Post by thread:Mixed use PORTB with Interrupt on Change

Hello, I'm using a 16F873 in an application and plan to use PORTB as 5 inputs, 3 outputs with the inputs being bits 3-7 to take advantage of the interrupt on change feature. The docs caution against reading the port which is understandable. Does the PIC do a read-alter-write when executing a BSF or BCF instruction or does it have the ability to set or reset a bit directly? Does using PORTB in this mixed mode screw up the interrupt on change? Thanks, howard Howard McGinnis spam_OUThmcginniTakeThisOuTdigital.net Electronic Visions, Inc. 1650 Barrett Drive Rockledge FL 32955 (321) 632-7530 http://ddi.digital.net/~hmcginni .....mcginnisKILLspam@spam@e-visions.com

Howard, The PIC does a "read-alter-write". My best advice would be to use a "shadow" register. Set that corresponding bit in the "shadow" register then move it into that port register. Best regards, Steve Steven Kosmerchock Radio Frequency Systems Phoenix Arizona USA http://www.rfsworld.com steve.kosmerchockKILLspamrfsworld.com http://www.geocities.com/researchtriangle/lab/6584 {Original Message removed}

Howard.... I don't think the docs mean not to read. You couldn't do anything then. See page 31 of the current f873 data sheet. Each time you test a bit it must read the D flip-flop to know what the state of the INPUT is. A change flag is set if and only if the change DFF is EXOR with the previous value at that pin....I think. Agree the data sheet is somewhat confusing. To make sure, I'd simply try each case, writing a simple program, and changing the various inputs. The convoluted docs will then become clear...i.e. writer's intent. Phil. On Wednesday, May 10, 2000 9:50 AM, Howard McGinnis [SMTP:.....hmcginniKILLspam.....DIGITAL.NET] wrote: > Hello, > > I'm using a 16F873 in an application and plan to use PORTB as 5 inputs, 3 > outputs with the inputs being bits 3-7 to take advantage of the interrupt > on change feature. The docs caution against reading the port which is > understandable. > > Does the PIC do a read-alter-write when executing a BSF or BCF instruction {Quote hidden}> or does it have the ability to set or reset a bit directly? Does using > PORTB in this mixed mode screw up the interrupt on change? > > Thanks, > howard > Howard McGinnis > EraseMEhmcginnispam_OUTTakeThisOuTdigital.net > Electronic Visions, Inc. > 1650 Barrett Drive > Rockledge FL 32955 > (321) 632-7530 > http://ddi.digital.net/~hmcginni > mcginnisspam_OUTe-visions.com

Shadow register won't help. Even the MOV instructions are `READ/PROCESS/WRITE, and the 'read' part will update the change bits. Been there. Have the flame marks. This is a fundamental problem with the 'PORTB change' function. You can't use the port for ANYTHING else (except one time setup) if you want to use PORTB change at high rates in/out and have NO missed changes. Had to keep a shadow INPUT register and checked for changes -after- any write to B (with a verify read), and called a special version of the ISR to deal with non-interrupt-generating changes. The special version got called about 1% of the time at my int rates (1kHz) showing that I had indeed been loosing pulses. (CCPx were used in another parts of the app) "Kosmerchock, Steve" wrote: {Quote hidden}> > Howard, > The PIC does a "read-alter-write". My best advice would be to > use a "shadow" register. Set that corresponding bit in the > "shadow" register then move it into that port register. > > Best regards, > Steve > > Steven Kosmerchock > Radio Frequency Systems > Phoenix Arizona USA > http://www.rfsworld.com > @spam@steve.kosmerchockKILLspamrfsworld.com > > http://www.geocities.com/researchtriangle/lab/6584 > > {Original Message removed}

Phil Anderson <KILLspampandersnKILLspamidir.net> wrote: > [If Port B can't even be read without screwing up the > interrupt-on-change hardware,] you couldn't do anything then. Howard's correct, Phil; ANYTHING you do to port B (read OR write) in your main program has the potential to interfere with proper operation of the interrupt-on-change feature. The ONLY 100%-reliable thing you can do with that feature is use it to wake up from sleep mode. -Andy === Andrew Warren - RemoveMEfastfwdTakeThisOuTix.netcom.com === Fast Forward Engineering - San Diego, California === http://www.geocities.com/SiliconValley/2499

Yes BSF/BCF do use read-alter-write, and yes this will screw up the int on change stuff if the timing of things is just right (or wrong in this case :-)). At least it screwed things up for me! Anyone got any good solutions for this problem. I managed to work round it by arranging things so that I only need to write to the port during a window when I know that there will be no changes on the input side of things (or at least there shouldn't be!). Not sure what happens if you use some sort of shadow register and then do a byte wide write to the port. I've not tried this anyone know if it helps? One last thing according to my copy of the 87X spec (rather old I'm at work at the moment), RB3 does not have int on change only RB4-RB7. Andy {Original Message removed}

> Van: Andy Shaw <spamBeGoneandysspamBeGoneSCO.COM> > Aan: TakeThisOuTPICLISTEraseMEspam_OUTMITVMA.MIT.EDU > Onderwerp: Re: Mixed use PORTB with Interrupt on Change > Datum: donderdag 11 mei 2000 10:40 [Snip] > One last thing according to my copy of the 87X spec (rather old I'm at work > at the moment), RB3 does not have int on change only RB4-RB7. You could try to use RB0. The edge-select for an interrupt will enable you to see any change. You can't configure it as fast as you can RB4 thru 7, but it'll work ... Regards, Rudy Wieser

Maybe a workaround: Put the signal on two input pins. The first direct, the second via RC combination (1k/1n). So at least one change comes through. Gerhard Andrew Warren schrieb: {Quote hidden}> > Phil Anderson <RemoveMEpandersnTakeThisOuTidir.net> wrote: > > > [If Port B can't even be read without screwing up the > > interrupt-on-change hardware,] you couldn't do anything then. > > Howard's correct, Phil; ANYTHING you do to port B (read OR > write) in your main program has the potential to interfere with > proper operation of the interrupt-on-change feature. The ONLY > 100%-reliable thing you can do with that feature is use it to > wake up from sleep mode. > > -Andy >

Howard.... I don't think the docs mean not to read. You couldn't do anything then. See page 31 of the current f873 data sheet. Each time you test a bit it must read the D flip-flop to know what the state of the INPUT is. A change flag is set if and only if the change DFF is EXOR with the previous value at that pin....I think. Agree the data sheet is somewhat confusing. To make sure, I'd simply try each case, writing a simple program, and changing the various inputs. The convoluted docs will then become clear...i.e. writer's intent. Phil. On Wednesday, May 10, 2000 9:50 AM, Howard McGinnis [SMTP:hmcginniEraseME.....DIGITAL.NET] wrote: > Hello, > > I'm using a 16F873 in an application and plan to use PORTB as 5 inputs, 3 > outputs with the inputs being bits 3-7 to take advantage of the interrupt > on change feature. The docs caution against reading the port which is > understandable. > > Does the PIC do a read-alter-write when executing a BSF or BCF instruction {Quote hidden}> or does it have the ability to set or reset a bit directly? Does using > PORTB in this mixed mode screw up the interrupt on change? > > Thanks, > howard > Howard McGinnis > EraseMEhmcginnidigital.net > Electronic Visions, Inc. > 1650 Barrett Drive > Rockledge FL 32955 > (321) 632-7530 > http://ddi.digital.net/~hmcginni > RemoveMEmcginnisEraseMEEraseMEe-visions.com