Dancer Speed follower circuit before i build one

[Jamie]

This is the scenario:
Lets say we have a catenary arm with a pot
on it that monitors the catenary position. This
arm has a chive (roller) where we wrap wire around
it to serve as a pully. As the wire starts to be pulled
away from the machine, this arm will move thus moving the
POT which signals the drive to give the reel of wire a
push to maintain the catenary arm in a desired position.

Now, we use regen drives for this application (Penta KB series)
to be exact, and here is the problem..
The load on the shaft is heavy and what happens is when the
catenary arm moves in position to signal the drive, there may be
a delay of response due to the extra swing load of weight and
thus the arm might end up in a position where high bias voltage is
now at its point and when the shaft finally gets going in the other
direction, it goes to fast! And thus, this extra weight generates
a lot of it's own momentum. (Swing load).

This is what i want to know before i go and build it. cause if
it's on the market i would rather buy it.

I need to detect the speed of which the catenary arm is moving,
not the reel speed! I do not want to employ encoders on the shaft
because of expense reasons.
It must generate a negative response to what the POT wants the
drive to do as the POT starts to change position and as soon as the pot
stops, the POT's real signal will be reported at the position it is
currently in.
I need this type of control to not allow the drive to accelerate so
fast that the catenary arm whiplashes.
My thoughts on this for ideal thinking, is something like a voltage
follower that mirrors the POT's signal, a capacitive system will create
an inhibiting signal as the voltage changes but not inhibit when the
voltage from the POT is stable or changing very slowly.

The end results is, the POT's signal to the drive will change slowly
to get the catenary arm into position but not too slow where it will
cause it to stall in position. Lead, Lags, Delays etc found on most
Dancer and drive controls does not do what i need, due to the fact that
a lot of these circuits tend to memorize last time effects and just
compound the problem.

I hope i have made this clear enough, if there something on the market
in like a mini board form that runs off lets say the REF's supply
voltage of around 10 or 15 volts that conditions the signal it would be
nice..
Other wise, i will have to make my own debouncer circuit..

P.S.
we have already tried outboard dancer boards like the old Reflex series
and the drive it self has an extensive set of controls which serve their
purpose just fine how ever, none of them give me this kind of control.

I have seen this in eurotherm digital drives where the max speed can
be retarded by the rate in which the POT value changes. The faster the
POT changes, the slower the drive/motor attempts to go.
but in our case, we are using non-digital drives ( a lot of them) with
Armature feed back!..
Thanks for any input.

 

[Tim Wescott]

This is the scenario:
Lets say we have a catenary arm with a pot
on it that monitors the catenary position. This
arm has a chive (roller) where we wrap wire around
it to serve as a pully. As the wire starts to be pulled
away from the machine, this arm will move thus moving the
POT which signals the drive to give the reel of wire a
push to maintain the catenary arm in a desired position.

Now, we use regen drives for this application (Penta KB series)
to be exact, and here is the problem..
The load on the shaft is heavy and what happens is when the
catenary arm moves in position to signal the drive, there may be
a delay of response due to the extra swing load of weight and
thus the arm might end up in a position where high bias voltage is
now at its point and when the shaft finally gets going in the other
direction, it goes to fast! And thus, this extra weight generates
a lot of it's own momentum. (Swing load).

This is what i want to know before i go and build it. cause if
it's on the market i would rather buy it.

I need to detect the speed of which the catenary arm is moving,
not the reel speed! I do not want to employ encoders on the shaft
because of expense reasons.
It must generate a negative response to what the POT wants the
drive to do as the POT starts to change position and as soon as the pot
stops, the POT's real signal will be reported at the position it is
currently in.
I need this type of control to not allow the drive to accelerate so
fast that the catenary arm whiplashes.
My thoughts on this for ideal thinking, is something like a voltage
follower that mirrors the POT's signal, a capacitive system will create
an inhibiting signal as the voltage changes but not inhibit when the
voltage from the POT is stable or changing very slowly.

The end results is, the POT's signal to the drive will change slowly
to get the catenary arm into position but not too slow where it will
cause it to stall in position. Lead, Lags, Delays etc found on most
Dancer and drive controls does not do what i need, due to the fact that
a lot of these circuits tend to memorize last time effects and just
compound the problem.

I hope i have made this clear enough, if there something on the market
in like a mini board form that runs off lets say the REF's supply
voltage of around 10 or 15 volts that conditions the signal it would be
nice..
Other wise, i will have to make my own debouncer circuit..

P.S.
we have already tried outboard dancer boards like the old Reflex series
and the drive it self has an extensive set of controls which serve their
purpose just fine how ever, none of them give me this kind of control.

I have seen this in eurotherm digital drives where the max speed can
be retarded by the rate in which the POT value changes. The faster the
POT changes, the slower the drive/motor attempts to go.
but in our case, we are using non-digital drives ( a lot of them) with
Armature feed back!..
Thanks for any input.

This is a solved problem. In fact, it sounds very much like you're
asking how to design a PD controller with your dancer arm as an input.

Many motor drives have built-in PID controllers; you can often dial in
the gains without having to get an external board. I don't know if this
is possible with your particular drive, however. I have looked for
'just' PID controllers, but there don't seem to be any available that
operate at this sort of speed. In theory you could use a PLC, but they
tend to operate much too slow for motion control.

I can tell you all about building a custom controller for any given
application, but I am very weak indeed when it comes to knowing what
parts to buy off the shelf to do this sort of job. So I'm taking the
liberty of cross-posting question onto sci.engr.control, where there are
people who _do_ know the answer to this sort of question.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[Ian Macmillan]

The root of the problem is the inertia of the dancer the weight of which is
being used to set the winding tension. Part of the answer is to make the
dancer as light as possible and apply the required force by means of a
pressure controlled air cylinder, or even by using a very long spring.

The more technical solutions have been covered by other responders.

All the best
Ian Macmillan

 

[jasen]

This is the scenario:
Lets say we have a catenary arm with a pot
on it that monitors the catenary position. This
arm has a chive (roller) where we wrap wire around
it to serve as a pully. As the wire starts to be pulled
away from the machine, this arm will move thus moving the
POT which signals the drive to give the reel of wire a
push to maintain the catenary arm in a desired position.

Now, we use regen drives for this application (Penta KB series)
to be exact, and here is the problem..
The load on the shaft is heavy and what happens is when the
catenary arm moves in position to signal the drive, there may be
a delay of response due to the extra swing load of weight and
thus the arm might end up in a position where high bias voltage is
now at its point and when the shaft finally gets going in the other
direction, it goes to fast! And thus, this extra weight generates
a lot of it's own momentum. (Swing load).

This is what i want to know before i go and build it. cause if
it's on the market i would rather buy it.

sounds like youre looking for proportional-integral-differential
feedback: google for PID

 

[Jamie]

sounds like youre looking for proportional-integral-differential
feedback: google for PID

I know what PID is, I how ever, always thought it was
Proportional, Integral, Derivative ?
I also know what i want exist in Digital controllers.
I have already come up with something using an ATtiny
with ADC input, PWM output., I was really looking for
some ready made units to throw in because the boss does not
like custom electronics made in house. He is afraid
of un-availability In case for I for example may move on
to a non negotiable place. He also does not like to be at the
hands out side Electronics Engineers due to past experience.

 

[Jamie]

In another life we had these damn things...you can fiddle with PID
loops and stuff all you want, which is what we did.You have sereral
error terms, which interact to cause you misery. Here's what we
finally did to get some peace.

The dancer pot was replaced with a gray-code absolute encoder, and
connected to a GE DV-series drive (reel drive) operating in position
control mode, namely "control the position of that f#$#@ dancer." It
has a PID loop, but you don't have to fiddle with it. Later on, we
added an ultrasonic sensor to measure the diameter of the material on
the reel and used that signal to tweak the gain of the system as the
reel (and therefore the "gear ratio") increased/decreased. That was in
1997 and it runs still today. If you do that, do yourself a favor and
write down a calibration procedure at the build and keep it forever.

There is some minor stuff to do, such as setting speed limits and
faults such as a material break. The install is costly, but if you ask
me it is cheap at twice the price.

Seems better to think in servo terms rather than speed control terms.

Thanks for your input. I'm glad to read the views of others that
understand exactly where I am coming from
Using any sort of encoder set up is out of the question due to cost
over head. This project originally was suppose to be quick and dirty
with very little unexpected undesirable results. The project was done by
a contractor due to the work load we already have in the plant.
After a few Penta Drives (KB regen series) going belly up on initial
power up of the line. I started to curl my brows a little bit at the
contractor's abilities. I say this because, the week before this company
diagnosed a problem on one of our lines that coast us $7k which
didn't solve the problem at hand! It was one of our young engineers that
suggested to have them look at the problem when i had already give them
my opinion of what it was, and it sure wasn't a drive, which is what
they replaced.
Any ways, after they left that day, i decided to inspect a couple of
the drives that burned up.. What i found was soot in side with no FUSE
protection at all per drive. The only protection they were relying on
was the main line fuses that feed all the drives on that circuit, that
issue has been corrected.
Now the part with the Dancer which i have come up with something now
and testing it. I know an encoder be it an incremental of Gray code
would do the job how ever, like i said COST!.

What I'm doing is using an AtTiny uC with ADC inputs and PWM
outputs to regenerate the +/- volts.

I use one ADC input that has a POT on it to adjust my max dancer speed
movement before the circuit starts to retard the ref signal. One ADC to
read the Dancer POT. One Input to select 0 or 5 volts as a center ref.
I was able to put it on a mini board and stick it in side the NEMA 4x
box using the 15V +/- reference pot leads as the supply to operate it.
I'm using the internal OSC of this AVR so that i have extra legs to
play with. I have 2 legs more to play with for other options as this grows.
It seems to work , we'll see in actual long term use.

 

[Steve Cothran]

In another life we had these damn things...you can fiddle with PID
loops and stuff all you want, which is what we did.You have sereral
error terms, which interact to cause you misery. Here's what we
finally did to get some peace.

The dancer pot was replaced with a gray-code absolute encoder, and
connected to a GE DV-series drive (reel drive) operating in position
control mode, namely "control the position of that f#$#@ dancer." It
has a PID loop, but you don't have to fiddle with it. Later on, we
added an ultrasonic sensor to measure the diameter of the material on
the reel and used that signal to tweak the gain of the system as the
reel (and therefore the "gear ratio") increased/decreased. That was in
1997 and it runs still today. If you do that, do yourself a favor and
write down a calibration procedure at the build and keep it forever.

There is some minor stuff to do, such as setting speed limits and
faults such as a material break. The install is costly, but if you ask
me it is cheap at twice the price.

Seems better to think in servo terms rather than speed control terms.

 

[Jerry Avins]

Jamie wrote:

...

Now the part with the Dancer which i have come up with something now
and testing it. I know an encoder be it an incremental of Gray code
would do the job how ever, like i said COST!.

The encoder is cheap compared to a day of lost production. We tried to
do Iraq on the cheap, and look where it got us.

...

Jerry

 

[Paul M]

Thanks for your input. I'm glad to read the views of others that
understand exactly where I am coming from
Using any sort of encoder set up is out of the question due to cost
over head. This project originally was suppose to be quick and dirty
with very little unexpected undesirable results.

Regen drives often have pots to adjust min speed, max speed, max accel
and max decel. They often also have a "comp" adjustment which controls
boost or sag in the response. Better regen drives have PID built in. I
have found that you can often get a dancer control stable by adjusting
the max accel and decel to match the inertial load. Cut off all but
the proportional portion of PID and adjust it as wide as possible eg
min speed at pot arm travel high and max speed at pot low. It should
work.

 

[Jamie]

Regen drives often have pots to adjust min speed, max speed, max accel
and max decel. They often also have a "comp" adjustment which controls
boost or sag in the response. Better regen drives have PID built in. I
have found that you can often get a dancer control stable by adjusting
the max accel and decel to match the inertial load. Cut off all but
the proportional portion of PID and adjust it as wide as possible eg
min speed at pot arm travel high and max speed at pot low. It should
work.

Yes, these drives have all of those controls how ever, they do not
correct for the problems totally.
These pay off units use a catenary arm that has a short travel to
them. If it wasn't for this short travel, we wouldn't be having so many
problems.
We can balance the units fine on lite loads but on heavy loads, they
start to whiplash due to the swing load that is taking place. making the
DB (dead band) at it's widest helps the problem greatly other wise, you
get out of control rocking, because the drive is regening and we need
this function to maintain tension at idle.

Any ways, the Sampling of readings using the AVR seems to work in
detecting the rate of POT value changing. it's kind of strange looking
when you watch this in action. The Catenary moves about 2 inches back
and forth slowly as it runs, but no jerking and slamming the arm on it's
stops. We might apply this same little unit on some over head
accumulators so that the operator does not have to manual nurse the
dollies in place for dance mode after a real chance on the fly.

 

[Mark]

I know what PID is, I how ever, always thought it was
Proportional, Integral, Derivative ?
I also know what i want exist in Digital controllers.
I have already come up with something using an ATtiny
with ADC input, PWM output., I was really looking for
some ready made units to throw in because the boss does not
like custom electronics made in house. He is afraid
of un-availability In case for I for example may move on
to a non negotiable place. He also does not like to be at the
hands out side Electronics Engineers due to past experience.

Would an accelerometer mounted someplace to give you a lead term be
helpfull?
Mark

 

[Paul M]

Yes, these drives have all of those controls how ever, they do not
correct for the problems totally.
These pay off units use a catenary arm that has a short travel to
them. If it wasn't for this short travel, we wouldn't be having so many
problems.
We can balance the units fine on lite loads but on heavy loads, they
start to whiplash due to the swing load that is taking place. making the
DB (dead band) at it's widest helps the problem greatly other wise, you
get out of control rocking, because the drive is regening and we need
this function to maintain tension at idle.

Any ways, the Sampling of readings using the AVR seems to work in
detecting the rate of POT value changing. it's kind of strange looking
when you watch this in action. The Catenary moves about 2 inches back
and forth slowly as it runs, but no jerking and slamming the arm on it's
stops. We might apply this same little unit on some over head
accumulators so that the operator does not have to manual nurse the
dollies in place for dance mode after a real chance on the fly.

I don't know what your process is, but I have never seen a system that
works well by tensioning with a drive. I would abandon that idea and
tension with a weighted roller arm, using the drive to maintain the
arm position only. You can vary the tension by using a movable weight
on the tension arm.

 

[Steve Cothran]

.. I would abandon that idea and

tension with a weighted roller arm, using the drive to maintain the
arm position only. You can vary the tension by using a movable weight
on the tension arm.

Yeah, and even better if he festooned the stuff on weighted rollers.

 

[Anthony]

I don't know what your process is, but I have never seen a system that
works well by tensioning with a drive. I would abandon that idea and
tension with a weighted roller arm, using the drive to maintain the
arm position only. You can vary the tension by using a movable weight
on the tension arm.

hrm.. There are tons of them that work just dandy in the paper industry.

--
Anthony

You can't 'idiot proof' anything....every time you try, they just make
better idiots.

Remove sp to reply via email

 

[Stanislaw Chmielarz]

hrm.. There are tons of them that work just dandy in the paper
industry.

What do You think about such tension control I found in
flexographic printer controlled by Digitrac by MAGPOWR?

 

[Paul M]

hrm.. There are tons of them that work just dandy in the paper industry.

You are right, I should not have made such a blanket statement. Let me
qualify this a little. Using the controls and motors he indicated
and/or I am assuming, the systems I have seen do not work as well as a
dancer roll simply because each part of the system has to be matched
exactly to the characteristics of parts in the system which change
with speed and material. Inertia varies with product weight. Product
properties differ. If you want to make a line flexible, then it needs
to be able to compensate for those things. Dancers can do this because
they make slack in the web while tensioning at the same time. Take a
hypothetical line that rolls out dough. As the does is calendared and
pulled to a roller conveyer it stretches. A dancer roll is actually
measuring that stretch and adjusting the roll speed as product varies.
The arm can be adjusted to pull on the web only slightly where a motor
tensioning system must overcome its on friction, which varies as the
machine ages and bearings get dirty. Adjusting the controls so they
work today will not insure they work next week.

So on a line that needs minimum tension and the material properties
vary, a dancer works much better.

I am working now on designing a new type of tensioning system for a
machine that wraps a material under tension onto a spool using a regen
drive system. In this situation I am looking for obtaining consistent
high tension values. I believe that this will work better than the
systems out there now. It's ironic that I would be working on
designing a drive tension system and make the statement that they
don't work well, but it really depends on the product properties. The
paper industry needs either a follower system with no tension or it is
tensioning a product with little to no stretch.

 

[Jerry Avins]

You are right, I should not have made such a blanket statement. Let me
qualify this a little. Using the controls and motors he indicated
and/or I am assuming, the systems I have seen do not work as well as a
dancer roll simply because each part of the system has to be matched
exactly to the characteristics of parts in the system which change
with speed and material. Inertia varies with product weight. Product
properties differ. If you want to make a line flexible, then it needs
to be able to compensate for those things. Dancers can do this because
they make slack in the web while tensioning at the same time. Take a
hypothetical line that rolls out dough. As the does is calendared and
pulled to a roller conveyer it stretches. A dancer roll is actually
measuring that stretch and adjusting the roll speed as product varies.
The arm can be adjusted to pull on the web only slightly where a motor
tensioning system must overcome its on friction, which varies as the
machine ages and bearings get dirty. Adjusting the controls so they
work today will not insure they work next week.

So on a line that needs minimum tension and the material properties
vary, a dancer works much better.

I am working now on designing a new type of tensioning system for a
machine that wraps a material under tension onto a spool using a regen
drive system. In this situation I am looking for obtaining consistent
high tension values. I believe that this will work better than the
systems out there now. It's ironic that I would be working on
designing a drive tension system and make the statement that they
don't work well, but it really depends on the product properties. The
paper industry needs either a follower system with no tension or it is
tensioning a product with little to no stretch.

Do you recall the vacuum columns on the old reel-to-reel computer tape
drives? Constant negative pressure in the column maintained constant
tape tension no matter which reel held the most tape, and photocells
sensed the loop positions, adjusting the servoed speed of the reel motors.

Jerry

 

[Jamie]

You are right, I should not have made such a blanket statement. Let me
qualify this a little. Using the controls and motors he indicated
and/or I am assuming, the systems I have seen do not work as well as a
dancer roll simply because each part of the system has to be matched
exactly to the characteristics of parts in the system which change
with speed and material. Inertia varies with product weight. Product
properties differ. If you want to make a line flexible, then it needs
to be able to compensate for those things. Dancers can do this because
they make slack in the web while tensioning at the same time. Take a
hypothetical line that rolls out dough. As the does is calendared and
pulled to a roller conveyer it stretches. A dancer roll is actually
measuring that stretch and adjusting the roll speed as product varies.
The arm can be adjusted to pull on the web only slightly where a motor
tensioning system must overcome its on friction, which varies as the
machine ages and bearings get dirty. Adjusting the controls so they
work today will not insure they work next week.

So on a line that needs minimum tension and the material properties
vary, a dancer works much better.

I am working now on designing a new type of tensioning system for a
machine that wraps a material under tension onto a spool using a regen
drive system. In this situation I am looking for obtaining consistent
high tension values. I believe that this will work better than the
systems out there now. It's ironic that I would be working on
designing a drive tension system and make the statement that they
don't work well, but it really depends on the product properties. The
paper industry needs either a follower system with no tension or it is
tensioning a product with little to no stretch.

I have corrected the problem by using an AVR "atmini" processor , i
simply pass the Pot Analog in, process it, pass the results back out to
recreate the analog signal but only conditioned, to correct for whiplash
control.
The system we have only uses a catenary arm roller with aprox 15" in
length and has no more than aprox 70 degrees max in swing.
It would have been simpler if we had an accumulator, but that isn't
the case here.
I gotta say, this has turned out much better than I thought it would
have. I was able to get the unit to track it so even at faster speeds,
the unit would condition the signal to force the arm into the near 0
volt position of the POT.
For most types of dancer controls, usually faster speeds means the
resting point of the arm will tend to be in different places at various
speeds.
Yesterday, we applied this circuit to an over head accumulator, to
control the fill speed until it reached resting dancing position. By
using a precision high resolution multiturn pot, we were able to get the
processor board to monitor and bias the slip clutch drive to hold a
crawl speed to where the operator could simply walk away from it with
out the
need to nurse it using only a POT and no encoders for speed control
etc..
I'm very pleased with what i was able to do with this tiny little
8 pin uC..

 

[Paul M]

Do you recall the vacuum columns on the old reel-to-reel computer tape
drives? Constant negative pressure in the column maintained constant
tape tension no matter which reel held the most tape, and photocells
sensed the loop positions, adjusting the servoed speed of the reel motors.

No I never got to be around any of that gear except for what I
happened to pick up in a lot of surplus. I used to buy stuff from
government and university surplus auctions. Some of the stuff I
rebuilt and used in my labs or savaged for parts to build custom lab
equipment. I've still got a junk yard in a warehouse I take from.
Saved a lot of money on variac autotransformers. I once bought a bomb
calorimeter and a microwave spectrum analyzer for $5 bucks. Somewhere
in there were some old tape drives but never tried to do anything with
them. I could never have afforded the equipment otherwise.

 

[Paul M]

I have corrected the problem by using an AVR "atmini" processor , i
simply pass the Pot Analog in, process it, pass the results back out to
recreate the analog signal but only conditioned, to correct for whiplash
control.

That's great. AGM makes a similar signal conditioner which I used to
trim a follower system out. I hope it continues to work well for you.