Coil winders Ideas

[David Schwab]

Here's a couple of sketches. As per the first ideas, there's not much there but the motors, a belt and a couple of pulleys.

http://geofex.com/FX_images/coilwinder.pdf

The things that have to be fabbed are the bobbin holder, eyelet on the belt and the tensioner.

Of course, now the operation of the mess is A Simple Matter Of Programming.

That looks good.

Do you know any good ways to de-spool the wire?

[R.G.]

If not managing wire for non-kinking by hand, I would make a long, skinny cone for it to de-spool over. Polyethylene should work. The idea is to have it spill over the side of the spool onto a low-friction surface that supports the inside of the wire loops and keeps them from collapsing until the diameter gets down small enough and the loop width gets long enough to keep them from overlapping into a kink.

Maybe a polyethylene sheet welded into a cone with a polyethylene spike on top to support the loops.

[Ampclutz]

Hi David,

Thanks for the response.

CNC machines use numbers. So, using the one number you gave (1500 windings) The only coil a cnc machine could wind would be a coil with 1500 windings stacked exactly on top of each other. I don't really think that would be a very practical wind! My guess is it would be fun to watch on someone else's machine!

From David's post:

"I wind with a back and forth motion... "

As you can guess, a CNC machine won't know what to do with that information. Too vague. If you want to wind a coil with a cnc machine you need to specify: number of winds, width of coil, and wire spacing. That is the minimum information and would need some interpretation to make a usable winding program.

Want to try again with more information? You can get as complex as you want to but there is a certain minimum level of telling the machine what to wind. You haven't met the minimum requirement yet...


Hope this helps,

AC

[R.G.]

Let's put it into perspective.

David - how many turns does the coil do while you traverse the coil width with the wire? That's probably not constant, so how about min and max?

The minimum traverse rate that ought to be used is something less than the wire diameter divided into the coil width. For instance, #43 with single enamel insulation is about 0.0025 diameter over insulation (source; MWS Wire Industries; this varies per insulation build and manufacturer). So if you traverse less than 0.0025 per turn of the coil, the wire piles up and spills over in one place instead of laying on in a layer. If you traverse exactly 0.0025 per turn of the coil and the wire really is that diameter, you get a perfect lay per layer. If you traverse more than 0.0025 per turn, you get varying degrees of scatter. If you do traverses as fast as a few turns per traverse, you get into pi-section coil territory.

So what's a good range of coil turns per traverse?

[David Schwab]

Gee... I really don't know! I've never measured, and I supposed there's a certain amount of randomness involved.

My winder is rated at 725-rpm maximum speed. I usually wind with the speed control about half way.. sometimes a little faster, sometimes a little slower. As is typical of hand winding, I often change the speed as I'm winding. If I get a good groove going I'm be up near 3/4 of the full speed or more. Faster than that and I can't feel enough control over the wire.

My traverse pitch is about 2 times per second, maybe a little slower, so each second I have moved my hand from left to right twice. But that varies as well. I'm watching the coil the whole time and trying to be uniform. If it looks like I'm piling up too much wire on the edges, or the middle, I'll compensate for that. I'll stay in one spot for a while, or move slower. I squeeze the wire between two fingers firmly. If I squeeze harder the winder slows down.

I'm winding into a space 5/16" wide. I use different gauge wire, and different numbers of turns depending on the pickup. 1500 turns of 42 is a starting point for one design. I also use 43, and I'm experimenting with 36.

I'm sure there's a wide range of variances from pickup to pickup, but they sound quite consistent.

Obviously with a controlled winder things can be done more uniformly.

[R.G.]

Hmmm...
1500t, 400rpm, that's 3.75minutes, 225 seconds. That's 6.667 turns/second

Reversing twice per second, that's a traverse every three and a fraction turns.

Are you sure you reverse twice a second? That seems awfully fast to me. You may, it just seems fast when I try to duplicate the motion.

#42 is about 0.003 diameter, so a level layer traverse is 5/16" divided by 0.003, or 104 turns if you can do it perfectly.

[Ampclutz]

David,

Ahha! You're getting to the real information. So, bobbin width (.3125 inches) divided by wire diameter (.0025 inches) equals 125. Therefore: If you tell the computer to go 125 winds while commanding it to traverse one bobbin width the wire will be laid down side by side. More than one traverse per 125 winds and it will be more "scattered" (This isn't actually a scatter wind though)... Less than one traverse per 125 winds and the windings will pile up on each other. What you have to do is tell the computer exactly what to do. This is a basis of design you can work with... You can wind anything you want to using ratios of winds to traverse. There is no random... (Well I guess there could be randomness but then the the pickups wouldn't be the same would they...)

Does this help?

AC

[R.G.]

It also brings up the need for some measurement. If you're trying for a perfect lay (down boys, that's what a layer wound coil is called) you have to measure the actual wire you use. #42 bare is 0.0025. #42 with insulation is 0.003 if you get "normal" heavy enamel magnet wire. If you got single enamel or some of the films, you could get 0.0026. This matters when you are getting 100+ turns per layer and you could get two extra turns per layer.

However, the wider your scatter is, the less this matters. Here's some stuff I found on #42 from MWS.

Values listed below are derived from the NEMA MW1000-1997 Standard or MWS Wire Industries internal inspection criteria
Insul Bld Insul Min. Insul Nom. Insul Max.
Single 0.0026 0.0028 0.0030
Heavy 0.0028 0.0030 0.0032
Triple 0.0030 0.0032 0.0035
Quad 0.0032 0.0034 0.0036

Notice that your layers will all be one wire deep, but only contain the number of turns that your traverse allows. Wider scatter makes for fewer turns in the same volume, more separation between wires, and therefore both lower self capacitance and lower inductance as well as lower signal strength.

[Ampclutz]

RG,

How do you plan to drive your motors? What sort of user interface do you plan to use? Do you plan to program in G-code or some other system? Run it on a PC or other...

AC

[David Schwab]

Are you sure you reverse twice a second? That seems awfully fast to me. You may, it just seems fast when I try to duplicate the motion.

Am I sure? No. But what I did was start a stop watch and then moved my hand about the way I do and timed it. It was a bit slower than that, but it's hard to say. Also I wasn't actually winding anything.

Might be about once per second or a bit slower??

[David Schwab]

David,

Ahha! You're getting to the real information. So, bobbin width (.3125 inches) divided by wire diameter (.0025 inches) equals 125. Therefore: If you tell the computer to go 125 winds while commanding it to traverse one bobbin width the wire will be laid down side by side. More than one traverse per 125 winds and it will be more "scattered" (This isn't actually a scatter wind though)... Less than one traverse per 125 winds and the windings will pile up on each other. What you have to do is tell the computer exactly what to do. This is a basis of design you can work with... You can wind anything you want to using ratios of winds to traverse. There is no random... (Well I guess there could be randomness but then the the pickups wouldn't be the same would they...)

Does this help?

Oh sure. I doubt the way I'm winding is having a profound effect on the pickup, other than not laying the wire down in neat little rows. I'm just attempting to evenly distribute the wire, while trying to scatter it up a bit.

But I couldn't get a machine wound coil by hand it I tried! Not unless I turned the bobbin by hand as I guided the wire.

I would think it would be interesting to try various patterns and see what they sound like. Then when you find something you like, you have it. I find the repeatability factor very appealing.

But then I have yet to wind two of these and have them sound very different, unless I change wire gauge or turn count, or some physical thing.

And even then it's subtle.

[R.G.]

How do you plan to drive your motors? What sort of user interface do you plan to use? Do you plan to program in G-code or some other system? Run it on a PC or other...AC

For my purposes, I'll probably write a simple program that inputs a total turns and a traverse rate, as well as accelerate and decelerate profiles and run that out a parallel port. This is something I can run on an old, otherwise useless PC laptop I have.

With either a stepper control chip or a bit of CMOS logic the motors will accept step/direction commands, so they'd work directly from most of the CAM programs out there. You could get as fancy as you like to do the programming for. For me, the choice of (almost) infinitely variable and programmable traverse per coil turn is good enough. I suspect it's as capable as 99% of the other winders that exist.

This gets back to one of my earlier ideas - once you get to computer controlled, you can make it as fancy - or unfancy - as you like. Full bore G-code generation is a nice idea, but probably requires more end effort than I need. But if I was making experimental pickups and wanted something like wave windings or sectionalized pickups, or even multifilar windings with tight top-to-bottom coil coupling for some reason, the same mechanics would do it. You'd just have to spend more time programming, or get a fancier up-front program.

[MKB]

For my purposes, I'll probably write a simple program that inputs a total turns and a traverse rate, as well as accelerate and decelerate profiles and run that out a parallel port. This is something I can run on an old, otherwise useless PC laptop I have.

With either a stepper control chip or a bit of CMOS logic the motors will accept step/direction commands, so they'd work directly from most of the CAM programs out there. You could get as fancy as you like to do the programming for. For me, the choice of (almost) infinitely variable and programmable traverse per coil turn is good enough. I suspect it's as capable as 99% of the other winders that exist.

This gets back to one of my earlier ideas - once you get to computer controlled, you can make it as fancy - or unfancy - as you like. Full bore G-code generation is a nice idea, but probably requires more end effort than I need. But if I was making experimental pickups and wanted something like wave windings or sectionalized pickups, or even multifilar windings with tight top-to-bottom coil coupling for some reason, the same mechanics would do it. You'd just have to spend more time programming, or get a fancier up-front program.

Hi all, I'm a little late to this great thread, but it has been a terrific read!! I have done a bit of robotics with stepper motors at my work, and y'all are working me up to do a winder with steppers.

Concerning linear steppers, I used this one in a project: McMaster Carr #8122K8, it's called a Through Motor Linear Actuator. It is basically a stepper motor with a jackscrew through the center, the motor rotates the nut that then is translated to linear motion with the screw. It allows faily fine resolution, and isn't too expensive. But you'll have to use some sort of linear table with it as the screw isn't too stable in the nut. It also has some speed limitations which may or may not be a problem if you are controlling traversal.

Has anyone found a good cheap stepper motor controller that allows RS-232 input to control the motor? I used this one at my work: http://www.pontech.com/products/stp100/index.htm
This worked perfectly; you can vary the speed of the motor very easily, change to forward and reverse motion, and easily read where the motor location is. It's almost too simple to use. Unfortunately they cost $160 each, too spendy for a home project. I'd LOVE to find something like this at around $30 each...

[R.G.]

Hi all, I'm a little late to this great thread, but it has been a terrific read!! I have done a bit of robotics with stepper motors at my work, and y'all are working me up to do a winder with steppers.

Ah, good. You're drafted. You're writing the drivers...

Concerning linear steppers, I used this one in a project: McMaster Carr #8122K8, it's called a Through Motor Linear Actuator. It is basically a stepper motor with a jackscrew through the center, the motor rotates the nut that then is translated to linear motion with the screw. It allows faily fine resolution, and isn't too expensive. But you'll have to use some sort of linear table with it as the screw isn't too stable in the nut. It also has some speed limitations which may or may not be a problem if you are controlling traversal.

Yeah, that was kind of my first attempt, but with hardware store threaded rod. The problem with all the commercial linear actuators I found was that they were expen$ive. One live and one dead stepper are cheap, as are the timing belt pulleys. The stepper is $5-$15, the pulleys are $3 or so, and the timing belt is another $5 maybe. The rest of it is just hardware store junk. The nice thing about the timing belt is that it can become it's own linear carriage for the short distances and light loads of a pickup winder.

Has anyone found a good cheap stepper motor controller that allows RS-232 input to control the motor? I used this one at my work: http://www.pontech.com/products/stp100/index.htm
This worked perfectly; you can vary the speed of the motor very easily, change to forward and reverse motion, and easily read where the motor location is. It's almost too simple to use. Unfortunately they cost $160 each, too spendy for a home project. I'd LOVE to find something like this at around $30 each...

Sure. PIC microcontroller. They accept RS232, and can put out motor phases directly.

But there's an issue. You really want a motor controller that issues motor phases based on step and direction commands from a motor controller. If you do that bit of virtualization, there are freeware programs to let you go to full-bore CNC driving for the thing.

Go look at the Linistepper for a simple and relatively cheap source of a microstepping driver. I was planning to use the linistepper to get from 0.015" per step down to 0.001" or less per step.

[MKB]

Go look at the Linistepper for a simple and relatively cheap source of a microstepping driver. I was planning to use the linistepper to get from 0.015" per step down to 0.001" or less per step.

Thanks for the link, that looks about right at first blush. I'll dig more into it this weekend.

As far as writing the code, we use at work a development system called TestPoint; it's basically a cheap far less functional version of LabView. In this system it is easy to develop code to control such a board, and makes it even easier to get a nice GUI to access it all. If this controller is like the Pontech one, I can see how it would be easy to write a program that would allow automation of the entire setup, allow speed control of the winder and traverse, direction control, acceleration/deceleration of the winder, and even count # of turns of the winder. Plus TestPoint allows unlimited distribution of executables. My only limit would be the time available to do this (lotsa irons in the fire). But what a fun project!!!! I'm very interested...