Diameter measurement of a wire

[MS15]

Hi,
I want to measure the diameter of a continous wire that I pull through an object. Measuing the diameter from just one angle would be totally fine. My main problem is that I only found high prices measurement systems. What I need is low budget. I know that I cannot expect wonders in accuracy, but that is fine.

These are the specifications:
Wire diameter: Between 1.5 and 3 mm
Sampling rate can be low, 10 Hz would be enough
Measuring accuray of 0.1 µm

I thought about building my own measurement system, but I am not sure how to proceed this. I read the most about CCD/CMOS sensors , I think the hardest part would be the light? The wire will slightly vary in its position within the measuring room (up to 3 mm in each direction I guess), so the light should probably come as a line and hit the sensor at each height in a straight 90° angle, is this right?

I would prefer a simple solution that can be bought and used out of the box. Do you know any for low budget? Otherwise I have to think about creating my own. I would like to retrieve the data using an arduino.

Thanks for your help.

[mzzj]

Hi,
.. What I need is low budget. I know that I cannot expect wonders in accuracy, but that is fine.

These are the specifications:
Measuring accuray of 0.1 µm

That's going to be some problem.. unless you mixed millimeters and micrometers

[Zucca]

http://www.protonproducts.com/products/dg-series/dg-series/




Not easy for DIY.

BTW, Metrology is not the right section. You should go in projects.

[Conrad Hoffman]

Measuring anything but optics (where interferometry is standard) to 0.1 um is tough, and near impossible if it's moving. That's why the commercial systems are so expensive. This doesn't seem out of place in the metrology section, as it's definitely a challenging metrology problem. Here's a crazy idea- pull the wire through a precise tube with close clearances and measure the capacitance between the wire and the tube. There's be guiding issues, but it might lend itself to a DIY effort. Or, run the wire between a surface and a guided roller and measure the roller position with a capacitance gage. Again, not hard to DIY. Trying to duplicate one of those rotating polygon systems would be a long and involved project.

[T3sl4co1l]

What tolerance of the wire's chemical and mechanical properties?

You might use eddy current sensing, but probably not to so many decimals.  Eddy current sensing is sensitive to composition, resistivity and stress condition (work hardening) (affecting resistivity and permeability).

A capacitive sensor might be okay, but the wire needs to be positioned even more precisely than for eddy current sensing.

If the density and elastic modulus is consistent, the wire could be secured between two rigid points and its mechanical resonance measured.  This could be done in continuous motion with some difficulty.

Tim

[HighVoltage]

I once visited a company that makes thin copper wires and they have to
continuously watch the thickness of the wire down to +/- 0,15 µm.
This is done by a laser measurement system. I am just not sure about the manufacturer.

[mycroft]

You can try a Linear Photodiode Array. Take a look at https://forum.pjrc.com/threads/39376-New-library-and-example-Read-TSL1410R-Optical-Sensor-using-Teensy-3-x for details. They claim sub-micron resolution using pixel interpolation.

[IconicPCB]

A linear camera is the right way to go about it. It s the basis of many a commercial instrument often found in wire drawing production lines.

Choice of the number of cells in the array will govern the final resolution of the measurement system.  With current technology it should not present much of a challenge to acquire the reading.
You will however have to run a monochromatic source in order to make a sensible measurement.
You will have to be able to calibrate the set up.

[Henrik_V]

Wire between two wheels (hardened steel, good ball bearings) , one wheel on a spring loaded (LDVT) length sensor.

[branadic]

Even though it's not metrology have you tried to use a fork light barrier?

-branadic-

[EmmanuelFaure]

Wire between two wheels (hardened steel, good ball bearings) , one wheel on a spring loaded (LDVT) length sensor.

+1

[mzzj]

Wire between two wheels (hardened steel, good ball bearings) , one wheel on a spring loaded (LDVT) length sensor.

I feel like people don't really grasp the scale of this problem. Best commercially available ABEC-9 ball bearings have guaranteed runout of  2,5 um. That's 25 times worse than wanted accuracy!

Even with very small contact force it would turn out to be also a wire hardness test. (for hardened gauge blocks see page 26 https://www.nist.gov/sites/default/files/documents/calibrations/mono180.pdf )

Teslacoil's idea is also neat but just like he says I think it would be more of material property measurement than diameter.

[T3sl4co1l]

Teslacoil's idea is also neat but just like he says I think it would be more of material property measurement than diameter.

Thanks.  Yeah, the optical method posted much earlier is probably the best bet, but mind that it needs to be done with quite good optics as 0.1um is a quarter wave in the blue-green part of the spectrum.  Results will be... interesting with white light (add some more image processing, and use the fringes to your advantage?).  A laser will still require a solution that takes diffraction into account.

Looks like there are also well-known interferometric methods to measure cylinders.  This method may not be effective if the surface finish is poor or contaminated.  (If the surface is rough, a longer wavelength could still be used, in which case you do need a more sensitive phase detector.)

Tim

[IconicPCB]

Commercial equipment uses linear CCD camera with monochromatic orange ( neon)? source.

[Henrik_V]

Wire between two wheels (hardened steel, good ball bearings) , one wheel on a spring loaded (LDVT) length sensor.

I feel like people don't really grasp the scale of this problem. Best commercially available ABEC-9 ball bearings have guaranteed runout of  2,5 um. That's 25 times worse than wanted accuracy!

not fixed to a ball bearing, how a bout a bush bearing?
And what do you get? A nice periodic trace   Slow solution is a low pass , or couple both wheels (chain with 3 wheel and an encoder .. lockin )

Shure a optical solution will beat it, but will it be that simple and cheap?
 
Another idea ...  a defined length of that wire with a defined tension (wheels and weigths ) give a diameter related resonating string   It's always nice to convert a problem to a counting one 

And I would try a 4 point resistive approach.. just out of curiosity

[T3sl4co1l]

Another idea ...  a defined length of that wire with a defined tension (wheels and weigths ) give a diameter related resonating string   It's always nice to convert a problem to a counting one 

Oh, I forgot about tension... I was thinking do it without stress, using the material's elastic modulus and mass to set the resonance.  See:

If the density and elastic modulus is consistent, the wire could be secured between two rigid points and its mechanical resonance measured.  This could be done in continuous motion with some difficulty.

That makes it even worse.

So then what, do it between two tensioners?  Man... the sources for error are beginning to boggle the mind on this one.  Okay, nevermind this, then.

And I would try a 4 point resistive approach.. just out of curiosity

That's not actually bad.  Obviously, it still depends on material properties: temperature, composition, stress state (work hardening increases resistivity), and that your probing length is consistent (or measured simultaneously, to get a ratiometric measurement).

Note that you cannot get a perfect cylindrical electric field, because you can only make point contact to the side of the element.  At best, you get a circular contact from probes all around, but this gives a cone shaped electric field at the end.

The length of that cone (or for a single point, a more blobby field shape, or any other superposition of them), will be dependent on the width of the wire.  You must solve for this field in order to get the correct equivalent length.  Even if the contacts are a meter apart, a 0.01% accurate measurement requires 0.1mm accuracy on those contact points!  The diameter is much more than this tolerance, therefore the field at the contact points must be known, or calibrated out!

By the way, you definitely wouldn't use mercury for a continuous, surround contact: it will inevitably wobble and bounce with the wire's motion, moving the point of contact; and its resistivity is ten times that of iron, so you are actually making the electric field condition that much worse!  (Nevermind the risk of contamination. )

Tim

[Henrik_V]

You don't need a surrounding contact... repeatability (as a mean over an acceptable time) is fine
And your point contact uncertainty is relative to the length you measure ..  3mm to 0.1µm is 30 000 parts.. not tooo worse... even if it's a square relationship ...
Temperature is always a challenge ..
I remember qualifying mechanical measurements  on an 'endless' steel band some tens of µm thick ... 20 years ago ...  in a former live ...

And for a d=3mm wire  0.1 µm on a production floor (not a clean room fab) is a pious hope, 
on the other hand: a well specified copper wire .. mmmh  .. could save money if you have tons daily .. but then a (keyence/µ-epsilon) optical system has a fast ROI 

     

[Vtile]

Use the idea from herr. Thomson. Shoot it with laser and measure the shadow from the wall ten meters away.