VFD , motor , PLC (start /stop/position application)

Peter Nachtwey said:

Our customers do it all the time. One must only do a simple calculation each scan ramp distance =(Velocity^2)/(2*decelertion rate). When the distance to go is less than or equal to the deceleration rate then the axis must start decelerating. Sometimes it is desirable to start ramping down one time period earlier so the equation becomes:
ramp distance =(Velocity^2)/(2*decelertion rate) + velocity*scan time.

Click to expand...

Pardon my ignorance, but I was under the impression that the equation for that integral was V^2/d . Where did the 2d come from?

Eeek!! Sacrilege!

Originally posted by Peter Nachtwey:

One must only do a simple calculation each scan ramp distance =(Velocity^2)/(2*decelertion rate).

Click to expand...

What? No jerk? No snap? No crackle? That better be one STIFF assembly.

Just kidding. I do it, too.

CroCop, the base integral is:

D = 1/2 * a * t²

with:
D = distance traveled during accel
a = accel (or in this case decel) rate
t = time of acceleration

For a linear acceleration the t above can be redefined as V/a

with:
V = final velocity (for accel) or initial velocity (for decel)
a = accel or decel rate

Substituting that into the above equation results in

D = 1/2 * a * (V/a)²
D = 1/2 * a * V²/a²

D = V²/2*a


Also, I disagree that the OP needs a servomotor. He almost certainly needs a better drive. He may need a bigger motor. But he really hasn't specifically said anything that should lead anyone to believe he has an inertia acceleration issue. Granted, he made that comment about immediate stops. But what he really needs here is a position profiler to give him a position command his system has a chance of following, not necessarily a drive system that will allow him to closer match an impossible command.

Keith

Excellent

I was lazy and looked up this:
http://hyperphysics.phy-astr.gsu.edu/Hbase/mot.html#mot5

I like this hyperphyics site.

What? No jerk? No snap? No crackle? That better be one STIFF assembly.

Click to expand...

You are right, v^2=2AD assumes the initial velocity is constant. No acceleration, jerk, snap, crackle or pop allowed.
The math for higher order systems is very complex. I worked it out with the help of Mathcad.

How are the 5th order polynomials coming? One of our hardware engineers worked out the equation for clutching using a 4th order polynomial in a fly shear application.

Originally posted by Peter Nachtwey:

How are the 5th order polynomials coming? One of our hardware engineers worked out the equation for clutching using a 4th order polynomial in a fly shear application.

Click to expand...

Not real well. I've gotten lazy with spring here. As much as I like 5^th order polynomials and all, playing with the kids outside and playing fastpitch seem like better options at this point. But sooner or later my curiosity will get the better of me and I'll dig back in.

Keith

kamenges said:

CroCop, the base integral is:

D = 1/2 * a * t²

with:
D = distance traveled during accel
a = accel (or in this case decel) rate
t = time of acceleration

For a linear acceleration the t above can be redefined as V/a

with:
V = final velocity (for accel) or initial velocity (for decel)
a = accel or decel rate

Substituting that into the above equation results in

D = 1/2 * a * (V/a)²
D = 1/2 * a * V²/a²

D = V²/2*a


Keith

Click to expand...

All right. I'd messed up the integral, that's what I get for not paying attention in Calc.

Good thing I'm back in school next semester, LOL.

Peter Nachtwey said:

Yes, lots. I used to work for a company that installed the mainly the scanning veneer lathe optimizer. Now we sell to these companies. I was quite literally thrown to the wolves when it came to motion control but the systems were well designed so scaling and tuning the motion controllers was easy. I didn't know it could be hard. It all seemed intuitive so here I am 25+ years later.

I have movies .wmv and .mp4 file on our ftp site.
ftp://ftp.deltacompsys.com/public/movies
These are provided by my customers.


Our customers do it all the time. One must only do a simple calculation each scan ramp distance =(Velocity^2)/(2*decelertion rate). When the distance to go is less than or equal to the deceleration rate then the axis must start decelerating. Sometimes it is desirable to start ramping down one time period earlier so the equation becomes:
ramp distance =(Velocity^2)/(2*decelertion rate) + velocity*scan time.

Click to expand...

I need to start looking at the age of the posts, I would have saved my fingers.

I understand that stopping at a position with varying speeds is common in motion control. I recently commissioned my 5th RMC 100 and as a matter of fact this system needed to do just that.

The original poster was trying to use a ramp function in his PLC along with decel times in his drive. The two wouldn't have worked together.

Which brings me to a question for the experts. Whenever I have set up the accel-decel times for a VFD I have always done so based on the physics of the drive and motor. If it takes 5S to accel and decel the biggest load, that is what I set the ramp rates to.

Would it make sense to set the accel-decel rates to 0 and let the controller (in this case the PLC) to control the signal to the drive? Using either a ramp function or the formulas posted above this would seem the way to make a "poor man's motion controller"

I understand there are differences with a true servo system such as motor inertia, scan times, etc....

I am intereseted in responses from anyone who has tried this or from the experts who can show me the errors of my thinking.

thanks

You bring up an interesting questions.

allscott said:

Would it make sense to set the accel-decel rates to 0 and let the controller (in this case the PLC) to control the signal to the drive? Using either a ramp function or the formulas posted above this would seem the way to make a "poor man's motion controller"

thanks

Click to expand...

I would think about it for a second or two if the required accuracy wasn't to high and the ramps can be gentle. I would do some calculations first just to see if there was a chance it would work. There are some questions to be answered first.
1. What is the effect of closing the loop at 10 milliseconds as opposed to 1 millisecond.
2. How will the jitter in closing the loop on a PLC affect the motion.
3. How does the type of ramp affect the acceleration and deceleration?
4. How does one implement the target generator so velocity and acceleration feed forwards can be used?

I am surprized no one has ever asked me question #1.

Any time I have any controller issuing speed commands to a drive I set the accel/decel rated to the minimumn value the drive will allow and do all the ramping in the controller, typically a plc. I just like not having to worry about another accel/decel rate influencing the system if something changes.

Your first point about setting the accel/decel rate the the physical limits of the system isn't such a bad thing either. However, with modern drives, the torque/current limits will just as likely cause velocity limiting as the velocity ramp does if you actually do set the accel/decel rated to the performance limit.

On most of the machines we build the mechanical system is designed with enough safety factor built in that the motor can't break the mechanicals no matter how hard it pushes. In those cases when we need to do a very fast stop I will set the speed command directly to zero and let the drive torque limit software control the decel rate.

Keith

I recently commissioned my 5th RMC 100

Click to expand...

Allscott what are you useing these on? Only installed 1 on an edger set works. It was loads of fun though, only time i`ve ever been involvled with real motion control.

Thomas Sullens said:

Allscott what are you useing these on? Only installed 1 on an edger set works. It was loads of fun though, only time i`ve ever been involvled with real motion control.

Click to expand...

Before I snapped I had been working in the steel industry for the past several years. In the plants I have been in almost everything was hydraulic.

I installed the first one to replace an obsolete Rexroth controller on a flying cutoff saw. The saw had two saw blades that were moved in and out with hydraulic cylinders at very high speeds and with pinpoint accuracy.

The second two were on a couple of steel sheet shears. This was a fairly simple application where two large cylinders needed to be moved in sync with each other with vastly changing loads on each. This had previously been done with bosch controllers which just never seemed to work right.

The fourth was adjusting the welding gap on a pipe mill. I won't go into detail as it is difficult to explain.

The last was for feed rate control on a pipe bevelling machine. Picture a big milling machine on it's side, the delta replaced what was basically a pressure compensated flow control for moving the cutting head in and out.

There is almost no limit for this technology, if you have a hydraulic cylinder that you want to move accurately this is the way to do it. The nice thing about the Delta controllers is that you don't have to be a hydraulic or controlls genius to set them up.

Delta controllers is that you don't have to be a hydraulic or controlls genius to set them up.

Click to expand...

I guess this what i really enjoyed the setworks job with this controller. They make some of us appear alot more technical than we are! Maybe one of these days we`ll build a piece of equipment that needs this kind of control. Thought we had last year, but after talking with Peter he explained another way to do what we were trying to achieve. It was regulating clamping preasure, but through a rotating union hooked on the back of a 6" cylinder where the cylinder turns with the shaft. Still think it would`ve worked betteeer with rmc100! Oh well some people are just not good salesmen!!!

Peter,

Thanks for the videos. It felt like home hearing and watching a lathe run again. I take it you did the positioning for the XY charger? The veneer sorter/stacker was very cool too. I'm not sure what the exact process was there, but it was a lot slower than what I am use too on the green end. Still a very nice, smooth application. The sharp chain and verticle saw mill was just awsome. I've been around some of that too. I don't think it could be described without actually seeing what all happens there. Very cool stuff. I couldn't quie make it out, but does the verticle saw do the same job as a curve saw? Or is this the "opti-mill" saw that makes the cants? That mill looked very familure. Even the sorter infeed looked like the one I have been around.

Thanks again,

Robert

plchacker said:

Peter,

Thanks for the videos. It felt like home hearing and watching a lathe run again. I take it you did the positioning for the XY charger?

Click to expand...

Yes, I worked for Applied Theory which is now part of USNR. Applied Theory provided controls for Durand ( now Raute Wood ) and Premier lathes. That was 20+ years ago. I was told the lathe in the video has over 34 axes of motion or 5 RMCs.

plchacker said:

The veneer sorter/stacker was very cool too. I'm not sure what the exact process was there, but it was a lot slower than what I am use too on the green end.

Click to expand...

This is an automatic lay up line built by Raute Wood ( Lahti Finland, Vancouver, BC Canada) and the site is in Louisiana east of Shreveport. Glue is applied to the inner sheets and they are over laided on top each other to make a big board below that is 30-50 ft long and 4 inches thick. It then goes to a press and fused into a big board that then can be cut into very long and big beams. It is a process that makes big ones out of small ones. There are two parts to the machine. The upstream part aligns the sheets so the line up with the laser line. There are two photo cells that determine the skew. The controller then speeds up or slows down some of the belts so the sheet is aligned. This uses a technique called super-imposed moves which I don't think has ever been discussed on this forum. The second part, which is shown first, shows the gates opening and closing. The belts must go in an opposite and equal direction so the sheets are not moved or dragged due to friction. The sheets fall below on the big wood 'billet' that you can see move if you watch carefully. Every thing you see above, the motion and the sequencing is all done by the controller. This application uses 15HP servos. The line is actually moving faster than designed and the controller is often sending + or - 10 volts to the drives.

plchacker said:

Still a very nice, smooth application. The sharp chain and verticle saw mill was just awsome. I've been around some of that too. I don't think it could be described without actually seeing what all happens there. Very cool stuff. I couldn't quie make it out, but does the verticle saw do the same job as a curve saw?

Click to expand...

Trees aren't straight and the best way to cut them is to follow the grain which is a curve. The scanner-optimizer downloads a cam table to the RMC for each piece of wood and each axis. There are at least 3 sets of axes working here moving the top and bottom chippers and saws. Each cam is geared to the feed chain so the cam action will slow down if the feed chain slows down.

plchacker said:

Or is this the "opti-mill" saw that makes the cants? That mill looked very familure. Even the sorter infeed looked like the one I have been around.

Click to expand...

This system was done by Porter Engineering, also of British Columbia. There is also a company called Optimil in British Columbia which often makes the machinery for Porter Engineering but I believe this machinery was made by Newnes also from British Columbia.

Thanks again Peter,


Without taking over the thread, our Spray-Line was built in BC as well, but was much different. I never did find out exactly who built the machine the first time. We rebuilt it and added new controls. Fun stuff. Gray Code encoder was my bestest friend.

Premier Gear builds a fine Lathe. Coe, too. I haven't worked with a Raute Wood. I'll take wood products over Pupl/Paper any day.