Tensioner Idea

[R.G.]

Beware sprung mass and intertia (linear and rotational).

The bobbin is quite oblong, so the velocity of the wire must vary periodically and by a factor if the bobbin rpm is more or less constant.

If thin wire is pulled by the bobbin over an overly massive pulley or pinch-roller assembly, the wire will break because it isn't strong enough to accelerate the sprung weight (and rotational inertia).

Also, once started moving, the pulley et al will tend to keep going, and so will overrun the bobbin, causing loose turns.

Always good advice. The long, springy "bobber" on most tensioners is intended to provide a high-compliance end for the variation in wire speed, I think. What that does is to allow the tension to vary a lot around an average tension as the speed changes.

The speed variation and resulting tension variation is one of the issues I've been pondering. One of my motivations for using a stepper based coil spinner in my coil winder idea was to let me do things like pre-distorting the winding speed of the coil twice a rotation to partially linearize the wire speed. Using an active drive on a rotation-by-rotation basis would let you do that. I haven't done any sims yet, but I believe that the wire's linear speed through a traverse is probably similar to a mildly distorted full wave rectified sine wave. Changing the coil rotation to predistort some of that out should be feasible and should get you much closer to fixed wire speed.

There is probably a mechanical linkage that will do that, perhaps a variation of the Geneva mechanism. That's another I haven't worked out yet. Programming a stepper to undistort a random waveform is easier than carving a set of metal cams.

If the variation in coil rotational velocity is close enough, tensioners become trivial, and very little tension compliance is needed.

I've actually been sidetracked by tensioner idea number ... three, was it?
Two nylon door-track wheels from Lowes home improvements. These come two to a pack, with included ball bearings, for $5. The bearings fit a T nut in a plywood substrate. The bolt is just an axle. The two rollers are glued together so they rotate as a unit.

A loop of magnet wire goes around the outer one. A loop of cord with a felt shroud goes around the inner one. The cord is pulled on by a $4 surplus solenoid run by a linear current drive, which is driven by a servo amp (TDA2030, $3") fed by a tension sensor (music wire, roller or glass bead and bend-y arm). Cuts the mass and rotational inertia down, easier to thread, and is cheaper and easier to machine.

We'll see. I bought parts on the way home.

Also beware mechanical resonances. The wire pulsates twice per turn, so at 1200 rpm, the pulsation frequency is (1200/60)(2)= 40 Hz. If the mass-spring resonant frequency of any part of the tensioner is not well above this frequency, when the resonance is hit, the resonant part will flail wildly, and may physically break off. Some mechanical damping is also helpful in controlling resonances.

In general, keep things very light, so the wire is able to dominate.

Yeah, good advice again.

I have some math to do on the predistortion of the coil winding. That will make this easier.

[R.G.]

I found the math, I think. The kinematics are almost those for a half-cycle of a piston-rod-crank in an engine. The only difference is that the "rod" changes in length during the "stroke" or half-rotation and the wire velocity never goes negative, but restarts a new half-cycle.

It's a distorted sine, faster acceleration up front, flattened to the rear, but starting at zero and ending at zero.

The flat side of a pickup must cause a zero-velocity point in the wire at the instant that the wire lies flat and before the new end starts pulling wire.

Real pickups aren't lines, so the math isn't perfect, but it does tell us a few things.

1. The peak velocity is at about 70 degrees after a new turn starts.
2. The peak velocity depends on the length from the bobbin edge to the last support before the wire is free. The further away the support is, the less peaky the velocity is. The closer, the more peaky. For a 3" bobbin width and a 2" spacing of feed to the bobbin edge, the peak is about 1.4-1.7 times the average velocity.

Predistorting the bobbin rotational velocity may help some, but it won't do it all with those zero-velocity points twice per turn.

[Joe Gwinn]

Always good advice. The long, springy "bobber" on most tensioners is intended to provide a high-compliance end for the variation in wire speed, I think. What that does is to allow the tension to vary a lot around an average tension as the speed changes.

The "bobber" is actually called a "dancer".

The speed variation and resulting tension variation is one of the issues I've been pondering. One of my motivations for using a stepper based coil spinner in my coil winder idea was to let me do things like pre-distorting the winding speed of the coil twice a rotation to partially linearize the wire speed. Using an active drive on a rotation-by-rotation basis would let you do that. I haven't done any sims yet, but I believe that the wire's linear speed through a traverse is probably similar to a mildly distorted full wave rectified sine wave. Changing the coil rotation to predistort some of that out should be feasible and should get you much closer to fixed wire speed.

There is probably a mechanical linkage that will do that, perhaps a variation of the Geneva mechanism. That's another I haven't worked out yet. Programming a stepper to undistort a random waveform is easier than carving a set of metal cams.

If the variation in coil rotational velocity is close enough, tensioners become trivial, and very little tension compliance is needed.

Way too complicated. Nor is cyclic acceleration and deacceleration of the bobbin going to yield nice windings. Too much flailing.

I've actually been sidetracked by tensioner idea number ... three, was it? Two nylon door-track wheels from Lowes home improvements. These come two to a pack, with included ball bearings, for $5. The bearings fit a T nut in a plywood substrate. The bolt is just an axle. The two rollers are glued together so they rotate as a unit.

A loop of magnet wire goes around the outer one. A loop of cord with a felt shroud goes around the inner one. The cord is pulled on by a $4 surplus solenoid run by a linear current drive, which is driven by a servo amp (TDA2030, $3") fed by a tension sensor (music wire, roller or glass bead and bend-y arm). Cuts the mass and rotational inertia down, easier to thread, and is cheaper and easier to machine.

The fear is that the rollers weigh too much.

I would suggest reading some patents on tensioners. There were a number of threads on the old AMPAGE forum in 2003 or so. Look for "tensioner", "winding speed", and the names of the manufacturers.

Also, 3,707,269; 4,912,274; and 4,526,329.

[R.G.]

I took a look.

I have reinvented bits of a couple of those patents, but with a different electronic servo on the braking spool. I didn't see any optical servo mechanisms, which I consider to be "known to one skilled in the art" and the mass of the spools is a detail of the building.

I'll see how the thing works.

[mbetter]

I'm usually not one to revive a zombie thread but there were some ideas here that got me thinking. Has anyone ever investigated the possibilities of using a piezo actuator to control tension?

I know next to nothing about the converse piezoelectric effect (in fact I just got those words from Wikipedia) but it seems to me that the idea holds in principle: you have an eyelet or wire guide nozzle with a diameter such to just fit the wire, a felt pad and a piezo crystal. Current is applied to the crystal, crystal expands, pad is forced against wire, voila, tension.

I'm probably missing something here, does anybody have any thoughts?

Thanks,

Mike

[Joe Gwinn]

I'm usually not one to revive a zombie thread but there were some ideas here that got me thinking. Has anyone ever investigated the possibilities of using a piezo actuator to control tension?

I know next to nothing about the converse piezoelectric effect (in fact I just got those words from Wikipedia) but it seems to me that the idea holds in principle: you have an eyelet or wire guide nozzle with a diameter such to just fit the wire, a felt pad and a piezo crystal. Current is applied to the crystal, crystal expands, pad is forced against wire, voila, tension.

I'm probably missing something here, does anybody have any thoughts?

It may be difficult to get enough travel from a piezo actuator.

[David King]

I have a good friend who spent a summer building micro-manipulators for electron microscopes using piezo "motors". You can stack them up to increase travel. They need relatively high DC voltages 100-400V to get them moving. I think it would be easier to work with a solenoid or linear motor that can be controlled at much lower voltages. You really shouldn't need the kind of resolution that piezos give you and the price isn't exactly cheap.

I still think a friction wheel that the wire goes around one turn, that's directly driving a small DC motor which is, in turn, dynamically braked with a variable resistor circuit via feedback loop, would be the most straightforward approach. It's main virtue is that the friction is in the motor, not on the delicate insulation of our magnet wire.

One other concept I've been thinking about is to use an oblong, bobbin shaped, friction wheel that's geared to the main winding shaft so that it pulls the wire pre-tension and equalizes the uneven pull of the bobbin being wound. Hard to visualize how this would work exactly but I'm sure there's a patent that covers it somewhere. -Basically we want the wire coming off our spool at a constant velocity so we can tension it steadily to minimize the changes in friction that would otherwise happen (generally friction is highest in the initial, still state, once the wire is moving the coefficient of friction drops and stays relatively constant). We can hopefully eliminate the dancer mechanism and use it merely as a tension gauge and a part of our feedback loop for controlling the tension.

Sorry I can't write this out any more clearly, my brain insists on imposing obfuscation at every turn.

[David Schwab]

That's an interesting idea.

I keep forgetting I have one of those Lego Mindstorm kits here that I've been wanting to mess with to see if I can make a programable traverse. You can build a scanner with it, so it should work.

[Bruce Johnson]

I'll throw a few other bits into the thought pile on wire tensioning mechanisms.

Remember that this wire we're working with is quite fragile. Copper will work harden and snap from being bent sharply or repeatedly. And the insulation on the wire is effectively a few ten-thousandths of paint.

When I was roughing out the design of my winder, one of the principles that I followed was that, from spool to bobbin:

1.) The wire should be bent as few times as possible
2.) Any bends should be as large a radius as is practical
3.) Everything should be rolling, not rubbing, with the one exception of the friction point that creates the tension.

To my thinking, any time that you bend the wire, you're increasing the risk of cracking the insulation, stripping off chunks of insulation, or breaking the wire. The tighter the radius that it bends over, the higher the risk.

So, if you're making up a mechanism to precisely control the tension, in order to get a high level of consistency, make sure that the mechanism itself isn't microscopically damaging the wire and causing other problems and inconsistencies.

On my machine, I kept it dead simple. The wire comes off the end of the spool and goes straight through a friction tensioner that's in line with the axis of the spool. Then it wraps about 160 degrees around a 1 1/2" dia Delrin pulley that spins with very low friction and inertia, and right on to the bobbin. For my tensioner, I used small pieces of D-shaped silicone rubber weatherstripping seal material. It's smooth and slippery, and soft enough that I get good resolution in adjusting the tension. It seems tough enough, too. In two years' worth of winding, the wire hasn't cut any noticeable groove into it.

[mbetter]

Has anyone tried using a cam geared to the main drive shaft to modulate the force applied to a stationary friction (felt or rubber?) tensioner?

[voodoochild]

I use a CNC for winding. I tried a couple of tensioners and they applied way too much tension for my liking. I had to back them off completely or they would wind too tight. I went to felt after that and it worked great but it was super hard to keep the tension consistent. Finally I came up with a method using several (4 or 5 depending on the pickup) 1/2'' nylon dowels staggered in a row. I moved them closer, further apart, up and down, etc until I got the right tension. Basically the wire weaves through the dowels and they create drag or tension. More or less depending on the position and number of dowels. Then the wire feeds through felt just to keep tension on the wire when the machine stops and starts. This is most simple and accurate tensioning device I have ever used. I don't have to worry about it wearing out, malfunctioning, etc.

[David King]

That seems very functional and the simplicity is worth a lot. I can see how you could decrease tension through the wind by slowly moving the posts apart from start to finish with a series of scissor/pantograph links between the posts and a leadscrew that pulled them closer or further apart.

[voodoochild]

That seems very functional and the simplicity is worth a lot. I can see how you could decrease tension through the wind by slowly moving the posts apart from start to finish with a series of scissor/pantograph links between the posts and a leadscrew that pulled them closer or further apart.

It definately works for me and I truly believe that simple is best. It's funny how with just about everything, I started out simple, then I worked to get as complex as possible, now I'm back to being simple again. I don't have any issues with it breaking down the insulation or anything like that.

[Achiles]

Tom Anderson uses a similar tensioner ,
i saw the tensioner when i visited Ton (1999).
2 rows of metal dowels .The wire goes in a zig-zag way.
You control the tension by contriling the distance of the rows.

[chevalij]

That's a standard type of tensioner used in places that work with yarn, cord etc.