Flying Shear Controller Cutting Acuuracy

Hi All, I Have a question, I am very thank full to you if help me,

I designed a flying shear controller based on fpga and microcontroller,

but the cutting error is more than 5 mm,

i want to reach cutting error to less than 1 mm,


{

The attached control block diagram is as same as control techniques documents,

best possible sample time of main loop is 2 ms,

all calculations are floating point for ease of operation // it seams it is better to change to integer for higher man loop speed
}

The main question : the control loop is optimum or not?
Thanks again for your valuable time

So you mean that most of all control is done inside the ac drive ?

The block shown appears to assume this is done in a Control Techniques Unidrive, probably on the motion module. I can't tell if this is the actual platform or if this diagram is just a model for a controller implemented on a different platform. I'm pretty sure Control Techniques has a canned flying shear implementation that you can get from them. That might be the way to go.

Is your error always in one direction or is it bipolar? Where in the profile does the error occur and what is the shape of the error value relative to the move profile?

A couple things catch my eye:
1) Are you sure your Kvff value is set to the optimal value?
2) There is no acceleration feed forward torque term from what I can see. This would be a very good thing to have.

Keith

widelto said:

So you mean that most of all control is done inside the ac drive ?

Click to expand...

No, There is a hardware based on FPGA And Microcontroller.

if done with microcontroller the scantime is way faster thus accuracy is far better.
get the speed from external encoder, make in your controller a array with speeds, and encoderdistances the shear needs to have to cut on correct position.
lets assume runningspeed is 1m/s
so the scantime of 2 ms will give you a max accuracy of 3 mm.
with a correct calibrated arduino you can get to the resolution of the encoder used.
The motor will need a certain time to speed up,
so the table has for 0.5 a value for example of 12 emcoder
for 1 m/s 18
for 2 m/s 32 steps etc.

kamenges said:

The block shown appears to assume this is done in a Control Techniques Unidrive, probably on the motion module. I can't tell if this is the actual platform or if this diagram is just a model for a controller implemented on a different platform. I'm pretty sure Control Techniques has a canned flying shear implementation that you can get from them. That might be the way to go.

Is your error always in one direction or is it bipolar? Where in the profile does the error occur and what is the shape of the error value relative to the move profile?

A couple things catch my eye:
1) Are you sure your Kvff value is set to the optimal value?
2) There is no acceleration feed forward torque term from what I can see. This would be a very good thing to have.

Keith

Click to expand...

Thank you Keith for your response

It seems the error always is in one direction, for your other questions, maybe my calculation is the source of error,

PP_Math=Proportional_Gain/100.0; //kp

Car_Vel_FF_Math=2048.0/(Max_Real_Carg_Velocity_float/(60.0)*Line_Pulse_m*Sample_Time_ms*Line_Average_Number/1000.0);//Kvff

for example if max carriage velocity is 110.9 meter/min
line-pulse-m is 72150 pulse/meter
sample_tmie = 2ms
Line_Average_Number = 10 that means line speed averaged at 20 ms then we have Car_Vel_FF_Math = 0.76786

Car_Vel_FF_Term =(int)(Filter_Line_Velocity * Car_Vel_FF_Math);
if(Car_Vel_FF_Term>=Max_Output)Car_Vel_FF_Term=Max_Output;
Pos_Error=Product_Length - Desired_Length; // in pulses
Proportional_Term = (int)(Pos_Error * PP_Math * 1.0);
if(Proportional_Limit==0)Proportional_Term=0;
else if((Proportional_Term>Proportional_Limit)&&(Proportional_Term>0))Proportional_Term=Proportional_Limit;
else if((Proportional_Term<-Proportional_Limit)&&(Proportional_Term<0))Proportional_Term=-Proportional_Limit;
float_Car_Cruise_Output =(int)(Proportional_Term+Car_Vel_FF_Term);
if(float_Car_Cruise_Output >= Max_Output) float_Car_Cruise_Output = Max_Output;

shooter said:

if done with microcontroller the scantime is way faster thus accuracy is far better.
get the speed from external encoder, make in your controller a array with speeds, and encoderdistances the shear needs to have to cut on correct position.
lets assume runningspeed is 1m/s
so the scantime of 2 ms will give you a max accuracy of 3 mm.
with a correct calibrated arduino you can get to the resolution of the encoder used.
The motor will need a certain time to speed up,
so the table has for 0.5 a value for example of 12 emcoder
for 1 m/s 18
for 2 m/s 32 steps etc.

Click to expand...

Thank you shooter,
data acquisition from encoders is done by fpga, and control loop is run in micro, because of my floating point calculation total control loop run time is 1.2 msec and I closed the loop at every 2 msec, do you say if the control loop scan time be less than 0.5 ms the drive and motor can respond?

I don't have enough information about your system to know if your velocity feed forward multiplier is correct or not. However, assuming that the Filter_Line_Velocity is truly a velocity (units/sec or units/min) there seem to be some numbers in your calculation I don't quite understand.

You indicate that the error is always one direction. Which direction?

What is developing your position command value? Are you confident that is correct?

Keith

kamenges said:

I don't have enough information about your system to know if your velocity feed forward multiplier is correct or not. However, assuming that the Filter_Line_Velocity is truly a velocity (units/sec or units/min) there seem to be some numbers in your calculation I don't quite understand.

You indicate that the error is always one direction. Which direction?

What is developing your position command value? Are you confident that is correct?

Keith

Click to expand...

Always the error is more than which is preset, for example the desired length is 6000 mm but actual is 6005 or 6006.
My main question is if Kff * Vline + kp * error that error is difference in actual and desired length is adequate for position loop or not? integral term is needed or not? what is the best algorithm to have same velocity at desired length,
is 2 msec cycle time adequate for position loop or not?

Rahni: If I were you i would use something like this:
http://www.motrona.com/cutter_control_1.html

widelto said:

Rahni: If I were you i would use something like this:
http://www.motrona.com/cutter_control_1.html

Click to expand...

Yes, I have used many idea from motrona FS150 flying shear controller in my own controller, but thing that I don't know is the best approach for control algorithm!!!

My main question is if Kff * Vline + kp * error is adequate for position loop or not? [error is difference in actual length and desired length]. integral term is needed or not? what is the best algorithm to have same velocity at desired length,
is 2 msec cycle time adequate for position loop or not?

i don't know much the rocket formula


are you using ac motor or servo motor?
what drive you used?

osmanmom said:

i don't know much the rocket formula


are you using ac motor or servo motor?
what drive you used?

Click to expand...

DC Motor / SANETRNO DCREG4 with tachometer feedback

what sensor detecting the position?