Coil Winder

[Gus]

Chris

How many amps max is the motor at full working voltage? Motor looks like the kind used in old data tape machines. Is that a tach on the other end?

Once you get the unit working I think a Nice DIY PWM control might be something to try. You could have current limiting and much better torque control at low RPMs you could even install a resistor brake.

I do like the light bulb setup nice and simple.

A 555, maybe a gate drive circuit and some power mosfets could be the start. Some small microcontrollers have a PWM out pin or two IIRC and could also handle the stepper drive wire transverse and the counting of the turns etc.



Gus

 

[CJ]

Thanks for the ideas Gus!
Could ellminate some complicated control hassles.

I made my first mandrel today, holds a 75 EI bobbin. Bought some 3/8 " drill stock, hacked me off a chunk, tapped a # 8/32 in the end for a thumbwheel, machined some plastic to hold the bobbin, and here ya go:

http://vacuumbrain.com/The_Lab/Winder/neve_mandrel.jpg

The round stock had to be pressed into the plastic with a vice, so I do not think it will slip. I can build a mandrel for all the popular bobbins and make setups pretty quick. I wanted the large flat side on the plastic so I would have somewhere to tape the loose leads while winding.


http://vacuumbrain.com/The_Lab/Winder/m75.jpg

 

[CJ]

Winders grab more tape than anybody on earth. Thats what they do all day, lay down insulation and tape it up. That's all I used to hear, a machine stopping and the sound of a Scotch Model 96 tape machine getting tugged on, a zillion times a day. So having a good tape machine is very important, especially when you need one hand to hold the wire or the insulation.

But, the #96 now lists for $273 !!! Sure, it's a nice definite length tape dispenser, and anybody who's anybody in the tranny biz has a ton of them sitting around with tape scraps all over them. I did find one as a "buyitnow" on ebay for 49 bucks, but I don't go there no mo.

Soooooo, I figured it was DIY time. Here is a wheel being done the "too cheap and lazy to run out for a hole saw and arbor" method. That 3/8 " drill rod is really coming in handy.

http://vacuumbrain.com/The_Lab/Winder/tape_1.jpg

Here are the parts ready to screw and glue. A roll of poly tape seen on the newly fabbed wheel.

http://vacuumbrain.com/The_Lab/Winder/tape_2.jpg

Here is the DIY tape dispenser sans cutoff blade, which will be liberated from some poor unsuspecting office geek while they are on their 3 martini lunch break.
I put a couple of dividrs in there so I can run three rolls of tape off of one dispenser. Probably 1/4 and 3/8 Permecel, and some 3/8 glass cloth.
How many winders have a tape machine that matches the winder also? :wink:

http://vacuumbrain.com/The_Lab/Winder/tape_4.jpg



So, 2 and 1/2 hours to DIY the equivalent of three Scotch #96 tape dispensers at 273 ea. plus tax and shipping (7 lbs a piece) equals about 840 dollars savings. My banker will like that.

So since we love our DIY at 840 a day, we have $5,880 a week, $23,520 a month, or $282,240 a year, or $2,282,240 every ten yeras and I am getting tweaked so going home for a load of the sticky icky, then some Spyro Gyra, then Letterman. I'm out!

cj

 

[CJ]

Still in the info gathering stage as far as working out the wire traverse details.

I thought I would pass along this link I found which is very cool for anybody messing with motors and stuff:

http://www.compumotor.com/catalog%5Feng%5Fref.htm

cj

 

[Adadan]

That's a great link for stepper info. If you want to look at the stepper stuff I did in school with cheap logic chips, let me know. I used a hybrid stepper, with the simplest drive--full step, two phases on, with a fairly fine stepping motor, which gave me plenty of resolution for my task (didn't need to get into mico-stepping, thankfully). I think I know roughly which box those notes are in. . .

FWIW, in my experience the microcontroller approach is much more flexible/versatile in the long run, but more time consuming and expensive initially. Probably worth the extra investments if you want the ability to easily reconfigure stuff later.

On the other hand, if you want quick & dirty, I think I did all the paper design and breadboarded the test stepper driver for my school project in one evening.

 

[CJ]

Holy Cow, look at all the forums on this CNC geek hangout site:

http://www.cnczone.com/forums/

 

[CJ]

Wow. I was handed a link to a wire tensioner. Far from a piece of felt in between a thumbwheel, this guy adjusts the tension depending on the winding rate. I always wonderd if tensioner neededto be tweaked durring startup and regular winding speed, could help keep things unifiorm when dealing with ultra light gauge wire.

http://www.azonicproducts.com/stepper.htm

 

[CJ]

OK, any math geeks out there?

Since 99 percent of all audio transformers are wound on a square or rectangular bobbin, I am trying to figure out what's going with the wire as far as linear speed is concerned. I am looking at a square bobbin for now so I don't get any brain bubbles.

As the corner of the bobbin comes up during rotation, the wire speed and location changes. This can make life interesting when designing a wire tensioner. If you feel the wire between your fingers while winding a bobbin, you can feel this oscilation. Most tensioners handle this with a "dancer", which is kind of like a low mass spring loaded tone arm type thingy that sits out above the tensioner..This acts kind of like a shock absorber . It's purpose is to try and keep a constant tension on the wire, even though all heck is breaking loose, especially at high RPM's. But this arm will have mass, which is bad, because this mass can break fine wire when really spinning the mandrel.

A square bobbin, at high speeds, can actually hammer the copper wire and distort it, which can weaken it and cause a possible failure down the line. You can really hear this slap when someone is winding a big rectangular coil, say 1 by 2 feet. A little slap is ok, as it tells you that a nice tight wind is being done. But too much slap can be bad, not only by harming the wire, but by giving the copper too much upward momentum, which actually can lift the wire off the bobbin momentarily. This usually manifests itself as a bowed out winding, which means less fill. Every coil has a little bow in it, but too much is bad. A slow winding rate is one answer, but for #50 wire....it takes a long time to wind a coil.

Anyway, I came up with the weird idea of changing the winding speed as the bobbin rotates, thus keeping a constant tension on the wire without a fancy tension arm. This could be done by writing a program for a stepper motor, but I need an equation first. I am not sure how to approach this math, either vectors or polar coordinates seem the most obvious.

So if anybody can give me an idea or two getting an equation, that would be cool. I guess I am looking for the linear speed of the magnet wire as it is being wound on a square bobbin. I am currently trying a brute force approach on graph paper, hoping this will get me going in the rifgt direction.

Here is a lame pic of what I am talikng about, this is a bobbin goint thru it's rotation, and the result to the wire as far as location. I am guessing this will track out as some sort of sin function or catenary.
Thanks!
cj

 

[gyraf]

Seems like a very good idea (and question) indeed. Sorry I can't help out here - but I'm watching the process from the sideline :shock:

Jakob E.

 

[Learner]

[quote author="CJ"]OK, any math geeks out there?

Since 99 percent of all audio transformers are wound on a square or rectangular bobbin, I am trying to figure out what's going with the wire as far as linear speed is concerned. I am looking at a square bobbin for now so I don't get any brain bubbles.

As the corner of the bobbin comes up during rotation, the wire speed and location changes. This can make life interesting when designing a wire tensioner. If you feel the wire between your fingers while winding a bobbin, you can feel this oscilation. Most tensioners handle this with a "dancer", which is kind of like a low mass spring loaded tone arm type thingy that sits out above the tensioner..This acts kind of like a shock absorber . It's purpose is to try and keep a constant tension on the wire, even though all heck is breaking loose, especially at high RPM's. But this arm will have mass, which is bad, because this mass can break fine wire when really spinning the mandrel.

A square bobbin, at high speeds, can actually hammer the copper wire and distort it, which can weaken it and cause a possible failure down the line. You can really hear this slap when someone is winding a big rectangular coil, say 1 by 2 feet. A little slap is ok, as it tells you that a nice tight wind is being done. But too much slap can be bad, not only by harming the wire, but by giving the copper too much upward momentum, which actually can lift the wire off the bobbin momentarily. This usually manifests itself as a bowed out winding, which means less fill. Every coil has a little bow in it, but too much is bad. A slow winding rate is one answer, but for #50 wire....it takes a long time to wind a coil.



Anyway, I came up with the weird idea of changing the winding speed as the bobbin rotates, thus keeping a constant tension on the wire without a fancy tension arm. This could be done by writing a program for a stepper motor, but I need an equation first. I am not sure how to approach this math, either vectors or polar coordinates seem the most obvious.

So if anybody can give me an idea or two getting an equation, that would be cool. I guess I am looking for the linear speed of the magnet wire as it is being wound on a square bobbin. I am currently trying a brute force approach on graph paper, hoping this will get me going in the rifgt direction.

Here is a lame pic of what I am talikng about, this is a bobbin goint thru it's rotation, and the result to the wire as far as location. I am guessing this will track out as some sort of sin function or catenary.
Thanks!
cj

[/quote]

Just a thought on sourcing constant tension to the square bobbin, may be you can find an alternative using a combination of pulley wheels which can obsorb the tension but also pull enough tension to the winding. You can probably use the stepper motor controlled by a timer to control a left to right oscillating mechanism to move the bobbin.

Sometimes the simplest approach can be the most effective and time efficient, with a spontaneous result to allow for fast adjustments.

Looking forward to see some christmas transformers :grin:

 

[Adadan]

I'm not really much of a physicist, so I'm sure somebody else can do me one better here. Looking at it very simply, the linear speed of a point at some distance from the center of a rotating object is given by

(speed) = (angular speed in radians/sec) * (distance from center to point)

So if you're looking at your motor speed in RPM you can multiply by pi/30 to get angular speed in radians/sec. (The time units aren't actually all that important as long as you're consistent, but radians/sec is common). Angular speed is just a measure of the rotating speed of your system, regardless of the position on the system (ever see that Calvin & Hobbes where Calvin's wide-eyed in the middle of the night after his dad points out that two points at different distances from the center of an LP are moving at different speeds?)

Then the problem is that the point where the wire contacts your square or rectangular bobbin is at a different distance from the center of the mandrel/bobbin as it turns. Even for a square, with sides of length '2a' for example, length from the center to a flat is 'a' and the length from center to a corner is 'a*(root 2).' (I wish I had an equation editor. . .)

Plus, as the coil is built up this distance obviously changes--the difference between two successive layers probably isn't that much, but the difference between the first layer and the final layer is going to be significant.

For a square bobbin and a first attempt, you may be able to just figure an appropriate speed for these two points (flat and corner) and then program the motor for a linear change between the two "extremes." For a rectangular bobbin it seems trickier, because the transistion from the "shorter flat" to corner vs "longer flat" to corner will be different. I've never wound a transformer, but it seems that most bobbins tend to be square-ish rather than drastic rectangles, so the simple approach may be workable.

To take into account the change in distance as you get build onto the coil, you could steadily ramp down the speed of the motor with increasing turns.

However, this is just the linear speed and ignores direction, which is significant if you care about the actual tension on the wire, and which obviously also changes with time. To do the velocity you need to look at direction, which is definitely a vector problem.

 

[EZ81]

Adadan, it can be done using only trigonometry, but the math gets rather ugly...
maybe i get something useful out of the derivative of SQRT( d² - (a²+b²)(2 sin²(phi) -1) ) with respect to t mess i scribbled together on the train ride home....

CJ, i think you need a pretty powerful stepper to do the necessary accelerating etc. at useful winding speed with a nearly full bobbin.

;Matthias[/i]

 

[Gus]

You could sense the four corners with four hall effects etc.,or even use an absolute encoder

 

[CJ]

Yes, momentum is a problem with a full spool of copper. Plus the motor's mass.

I realized that that it is an easier math problem if you put the wire feeder out at an infinite point in space, so you can dis-regard the vertical motion of the wire. I can see that the distance from the bobbin to the feeder pulley will actually affect the linar speed. I brute forced this problem on some polar graph paper and plugged in a math table into Excel and got a crude graph.

Used cos b - cos a, in 5 degree increments... ada might be right, might need some calculus here. And as the wire gets spooled onto the bobbin, the shape changes from a square towards a circle, so the linear speed graph is going to get tweaked. I guess this is why they put a small shim on the sides of a former sometimes, to give the winding a little bow, keeps the core tube from collasping due to the Greek arch effect, keeps wire slap down also. You can imagine this as a car driving down the road with square tires, and the road is the wire. A little curvature in the tire would help greatly. I think a flat surface can generate a cracking of the whip effect at hight speeds, which can be destructive. Interesitng stuff.

I guess a rectangular bobbin would have the same graph as a square one, only two sections of the linear speed curve would be higher than the short side of the bobbin curve.

I will probably end up with a constant speed motor, but this is good to know what is going on with the wire.

With the square bobbin, you only need to do the math for pi/2 radians (90 degrees) as the cycle repeats itslf, so this is like a piecewise function.

Here is the speed graph. Could be the top of a sin curve, or some other function.

http://vacuumbrain.com/The_Lab/Winder/wirespeed1.jpg

 

[Adadan]

CJ,

This is an interesting problem--the more I think about it the more interesting it gets. Your curve made it clear that what we're looking at is the shape of the circular arc traced by each corner of the square as it turns. (I was trying to picture this before, but actually having your picture made it way easier.)

For a rectangle it would be just as you said, since each segment is an arc that is a smaller/larger section of a circle with the same radius. (For the square, each segment is an arc that is the same section of a circle with the same radius.) I'll have to look at a book or two later. . . I don't actually think this will require calculus. . .

I guess I wasn't that clear earlier when I said you needed to use vectors--I meant if you are worried about velocity (direction as well as speed) you need vectors (if you were trying to figure out actual wire tension, for example).

As Gus pointed out, even if you end up with the correct equation, if you decide to try it you'll need a "corner sensor" or some other way to sync your microcontroller to the actual position of the bobbin. . .

 

[CJ]

Yes, interesting problem.
I gave up for a few minutes to work on another aspect. I think this is the way to figure out the angle of pickup for the bobbin corner for various size bobbins. I can not imagine a bobbin being anymore than twice as tall as wide so I just did a few ratios. You just divide the length of one side of the bobbin by the other, take tha inverse tangent of that ratio, and you have one angle. Then just subtract that angle from 90 degrees to get the other pickup angle.

Here is some chicken scratch:

http://vacuumbrain.com/The_Lab/Winder/bobbin_pickup.jpg

I have heard people say that when you get the right bobbin size, the right wire size, the right layer length, and the right winding speed and tension, that you end up with a really nice transformer. Happens about one out of a hundread. UTC A-10 comes to mind, Triad HS-29 also. Really nice windings.

I downloaded a trial version of Mathematica to try and help simulate what is going on, but was too caffeinated to read a 1 zillion page manual to add two and two, so I removed the ten billion megs off the machine and went back to caveman math. Here is a 1 by 2 ratio bobbin on some polar paper. The pickup points are different, but each one takes the wire 90 degrees.

http://vacuumbrain.com/The_Lab/Winder/rectangle_a.jpg

The green corner takes over when the bobbin rotates 90 degrees.

http://vacuumbrain.com/The_Lab/Winder/rectangle_b.jpg

 

[PRR]

> As the corner of the bobbin comes up during rotation, the wire speed and location changes.

Yeah, and until you get that million-core contract that lets you hire 20 dummies and 3 wizz-kids to keep up with demand, I think the economic answer is to wind slower. What you are proposing is to rev it up until it busts, then find a way to rev higher.... this may be OK for drag-racing, but we be doin fine audio here, dude. What is your hurry?

There is a clever mechanical monkey-motion to generate that speed wobble. I just don't know what it is.

Here is a stupidly effectve way to do it (sorry, it negates much that you already built):

The round doughnut runs at constant RPM. It rubs on the edge of the bobbin or an equivalent shape on the bobbin spindle. Since even fine wire won't make a sharp corner, the bobbin-edge should not have sharp corners, rather rounded about as the wire's comfortable bend. However if the drive edge is a little sharper than the wire curve, the bobbin slows a bit at the corners, which may be desirable.

The round doughnut must bounce up and down without losing contact, and without wobbling sideways. Its shaft does not have to be exactly parallel to the bobbin spindle, it could be a long swing-arm.

The wire speed will rise (about double) as the coil builds up. This may be slow enough you can just turn-down RPM with a spare hand, or when you break for tape. (Really dumb would be to power the tire with a battery that runs-down as fast as the coil bulds-up....)

The doughnut is logically a model car/airplane tire, available in any size and various tractions.

Alternately, the tire could run ON the winding. Then wire speed IS constant. However there may be a tendency for the wire to run at the right speed and the bobbin to run slower, throwing a big snarl back behind the bobin.

Yet another trick is an un-driven tire running on the winding, turning a tach, and a servo to the bobbin spindle motor that tries to keep tire RPM constant.

In toothed gearing, I think you need two heart-shaped gears to generate the speed wobble for 90 bobbin degrees in one heart-gear turn, and a 1:4 gear-up to make that happen 4 times per bobbin turn. That answers the square case; rectangles are tough. Maybe two double-heart gears. This is REALLY not practical without a government contract.

The bobbin is always near-square because we want a maximum of iron area inside a minimum length of copper. A circle is ideal, and big power iron uses various tricks to get round. Most audio is done with stock stampings all one size so square is the best we can do. Sometimes practical annoyances lead to a rectangle core, but if it is far off from a square it is far off from the optimum area/length ratio and a bad design. 2:1 is not a disaster, 10:1 would kill you with high resistance for low inductance.

 

[Swedish Chef]

I'm not sure if I should really open my mouth after a PRR post, but re: the tyre/ bobbin combo, if there is a problem of traction of the tyre on the bobbin, would it be possible to have a "Dummy" bobbin on the same shaft as the "Real" bobbin? You could then dedicate this to have max traction without worrying about avoiding the winding.

OR if you could spring tension the drive shaft in the y-plane, your dummy could be a "square" cog with the same aspect ratio as the bobbin, which is driven off a similarly toothed cog...

This will still not help with the prob of speed change as the turns accumulate.

Is this helpful? :?

chef

 

[babyhead]

I have been following this thread with curiosity... and been thinking about your problem all day. A little intimidated here, but I just want to throw a laymans idea into the forum.

How much of the corners of the bobbin problem could be rectified by:

a) spring isolate the final tension point of the wire and

b) setting up a sprocket system similar to a sewing machine. A 4:1 sprocket with a piston to raise and lower our final wire guide to the beat of the winder and to help absorb corner shock.

I was thinking about how I would keep an even tension if I were to do it by hand and then thought about how to mechanize such. #a might be enough to null the curve, with the right tension.

But alas- I no PhD... just a music degree. ii V I? Bach?

 

[PRR]

> spring isolate the final tension point of the wire

He addressed that somewhere up-up-thread. A "dancer" always has mass and when you crank the system to the limit, that mass breaks the wire.

If you look at mainframe computer tape drives, they ran 1" tape in a loop in a 1.010" * 4" box, and pulled a small vacuum below the loop. This is equivalent to a dancer with mass equal to half the tape in the loop, no mechanical monkey-motion mass. But the mass/diameter ratio of wire is so much higher that I doubt you could get low-leakage operation and even a little mass might be enough to break wire. Maybe an air-jet to apply lateral force on the wire without adding mass. As if this operation was not already loud enough.

> setting up a sprocket system similar to a sewing machine.

That was my thought: a Singer has MUCH more thread-jerk than a square bobbin. And while there is a dancer in the system, mostly it is that arm in front jerking up and down slightly out of phase with the needle poking in and out of the fabric. What is that motion? Nevermind, it isn't similar enough to ours to be directly useful.

> raise and lower our final wire guide

Assuming the wire feeds horizontally: the up/down of the bobbin is not really the problem. The problem is that the radius of the bobbin varies from 1.0 to 1.4 and back, so (for constant bobbin RPM) the wire velocity must vary 1.0:1.4 and back four times for each bobbin turn. If we wind very slow, we can let the feed spool RPM vary 1.0,1.4,1.0,1.4,1.0,1.4..., but the feed spool is high-mass and we can't do that quick simply by yanking on the wire.

We could use a Singer arm if we found the right motion. If we had a crank geared-up 1:4, a connecting rod and crosshead, and a pulley to allow the wire to pass straight or in a V... pull the wire to a V when the wire is in line with a flat, let it go straight as the corner comes up. The wire enters at constant linear speed, but the varying path length forces it to leave at varying linear speed. And some straightedge & compass work will quickly tell the depth of V to get the right ratio for the flat and the corner. But is it a good match at the in-between points? I doubt a sine crank is close enough. It might work better if the roller overshot the wire so it ran straight a large fraction of the time, but that is clearly not right.

Here's the no-math way to plot the V-pulley path: set up a bobbin, and a round "spool" with the same circumference. (For a 1"x1" bobbin, a 4"/Pi= 1.273" round spool.) Gear them solidly at 1:1 ratio. (You could put them on the same spindle, go out to a fixed pulley and back.) Near the bobbin, mount two fixed pulleys. A third pulley between pulls down with a weight. As you slowly crank the spindle, trace the position of that pulley for each spindle angle. Whittle a cam in that shape.