# Measuring a very fine wire with 10 µm precision

Stopping exactly on the far lip is required in order to properly reverse to the new layer

But unless the error was such that you had accumulated an integral number of wire thicknesses, that's not going to help.

Let's say it nominally holds 100 turns along the length. You measure that the wire is a bit thicker say, and by the time you get to where the 99th row would be, you think Oops, we've actually got 99.5 thicknesses. So what do you do? You can't fit another full turn into 0.5 of turn's gap, so you can't stop exactly on the far lip anyway... not in terms of wire widths.... so you reverse with half a wire width gap.

(That made sense to me in my head- might need a drawing to describe it but I cba doing that right now.)

Point is, I'm pretty sure there are best-practices for doing this commercially, and I don't think this is one of them with all due respect.

Boots507: Stopping exactly on the far lip is required in order to properly reverse to the new layer and keep a tight packing without overlaps.

By measuring the wire thickness you could estimate the number of turns needed but I don't see how you'd ever get it exact.

I don't know whether it would be practical with wire that fine, but I wonder whether it might be more effective to detect the outer diameter change as the wire hits the lip at the end of the former, rather than try to [u]predict[/u] when it will hit? If you used a floating pulley, you'd need to detect a movement equal to the thickness of your wire.

Boots, me thinks you're approaching this problem from the wrong direction.

Also, copper wires are force-drawn/pulled through a die at the factory, so they're pretty much the same diameter. Being able to detect variations of 0.001mm won't do squat to what you're trying to accomplish. It's immaterial.

I would concentrate my efforts on a mechanism for consistent tight winding, and accurate inductance values. If the wire ends in the middle of the coil (to achieve the correct inductance value), then so be it. Tape the wire at that spot, then run a few inches of flying lead wires, solder to the transformer terminals (or solder to a thicker flying lead wire). The wire doesnt necessarily have to end at the lip. -- thats how commercial audio transformer makers do it.

EDIT: The core you use for your transformer will also have variations in their permeability. Even if you got your wire diameter detector correct (which I doubt), the next core you grab may be slightly off, and the next transformer you wind will not end up exactly at the lip (to achieve the correct inductance value)... so now what? You're back to where you started. You cannot predict where the wire should end because you'd have variations in the size and permeability of your core (if you care about getting the right inductance value).

I try to go back to the original question, how can I measure the wire with that precision ? I don't think there is a practical mechanical way to achieve such precision. So what about a digital microscope ? http://www.leica-microsystems.com/products/digital-microscopes/ well I don't know your budget (there are cheaper options), but the idea is to "enlarge" the wire, and measure the variation of the image. I mean, measure the difference of light transmittance ..... Maybe it's just an weird idea but I have no other.

Lets try to design the sensor you need,

## design 1: - conductive wire

imagine a V shaped sensor that guides the wire. depending on the thickness of the wire it will be higher or lower in the V shape make the sides of the V of a conducting material with known (relative to wire high) resistance The place of the wire will change the resistance of the V as the level where the wire is short-cuts the current.

place a laser let the wire move in front of it. @10uM a known amount of light will pass above and below the wire make that minimal @12 uM - you need mirrors at exact angles place a TSL235 in the laser bundle - this convert intensity to frequency. when the diameter varies the intensity will vary

laser >-------------------- | - - - - - - [sensor]

now shoot!

Best option is to use an optical microscope. "Seeing" 1um is not hard at all. Get a decent CCD camera from Pixelink or Basler or any other brand with a 1" sensor and a pixel size of 2-3um. You'll also need a microscope and some basic image processing to automatically identify the wire (easy when looking for a straigth line on a uniform backgound) It shouldn't cost more than \$1000.

Hello, I think the weapon of your choice would be a so called 'micrometer' :P http://en.wikipedia.org/wiki/Micrometer

Plus: If you want to measure in such small dimensions, be aware of the fact that even the warmth of your hand and also the atmosphere around the wire has an impact on the extension of the wire. I'm going with vasquo's oppinion.

//edit: With what I want to say: If you're going to measure a wire by applying voltage, it will expand slightly due to warming up slightly (electrons colliding inside the copper), so I assume you can basically forget that...

osterchrisi: Hello, I think the weapon of your choice would be a so called 'micrometer' :P http://en.wikipedia.org/wiki/Micrometer

No I don't think so, resolution and precision are not enough maybe something like this ... http://www.keyence.com/services/download.php?file=im6000_ka.pdf&fs=IM-6000&ws=none&img=oc_cmm_lp.jpg&layout=d2d&lil_id=1309212783&lil_ly=aln-l_tfsz-20_dfsz-16&aw=gagooglekaim899063b&gclid=CLC97ta9-rQCFUFb3god9Q8AvQ but again, the budget is ? I think szangvil and robtillaart have the point.

robtillaart: Lets try to design the sensor you need,

## design 1: - conductive wire

imagine a V shaped sensor that guides the wire. depending on the thickness of the wire it will be higher or lower in the V shape make the sides of the V of a conducting material with known (relative to wire high) resistance The place of the wire will change the resistance of the V as the level where the wire is short-cuts the current.

You do know the wire is insulated (and not BARE). It's a "magnet" wire, the kind used to make coils. If you strip the insulation just so you can measure it's resistance, well, you just made your coil useless.

I think the OP needs to visit an audio transformer manufacturer so he can see how it's done. These guys don't measure their wire diameters to micron precision or use lasers or microscopes as they wind. But they do measure their inductance and DC resistance values. They don't care where the loop ends. Overwind a few extra turns, measure, unwind, measure, tap it, then tape it. Done.

Have you guys seen the insides of an audio transformer?

robtillaart: The place of the wire will change the resistance of the V as the level where the wire is short-cuts the current.

I don't think that would work, because the wire is insulated. Using a camera and microscope seems like the most likely approach to me, but the other option would be to create a rolling version of a standard wire gauge (using a pair of opposed precision ground rollers with a slight taper, and angled so that they tended to pull the wire away from the narrow end of the V as it rolled between them). The height of the wire in the gauge would tell you the thickness of the wire to whatever precision you wanted, as long as your rollers were sufficiently accurate.

For initial optical testing, there seems to be inexpensive microscope cams up to advertised 800x power.

If we’re winding inductors here, the inductance of a coil is proportional to the number of turns squared. If the aim is to produce an inductor with a specific inductance, then the number of turns would be the important parameter to control. There should be little effect if things get scrambled a bit at the ends.

Thinking further, the concept of measuring one parameter (wire diameter) and calculating the number of turns based on that is flawed. It is an open-loop system and doesn't take a myriad of other variables into account, therefore the calculation will never be perfect. What if the next batch of forms are slightly different in size? What if the standard deviation of same changes? Etc. etc.

A better approach would be to somehow sense when the winding reaches the end of the form rather than calculate it. This is a feedback mechanism that allows for all those other variables (including the ones we haven't thought of) and doesn't require constant measurements of the input (wire). Maybe it's the angle of the wire from the supply point, or maybe tension, etc.

I can buy any number of products that are neatly wound on their spools, e.g. string, thread, wire, solder, etc. There must be some fairly common mechanism to do this that you just need to discover. The wheel has already been invented here. The question I would ask is how are all these other products wound on their spools neatly, not how can I invent an original contraption to do it.

Hello all, I want to thank you for your participation in this thread so far. My name is Dan and am Boots507's colleague. I would like to better explain our motivation for creating "perfectly" wound coils. In Non-Destructive Testing (eddy-current), electromagnetic coils interact with conducting samples via induction processes. Theoretical models can be compared to experimentally acquired data in order to infer geometrical and electrical properties of the sample. However, this approach requires integrating the electric filed over the cross-section of the coil. Thus, these solutions are highly sensitive to the quality of the coil. More explicitly, they assume that the coil has a uniform turn density and a rectangular cross-section. Therefore, we need a machine that creates ideally stacked coils.

An exact solution gives the theoretical inductance of a coil based on inner diameter, outer diameter and number of turns (uniform turn density and rectangular cross-section). Coils that I have wound very carefully by hand agree almost perfectly, whereas sloppily wound coils deviate significantly.

Hi Dan, While in the past I have wound a few coils with AWG 44 wire,ever none needed that type of precision..My gut feeling is this may be almost impossible . I guess I do not fully understand your problem but just to throw this out. Why can you not wind your coil as precise as you can in practice, and later in some uniform test, add or subtract some correction factor so that each coil results in a uniform result.All the previous posts pointed out the pitfalls and problems in trying to get absolute winding results.. Anyway, Goodluck jolphil

dan38711: Coils that I have wound very carefully by hand agree almost perfectly, whereas sloppily wound coils deviate significantly.

Where on this spectrum from perfect to useless do ordinary machine-wound coils using simple turn counting fall? Would it not be sufficient to machine-wind one layer of the coil, manually pause it at the end of the layer and note the number of turns, and then program the winder to use the same number of turns for the other layers? This could be a very simple process if your CNC controller was designed to support it, and without needing any exotic measuring techniques. It seems unlikely to me that you would need to deal with variations in the wire thickness within a single spool, and I expect that once you have got it set up for one coil you could then continue to wind multiple similar coils with the same settings.

PeterH: Would it not be sufficient to machine-wind one layer of the coil, manually pause it at the end of the layer and note the number of turns, and then program the winder to use the same number of turns for the other layers?

PeterH,

To me and Dan this seems likes a very good approach to the problem. Although it adds a step in the calibration of the spooling machine to get the number of turns (and consequently the theorical diameter); it seems reasonable to think that from that point on a multitude of coils could be wound using the setting acquired during the calibration run.

I am going to give it a try tonight and report my findings as soon as possible.

Thanks for the idea

Roughly speaking, what is the production volume required? Tens, hundreds, thousands ... ? Also production rate, how many per day, hour, or whatever.

JimboZA: This is a nice diagram explaining the difference between accuracy and precision

In machining we would specify that precision as 'tolerance'. Every cut is to some measure with some tolerance (like +/-0.1 or maybe +0.1/-0.0).

GoForSmoke:

JimboZA: This is a nice diagram explaining the difference between accuracy and precision

In machining we would specify that precision as 'tolerance'. Every cut is to some measure with some tolerance (like +/-0.1 or maybe +0.1/-0.0).

But even the device(s) used to measure that such 'tolerance' is within specification or not is subject to a variation in precision. Accuracy is a poorly understood property in the real world.

Lefty