Hi Alexey,

 

Yes that sounds similar to what others have done such as User Bengt with a VFD.

 

The first step is to see if you can control the raw speed with the PWM.  Use the PWM1KHz.c example to see if it is possible to control the speed.

 

There is no direct support of PWM output mode.  A few lines of C can be used to write the output of an axis channel to a PWM.  You can then operate the spindle as a closed loop servo where the axis channel will adjust the output (PWM) so that the encoder feedback moves at the commanded speed.  See the OutputToPWM.c example (but remove the additions of 128 because you have direction and magnitude rather than anti-phase control).

 

Are you using Mach3 or KMotionCNC or ?

 

Regards

TK

Group: DynoMotion	Message: 5268	From: Tom Kerekes	Date: 6/19/2012
Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu

Hi Alexey,   I like to do things step-by-step.  I asked for you to test the PWM first.  Have you done this?   The next step would be to test the encoder.  Does it count?   Regards TK Group: DynoMotion Message: 5269 From: Alexey Volkov Date: 6/19/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Ok, I would try the PWM ASAP. Yes, the encoder does counts. Best regards, Alexey Sent from my iPhone On 2012-06-19, at 8:31 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   I like to do things step-by-step.  I asked for you to test the PWM first.  Have you done this?   The next step would be to test the encoder.  Does it count?   Regards TK Group: DynoMotion Message: 5277 From: Alexey Volkov Date: 6/20/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hello Tom The direct PWM control is working now, direction control and on/off is also working, as an encoder. The PWM output is quite non-linear 8-(. Shoud I fix it or the encoded feedback would take care of it? What is next? Best regards, Alexey Sent from my iPhone On 2012-06-19, at 8:31 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   I like to do things step-by-step.  I asked for you to test the PWM first.  Have you done this?   The next step would be to test the encoder.  Does it count?   Regards TK Group: DynoMotion Message: 5278 From: Tom Kerekes Date: 6/20/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey,   You forgot to state how non-linear it is. You might try different PWM frequencies to see if it is more linear.   The next step would be to try to servo.    I think you could use Bengt's VFD C Program:   http://f1.grp.yahoofs.com/v1/0DniT23gZGM4hZO1Tas7mIoS3IhWIknbjZnImwc9H389n3wMLTYgqneXqQW6pgcj_TNpqgLZINfgIerqGyaTCg/Bengt%27s%20CombiMill/PWMtoVFD/OutputToPWMandDirGearV6.c   and just ignore or strip out the Hi/Lo speed range stuff.  You didn't describe how your enable or direction control works. You may need to modify things for your system   Have the C Program running, configure an axis as Encoder Input, No Output, no filters, small P gain 0.01, Max Error 100000, Max Output 255 (or less), then on the step response screen Zero the axis, Enable the Axis, then make a move.   You may need to reverse the direction control or encoder count direction (InputGain0) to have negative feedback rather than positive feedback run away.   Regards TK   Group: DynoMotion Message: 5279 From: Alexey Volkov Date: 6/20/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hello Tom It is quite non-linear right now but I would try to fix it with common-collector buffer. More like progressive correlation: PWM values had to be multiplied by 2 if I want to stay linear. Different PWM frequency pre-scaler does not affect the linearity too much. I think what the motor driver itself does affect the integrator.  The "enable" and "direction" control is just 2 bits (IO27 enable, IO28 CCW). I even don't have to stop the PWM to reverse the direction, just flipping the bit and motor drive does the rest. By the way, how the encoder resolution comes into the work? Best regards, Alexey Sent from my iPhone On 2012-06-20, at 2:14 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   You forgot to state how non-linear it is. You might try different PWM frequencies to see if it is more linear.   The next step would be to try to servo.    I think you could use Bengt's VFD C Program:   http://f1.grp.yahoofs.com/v1/0DniT23gZGM4hZO1Tas7mIoS3IhWIknbjZnImwc9H389n3wMLTYgqneXqQW6pgcj_TNpqgLZINfgIerqGyaTCg/Bengt%27s%20CombiMill/PWMtoVFD/OutputToPWMandDirGearV6.c   and just ignore or strip out the Hi/Lo speed range stuff.  You didn't describe how your enable or direction control works. You may need to modify things for your system   Have the C Program running, configure an axis as Encoder Input, No Output, no filters, small P gain 0.01, Max Error 100000, Max Output 255 (or less), then on the step response screen Zero the axis, Enable the Axis, then make a move.   You may need to reverse the direction control or encoder count direction (InputGain0) to have negative feedback rather than positive feedback run away.   Regards TK   Group: DynoMotion Message: 5280 From: Tom Kerekes Date: 6/20/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey,   After you get the Spindle Axis working as a Servo Axis you will then be able to Jog the axis accurately at any number of encoder counts/sec that you wish.  The encoder resolution will affect the conversion factor that you will need to use to convert from RPM to encoder counts/sec.  For example if the encoder counts/rev is 4000 then 1000RPM would be:   1000 rev/min / (60 sec/min) x 4000 counts/rev = 66667 counts/sec   so the conversion factor from RPM to counts/sec would be 66.667   Regards TK     Group: DynoMotion Message: 5281 From: Alexey Volkov Date: 6/20/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Attachments : Hello Tom Thanks for the explanations. I would try out the servo mode ASAP I have found out the correlation of the PWM output to the Spindle RPM It looks like  PWM_OUT =  where x is desired speed. Is it possible to use kflop filters for this? Best regards, Alexey On Wed, Jun 20, 2012 at 3:44 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   After you get the Spindle Axis working as a Servo Axis you will then be able to Jog the axis accurately at any number of encoder counts/sec that you wish.  The encoder resolution will affect the conversion factor that you will need to use to convert from RPM to encoder counts/sec.  For example if the encoder counts/rev is 4000 then 1000RPM would be:   1000 rev/min / (60 sec/min) x 4000 counts/rev = 66667 counts/sec   so the conversion factor from RPM to counts/sec would be 66.667   Regards TK     Group: DynoMotion Message: 5283 From: Tom Kerekes Date: 6/21/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey, See if you can get Bengt's C Program to work for you. I would need to plot your cubic equation over your intended range of RPM to see how non-linear it is.  But if I plug in 0 RPM then the PWM value would be 30 which is ok/good that we are able to stop with a small or zero PWM value. If you were to run open loop you could just put this cubic equation right in your C code to convert commanded RPM to a fixed PWM value.  But this probably isn't necessary if you intend to run closed loop. Regards TK From: Alexey Volkov <erry321@...> To: DynoMotion@yahoogroups.com Sent: Wednesday, June 20, 2012 9:00 PM Subject: Re: [DynoMotion] KFlop PWM to 0-10V converter + 400 lines encoder: how to configure output?   Hello Tom Thanks for the explanations. I would try out the servo mode ASAP I have found out the correlation of the PWM output to the Spindle RPM It looks like  PWM_OUT =  where x is desired speed. Is it possible to use kflop filters for this? Best regards, Alexey On Wed, Jun 20, 2012 at 3:44 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   After you get the Spindle Axis working as a Servo Axis you will then be able to Jog the axis accurately at any number of encoder counts/sec that you wish.  The encoder resolution will affect the conversion factor that you will need to use to convert from RPM to encoder counts/sec.  For example if the encoder counts/rev is 4000 then 1000RPM would be:   1000 rev/min / (60 sec/min) x 4000 counts/rev = 66667 counts/sec   so the conversion factor from RPM to counts/sec would be 66.667   Regards TK     Group: DynoMotion Message: 5286 From: Alexey Volkov Date: 6/21/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hello Tom I just have finished implementing the PWM control into my init file. So far so good, the spindle start is a little bit rough (it overshoot, slows down and goes slow to the set speed) but the speed stabilizes in 4-5 seconds. I have played with P and D, right now P=0.01 and D=0.01. It is the best parameters I have found so far. Is it possible to use float for PWM output to increase the resolution? The speed has minor oscillation, but hardly noticeable and not critical right now, but I have feeling what it mostly because low PWM resolution due integer type used and non-linearity  of the 0-10v converter. So, my understanding, right now I have to write a bunch of small programs for the different M-codes related to the spindle control.  Best regards,  Alexey Sent from my iPhone On 2012-06-21, at 9:14 AM, Tom Kerekes <tk@...> wrote: Hi Alexey, See if you can get Bengt's C Program to work for you. I would need to plot your cubic equation over your intended range of RPM to see how non-linear it is.  But if I plug in 0 RPM then the PWM value would be 30 which is ok/good that we are able to stop with a small or zero PWM value. If you were to run open loop you could just put this cubic equation right in your C code to convert commanded RPM to a fixed PWM value.  But this probably isn't necessary if you intend to run closed loop. Regards TK From: Alexey Volkov <erry321@...> To: DynoMotion@yahoogroups.com Sent: Wednesday, June 20, 2012 9:00 PM Subject: Re: [DynoMotion] KFlop PWM to 0-10V converter + 400 lines encoder: how to configure output?   Hello Tom Thanks for the explanations. I would try out the servo mode ASAP I have found out the correlation of the PWM output to the Spindle RPM It looks like  PWM_OUT = <image.png> where x is desired speed. Is it possible to use kflop filters for this? Best regards, Alexey On Wed, Jun 20, 2012 at 3:44 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   After you get the Spindle Axis working as a Servo Axis you will then be able to Jog the axis accurately at any number of encoder counts/sec that you wish.  The encoder resolution will affect the conversion factor that you will need to use to convert from RPM to encoder counts/sec.  For example if the encoder counts/rev is 4000 then 1000RPM would be:   1000 rev/min / (60 sec/min) x 4000 counts/rev = 66667 counts/sec   so the conversion factor from RPM to counts/sec would be 66.667   Regards TK     Group: DynoMotion Message: 5289 From: Tom Kerekes Date: 6/21/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey,   If you posted some plots we might be able to offer ideas on how to improve things.  You might want to add a low pass filter to smooth the output.  You might want to add ~25 PWM counts to the Output if it takes a PWM value of 30 before it starts to move.  Otherwise with a P Gain of 0.01 an error of 3000 counts will build up before enough output to cause any motion.  If you do add 25 to the PWM then you should decrease the Max Output to 255-25 = 230 so that the max PWM is never exceeded.   The PWM is inherently an integer value (a counter).  With an 8-bit PWM each count is ~ 0.4% of full range.   There is actually a means of using a Step/Dir Generator as a PWM where by the pulse is fixed and the frequency varies to get a different duty cycle.  This has a potential of higher resolution on the low end of duty cycle.  However it involves some math.  See the StepAsPWM.c example.   There are already written C programs for controlling the Spindle by Jogging an Axis.  See the C:\KMotion429\C Programs\SpindleUsingJogs directory.   Regards TK         Group: DynoMotion Message: 5290 From: Alexey Volkov Date: 6/21/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hello Tom How do I add the base to PWM? Should I use the offset or I can just add it inside the cycle? Regarding the PWM linearity, I would try to build convertor with the better parameters. I am using the optocoupler to drive the convertor. It may affect the linearity also due the turning on/off timing difference. Thanks for the help and for the explanations. Kflop is a great controller, it is just too much can be done, so it is really hard to find the information. It is nowhere in documentation I can find how to hook up the PWM to the axis. Thanks a lot and best regards, Alexey P.S. I still have a way to go with the machine, more questions coming. Sent from my iPhone On 2012-06-21, at 12:17 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   If you posted some plots we might be able to offer ideas on how to improve things.  You might want to add a low pass filter to smooth the output.  You might want to add ~25 PWM counts to the Output if it takes a PWM value of 30 before it starts to move.  Otherwise with a P Gain of 0.01 an error of 3000 counts will build up before enough output to cause any motion.  If you do add 25 to the PWM then you should decrease the Max Output to 255-25 = 230 so that the max PWM is never exceeded.   The PWM is inherently an integer value (a counter).  With an 8-bit PWM each count is ~ 0.4% of full range.   There is actually a means of using a Step/Dir Generator as a PWM where by the pulse is fixed and the frequency varies to get a different duty cycle.  This has a potential of higher resolution on the low end of duty cycle.  However it involves some math.  See the StepAsPWM.c example.   There are already written C programs for controlling the Spindle by Jogging an Axis.  See the C:\KMotion429\C Programs\SpindleUsingJogs directory.   Regards TK         Group: DynoMotion Message: 5291 From: Tom Kerekes Date: 6/21/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey,   Yes something like      FPGA(IO_PWMS+0) = NewOutput + 25;  // put the speed correcting for offset Thanks for your patience.   Regards TK   Group: DynoMotion Message: 5292 From: Alexey Volkov Date: 6/22/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hello Tom I told you, I would have a lot of questions. I have a questions about the user variables and variables. I have been using simplified versions of the of the SpindleUsingJogs example programs. The SpindleMillJog.c  program uses #1 variable and #97, #98, #99 user variables. So, to set it as the one to be executed as a SXXX g-code command I have choose thread #2 and Var 1. It works. It successfully extract RPM information and even can put the data into user variable #97 and some more information in #98 and #99. Question is:   How can I access to those user variables from another program running in another thread or after? Or what I should do, declare. etc. to make those variables available to the other programs? I have been trying to transfer the speed into SpindleOnCWJog.c, no success so far. It acts as #97, #98, #99 are empty. What I am doing wrong? Best regards, Alexey On Thu, Jun 21, 2012 at 3:20 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   Yes something like      FPGA(IO_PWMS+0) = NewOutput + 25;  // put the speed correcting for offset Thanks for your patience.   Regards TK   Group: DynoMotion Message: 5293 From: Tom Kerekes Date: 6/22/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey, The persist.UserData variables are global and will persist after a Thread terminates.  So I don't understand what the problem might be. Just assign the SpindleOnCWJog.c program to the M3 variable and it should work. You might add printf statements to debug what is going wrong. Regards TK From: Alexey Volkov <erry321@...> To: DynoMotion@yahoogroups.com Sent: Friday, June 22, 2012 3:07 PM Subject: Re: [DynoMotion] KFlop PWM to 0-10V converter + 400 lines encoder: how to configure output?   Hello Tom I told you, I would have a lot of questions. I have a questions about the user variables and variables. I have been using simplified versions of the of the SpindleUsingJogs example programs. The SpindleMillJog.c  program uses #1 variable and #97, #98, #99 user variables. So, to set it as the one to be executed as a SXXX g-code command I have choose thread #2 and Var 1. It works. It successfully extract RPM information and even can put the data into user variable #97 and some more information in #98 and #99. Question is:   How can I access to those user variables from another program running in another thread or after? Or what I should do, declare. etc. to make those variables available to the other programs? I have been trying to transfer the speed into SpindleOnCWJog.c, no success so far. It acts as #97, #98, #99 are empty. What I am doing wrong? Best regards, Alexey On Thu, Jun 21, 2012 at 3:20 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   Yes something like      FPGA(IO_PWMS+0) = NewOutput + 25;  // put the speed correcting for offset Thanks for your patience.   Regards TK   Group: DynoMotion Message: 5294 From: Alexey Volkov Date: 6/22/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu HelloTom I works now. Did not changed anything. One more question: I have some jitter on the encoders, (the encoders have very low resolution of 100ppr, just to detect the stall and steps skipping) Is it possible to apply small filtration on the encoder inputs to reduce the motor chatter? Best regards,  Alexey Sent from my iPhone On 2012-06-22, at 3:26 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey, The persist.UserData variables are global and will persist after a Thread terminates.  So I don't understand what the problem might be. Just assign the SpindleOnCWJog.c program to the M3 variable and it should work. You might add printf statements to debug what is going wrong. Regards TK From: Alexey Volkov <erry321@...> To: DynoMotion@yahoogroups.com Sent: Friday, June 22, 2012 3:07 PM Subject: Re: [DynoMotion] KFlop PWM to 0-10V converter + 400 lines encoder: how to configure output?   Hello Tom I told you, I would have a lot of questions. I have a questions about the user variables and variables. I have been using simplified versions of the of the SpindleUsingJogs example programs. The SpindleMillJog.c  program uses #1 variable and #97, #98, #99 user variables. So, to set it as the one to be executed as a SXXX g-code command I have choose thread #2 and Var 1. It works. It successfully extract RPM information and even can put the data into user variable #97 and some more information in #98 and #99. Question is:   How can I access to those user variables from another program running in another thread or after? Or what I should do, declare. etc. to make those variables available to the other programs? I have been trying to transfer the speed into SpindleOnCWJog.c, no success so far. It acts as #97, #98, #99 are empty. What I am doing wrong? Best regards, Alexey On Thu, Jun 21, 2012 at 3:20 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   Yes something like      FPGA(IO_PWMS+0) = NewOutput + 25;  // put the speed correcting for offset Thanks for your patience.   Regards TK   Group: DynoMotion Message: 5295 From: Tom Kerekes Date: 6/22/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Hi Alexey,   I assume you are running closed loop Steppers?  A plot always helps us understand what you are referring to.  But I assume it is ~1 count of servo dither?  You might be able to reduce it by using the Dead Band (Range 1 ~ 2 counts, gain of 0.0 ~ 0.01).  Otherwise you can essentially run open loop by setting the PID gains to zero and still detect following errors.   Regards TK  Group: DynoMotion Message: 5296 From: Alexey Volkov Date: 6/22/2012 Subject: Re: KFlop PWM to 0-10V converter + 400 lines encoder: how to configu Thanks, Tom Yes, it is a closed loop. I would try the dead band. Best regards, Alexey Sent from my iPhone On 2012-06-22, at 4:18 PM, Tom Kerekes <tk@...> wrote:   Hi Alexey,   I assume you are running closed loop Steppers?  A plot always helps us understand what you are referring to.  But I assume it is ~1 count of servo dither?  You might be able to reduce it by using the Dead Band (Range 1 ~ 2 counts, gain of 0.0 ~ 0.01).  Otherwise you can essentially run open loop by setting the PID gains to zero and still detect following errors.   Regards TK