High-pass Filter for S7 position control

[Cuke]

Hello

does any one know or have a source file for a high-pass filter for S7 to help with high speed position control over-shot

Thanks in advance

Cuke

 

[Peter Nachtwey]

I don't have anything for S7 specifically.

ftp://ftp.deltacompsys.com/public/mcd/Butterworth Hi Pass.htm
ftp://ftp.deltacompsys.com/public/mcd/Bessel High Pass.htm

You can convert this to S7 without to much difficulty.

So what does a high pass filter have to do with overshoot?

 

[Cuke]

If i create an acceleration actual measuerd from my moving axis and use this to act on my output from the loop to limit the output or increase, but use the filter to only act on a fequency I could then increae my gains to reduce the lag??? A Question do spline curves help a system in motion or say with linear?

 

[Peter Nachtwey]

I think what you are looking for is a differentiator which should be the D part of a PID.

Use feed forwards to reduce lag. Velocity feed forward reduce lag while at constant velocity. Acceleration feed forwards reduce lag while accelerating and decelerating.

S-curves should provide better results but they also require higher accelerations rates if the ramp is to be completed in the same time. S-curves reduce the amount of oscillation at the end of the ramp.

 

[Cuke]

Thank you for the help

We are going to be starting up today I hope.So I'll see what's the situation, the machine is on Linear rails and beeing controled by hydraulic servo valves, previously it was on tie bars so I had lots of help from the mechinacal structure to slow down

Is it normal to use D on high speed servo hydraulic system? What are the pit falls if any?

Another question, what is an optimal scan time for the update of a postion control loop traveling at say 1500m/s over a distance of 300mm that including ramp and brake

sorry for all the questions

Thanks in advance

Cuke

 

[Peter Nachtwey]

Is it normal to use D on high speed servo hydraulic system? What are the pit falls if any?

YES!!!! However, it depends on the quality of the motion controller and the feedback. It is VERY important that the feedback be sampled on regular intervals. A PLC cannot do this. We sample 1024 times a second using a FPGA so there is no software latency. This guarantees the sampling in done EXACTLY every 1/1024 of a second.

Another question, what is an optimal scan time for the update of a postion control loop traveling at say 1500m/s over a distance of 300mm that including ramp and brake.

Have you calculated the acceleration and deceleration rates required to achieve your goals? I have. You are starting to push the limits.
Check this out.

ftp://ftp.deltacompsys.com/public/PDF/Mathcad - SineMove SI.pdf

I am assuming your 1.5 m/s is a peak speed. This move will take .4 seconds I am also assuming a sinusoidal type of ramp because the motion profile constains only one fundamental frequency. A trapezoidal move will ring after accelerating and decelerating.

Don't be surprise if you have difficulties with this project.

 

[Cuke]

I'll let you know how we get on, I've also tried this with softplc and managed to get updates of 1ms on my loops that's doing 8 axis this one is PLC (normally 2ms) so we'll see.

Most of my stuff is Linear and I calculate before doing the trajectory
for saftey but I've mangaed to put a spline curve with ratio points to the other points and we'll try that( use percent from target to calculate the other points as the target changes rather than fixed??)

Any way great talking to you and once again thanks
Cuke