Hi friends, I am working on new pharmaceutical project in that I need to control the three DC motor at constant speed. I have worked in motion control project on Silab 8051 based processor. Please suggest me for, 1) good "PID" book, tutorials, example etc.. 2) micro-controller & motor driver ckt or dev / evl kit. (processor I prefer Silab 8051 or cortex-m3 ) Thanks in advance, Kishore.
need suggestion for PID in DC motor controlling
Started by ●December 4, 2009
Reply by ●December 4, 20092009-12-04
On Dec 4, 3:56=A0am, kishor <kiis...@gmail.com> wrote:> Hi friends, > I am working on new pharmaceutical project in that I need to control > the three DC motor at constant speed. > I have worked in motion control project on Silab 8051 based > processor. > > Please suggest me for, > 1) good "PID" book, tutorials, example etc.. > 2) micro-controller & motor driver ckt or dev / evl kit. (processor I > prefer Silab 8051 or cortex-m3 ) > > Thanks in advance, > Kishore.http://www.engin.umich.edu/group/ctm/examples/motor/PID2.html Look around the CTM PID website because there is more info. You will be pushing your 8051 to the limits. I would look at a TI starter kit. I recently bought one for $400 that included a TI Piccolo DSP and two small motors. It has Code Composer and source code so it will run out of the box. The TI kit is using PWM outputs. Peter Nachtwey
Reply by ●December 4, 20092009-12-04
On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote:> Hi friends, > I am working on new pharmaceutical project in that I need to control the > three DC motor at constant speed. I have worked in motion control > project on Silab 8051 based processor. > > Please suggest me for, > 1) good "PID" book, tutorials, example etc.. 2) micro-controller & motor > driver ckt or dev / evl kit. (processor I prefer Silab 8051 or cortex-m3 > ) >The one by Åström and Hagglund is supposed to be good. For general control theory in an embedded system you may be interested in mine; see http://www.wescottdesign.com/actfes/actfes.html. An 8051 isn't much processor for this sort of thing -- you'll probably end up with the sampling rate limited by the chip, and the control quality limited by the sampling rate. You'd really rather be up against the limits of performance of the plant, instead. So I'd recommend that you look to that Cortex part instead... -- www.wescottdesign.com
Reply by ●December 5, 20092009-12-05
On Fri, 04 Dec 2009 17:21:11 -0600, Tim Wescott <tim@seemywebsite.com> wrote:>On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote: > >> Hi friends, >> I am working on new pharmaceutical project in that I need to control the >> three DC motor at constant speed. I have worked in motion control >> project on Silab 8051 based processor. >> >> Please suggest me for, >> 1) good "PID" book, tutorials, example etc.. 2) micro-controller & motor >> driver ckt or dev / evl kit. (processor I prefer Silab 8051 or cortex-m3 >> ) >> >The one by �str�m and Hagglund is supposed to be good.It is. If you are talking about "PID Controllers: Theory, Design, and Tuning," 2nd ed., which I have here and have read and enjoyed.>For general >control theory in an embedded system you may be interested in mine; see >http://www.wescottdesign.com/actfes/actfes.html. > >An 8051 isn't much processor for this sort of thing -- you'll probably >end up with the sampling rate limited by the chip, and the control >quality limited by the sampling rate. You'd really rather be up against >the limits of performance of the plant, instead.The SiLabs 8051 may actually be the only family sporting a 16-bit 1MHz SAR ADC with DMA (necessary) controller to transfer to memory. I'm using it, now. Works fine, though their DMA docs leave something to be desired. (I can fill in details.) I require about 600kHz rate and am getting excellent results. Processing power is pretty good, but I admit having written much in assembly just to make absolutely sure. The fast data rate is sparsely used for the two PID loops I'm running, which are in c. But their rates are much slower, so it's no problem. And it's not motor control.>So I'd recommend that you look to that Cortex part instead...Perhaps. But I wouldn't abruptly set aside the SiLabs parts, either. They perform pretty well. (You know more about motor control apps than I do, though.) Jon
Reply by ●December 5, 20092009-12-05
On Dec 4, 9:37=A0pm, Jon Kirwan <j...@infinitefactors.org> wrote:> On Fri, 04 Dec 2009 17:21:11 -0600, Tim Wescott <t...@seemywebsite.com> > wrote: > > >On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote: > > >> Hi friends, > >> I am working on new pharmaceutical project in that I need to control t=he> >> three DC motor at constant speed. I have worked in motion control > >> project on Silab 8051 based processor. > > >> Please suggest me for, > >> 1) good "PID" book, tutorials, example etc.. 2) micro-controller & mot=or> >> driver ckt or dev / evl kit. (processor I prefer Silab 8051 or cortex-=m3> >> ) > > >The one by =C5str=F6m and Hagglund is supposed to be good. > > It is. =A0If you are talking about "PID Controllers: Theory, Design, and > Tuning," 2nd ed., which I have here and have read and enjoyed. > > >For general > >control theory in an embedded system you may be interested in mine; see > >http://www.wescottdesign.com/actfes/actfes.html. > > >An 8051 isn't much processor for this sort of thing -- you'll probably > >end up with the sampling rate limited by the chip, and the control > >quality limited by the sampling rate. =A0You'd really rather be up again=st> >the limits of performance of the plant, instead. > > The SiLabs 8051 may actually be the only family sporting a 16-bit 1MHz > SAR ADC with DMA (necessary) controller to transfer to memory. =A0I'm > using it, now. =A0Works fine, though their DMA docs leave something to > be desired. =A0(I can fill in details.) =A0I require about 600kHz rate an=d> am getting excellent results. =A0Processing power is pretty good, but I > admit having written much in assembly just to make absolutely sure. > The fast data rate is sparsely used for the two PID loops I'm running, > which are in c. =A0But their rates are much slower, so it's no problem. > And it's not motor control. > > >So I'd recommend that you look to that Cortex part instead... > > Perhaps. =A0But I wouldn't abruptly set aside the SiLabs parts, either. > They perform pretty well. =A0(You know more about motor control apps > than I do, though.) > > JonEveryone makes such a big deal about the PID routine when it is really a minor part of any motion controller. It is possible to execute PIDs very quickly but that alone isn't very useful. Running a PID with fast updates introduces problems that can't be easily solved with integer only CPUs. The reason for this is that the integrator coefficient are often too small to fit in a 16 bit integer. Likewise the derivative coefficients is often very big, so big that overflow becomes a problem. In short, the faster you want to run the PID the greater the dynamic range must be. This usually requires 32 bit math. I question the 600kHz rate on a 8051. We use SiLabs 8051s in some of our products but not for motion control. We have made a SIMPLE 8051 PID controller. It worked well but I wouldn't use one for controlling DC motors. Peter Nachtwey
Reply by ●December 5, 20092009-12-05
On Sat, 05 Dec 2009 09:02:23 -0800, pnachtwey wrote:> On Dec 4, 9:37 pm, Jon Kirwan <j...@infinitefactors.org> wrote: >> On Fri, 04 Dec 2009 17:21:11 -0600, Tim Wescott <t...@seemywebsite.com> >> wrote: >> >> >On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote: >> >> >> Hi friends, >> >> I am working on new pharmaceutical project in that I need to control >> >> the three DC motor at constant speed. I have worked in motion >> >> control project on Silab 8051 based processor. >> >> >> Please suggest me for, >> >> 1) good "PID" book, tutorials, example etc.. 2) micro-controller & >> >> motor driver ckt or dev / evl kit. (processor I prefer Silab 8051 or >> >> cortex-m3 ) >> >> >The one by Åström and Hagglund is supposed to be good. >> >> It is. If you are talking about "PID Controllers: Theory, Design, and >> Tuning," 2nd ed., which I have here and have read and enjoyed. >> >> >For general >> >control theory in an embedded system you may be interested in mine; >> >see http://www.wescottdesign.com/actfes/actfes.html. >> >> >An 8051 isn't much processor for this sort of thing -- you'll probably >> >end up with the sampling rate limited by the chip, and the control >> >quality limited by the sampling rate. You'd really rather be up >> >against the limits of performance of the plant, instead. >> >> The SiLabs 8051 may actually be the only family sporting a 16-bit 1MHz >> SAR ADC with DMA (necessary) controller to transfer to memory. I'm >> using it, now. Works fine, though their DMA docs leave something to be >> desired. (I can fill in details.) I require about 600kHz rate and am >> getting excellent results. Processing power is pretty good, but I >> admit having written much in assembly just to make absolutely sure. The >> fast data rate is sparsely used for the two PID loops I'm running, >> which are in c. But their rates are much slower, so it's no problem. >> And it's not motor control. >> >> >So I'd recommend that you look to that Cortex part instead... >> >> Perhaps. But I wouldn't abruptly set aside the SiLabs parts, either. >> They perform pretty well. (You know more about motor control apps than >> I do, though.) >> >> Jon > Everyone makes such a big deal about the PID routine when it is really a > minor part of any motion controller. It is possible to execute PIDs > very quickly but that alone isn't very useful. Running a PID with fast > updates introduces problems that can't be easily solved with integer > only CPUs. The reason for this is that the integrator coefficient are > often too small to fit in a 16 bit integer. Likewise the derivative > coefficients is often very big, so big that overflow becomes a problem. > In short, the faster you want to run the PID the greater the dynamic > range must be. This usually requires 32 bit math. > > I question the 600kHz rate on a 8051. We use SiLabs 8051s in some of our > products but not for motion control. We have made a SIMPLE 8051 PID > controller. It worked well but I wouldn't use one for controlling DC > motors. > > Peter NachtweyI rarely get away with 16-bit math for motor control -- I've had some geared axes where the important thing was to reliably run into a stop 'fast enough but not too fast'; for that I could get away with it. Just about everything else I do ends up using 32 bits. I suppose I was being a bit hasty rejecting the Scilabs part. The 'real' answer is to estimate the processing power you need for your application then evaluate the processors you like -- but often on a small production run the better answer is to get a Really Capable microprocessor that exceeds all your performance specifications by an order of magnitude -- this lets you concentrate on the problem to be solved, instead of spending effort (hence engineering money) to shoehorn the solution into a place where it barely fits. -- www.wescottdesign.com
Reply by ●December 5, 20092009-12-05
On Sat, 5 Dec 2009 09:02:23 -0800 (PST), pnachtwey <pnachtwey@gmail.com> wrote:>On Dec 4, 9:37�pm, Jon Kirwan <j...@infinitefactors.org> wrote: >> On Fri, 04 Dec 2009 17:21:11 -0600, Tim Wescott <t...@seemywebsite.com> >> wrote: >> >> >On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote: >> >> >> Hi friends, >> >> I am working on new pharmaceutical project in that I need to control the >> >> three DC motor at constant speed. I have worked in motion control >> >> project on Silab 8051 based processor. >> >> >> Please suggest me for, >> >> 1) good "PID" book, tutorials, example etc.. 2) micro-controller & motor >> >> driver ckt or dev / evl kit. (processor I prefer Silab 8051 or cortex-m3 >> >> ) >> >> >The one by �str�m and Hagglund is supposed to be good. >> >> It is. �If you are talking about "PID Controllers: Theory, Design, and >> Tuning," 2nd ed., which I have here and have read and enjoyed. >> >> >For general >> >control theory in an embedded system you may be interested in mine; see >> >http://www.wescottdesign.com/actfes/actfes.html. >> >> >An 8051 isn't much processor for this sort of thing -- you'll probably >> >end up with the sampling rate limited by the chip, and the control >> >quality limited by the sampling rate. �You'd really rather be up against >> >the limits of performance of the plant, instead. >> >> The SiLabs 8051 may actually be the only family sporting a 16-bit 1MHz >> SAR ADC with DMA (necessary) controller to transfer to memory. �I'm >> using it, now. �Works fine, though their DMA docs leave something to >> be desired. �(I can fill in details.) �I require about 600kHz rate and >> am getting excellent results. �Processing power is pretty good, but I >> admit having written much in assembly just to make absolutely sure. >> The fast data rate is sparsely used for the two PID loops I'm running, >> which are in c. �But their rates are much slower, so it's no problem. >> And it's not motor control. >> >> >So I'd recommend that you look to that Cortex part instead... >> >> Perhaps. �But I wouldn't abruptly set aside the SiLabs parts, either. >> They perform pretty well. �(You know more about motor control apps >> than I do, though.) >> >> Jon>Everyone makes such a big deal about the PID routine when it is really >a minor part of any motion controller. It is possible to execute PIDs >very quickly but that alone isn't very useful. Running a PID with >fast updates introduces problems that can't be easily solved with >integer only CPUs.Many integer-only CPUs are quite capable of wide dynamic range computations at fast update rates. For example, I use hand coded 24-bit and 48-bit accumulations on the 8051. You don't explicitly say it, but floating point hardware (if that's your meaning) isn't the dividing line here. It just makes it easier for some.>The reason for this is that the integrator >coefficient are often too small to fit in a 16 bit integer. Likewise >the derivative coefficients is often very big, so big that overflow >becomes a problem. In short, the faster you want to run the PID the >greater the dynamic range must be. This usually requires 32 bit math. > >I question the 600kHz rate on a 8051.That's not the PID rate, just the raw ADC operational rate. And the results are excellent for the purpose intended. The two PIDs I have are driven at the rate of 10Hz, which is nothing like 600kHz. I explicitly said that before, along with the fact that they aren't used for driving DC motors, and I sense you may have been confused when reading my words. But maybe you mean something quite directed, here, that I've missed in reading you.>We use SiLabs 8051s in some of >our products but not for motion control. We have made a SIMPLE 8051 >PID controller. It worked well but I wouldn't use one for controlling >DC motors.Can you talk about specifically why no SiLabs parts are adequate for controlling a DC motor? You brought up the issue here of 'integer only', so I might assume that broad brush is why. But it may not be your reason, which is why I'm asking. I'm interested in your thinking here. Jon
Reply by ●December 5, 20092009-12-05
On Sat, 05 Dec 2009 11:57:30 -0600, Tim Wescott <tim@seemywebsite.com> wrote:>On Sat, 05 Dec 2009 09:02:23 -0800, pnachtwey wrote: > >> On Dec 4, 9:37�pm, Jon Kirwan <j...@infinitefactors.org> wrote: >>> On Fri, 04 Dec 2009 17:21:11 -0600, Tim Wescott <t...@seemywebsite.com> >>> wrote: >>> >>> >On Fri, 04 Dec 2009 03:56:15 -0800, kishor wrote: >>> >>> >> Hi friends, >>> >> I am working on new pharmaceutical project in that I need to control >>> >> the three DC motor at constant speed. I have worked in motion >>> >> control project on Silab 8051 based processor. >>> >>> >> Please suggest me for, >>> >> 1) good "PID" book, tutorials, example etc.. 2) micro-controller & >>> >> motor driver ckt or dev / evl kit. (processor I prefer Silab 8051 or >>> >> cortex-m3 ) >>> >>> >The one by �str�m and Hagglund is supposed to be good. >>> >>> It is. �If you are talking about "PID Controllers: Theory, Design, and >>> Tuning," 2nd ed., which I have here and have read and enjoyed. >>> >>> >For general >>> >control theory in an embedded system you may be interested in mine; >>> >see http://www.wescottdesign.com/actfes/actfes.html. >>> >>> >An 8051 isn't much processor for this sort of thing -- you'll probably >>> >end up with the sampling rate limited by the chip, and the control >>> >quality limited by the sampling rate. �You'd really rather be up >>> >against the limits of performance of the plant, instead. >>> >>> The SiLabs 8051 may actually be the only family sporting a 16-bit 1MHz >>> SAR ADC with DMA (necessary) controller to transfer to memory. �I'm >>> using it, now. �Works fine, though their DMA docs leave something to be >>> desired. �(I can fill in details.) �I require about 600kHz rate and am >>> getting excellent results. �Processing power is pretty good, but I >>> admit having written much in assembly just to make absolutely sure. The >>> fast data rate is sparsely used for the two PID loops I'm running, >>> which are in c. �But their rates are much slower, so it's no problem. >>> And it's not motor control. >>> >>> >So I'd recommend that you look to that Cortex part instead... >>> >>> Perhaps. �But I wouldn't abruptly set aside the SiLabs parts, either. >>> They perform pretty well. �(You know more about motor control apps than >>> I do, though.) >>> >>> Jon >> Everyone makes such a big deal about the PID routine when it is really a >> minor part of any motion controller. It is possible to execute PIDs >> very quickly but that alone isn't very useful. Running a PID with fast >> updates introduces problems that can't be easily solved with integer >> only CPUs. The reason for this is that the integrator coefficient are >> often too small to fit in a 16 bit integer. Likewise the derivative >> coefficients is often very big, so big that overflow becomes a problem. >> In short, the faster you want to run the PID the greater the dynamic >> range must be. This usually requires 32 bit math. >> >> I question the 600kHz rate on a 8051. We use SiLabs 8051s in some of our >> products but not for motion control. We have made a SIMPLE 8051 PID >> controller. It worked well but I wouldn't use one for controlling DC >> motors. >> >> Peter Nachtwey > >I rarely get away with 16-bit math for motor control -- I've had some >geared axes where the important thing was to reliably run into a stop >'fast enough but not too fast'; for that I could get away with it. Just >about everything else I do ends up using 32 bits. > >I suppose I was being a bit hasty rejecting the Scilabs part. The 'real' >answer is to estimate the processing power you need for your application >then evaluate the processors you like -- but often on a small production >run the better answer is to get a Really Capable microprocessor that >exceeds all your performance specifications by an order of magnitude -- >this lets you concentrate on the problem to be solved, instead of >spending effort (hence engineering money) to shoehorn the solution into a >place where it barely fits.Good points. In my case, the cost of the SiLabs part was the same as a 16-bit SAR ADC and cost and size is _very_ important. Trade-offs to consider. What has surprised me here is that I had anticipated far more 'shoehorn'ing than I've actually had to do. So getting the application from zero (no code) to 100% of prototype target code was one month's time. A lot better than I'd hoped. And the remaining processor power is far more than I'd planned on. Which opens up some interesting possibilities regarding 'features' I would have opposed, prior to this experience. The processor runs (in my case) at 25MHz. The instructions, roughly speaking, require a clock count equal to their encoded byte count. A custom-written (by me, over the span of about 1 day though testing added some on the second day), 24-bit integer input (ADC accumulator results), 32-bit FP output ln() takes 420 clocks (memory serving) to execute (about 17 microseconds.) Output precision is monotonic to 21 bits... slightly less than the 23+1 that exists in the FP format. I had expected much worse performance. Jon
Reply by ●December 5, 20092009-12-05
On Dec 5, 10:34=A0am, Jon Kirwan <j...@infinitefactors.org> wrote:> Can you talk about specifically why no SiLabs parts are adequate for > controlling a DC motor? =A0You brought up the issue here of 'integer > only', so I might assume that broad brush is why. =A0But it may not be > your reason, which is why I'm asking. =A0I'm interested in your thinking > here.I am not picking on Silab parts. I wouldn't even use most 32 bit processors. I know you can make software 32 or 48 bit accumulators but it can't be done quickly either in programming time or execution time. The PID is such a small part of what a motion controller must do. What about the motion profile generator? That is the hard part if you need to process commands on-the-fly. The calculations get to be intensive. A lot depends on your definition of control. A PID holding a motor at a position is very limited. I say AMEN to Tim's statement about getting a microprocessor that has 10 times the estimated required processing power. Peter Nachtwey
Reply by ●December 5, 20092009-12-05
On Sat, 5 Dec 2009 16:39:07 -0800 (PST), pnachtwey <pnachtwey@gmail.com> wrote:>On Dec 5, 10:34�am, Jon Kirwan <j...@infinitefactors.org> wrote: >> Can you talk about specifically why no SiLabs parts are adequate for >> controlling a DC motor? �You brought up the issue here of 'integer >> only', so I might assume that broad brush is why. �But it may not be >> your reason, which is why I'm asking. �I'm interested in your thinking >> here. >I am not picking on Silab parts. I wouldn't even use most 32 bit >processors. I know you can make software 32 or 48 bit accumulators >but it can't be done quickly either in programming time or execution >time. The PID is such a small part of what a motion controller must >do. What about the motion profile generator? That is the hard part >if you need to process commands on-the-fly. The calculations get to >be intensive. A lot depends on your definition of control. A PID >holding a motor at a position is very limited. > >I say AMEN to Tim's statement about getting a microprocessor that has >10 times the estimated required processing power. > >Peter NachtweyOkay. Thanks for taking a moment to clarify. I agree with both of you about getting something grossly overpowered, as well. If the application space permits, tools are appropriate, etc. My experience is in cost-sensitive environments, not feature- sensitive, with significant pressure on what can be afforded for cpu speed. Such factors must be balanced _better_ than your competition to provide a wedge to improve odds at success. And it is better still if you can 'do the impossible.' :) Jon P.S. I'm broadly competent in numerical methods, transcendental and float implementations, and so on. So, where for others it is daunting it may be less so for me when implementing such custom routines. As I earlier pointed out, it took me a little over a day to start from zero to fully test an ln() function written in assembly, which was a central brick for the app. Others would find that a longer process, especially if unfamiliar. Available skill sets help determine what's a reasonable option and what isn't, as always.
