Small coil winder

[quantumeer]

I need to wind a few small coils, around 5 mm diameter and a couple of thousand turns of 49 or 50 gauge wire (0.030 to 0.025mm).  Normal coil winding tools are fine for stronger wire, but for wire with a breaking strength around 10 g, our bigger tools (pictured) don't have the finesse.  The gearing and inertia interfere with the delicate control neeeded, while the automatic tensioning and traverse don't work on this scale: finger feeding is best, so having a right-hand winding handle is unhelpful.


Quaint but useful Douglas coil winder for chunkier wire


Showing the winding handle, turns counter, dual-disc continuously variable traverse, dual (foward and reverse) leadscrews with automatic turnaround, and (top) wire supply tensioning brake

... but all I need is a little motorised spindle with instant control and a turn counter, the whole thing to fit under a microscope.  To emphasise the delicacy of the operation, here's why you don't lend your 50 gauge wire to just anybody:


Find the end! Actually that's easy because there are so many. Find the real end. Don't pull too hard or you'll have two more.

My first thoughts were of a geared DC motor, because the electronics would be easier, but then I would have to do something about a turns counter.  The motors I tried also had a fraction of a turn overrun past a dead stop, which is not ideal either.  So off to the backing store in search of a little stepper.


They don't make floppy drives like that any more.  A nice little stepper and buffer chip just waiting for a home.

So here is today's treasure, ready to build something.

[Bernd]

Hey Q,

That looks to be an interesting project. I've got a few those old drives laying around, I think. 

Bernd

[Brass_Machine]

This will be good to watch.

Eric

[quantumeer]

Thanks for your interest. The early versions of things like those drives, before their off the shelf components get replaced by a fully custom design, are a great source of bits.  Speaking of which, I have found enough bits to complete the electronic "design"



The display is a module with a 2-wire interface, which will save on solder and software compared to the ubiquitous LCD, but the real reason it has the honour of being part of this project is the same as the selection of the PIC controller: I found it in the useful box.  Almost anything would work.

Next, some soldering.
Mark

[doubleboost]

Interesting stuff
That spool of wire reminds me of the 15kg reel of mig wire i dropped 
John

[Bernd]

Q,

I take it that 16F88 is a uM processor. Gonning to do a how to program on that. If you do put that thread down in the Electronics and IC Programing of the forum. I've always wanted to know how you write a progam for such a small processor.

Bernd

[Anzaniste]

Will the system include a drive for the feed spool of wire to overcome a) friction and b) inertia?

50swg = .001" at 70,000psi tensile = break strength of 0.19 ounces, no wonder you've got to be careful!!!

[quantumeer]

You're right, there's no way you can pull the reel around by the wire.  However, if you take care of the end cheeks, wire will just slip off the end of the reel with almost no force; it imparts a negligible twist to the wire.  Here is a video of a nice little machine that does this with tension control:

but I'm not planning on making one of these.

Mark

[quantumeer]

Decided to do the wiring using solderable enamelled wire (the verowire or roadrunner system), because it's quick for one-offs, and more serviceable than you might think.  Can't say the same for the wiring pens though - overpriced and poorly made.


The repaired wire friction device works better than the original and only cost small change.


Wiring close up. I'll clean up the flux residue after verifying it works.

Next job, firmware. I'll be using it to verify the wires went to the right places at the same time as writing it.

Mark

[quantumeer]

I do software incrementally, to limit the number of things that can go wrong at each stage.  I'll try to show what I mean, without going into any actual code. Maybe as Bernd sugests, I could do that elsewhere.

I'll be running the growing code on the actual circuit, so that is one of the things that could be wrong.  First thing to do is get some output going.  Actually, driving that big display is too big a chunk to start with, but I can re-use code I wrote before so it is no trouble to try, and I'm feeling lucky.  If it fails, I'll drop back to making an LED flash.


I wish I could say that's the first time my firmware has greeted me like that, but I have a track record in this area:



Still, that's more success than failure.  A quick reversal of the character order in the code, and step 1 is done:


So the microcontroller is running, the in-circuit programming works, and the display is connected right. Good.

Step 2.  Plug in the pot that will be the speed control, add some code to read the voltage it produces, and send the reading to the working display to see if it looks sane.



Lucky again.  Twiddle the pot to see if the reading goes up and down, and it looks good.

Step 3.  Plug in the switch that will be used to zero the displayed count, and add some code to set the display to zero if the switch is closed.  Waahhh!  Display stuck at zero for both settings of the switch, and all settings of the pot.  Since I know everything else is working, it can only be the switch wiring or the code I just added.  Use a voltmeter to check the microcontroller pin that reads the switch.  Hum. 35 mV with switch open; should be 5V.  Solution after the break...



Something's pulling it down, but 35 mV is not low enough to be a short to ground. I did configure that pin as an input, didn't I?  Back to the code.  It turns out that I made a typo in the line that was supposed to do that.  The switch signal is connnected to an output that happens to be trying to be a logic zero.  Fix the typo, and now the switch works.

Step 4, the stepper motor.  Plug in the motor, and add a bit of code to do step/delay/step/delay... The motor twitches, but doesn't rotate.  I was sort of expecting that, since I didn't know the order of the windings.  A quick experiment with two wires from the power supply gives me the sequence of pulses that makes it go.  Reorder the wires in the motor plug and off it goes!

Step 5, unnecessary but can't resist.  Add code to use the value from the pot to set the stepper delay, and we have speed control.  Cute, but the real code comes later.  Time for bed, says Zebedee.

Mark

[quantumeer]

Another Sunday, so did the real code.  I decided to add a beeper to help keep track of the turns count without taking eyes off the coil: different beeps for tens, hundreds and thousands.  There was just enough room on the board for the additional resistor and connector.

Here is a video of the thing in action (not youtube, so I just put the URL)

http://www.ytlab.myzen.co.uk/instruments/finewinder/winder.avi


The code is 195 lines of C, translating into 1053 instructions in the PIC, which is therefore only 25% full.  Next, add the chuck and an enclosure.