Post by thread:[PIC] Open drain output from conventional output p

Yes, you can do that. Be aware that any Read-Modify-Write operation on the port - such as bcf & bsf - will update the output latch. So you must make sure the output latch for that pin is set to 0 before setting the TRIS for that pin to 0. At 09:12 AM 11/17/2006, you wrote: >In my circuit I need a couple of open drain outputs (for varying the >resistance to ground of a certain point by connecting resistors). However >the PIC I'm using (12f675) has no open drain outputs, so I was wondering >whether it would be possible to simulate this by switching the pin between a >high impedance input and a low output? Are there any flaws I've missed in >this idea? (yes I could use external FETs, but given the reasons for using >an 8-pin PIC, I'm also trying to keep component count down).

Chris McSweeny wrote: > In my circuit I need a couple of open drain outputs (for varying the > resistance to ground of a certain point by connecting resistors). > However the PIC I'm using (12f675) has no open drain outputs, so I was > wondering whether it would be possible to simulate this by switching the > pin between a high impedance input and a low output? Are there any flaws > I've missed in this idea? I don't think you'll have a problem. As you can see in the circuits of the port pins, a normal port differs from the open drain port basically in the added p-channel FET to Vdd. As long as you don't activate this FET (by setting the port to "high"), it should behave in the same way as the open drain output. I didn't check your PIC's data sheet, but one difference may be that the open drain output can handle higher voltages. As long as you stay below Vdd, this shouldn't make any difference, though. Gerhard

> However > the PIC I'm using (12f675) has no open drain outputs, so I > was wondering > whether it would be possible to simulate this by switching > the pin between a > high impedance input and a low output? Are there any flaws > I've missed in > this idea? (yes I could use external FETs, but given the > reasons for using > an 8-pin PIC, I'm also trying to keep component count down). This is standard practice, used for instance for I2C. Keep in mind that the voltage on the 'open drain' output must not be higher than your PIC's Vdd. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Thanks for the reminder - an easy thing to forget (and to the others for confirming it's a valid thing to do). I presume that if I use a 12f683 where the latches are set to low be default according to the datasheet, I will be safe in simply switching from input to output, but with a 12f675 where they are undefined I should make sure by doing a write first. Does a write actually force the data latches in this case, so I simply need to do bcf on GPIO immediately before doing a bcf on TRIS, and that will override any input value read? On 11/17/06, Kevin Timmerman <spam_OUTlistsTakeThisOuTpcmx.net> wrote: {Quote hidden}> > Yes, you can do that. Be aware that any Read-Modify-Write operation > on the port - such as bcf & bsf - will update the output latch. So > you must make sure the output latch for that pin is set to 0 before > setting the TRIS for that pin to 0. > > > At 09:12 AM 11/17/2006, you wrote: > >In my circuit I need a couple of open drain outputs (for varying the > >resistance to ground of a certain point by connecting resistors). However > >the PIC I'm using (12f675) has no open drain outputs, so I was wondering > >whether it would be possible to simulate this by switching the pin > between a > >high impedance input and a low output? Are there any flaws I've missed in > >this idea? (yes I could use external FETs, but given the reasons for > using > >an 8-pin PIC, I'm also trying to keep component count down). > > -

Actually having done another search, I've realised all the info I needed was already available (just need to search for the right terms), and that bcf on GPIO is generally a bad thing to do anyway! Sorry for being such a newbie (well actually I've been working with PICs for years, but not really doing anything very complicated). On 11/17/06, Chris McSweeny <.....cpmcsweenyKILLspam@spam@gmail.com> wrote: {Quote hidden}> > Thanks for the reminder - an easy thing to forget (and to the others for > confirming it's a valid thing to do). I presume that if I use a 12f683 where > the latches are set to low be default according to the datasheet, I will be > safe in simply switching from input to output, but with a 12f675 where they > are undefined I should make sure by doing a write first. Does a write > actually force the data latches in this case, so I simply need to do bcf on > GPIO immediately before doing a bcf on TRIS, and that will override any > input value read? > > On 11/17/06, Kevin Timmerman <listsKILLspampcmx.net> wrote: > > > > Yes, you can do that. Be aware that any Read-Modify-Write operation > > on the port - such as bcf & bsf - will update the output latch. So > > you must make sure the output latch for that pin is set to 0 before > > setting the TRIS for that pin to 0. > > > > > > At 09:12 AM 11/17/2006, you wrote: > > >In my circuit I need a couple of open drain outputs (for varying the > > >resistance to ground of a certain point by connecting resistors). > > However > > >the PIC I'm using (12f675) has no open drain outputs, so I was > > wondering > > >whether it would be possible to simulate this by switching the pin > > between a > > >high impedance input and a low output? Are there any flaws I've missed > > in > > >this idea? (yes I could use external FETs, but given the reasons for > > using > > >an 8-pin PIC, I'm also trying to keep component count down). > > > > --

Chris McSweeny wrote : > I presume that if I use a 12f683 where the latches are > set to low be default according to the datasheet, I will be > safe in simply switching from input to output,... Not at all. Besides of never trusting "defaults", the problem is this : Say you want pin-1 to be OC with an external pullup. You set pin-1 "low" and then "toggle" tris-1. All is fine so far, if there wasn't any other pins used ! Now say you also have another pin-2 that is an output. What will happen is that when tris-1 is "input" the pin-1 will be pulled high by the external pullup. As expected and by design when using an "open collector bus". Now if you bcf/bsf to pin-2, you have an R-M-W operation. The whole GPIO will be read (incl pin-1 that now will be read as "high"), pin-2 is set according to the bcf/bsf and finaly the whole GPIO is written back. This will now set pin-1 "high" !!! So the solution ot to always set pin-1 "low" right before setting tris-1 as an output. This gets a little more complicated if you have multiple OC pins with external pullup's, but in general itäs the same. Jan-Erik.

I was thinking more of the initial case on power up - though in fact I probably want to have them OC at that point anyway so it's not an issue. I've realised the problem with RMW cycles now anyway! I'll have multiple OC pins, but only want to set one low at a time, so it shouldn't be a problem, as I can just bo a bcf followed by a movlw, movf TRIS to switch the pin which is OC (and investigation shows the pull-up is only to ~0.2V anyway!) On 11/17/06, Jan-Erik Söderholm <.....jan-erik.soderholmKILLspam.....telia.com> wrote: {Quote hidden}> > Chris McSweeny wrote : > > I presume that if I use a 12f683 where the latches are > > set to low be default according to the datasheet, I will be > > safe in simply switching from input to output,... > > Not at all. Besides of never trusting "defaults", > the problem is this : > > Say you want pin-1 to be OC with an external pullup. > > You set pin-1 "low" and then "toggle" tris-1. > All is fine so far, if there wasn't any other pins used ! > > Now say you also have another pin-2 that is an output. > > What will happen is that when tris-1 is "input" the pin-1 > will be pulled high by the external pullup. As expected > and by design when using an "open collector bus". > > Now if you bcf/bsf to pin-2, you have an R-M-W operation. > The whole GPIO will be read (incl pin-1 that now will be > read as "high"), pin-2 is set according to the bcf/bsf and > finaly the whole GPIO is written back. This will now set > pin-1 "high" !!! > > So the solution ot to always set pin-1 "low" right before > setting tris-1 as an output. > > This gets a little more complicated if you have multiple > OC pins with external pullup's, but in general itäs the same. > > Jan-Erik. > > >

I really think the appropriate port bits can be set low once, during initialization using a movwf. Then you can do bsf and bcf TRIS all you want. The read/modify/write problem is only a problem when dealing with a PORT, and is due to external loading keeping a pin from moving immediately after a write. The TRIS register does not have this problem, so RMW on TRIS all you want. The 18 series gets around this problem by having a separate LATCH register for each port. When you read the LATCH, it will read what you last wrote to it, no matter what state the actual pin is. Harold {Quote hidden}> I was thinking more of the initial case on power up - though in fact I > probably want to have them OC at that point anyway so it's not an issue. > I've realised the problem with RMW cycles now anyway! I'll have multiple > OC > pins, but only want to set one low at a time, so it shouldn't be a > problem, > as I can just bo a bcf followed by a movlw, movf TRIS to switch the pin > which is OC (and investigation shows the pull-up is only to ~0.2V anyway!) > > On 11/17/06, Jan-Erik Söderholm <EraseMEjan-erik.soderholmspam_OUTTakeThisOuTtelia.com> wrote: >> >> Chris McSweeny wrote : >> > I presume that if I use a 12f683 where the latches are >> > set to low be default according to the datasheet, I will be >> > safe in simply switching from input to output,... >> >> Not at all. Besides of never trusting "defaults", >> the problem is this : >> >> Say you want pin-1 to be OC with an external pullup. >> >> You set pin-1 "low" and then "toggle" tris-1. >> All is fine so far, if there wasn't any other pins used ! >> >> Now say you also have another pin-2 that is an output. >> >> What will happen is that when tris-1 is "input" the pin-1 >> will be pulled high by the external pullup. As expected >> and by design when using an "open collector bus". >> >> Now if you bcf/bsf to pin-2, you have an R-M-W operation. >> The whole GPIO will be read (incl pin-1 that now will be >> read as "high"), pin-2 is set according to the bcf/bsf and >> finaly the whole GPIO is written back. This will now set >> pin-1 "high" !!! >> >> So the solution ot to always set pin-1 "low" right before >> setting tris-1 as an output. >> >> This gets a little more complicated if you have multiple >> OC pins with external pullup's, but in general itäs the same. >> >> Jan-Erik. >> >> >> -

Thanks for clarifying that. So as I originally suggested, with a 12f683 I can rely on the port latches being initialised to low and simply set the TRIS as appropriate. Jan - when I was suggesting doing a movlw, movwf TRIS, that was to switch one pin to low and another pin back to OC at the same time, though I could presumably just as well do bcf, bsf on the appropriate bits. I've not got a good idea what's going on the other end of the resistor - it is a current setting resistor in a boost constant current regulator. I guess thinking about it, it is more likely to be switched with a very low duty cycle when I'm measuring rather than a fixed low voltage - I shan't be relying on the value when it's switched anyway (though actually now slightly concerned it may go above VDD). On 11/17/06, Harold Hallikainen <@spam@haroldKILLspamhallikainen.org> wrote: {Quote hidden}> > I really think the appropriate port bits can be set low once, during > initialization using a movwf. Then you can do bsf and bcf TRIS all you > want. The read/modify/write problem is only a problem when dealing with a > PORT, and is due to external loading keeping a pin from moving immediately > after a write. The TRIS register does not have this problem, so RMW on > TRIS all you want. The 18 series gets around this problem by having a > separate LATCH register for each port. When you read the LATCH, it will > read what you last wrote to it, no matter what state the actual pin is. > > Harold > > > > I was thinking more of the initial case on power up - though in fact I > > probably want to have them OC at that point anyway so it's not an issue. > > I've realised the problem with RMW cycles now anyway! I'll have multiple > > OC > > pins, but only want to set one low at a time, so it shouldn't be a > > problem, > > as I can just bo a bcf followed by a movlw, movf TRIS to switch the pin > > which is OC (and investigation shows the pull-up is only to ~0.2V > anyway!) > > > > On 11/17/06, Jan-Erik Söderholm <KILLspamjan-erik.soderholmKILLspamtelia.com> wrote: > >> > >> Chris McSweeny wrote : > >> > I presume that if I use a 12f683 where the latches are > >> > set to low be default according to the datasheet, I will be > >> > safe in simply switching from input to output,... > >> > >> Not at all. Besides of never trusting "defaults", > >> the problem is this : > >> > >> Say you want pin-1 to be OC with an external pullup. > >> > >> You set pin-1 "low" and then "toggle" tris-1. > >> All is fine so far, if there wasn't any other pins used ! > >> > >> Now say you also have another pin-2 that is an output. > >> > >> What will happen is that when tris-1 is "input" the pin-1 > >> will be pulled high by the external pullup. As expected > >> and by design when using an "open collector bus". > >> > >> Now if you bcf/bsf to pin-2, you have an R-M-W operation. > >> The whole GPIO will be read (incl pin-1 that now will be > >> read as "high"), pin-2 is set according to the bcf/bsf and > >> finaly the whole GPIO is written back. This will now set > >> pin-1 "high" !!! > >> > >> So the solution ot to always set pin-1 "low" right before > >> setting tris-1 as an output. > >> > >> This gets a little more complicated if you have multiple > >> OC pins with external pullup's, but in general itäs the same. > >> > >> Jan-Erik. > >> > >> > >> --

> I really think the appropriate port bits can be set low once, during > initialization using a movwf. that depends on what your app is doing with the rest of the port > Then you can do bsf and bcf TRIS all you > want. The read/modify/write problem is only a problem when dealing with a > PORT and is due to external loading keeping a pin from moving immediately > after a write. that is one situation in which it occours yes the problem that is relavent to this discussion comes when you are using some outputs on a port as open collector ans some as normal outputs. lets say there is a normal output on A1 and a pin being driven as open collector by use of tris on A0 which is pulled high #init, all of port A as an output driving low CLRF PORTA CLRF TRISA #make A0 an input allowing it to float high BSF TRISA,0 #make A1 high, #unfortunately due to RMW this also makes bit 0 of PORTA high BSF PORTA,1 #make A0 an output again, we throught we would be making #it drive low but in fact we end up making it drive high BSF TRISA,0 of course if we used LATA rather than PORTA this code would be fine but unfortunately only the 18 series have LAT.

{Quote hidden}> >> I really think the appropriate port bits can be set low once, during >> initialization using a movwf. > that depends on what your app is doing with the rest of the port > >> Then you can do bsf and bcf TRIS all you >> want. The read/modify/write problem is only a problem when dealing with >> a >> PORT and is due to external loading keeping a pin from moving >> immediately >> after a write. > that is one situation in which it occours yes > > the problem that is relavent to this discussion comes when you are using > some outputs on a port as open collector ans some as normal outputs. > > lets say there is a normal output on A1 and a pin being driven as open > collector by use of tris on A0 which is pulled high > > #init, all of port A as an output driving low > CLRF PORTA > CLRF TRISA > > #make A0 an input allowing it to float high > BSF TRISA,0 > > #make A1 high, > #unfortunately due to RMW this also makes bit 0 of PORTA high > BSF PORTA,1 > > #make A0 an output again, we throught we would be making > #it drive low but in fact we end up making it drive high > BSF TRISA,0 > > of course if we used LATA rather than PORTA this code would be fine but > unfortunately only the 18 series have LAT. > I agree completely! Another way around this is to keep a mirror of the PORT in RAM. Do the bsf and bcf on the mirror, then copy it to the port. Harold -- FCC Rules Updated Daily at http://www.hallikainen.com - Advertising opportunities available!

Chris McSweeny wrote : > Thanks for clarifying that. So as I originally suggested, with a 12f683 I > can rely on the port latches being initialised to low and simply set the > TRIS as appropriate. No no no, that's not *generaly* so. It depends on what you "do" with the other pins on GPIO. I thought my description of the case where RMW will bite you was clear enough. And there is *no* difference between 12F629/675/683 in this case (besides of some default, but...) Or *any* other PIC16 or PIC18 (as long as we are talking about the PORTx registers). > Jan - when I was suggesting doing a movlw, movwf TRIS, that was to switch > one pin to low and another pin back to OC at the same time, though I could > presumably just as well do bcf, bsf on the appropriate bits. Depends on your timing requirements. I just thought you where changing *one* bit in TRISIO, and thought there was no need to mess with W in that case. > I've not got a good idea what's going on the other end of the resistor - it > is a current setting resistor in a boost constant current regulator. I guess > thinking about it, it is more likely to be switched with a very low duty > cycle when I'm measuring rather than a fixed low voltage - I shan't be > relying on the value when it's switched anyway (though actually now slightly > concerned it may go above VDD). OK, I though it was a "normal" pullup... Jan-Erik.