Post by thread:PID question.

Funny you should ask, Im working on the same project at the moment. The only problem Im having is the maths routines, they seem to be getting up to 64 bit. The actual calculation is- dMV=G[dE + E*T1*I + D* (dE/T2)] where- dMV = change in Manipulated Variable G = Gain (100/Proportional Band) E = Error (SetPoint - ProcessVariable) for reverse action, Error (ProcessVariable - Setpoint) for direct action dE = ErrorThisScan - ErrorLastScan T2 = Time between scans (in minutes) T1 = Time since error occured (in minutes) I = Integral time (R/M) D = Derivative time (minutes) There are other things to consider as well, integral windup, bumpless transfer from manual to auto, etc Anyone knowing any alterations to this equation or anything else that might make writing this code easier are most welcome to reply Justin Grimm spam_OUTreaperTakeThisOuTsouthwest.com.au ---------- {Quote hidden}> From: Mark Lezama <.....mfidelKILLspam@spam@TELCEL.NET.VE> > To: PICLISTKILLspamMITVMA.MIT.EDU > Subject: PID question. > Date: Friday, 6 March 1998 21:09 > > I have working with RTU designe base in PIC16c74. I have to control a > valve with a 4-20 ma signal, but I want to implement a digital PID > controller, anyone here have a PID code or can explain me how I have to > implement. > > Thanks in advance! > > Mark Lezama

On Fri, 6 Mar 1998 14:23:05 -0600 Calvin <.....tgoKILLspam.....CHIH1.TELMEX.NET.MX> writes: >Comment: > >PID controls are inherently analog. This does not mean they cannot be >implemented digitally, but it is generally a lot easier to do it with >one or >two op-amps. It depends on what the process is of course. For simple fast processes I'd agree that analog can work well and digital offers no advantage (I won't go as far as to say that PID is "inherently analog" though). For example, I have a Magnavox CD player (ca. 1984) that uses op-amps throughout the servo loops that keep the laser beam on track on the disk. I suspect newer players still do this process mostly analog, it's just buried in the big chips. Analog has the neat feature that you can just wire in some potentiometers and unpluggable capacitors and play with it until the control is "optimized." This can be hard to explain to the boss though. On the other hand, if the process is slow enough that the time constants in the integrator and differentiator run to seconds or minutes, analog components to do that can get unwieldy. Or if you want or need to add various nonlinear special cases ("fuzzy PID" ???) it may be best to go digital. _____________________________________________________________________ You don't need to buy Internet access to use free Internet e-mail. Get completely free e-mail from Juno at http://www.juno.com Or call Juno at (800) 654-JUNO [654-5866]

Thanks Justin. I will try to implement. I think is the problem is the # of bit I have to use a floting point rutine. BTW, in the formula I dont know what means "Time since error ocurred", maybe Is a fool question but I am trying to understand anything before implement the code. Mark Lezama. Justin Grimm escribis: {Quote hidden}> Funny you should ask, Im working on the same project at the moment. > The only problem Im having is the maths routines, they seem to be getting > up to > 64 bit. > The actual calculation is- > dMV=G[dE + E*T1*I + D* (dE/T2)] > where- > dMV = change in Manipulated Variable > G = Gain (100/Proportional Band) > E = Error (SetPoint - ProcessVariable) for reverse action, > Error (ProcessVariable - Setpoint) for direct action > dE = ErrorThisScan - ErrorLastScan > T2 = Time between scans (in minutes) > T1 = Time since error occured (in minutes) > I = Integral time (R/M) > D = Derivative time (minutes) > > There are other things to consider as well, integral windup, bumpless > transfer > from manual to auto, etc > > Anyone knowing any alterations to this equation or anything else > that might make writing this code easier are most welcome to reply > > Justin Grimm EraseMEreaperspam_OUTTakeThisOuTsouthwest.com.au > > ---------- > > From: Mark Lezama <mfidelspam_OUTTELCEL.NET.VE> > > To: @spam@PICLISTKILLspamMITVMA.MIT.EDU > > Subject: PID question. > > Date: Friday, 6 March 1998 21:09 > > > > I have working with RTU designe base in PIC16c74. I have to control a > > valve with a 4-20 ma signal, but I want to implement a digital PID > > controller, anyone here have a PID code or can explain me how I have to > > implement. > > > > Thanks in advance! > > > > Mark Lezama

No worries, Mark Time since error occured means- the time (in minutes) since the error was last zero i.e. PV = SP. The integral multiplies the 'GAIN constant' by the ERROR. The result is then multiplied by the 'INTEGRAL constant' (repeats/minute) and then multiplied by the TIME since the PV "last" equalled the SP. The TIME for this calculation should be in minutes or any part of. i.e. 2.039 minutes. As you can see the integral component corrects for the offset over time. I hope this helps Let us know of your progress Justin Grimm KILLspamreaperKILLspamsouthwest.com.au ---------- > From: Mark Lezama <RemoveMEmfidelTakeThisOuTTELCEL.NET.VE> > To: spamBeGonePICLISTspamBeGoneMITVMA.MIT.EDU > Subject: Re: PID question. > Date: Saturday, 7 March 1998 22:45 > > Thanks Justin. I will try to implement. I think is the problem is the # of > bit I have to use a floting point rutine. BTW, in the formula I dont know > what means "Time since error ocurred", maybe Is a fool question but I am > trying to understand anything before implement the code. > > Mark Lezama. > > Justin Grimm escribis: > > > Funny you should ask, Im working on the same project at the moment. > > The only problem Im having is the maths routines, they seem to be getting {Quote hidden}> > up to > > 64 bit. > > The actual calculation is- > > dMV=G[dE + E*T1*I + D* (dE/T2)] > > where- > > dMV = change in Manipulated Variable > > G = Gain (100/Proportional Band) > > E = Error (SetPoint - ProcessVariable) for reverse action, > > Error (ProcessVariable - Setpoint) for direct action > > dE = ErrorThisScan - ErrorLastScan > > T2 = Time between scans (in minutes) > > T1 = Time since error occured (in minutes) > > I = Integral time (R/M) > > D = Derivative time (minutes) > > > > There are other things to consider as well, integral windup, bumpless > > transfer > > from manual to auto, etc > > > > Anyone knowing any alterations to this equation or anything else > > that might make writing this code easier are most welcome to reply > > > > Justin Grimm TakeThisOuTreaperEraseMEspam_OUTsouthwest.com.au > > > > ---------- > > > From: Mark Lezama <RemoveMEmfidelTakeThisOuTTELCEL.NET.VE> > > > To: PICLISTEraseME.....MITVMA.MIT.EDU > > > Subject: PID question. > > > Date: Friday, 6 March 1998 21:09 > > > > > > I have working with RTU designe base in PIC16c74. I have to control a > > > valve with a 4-20 ma signal, but I want to implement a digital PID > > > controller, anyone here have a PID code or can explain me how I have to > > > implement. > > > > > > Thanks in advance! > > > > > > Mark Lezama

Also Microchip's AN531 app note on their website has some PID code in it. Good luck Justin Grimm EraseMEreapersouthwest.com.au ---------- > From: Mark Lezama <RemoveMEmfidelEraseMEEraseMETELCEL.NET.VE> > To: RemoveMEPICLISTspam_OUTKILLspamMITVMA.MIT.EDU > Subject: Re: PID question. > Date: Saturday, 7 March 1998 22:39 > > Yes, you are right. but I have and interface with a host computer, and the > parameters need to be change from a long distance. Is there any way to do this > with OP-amps? at lest in digital comuniation. > > Mark Lezama > > Calvin escribis: > > > Comment: > > > > PID controls are inherently analog. This does not mean they cannot be > > implemented digitally, but it is generally a lot easier to do it with one or > > two op-amps. > > > > Just a thoght > > > > Calvin > > > > {Original Message removed}