Winding technology.

knutabru · Aug 22, 2010

Hello.
I'm about to take on a project regarding winding up a thin thread on a drum in a single layer. So far it sounds simple...

But...

The thread does not like beeing touched, so the drum will have a square shape so the thread are thouched only 4 times per revolution-lenght.
The thread can be thight but does not tolerate beeing stretched. So the drum must vary the winding-speed so the perimeter of the square wheel has a constant speed.
The extruder that delivers the thread, will give me the "line-speed".

So far I have planned this:

HW:

S7-315T-2DP and Sinamics S120 (had a course in this some years ago)
Use servomotors with gears for the drum and traversing axis (the winding will be very slow, aprox. 1 rev./min)
Only use the encoders on the servomotors for the winding application. (keeping cost and complexity down)
Recomend the mechanics of the traversing-axis to consist of a ball-screw.

SW:

Use the "line-speed" from the extruder to control a virual axis as a master.
Set up a cam-connection from the virtual axis to the Drum-axis. The cam having a sine-function that repeats 4 times every revulotion, hopefully giving a constant perimeter-speed.
Set up a gear-connection from the Drum-axis to the traversing axis that lays the thread on the drum.

Comments so far are welcome.

From this set-up I have some questions that I hope you gurus can guide me with.

Should the virtual axis be set up as a linear or rotary axis, with/Without modulo?
The drum, same apply here, should that be set up as a linear or rotary axis, with/Without modulo?
As I understand, the cam must be synchronized so the sine-function corresponds with the cornes of the square. I believe this must be done with a ind. sensor looking at a flag on the drum. Will it be neccesary to sync at every revelution?

Is there something I have overlooked? (exept my spelling)

Best regards from Knut.

Peter Nachtwey · Aug 22, 2010

knutabru said:
[*]Set up a cam-connection from the virtual axis to the Drum-axis. The cam having a sine-function that repeats 4 times every revulotion, hopefully giving a constant perimeter-speed.

I think you have the right goal. I doubt a sine function will achieve it. Are you going to take into account the angle from horizontal of the thread?

[*]Should the virtual axis be set up as a linear or rotary axis, with/Without modulo?

I always set this up modulo so the virtual master index into the cam table repeats.

[*]The drum, same apply here, should that be set up as a linear or rotary axis, with/Without modulo?

Modulo.

[*]As I understand, the cam must be synchronized so the sine-function corresponds with the cornes of the square. I believe this must be done with a ind. sensor looking at a flag on the drum. Will it be neccesary to sync at every revelution?

I would use a 17 bit SSI absolute encoder mounted at the drum. In know Danaher makes one. No homing is required.

If you do use a quadrature encoder then you should only need home it once until power is off unless you use gears. If you don't get an integer number of counts per revolution you probably should reset the count every turn at least.
What I don't like about the flag idea is that there is always jitter.

DickDV · Aug 22, 2010

It seems to me a lot simpler to have a dancer running on the thread and control to deflection which would represent thread tension. There would not need to be much deflection of the thread at these speeds to detect tension.

You could even use the extruder speed as the line speed with the dancer being a trim on the line speed.

Not being a servo guy, maybe that is what modulo means in "servo-speak".

Peter Nachtwey · Aug 22, 2010

DickDV said:
It seems to me a lot simpler to have a dancer running on the thread and control to deflection which would represent thread tension.

Dancers react rather than predict. The control should be cam based but a dancer wouldn't hurt to help trim the control signal if necessary.

You could even use the extruder speed as the line speed with the dancer being a trim on the line speed.

That would be a good plan. Now what is the formula for the winder speed as a function of extruder speed? I love problems like this.

Not being a servo guy, maybe that is what modulo means in "servo-speak".

Modulo is not a servo term but a math term. 5 mod 3 = 2. If the system has an integer number of counts per revolution it is handy if the controller automatically resets the the count to 0 when the count reaches the unwind count or the counts per revolution. If the system has gears there usually isn't an integer number of counts per revolution and a hardware reset is required. In some cases there is a gear ratio of 13 to 11 and the system can unwind after 11 turns. The problem is that the cam table must be 11 revolutions long.

DickDV · Aug 23, 2010

Peter, the OP stated that the speed was only 1 rpm and only single layer. Under these conditions, wouldn't reactive control be adequate?

It seems to me that the dynamic aspects of this application are extremely low. But, then again, I usually avoid winder applications since I have a very good winder guy in my territory that I can refer these kinds of jobs to.

Manglemender · Aug 23, 2010

Peter Nachtwey said:
Dancers react rather than predict.

Whilst dancers are, by there very nature, "reactive" In most widing applications the dancer is only used as a trim for the predicted values. In this instance it sounds like a dancer might be a good idea to control tension in the thread with a the position being used by a PI controller to trim linespeed.

I don't think that Siemens are pushing the 317-T controller any more; you may wish to considder a D425 or a D435.

Winder speed% is: Linespeed% * Synchronous Diameter/Actual Diameter

Where Synchronous Diameter is the smallest diameter at which full speed can be achieved i.e. Linespeed = 100% and Winderspeed = 100%

Nick

Peter Nachtwey · Aug 23, 2010

DickDV said:
Peter, the OP stated that the speed was only 1 rpm and only single layer. Under these conditions, wouldn't reactive control be adequate?

Yes and I bet knutabru thought of using the dancer too. knutabru said the thread doesn't like to be touched. Then math is easy. At those speeds cam tables aren't required, the position, velocity and acceleration can be calculated on the fly.

knutabru · Aug 23, 2010

Manglemender said:
I don't think that Siemens are pushing the 317-T controller any more; you may wish to considder a D425 or a D435.

Winder speed% is: Linespeed% * Synchronous Diameter/Actual Diameter

Where Synchronous Diameter is the smallest diameter at which full speed can be achieved i.e. Linespeed = 100% and Winderspeed = 100%

Nick

The 315T and 317T are well alive and not planned for discontinuation yet. But sure, the Simotion D would also be a good choice for this since it integrates with the Sinamics S120. -If I only had the time to learn to use it...

Calculating the speed on the fly might be a good idea. I'll try that tomorrow in OB65.

Knut

knutabru · Aug 23, 2010

Peter Nachtwey said:
Yes and I bet knutabru thought of using the dancer too. knutabru said the thread doesn't like to be touched. Then math is easy. At those speeds cam tables aren't required, the position, velocity and acceleration can be calculated on the fly.

I have talked about using a dancer with the customer, but that sollution would not benefit the thread, as you mention. But as suggested, calculating the speed, on the modulo axis, would be a lot easier. -Perhaps syncing the axis at every revelotuion using a dedicated sync-input on the sinamics to avoid any chance of drifting in the modulo position.

Having used most of the day, only producing Cams that made the axis jump and shake, it seems more promising to just calculate the speed as suggested since the winding-speed are so slow.
-I'll try that.

Knut

L D[AR2P#0.0] · Aug 23, 2010

Am I missing something here ? - the extruder is the master so a Virtual master is not applicable. How will the extruder feed you the line speed ?

knutabru · Aug 23, 2010

L D[AR2 said:
Am I missing something here ? - the extruder is the master so a Virtual master is not applicable. How will the extruder feed you the line speed ?

I have not explained that detail good enough. The extruder are controlled by a PLC that, I was told, will give out a feed-speed for the winder to use as a line-speed. The extruder are a seperate machine delivered from a different company and consists, among others, of a small pump that deleivers the fluid that hardens and becomes the thread.

It was that speed I originally planned to us as a speedsetpoint for a virtual master in the winder.

Knut

shooter · Aug 24, 2010

a dancer can not be used, however a vacuum chamber is still possible.
look at an old taperecorder for the computer PDP they did have this.
you can change the tension by more or less vacuum.
You can also use a two, three sprong wheel.
look at how spaghetti is made.
It will be very hard to control the speed as you will need to speed up every corner, think of it as a sprocketchain wheel.
I would solve this by using a mechanical solution rather then rely on electronics.

knutabru · Oct 10, 2010

Hello again.

The project planning evolves gradually, the machine are now taking shape. I have required an extra encoder on the output wheel from the extruder that delivers the thread, so I can get an accurate line-speed to feed the winding machine.

As mentioned earlier by Mr. Nachtwety; a dancer would also help, I have started to look for a non-contact measuring of the thread, but have not been successful so far.

Does anybody know about an analogue sensor that can tell me for example the position of a 1 mm thick thread inside a frame? Either optically or with capacitive measuring.

Knut.

Peter Nachtwey · Oct 10, 2010

Wow, this is taking time.

How much tension can the thread withstand?

If you have a way of measuring torque at the motor then you can gear the winder to the lines speed with a slightly more than one to one ratio. Then you enable a torque limit so the winder doesn't pull too tightly.

leitmotif · Oct 10, 2010

knutabru said:
Hello.
I'm about to take on a project regarding winding up a thin thread on a drum in a single layer. So far it sounds simple...

But...

The thread does not like beeing touched, so the drum will have a square shape so the thread are thouched only 4 times per revolution-lenght.

The thread can be thight but does not tolerate beeing stretched. So the drum must vary the winding-speed so the perimeter of the square wheel has a constant speed.

The extruder that delivers the thread, will give me the "line-speed"..

QUESTIONS
1. You are calling out for single wrap on the take up drum.
OK now that you have a single wrap on it what do you do with it then? Is this more of a capstan where as you put it on you take if off?
2. I sort of comprehend why a square drum to avoid contact as much as possible. However I think you are going to have control issues with keeping constant peripheral speed on drum to match input string speed and prevent slacking or stretching on string. Of course this all depends on the drum speed at 1 RPM maybe not a problem but with more RPM I think problem will defintely increase.
3. What are dimensions of the string?
Dan Bentler

Winding technology.

knutabru

Member

Peter Nachtwey

Member

DickDV

Member

Peter Nachtwey

Member

DickDV

Member

Manglemender

Member

Peter Nachtwey

Member

knutabru

Member

knutabru

Member

L D[AR2P#0.0]

Lifetime Supporting Member

knutabru

Member

shooter

Member

knutabru

Member

Peter Nachtwey

Member

leitmotif

Member

Similar Topics