >think that the BCF/BSF commands are messing with my port states. When I run >this program my BS2 timesout on both the serin and serout. I did find a >problem in that once I set the _INSHAKE line high it doesn't go low. I added a >BCF instruction at all the exit points of the serin routine and the _INSHAKE >line still stays high all the time. If I reset the PIC it has a 50/50 chance >of being high or low all the time. I am going crazy about now. I don't know >why this part of the code is so difficult for me. > >I found a reference in a book I have by Mike Predko that talks about the >problem using BSF and BCF that a read takes place and if the line is pulled >high via another device, it will be set high after the read takes place. I >believe that exactly what one of my problems may be. I am not sure what the >simpliest way is to solve this problem because I use the BSF/BCF instruction >throughout the PIC serout routine which is timing sensitive. > >Am I to understand using the MOVLW AND MOVWF PORT,B or the AND (other) command >will not inadvertently alter the pin states? > >Should I rewrite all my routines to eliminate BSF/BCF instructions? > >What is the most efficient way to alter a single output line? For example if I >want to set portb,5 high what instructions can do this efficiently without >altering the other pin states? Maybe the AND command? Could someone provide an >example? > >By the way I am not sure what the term "bit banging" means. I think it means >bit manipulation using instructions such as RRF,RLF, AND, OR etc. and not MOV >or SWAP etc. Am I correct? > >Thanks > > Yes you have hit the nail on the head, BSF and BCF on a port are read write modify instructions in most porcessors i.e. They read the entire byte (Be this a port or a register), set or clear the bit and then write the byte back, but this is only a problem when you mix inputs and outputs (Normaly). If the pin is programmed as an output, the code will (In most processors) read the state of the write latch, however some do read the actual pin state. If I recall (Don't have a data book handy) the PIC will just change the state of the required pin and leave the others alone. ie. The PIC has boolean instructions and processing abilities However your AND type response is not the way to go, as it must read the state of all the pins (In and out) and then you write the result back. BSF PORTB,5 is the simplest way to set the pin Bit baning is just a name for sending out serial data bit by bit in software Quite simply I think that your problem is elsewhere Dennis -=====================================================================- Dennis Plunkett: Embedded Hardware, Software design NEC Australia ph 03 9264-3867 -=====================================================================-

> Van: <Nichole Petty> <spam_OUTPHXSYSTakeThisOuTAOL.COM> > Aan: .....PICLISTKILLspam@spam@MITVMA.MIT.EDU > Onderwerp: Serial Blues Continued > Datum: vrijdag 19 juni 1998 1:01 > [Cut] > I found a reference in a book I have by Mike Predko that talks about the > problem using BSF and BCF that a read takes place and if the line is > pulled high via another device, it will be set high after the read takes place. > I believe that exactly what one of my problems may be. I am not sure what > the simpliest way is to solve this problem because I use the BSF/BCF > instruction throughout the PIC serout routine which is timing sensitive. > > Am I to understand using the MOVLW AND MOVWF PORT,B or the AND > (other) command will not inadvertently alter the pin states? All instructions that change data with regards to allready existing data are prone to this effect. Those instructions have to read, modify and than write the data back to the origin. And Yes, BCF does a _Byte_ read, OR the byte with the appropriate bit-mask (2^Bit-Number) and writes the Byte back. So does allmost every instruction (exept instructions like CLRxxx, BTFSx, MOVF, etc.) But, now the good part, you only have to think about it when using _I/O_. > Should I rewrite all my routines to eliminate BSF/BCF instructions? No. Those you could replace with something like : MOVLW 0x01 << BitNumber IORWF PORTx and thus emulating a BSF instruction, but alas, the IORWF has got the same problem. The only way to get rid of this effect is to use the 'shadow-register' method. Take a register, modify that and than copy it to your I/O-Port. Example : BSF SHADOWB,5 MOVF SHADOWB,W MOVWF PORTB > What is the most efficient way to alter a single output line? For example > if I want to set portb,5 high what instructions can do this efficiently without > altering the other pin states? Maybe the AND command? Could someone > provide an example? Normally, when you define your in & outputs and do not change them anymore (the directions I mean :-) you won't have a problem. Only when you change a Input-pin to Output this will happen. Example: BSF STATUS,RP0 MOVLW B'00001111' ; Low 4 are Input, High 4 are Output MOVWF TRISB ;/ BCF STATUS,RP0 CLRF PORTB ; Output the code B'0000xxxx' ; No input is suplied at the low 4 pins. They are all 1 (High) BCF PORTB,6 ;Change Output to B'0100xxxx' This will work fine, until you change one of the Input-pins to Output. Why ? Ok Follow me... BCF is a Read-Modify-Write instruction, right ? Ok. Read PORTB. Result is ........ Yes, Its 00001111. (try it out !) Now Set bit 6 : 01001111. And write it back. The Output-buffer for the Low 4 pins has changed from 0000 to 1111 because if the state the inputs were in. If you now change pin 0 to Output it will be High, and not Low. > By the way I am not sure what the term "bit banging" means. I think it > means bit manipulation using instructions such as RRF,RLF, AND, OR > etc. and not MOV or SWAP etc. Am I correct? My definition of the term is that only BCF and BSF instructions are used for "bit banging" (changing I/O pin levels), because AND & OR are Byte-instructions (although they can change a single bit, but so can a INC or MOV instruction). But, as said, thats _my_ definition. > Thanks Your welcome. Greetz, Rudy Wieser

The bit operations work as a read modify write. If you try two consecutive bit set instructions it is possible that the second one will execute prior to the first ones effect on its port line can overcome line capacitance. By using a dummy register to perform these bit manipulations followed by a mov command to shove the mirrored register to the output port you can avoid the read modify write problems. Bit banging usually refers to brute force manipulation of I/O pins to perform something like serial communication instead of using a part with a build in USART to perform the operation in hardware while the software is doing something else. This allows a lower cost part to be used if you have instruction cycles to burn for "software virtual peripherals". At 07:01 PM 6/18/98 EDT, you wrote: >I found a reference in a book I have by Mike Predko that talks about the >problem using BSF and BCF that a read takes place and if the line is pulled >high via another device, it will be set high after the read takes place. I >believe that exactly what one of my problems may be. I am not sure what the >simpliest way is to solve this problem because I use the BSF/BCF instruction >throughout the PIC serout routine which is timing sensitive. > >Am I to understand using the MOVLW AND MOVWF PORT,B or the AND (other) command >will not inadvertently alter the pin states? > >Should I rewrite all my routines to eliminate BSF/BCF instructions? > >What is the most efficient way to alter a single output line? For example if I >want to set portb,5 high what instructions can do this efficiently without >altering the other pin states? Maybe the AND command? Could someone provide an >example? > >By the way I am not sure what the term "bit banging" means. I think it means >bit manipulation using instructions such as RRF,RLF, AND, OR etc. and not MOV >or SWAP etc. Am I correct? > >Thanks > > Larry G. Nelson Sr. L.NelsonKILLspamieee.org http://www.ultranet.com/~nr