I am looking at converting some winders over which are presently winding a
>string like product which runs at a constant line speed. I need to maintain
>a constant tension as the roll builds, but I want to do it without a
>feedback. I am presently doing it now with a DC drive which has a torque
>taper kit installed but I am really interested in AC drives low maintenance.
>As most would speculate the winder shaft must slow down as the roll builds
>and the torque must increase at the same time to maintain this constant
>tension. I am thinking that possibly there is an AC inverter designed for
>this application or a controller which could monitor the tension and
>feedback to drive using the motor current. Remember the whole idea here is
>to illiminate the DC drive and als the dancer. I know that it is possible to
>use say a load cell to feedback but this gets into the big bucks.I am trying
>to keep the costs down if possible. Any suggestions will be appreciated.
Why isn't tension directly proportional to torque, regardless of speed?
I have some doubts this is practical, unless the filament is quite strong, and the tension can be allowed to vary over a considerable range. The dancer is not only a feedback mechanism, but also acts as a buffer between the huge mass of the spool, and the constant velocity of the filament paying out from where it is produced. If you connect the spool to the source with no buffer in between, irregularities in the winding on the spool require abrupt changes in rotation speed of the spool to avoid wild fluctuations in the filament tension. If the filament comes out at less than, say, 1 inch/second, this may be quite doable. If it is coming out at hundreds of inches per second, it is very unlikely that you can make this work without a dancer. You could probably make it work with a dancer that doesn't feed back any information, letting the drive maintain tension as you suggest. A VFD should be able to do most of this, with some simple op-amp circuits to provide a decreasing speed command as the spool fills. but, that doesn't really sense, or control torque. Feeding back a dancer position would work a lot better at controlling the tension. A VFD and an AC induction motor should be fine as a low maintenance drive system.
The product is very strong and will run up to 300 feet per minute, thus the dancer does work well, however the winders we use here we would like to get rid of the dancers(they are troublesome and the pots are always a source of failure). As I had mentioned we do use a standard drive with a torque taper circuit card which works well, it basically tapers torque up as the speed of the roll decreases in a preset manor. I have been told that there is manufacturers using open loop VFD's for winding product such as ours. I do understand where you are coming at with the type of product and speed at which it runs. Would it be feasable to mount a roller on a load cell to which the netting rides over before it winds onto the roll, this load cell would feedback to a temperature/process controller such as an Omron E5AK where the load cell output could be dealt with through a PID loop then fed to the drive thus making the controlability easier. The controller would then be used to simply set the tension.
Seems to me that if your are open to VFDs, the cost of a load cells should not be too high. Definitely pots are troublesome and to be avoided. Tell me, how big is this application? I.e. 0.1 HP or 10,000? If the entire rig, with or without dancers, is mounted on a base that is capable of sliding ever so slightly and is restrained by a load cell with a 4-20mA output, a simple PID controller would be able to handle the job. It would have to be a fast analog job. No DCS. This will be a fast loop.
Is it the dancers, per se, that are a problem or is it the associated pots?
Walter thanks for the insight, you got me thinking even more here, when you were asking about the package I was trying figure what you were getting at, the winder indeed is a small package, 3/4 hp, I was thinking the product had to go over the load cell but having the winder rest against the load cell say on linear bearings would be a choice way of setting it up. When you say "no DCS" what does this mean, I am unfamiliar with that term. Also anyone know of some source of cheap load cells, again I am sorry for the term of cheap or cost being kept down, I used to work for my brother and we used load cells which were thousands of dollars.I am hoping that since the application is not that critical we could possibly use a chassis mount VFD like a KB unit and use a 1/16 DIN Omron controller and possibly a 200-400 dollar load cell, hook them all together and tune controller, then put it online and fine tune the loop. We were thinking of possibly using the accel/deccel to either quicken the response time or to nullify it, sort of like a filter.The dancers are just problematic in that the wheels we use tend to roll the product and cause a twist if the winder is not lined up properly, if the bearings in the wheels tighten the rolls will torpedo, if the weights on the dancer arm move again we torpedo the pots get wet sometimes or just fail, we started to use a 45 degree pot used for dancer applications with no stops which helped but over the years I have learned to hate dancer arms.
"Walter Driedger" wrote in message news:dX6Na.354611$ snipped-for-privacy@news2.calgary.shaw.ca...
A DCS (distributed control system) is a very large control system used in the process industries. It is used to control big things and has a maximum scan rate of about 1/4 second. I'm not familiar with the Omron controller but it would probably be OK. I would guess a 1/10 second scan rate would do.
If I may suggest, why don't you get rid of the pots alltogether by means of optics? For my degree I had to do a thesis whereby you unwind a fibre optic cable from one drum unto another under constant pressure - both drums had to reel under its own power as the fibre is too weak to take strain.
What I eventually did was to make a card with a grey code punched in it. This card was attached to the fibre via a grooved wheel running on the fibre and the card itself moved loosely through a slot. In the slot I had LEDs (4 LEDs as I used a 4-bit configuration) on the one side and obviously 4 light sensing transistors on the other side. The 4 light sensing transistor outputs went to the control circuit which controlled the unwinding motor - winding motor was assigned as a constant in order to minimise the variables. As the card was weighted with lead the whole aim of the exercise was to keep the angle of the fibre constant - it is directly proportional to the force exerted on the said fibre. Thus, from the one drum to the other the fibre formed a V with the weighted card and grooved wheel hanging at the lowest point of the V.
What was proofed with the exercise is that both motor types had to be similar as far as possible - start-up was a bitch as it had to be done slowly and different motor types have different start-up curves.
Unfortunately I eventually discarded this thesis information as I am employed in the South African mining industry and there is no need for it. Please remember me when you patent this idea and make lots of money.
We have a similar application. We wind expanded polystyrene foam as it is extruded. The foam sheet is about 1/8" thick and 30 inches wide. A "pull roll" pulls the material from the extruder and presents it to a winder, where rolls of about 48" in diameter are created. We try to maintain a constant tension on our web as it is wound. About 1 lb/linear inch or 30 lbs total. In the past we have used an ac motor and Safetronic drive with a winder option added to the drive. The first winder we implemented this system on, did not perform well. I think this was due in large part to the fact that 4 different people worked on this winder control system without a common understanding of the underlying design. At installation they ended up disabling the optional winding functionality and set it up like our prior winders which function "so so". The result was that management decided that we needed to add some feedback to the system. It will probably never be known if the system could have worked as designed.
For what it is worth I'll lay the concept out for you and would be interested in any comments on it. We used two pots. One to set the line speed, which was constant for a particular run and the other to fine tune the tension. If the linear line speed and winding motor rpm are known then the roll diameter can be calculated. The roll diameter will determine the required torque to maintain a specific web tension. With the proper scaling one pot acted at an indication of line speed. The operator would have to determine the appropriate setting for a specific product run. This pot effectively replaced a tach feedback from the upstream pull roll.
The second pot scaled the torque further so different sheet widths, thicknesses and types could be accomodated. The drive ran in torque mode so the winder was not trying to match speed. The speed reference pot setting did not have to be perfect. Both pots were not controling speed, only torque values. It the pots were a little off the tension would be a little off. Fortunately precise control of tension is not required for our application.
When I return to work I can send you the drive p/n. Good luck.
D.Miller
n news:be175q$nfg$ snipped-for-privacy@news.eusc.> I am looking at converting some winders over which are presently
Depends on tension and the mass of the winder but a linear bearing which needs to have insignificant stiction and friction after years of service and crud build up sounds like more of a maintenance problem than the dancer and pot you were trying to avoid.
Load cells are not cheap and they are fragile. They need mechanical overload stops which are impractical for such small deflections so you need some kind of sprung coupling or mounting system which collapses on overload.
If you know the material speed (there must be something in the system you can pick that up from, or perhaps it is constant) then you can deduce the winder diameter by measuring its speed.
If you can find an ac drive which provides that function then cool. Otherwise I would have thought a small PLC which takes in the speeds as analog or pulse rate and provides a torque request to the drive as analog or perhaps via a data link would be possible. I presume it takes a while to wind so you don't have to perform the calculations that frequently.
Now that NOSPAM has mentioned the problems with linear bearings a thought came to mind, why not just rubber mount the winder with the load cell directly bolted to the base frame and the winder head, since the load cell needs almost no deflection. Also the whole reasoning for avoiding the dancer is not only the high maintenance aspect but also the problems we have with our product going around the wheels and tracking.The winder drive which I use(Have 1 unit) works on torque taper, it works well but is limited on build up ratio, once you go beyond this ratio you are not controlling properly. I like this idea of load cell to controller to drive, controllers nowadays are very high tech with fuzzy logic and all and are not too expensive either. Just that damn load cell and conditioner is where I need to shave a little off of.
We have a measurement of rotational speed. We are talking about measuring linear tension. There is, at present, not measurement of linear speed. I'm not quite sure how to get it. The other posts arguing about whether the formula below is correct miss the point that we have, as yet, no measurement of linear speed.
He was just pointing it out and it is intertesting.
Work done (energy) is force applied * distance moved.
If you keep the rate of change of distance (speed) constant and you keep the force (tension) constant then the rate of change of energy (power) will also be constant.
Whatever system you choose to control tension if it is working properly the winder will be fed with constant power.
So measuring the power input to the winder is another possible feedback signal to servo from.
I doubt the drive and motor efficiency will be constant so measuring the drive electrical input power will not be accurate. It might be close enough ever tried measuring the input power on your existing systems?
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.