Dancer Systems

How do dancer systems work? This is going to be a bit of learning by doing,
I have spent a vast amount of time searching for material trying to learn about web handling, and
am picking up pieces, but still have a long way to go.

I have a machine that I am in the process of replacing the DC drives with new, and a PLC to interface between.

The existing system, has a dancer 'rate of change' I think. When the axle is enabled, the dancer is lifted to the desired position very quickly and smoothly, consistently.

I want to be able to do the same thing with the new system I install. I intend to swap the 'rate of change' device to a linear transducer so that I know the position of the dancer, and then implement a PID to control the initial dancer position.

Note that this is at zero speed, I need to be able to take up the tension in the web. The dancers are counter-acted pneumatically.

I know, that when the web is moving, I take the dancer position as a trim on the line speed reference. Which takes me onto my next question, speed reference setting.

I asked a similar question a while back on surface speed calculation, if my surface speed reference is 10m/min, my motor rated RPM is 3200, gearing ratio 4, reel circumference 3000mm, then, by my calcs, the axle reference should be 13rpm. I know I have to do a reel diamater calc, which I found in this thread

http://www.plctalk.net/qanda/showthread.php?t=19333&highlight=dancer+control

To do this calc, if I calculate the total length, based on the known values, then I can subtract the distance per second every second, then work backwards to get the decreasing diameter.

The reel calc stuff is a while away yet, what I want to do first is get the new system working where it is stable and in control.

I have seen some mosts mentioning a model, do I / how can I create one for my system?

I am trying to get as much theory and calculations done as practical, I have been in situations before, where it doesn't happen that way, but is normally a good place to start.

In answer to your first question, "How do dancers work", all the dancer systems I have come across use air pressure to set the dancer force which in turn sets the material tension. The control system keeps the dancer in the middle of it's travel and whilst the dancer is not moving, the material tension will be set by the dancer force. The actual material tension will depend on the material path around the dancer.

I haven't come across "a rate of change" only transducer on a dancer - what exactly is this ?

Concerning your modelling, what elements of the system are you intending to model. For example, if you want to feedforward the predicted torque demand to the motor controller you need to know the actual tension demand, the inertia of the roll/core/drive train, the instantaneous acceleration rate and the friction in the roll drive system.

I guess the question wasn't specific, this system is exactly as you mention, the tension against the web is controlled by the air pressure in the cylinders.

I haven't come across "a rate of
change" only transducer on a dancer - what exactly is this ?

Click to expand...

From what I can gather, and the terminology in the documentaion is 'coil plunger', it is basically a coil with a core that changes its position when the dancer moves. Taking readings on this, the voltage was steady, when moved, the voltage changed, dependant on the speed, the faster the movement, the greater the voltage.
This indicates it is some sort of positioner, then when the web starts moving, this is used as a trim with line speed.

So could I use a PID to control the initial dancer position?
How difficult is this type of system to tune? I read somewhere in a post, that the gains should be changed as the web size changes, is this to make the PID less/more responsive? How is this acheived, by calculations?

Once the web starts moving, I need to match the surface speed of the new reel to the line surface speed. Is my method of doing the surface speed reference the best way?

In terms of the modelling, I am not an engineer, so will have to research how to calculate those figures.

We use variable inducter coil here, but we also have a linear resistor on one press to. You set your pound per linear inch using the force on the puematic rams holding the dancer. Then the variable inducter or linear resistor to know the location of the dancer. Feed that info back into your pid to vari roll speed. Then you have to tune the pid to vari the roll speed to make the dancer stay in postion. These number will be a compromess between a light and heavy roll. I have never had to tune one of these loop so I can't anwser you question on hard hard they are to tune, but lean more toward fast reactive. Does you system also use a infeed dancer as well?


Hope that help some

jjsscram

Most of the dancer systems I have used have an LVDT (linear voltage displacement transducer) to determine dancer position. These LVDTs have an in-built circuit that produce a voltage proportional to position. I have never monitored the voltage on the coils of the LVDT - could it be that the processing of the coil signals in your system actually produces a voltage proportional to position ?

We have a VFFS(vertical form, fill, & seal) machine here with a pretty simple setup for tensioning the web. Basically, the unwind shaft for the web is controlled by a DC motor(directly coupled). The drive for the DC motor is controlled by a rheostat that is mounted at the pivot point of the dancer. In other words, the up and down motion of the dancer controls the speed of the unwind motor. As the web is pulled through the machine by the primary pull rollers and the tension increases and the dancer rises, the rheostat at the pivot point increases the unwind speed, and vice versa.

You may already be familiar with this setup, but thought I'd throw it out there. Also, you never really mentioned what kind of machine you're looking to implement this on, so your mechanical setup might be a constraint. I thought maybe the principle might provide some insight for you.

Good Luck,
Jeff

Most of the dancer systems I have used have an LVDT (linear voltage displacement transducer) to determine dancer position. These LVDTs have an in-built circuit that produce a voltage proportional to position.

Click to expand...

I am going to install something very similar, it is basically a linear device, that gives me a signal proportional to position. At maximum stroke, I get 10v, and at minimum stroke, I get 0v.

The biggest problem that I think I am going to have, is tuning this to get the dancer to hold at the SP reasonably quickly. ie take the tension at zero line speed.

Then when the web starts to move ( I am using tacho as line speed reference ), then match surface speed of reel to line speed. The way I am thinking of doing this, is take reel circumference, gear gear ratio, motor nominal rpm, then use these figures to calculate motor reference. Then use the dancer as a trim on top.

I think that matching the accel / decel rates with different reel sizes / weights could be a problem, but then, this is what the dancer is for?

Getting your system stable over the complete operating range of speed/diameters/widths with just the dancer depends on the range of your variables. I have found that one of the key factors is the motor controller speed loop gain. If you have a large change in inertia (say the roll starts at 1200mm diameter and is on a 100mm core) then I would expect you to have to modify the speed loop gain as a function of inertia (High gain with high inertia, low gain with low inertia). You should be able to establish your speed loop gains by running the roll drive only with various diameter rolls installed (rolls taped down) and the dancer trim disabled. You can also check your line speed following by running the machine like this and keeping the diameters fixed (temporarily). Once you are confident that the roll surface speed follows line speed with the dancer disabled, and the speed loop is stable over your diameter range, then enable the dancer and verify the correct operation of the dancer correction (again with the roll taped down). When you have done this, you will be ready to run material. I would start with the diameter fixed and do stop/starts to get your dancer gain correct, then enable the diameter calculation.

Here are my two cents, obviously the dancer controls 2 things, PLI (Tension) and either following roll, speed correction (most common) or leading roll speed cprrection.

As stated before, the aire pressure is set on the cylinders to a specific pressure which will give you PLI (mathmatically figured). This is set and generally does not change, unless the operator wants or needs to run with a different web tension.

As far as line speed, generally there is a master roll that is to run at the desired speed reference. Everthing follows that roll generally from the encoder feedback, not necessarily by speed reference (speed reference is sometimes used as a trim). On everthing that I have seen each roll follows the encoder feedback from the roll preceeding itself, working back towards the Master modified only by any dancers or tension monitoring rolls.

The roll that is after the dancer(back from the master)speed is set by the roll that is ahead of the master + or - the dancer position. In most cases (depending on line speed and web widths), the dancer position only changes the speed of the following roll by a percentage of the speed of the preceeding roll.

Zero Speed or very low speeds (depending on the type of web), frequently rolls are set in torue mode, with the exception of the master. A lot of times at startup the rolls are set to jog until it pulls a specific tension in a particular area and then it is switched to Torque Mode, then once the entire line is holding tension at zero speed the line can start. Either once it starts or at a low speed the drives are switched to speed regulation mode, so the line can ramp up. Even in torque mode, torque is adjusted by the dancer position and tension monitors. Once of the main reasons why you don't leave the rolls in torque mode, is because it is easy to get a run away if there is any slipage between the sheet and roll.

stupidav,
The system that you described sounds like each roll has its own drive. Another method of maintaining tension in this type of system, is to have each preceding roll slightly smaller or larger, depending on what way you look at it.

The system I am updating, is a gravure press, there is only one motor that drives all 5 rolls, linked by a drive shaft if you like.

The UW & RW are controlled by center drive DC Motors, I am replacing the drives as the current ones are past there best, plus the new ones are half the size of the current stack alone.

So with the DC drives, I should be able to get maximum torque at lower speeds.

There is a tacho that gives a line speed reference, when the web is moving, I will use this for my speed reference, which is calculated in PLC based on drive roll size, gear ratio etc.

I then plan to work this back to get the surface speed reference for the roll, again, based on gear ration and reel size.

The dancer will then be fed into the PID, and used as a trim on top of the speed reference.

Just something to think about. You have center drive motors you say. How are you planning to keep the RW speed constant? Typically on our web systems the dancer does not trim the line speed, it's only used on the unwind. If you're using center drive, you'll have twice the work trying to trim an Unwind and Rewind using the same speed reference, as one will be getting smaller while the other is getting larger. As I said, just something to think about.

russmartin

At present I am only relacing the UW drives, so the current setup is controlling the Rewind Speed. The RW also has a Dancer, on startup, the RW starts up, lifts the dancer to app mid position, and then, depending on opposing force at UW end, will hold steady. If no force, the Web creeps through the machine.

In terms of the Dancer trim, my plan was to take a % of line speed, and a % of the PID CV based on dancer position, and add the 2 together to get the reference speed required. The line speed is the dominant factor, app 90 - 95 &, with the PID % being about 5%.

Does mysound like the correct method? Or am I totally off track?

Yes. If the diameter calc is correct, 5% trim should be enough. Where will you be getting the thickness of the material from - measured or entered by the operator or is it fixed becasue it runs the same material all the time ?

There are app 5 different thickness's, and this has quite an effect on the reel size as you can imagine.

At this stage, I was going to attempt to take average, both on thickness, and length.

There is no HMI at the moment, there is plan to put one on at in the near future however.

The other big issue is the length, there is less with thicker material, and vice versa. This could possibly be more obtainable by using an average than the thickness.

To do this calc, if I calculate the total length, based on the known values, then I can subtract the distance per second every second, then work backwards to get the decreasing diameter.

Click to expand...

I posted this in how I plan to do the diameter calc. I think this should work. The big issue that I think that I will have, is maintaining dancer position at zero speed, without hunting occuring, then stable acc / dec rates.

I am starting on this tomorrow, so will see how I go.

You have 3 things to consider with a winder. Line speed reference, Diameter of winding roll (constantly increasing or decreasing), And dancer position. There will be a loop in the controller for each of these. Then all 3 are summed to give the drive an RPM speed reference. The three loops are typically external to the sho-nuff drive, then the drive has it's own speed and current regulators to deal with the final RPM reference.

In my experience, the operators don't like to see the dancer move while winding/unwinding. But it has to. here's why.

In an unwinder, first you need an accurate calculation of the starting diameter. This is done differently in different systems. But it is basically line speed ref / spindle RPM. This value along with line ref will match the surface speed of the unwinding roll to the print cylinder speed. As the roll unwinds, the diameter is constantly recalculated. For the diameter calculation to change, the spindle RPM must change, as the line ref is constant. The only way for the spindle RPM to change is dancer influence.

So you start with a diameter, then as the roll builds down, the RPM becomes too slow, which will pull the dancer, which will increase the spindle RPM, which will decrease the Diameter value, which will increase the spindle RPM, which will allow the dancer to recover to it's center position. If the PI gains in these separate loops are screwey, the dancer wags all over the place.

In the old analog systems, this can be done very smoothly and dancer movement is minimal. In the digital systems of today, the diameter range is traversed in steps. The better the diameter resolution, the smaller the diameter step, which means less dancer movement. But it still has to move to work.

Some of the better drives come with diameter and dancer functions built in, but study the manuals carefully. It's more of an art than a science to get a winder right. it's very easy to cause undue tension upsets in the web, and, as you know, in gravure printing, tension is everything.

Simon is right. The dancer should trim the total speed ref by about 5%, depending on the material you are winding--more if you're winding a rubber band, less if you're winding steel.

Also, dancers supply a variable signal to the controller that represents position. The type of transmitter you use depends on your system. But try to stay away from pots--they eventually wear out and cause escalating problems. I have used LVDT's and there is a device called a CVDT--Circular variable differential transformer--that is typically easier to mechanically fit to the application--and they last forever.