Design for a Homebrew Paint Thickness Tester

[StuartA]

I’d like to own a Paint Thickness Tester for use on car paintwork. There are some very cheap ones available on-line (using ultrasonics), but they almost certainly don’t have the resolution required to be very useful. My guess is that the ones I’d like to own cost more than I want to pay, but my gut instinct is that something adequate could be built at home for not very much money at all.

That lead me to search for detailed information on how they work, and ideally, a design to construct one.
The best I’ve been able to do is find a few interesting patents, which reveal quite a variety of designs, working at both AF and RF frequencies and measuring either frequency shift, or the extent of coupling via eddy currents.

I suspect that, starting from scratch, this could make an interesting but very time consuming project. I wonder if anyone has ever seen a design for a homebrew thickness tester?

S

[MosherIV]

Hi

Firstly, no I have not seen any home brew thickness testers.

I worked in the NDT test industry in the past, in ultrasonics.
The thickness meters have become commodity items, hence they have dropped in price.
The cheapest way to do thickness measurement is with ultrasonics, hence you will not find commercial solutions with rf.

You can certainly get resolution down to .01mm with a high ultrasonic freq, as far as I know.

If by "very cheap" you mean less than £80, yes I would avoid it. Anything decent will be at least that price. Check the unit can do the resolution (thickness) that you need first.

[StuartA]

Thanks for that insight. As you may know, there are ultrasonic testers on eBay for £10, but the claimed resolution is nowhere near what you need for paint. My impression is that anything less than ~£100 won't be worth having.

[Conrad Hoffman]

If you always shoot the same kind of paint, and are willing to do some tests for calibration, a capacitance measurement through the layer might do it. Put a 20 mm square plate or similar against the surface, then measure to the metal behind. Yeah, only works with metal, but the resolution should be quite good. The downside is the dielectric constant of the dry paint would have to be the same, and that might change with color and the degree of dryness. If it could do the job, the circuit could be a simple bridge with a calibrated pot.

[tunk]

Whey not get one of the cheap ones and see if you can hack or calibrate it?

[StuartA]

Given that you can buy a digital frequency counter for next to nothing now, the idea of looking at how an RF oscillator can be pulled by attaching some sort of capacitive or inductive probe alongside the tank circuit, and attempting correlate frequency shift with paint thickness, might work. I suspect that the devil would be in the detail.

Of course the simplest design of metal detectors which are around are based on having two oscillators, a reference oscillator and a similar one but with a probe, and they look at the beat frequency. Again, perhaps you could get a correlation between paint thickness and the beat frequency.

The cheap thickness meters on the market are ultrasonics and I have no background in that area. I'd feel a lot more confident about trying something based on RF measurements.

[duak]

Just a goofy idea here, but is thickness the desired characteristic to be measured - or is coverage, porosity  or something else more useful?  For example, is resistance (or conductivity) or the breakdown voltage a better indicator of the quality or better at locating thin spots or defects?

[Marco]

Capacitance will be far too finnicky for a general purpose device. Ultrasonic TDR using a chirp and looking for an impedance mismatch seems the best bet.

A search for ultrasonic chirp "cross correlation" didn't turn up any source code, I doubt anyone has made an open source device.

[MosherIV]

The theory is simple, create an unltrasonic pulse (chirp) and look at the returned signal.
The implementation however is massively complicated.

Your first problem is how to generate the utlrasonic pulse and couple it to the device under test?
Bear in mind that the same transducer that generates the pulse is also the same transducer that is used to get the echo!
As far as I know, no one outside the industry knows how to make the transducers.

The electronics for the pulse generation and echo receiver and processor is quite complex and difficult.

Then the analogue signal is digitised into a continuous ADC stream that needs to be processed for the peaks that denote a echo. The software must also deal with velocity of the pulse through the material (the paint and substrate) and then do the calculation for the thickness.

When I worked in the industry, I was part of a team of 8 engineers, mixed dissipline of electronics, software, mechanical and ultrasound expert!

[techman-001]

I’d like to own a Paint Thickness Tester for use on car paintwork. There are some very cheap ones available on-line (using ultrasonics), but they almost certainly don’t have the resolution required to be very useful. My guess is that the ones I’d like to own cost more than I want to pay, but my gut instinct is that something adequate could be built at home for not very much money at all.

That lead me to search for detailed information on how they work, and ideally, a design to construct one.
The best I’ve been able to do is find a few interesting patents, which reveal quite a variety of designs, working at both AF and RF frequencies and measuring either frequency shift, or the extent of coupling via eddy currents.

I suspect that, starting from scratch, this could make an interesting but very time consuming project. I wonder if anyone has ever seen a design for a homebrew thickness tester?

S

Yes, we built one around 1972. My memory of the details is a little hazy as it was so long ago and I was only a technician, it was designed by the chief engineer.

It was magnetic, so only worked on steel car parts.

The design was a small steel "E" core, like a transformer with two coils of wire, each coil wound around a leg of the "E".

It was calibrated using various thicknesses of non magnetic spacers such as paper or plastic between the gap from the "C" core and a small piece of steel which closed the gap in the "C" core providing a magnetic path.

I recall at the time, they tested the paint thickness of a  beautiful, deep and lustrous paint job on a Chrysler which measured 40 thousandths of a inch, and a commercial Holden utility with a plain white paint found to be 1.5 thou !

I'm guessing that it behaved exactly like a transformer in operation, was probably made from a small laminated transformer and so it should be trivial to make one

[m3vuv]

what about a banana gauge?,no electronics to go wrong too

[Kleinstein]

The capacitive way should be relatively simple to build, but the big unknown is the paint properties, that can change even with color.
There is no need to have a direct contacts to the metal, just a large area at the outside is enough and this can also shield the sensitive part. It would be at least useful for relative measurements, so to see if the paint thickness is uniform, e.g. to find hidden repaired areas.
The extreme version would use the painted area as the end of a cavity resonator and could in theory even measure the Er of the paint too.


The Ultrasonic way has a similar problem with the paint properties, though likely not as bad with different colors.#

The magnetic way sound quite plausible and possible, as essentially all colors of paint are magnetically not active. One way would be to to measure the inductance with a small U shaped core at a relatively low frequency, so that eddy currents (especially in the steel of the car) are not a big factor.  The other option would be to use a rather high frequency (e.g. some 13 MHz) and include the eddy currents. So the steel would be mainly seen as a conductive area, not as a high permeability.

[StuartA]

I've had some amazingly goods results from a modern, miniature 220uH inductor (one of the ones that looks just like a resistor).
Placing it on its side on plain steel, the measured inductance goes up to nearly 260uH, then sliding in pieces of 200um thick plastic sheet between the metal and inductor, the measured L falls back down again, useful out to at least 600um. It's quite insensitive to the presence of my hand. I'd say it probably gives me a resolution of ~10um.

The reason I want a gauge is for use in machine-polishing a 20 year old car, which has lots of swirl marks. You really do not want to polish through the clear-coat. You can only measure the clear-coat thickness using infrared, but those instruments are very, very expensive. So, I'm just probably looking at just setting myself an absolute limit of not polishing off any more than maybe 50um (and ideally, less), which should mean I'm still in the clear-coat.

[Kleinstein]

For just a relative measurement with clear paint, the capacitive way should also work. The clear paint should not vary that much in Er as it may happen with pigments. Measuring the capacitance change in the pF range is not that difficult.

For inductor way (low frequency), I would consider a U shaped core,  like some ferrite cores used in some common mode chokes with one part removed. The test would be using kind of a variable air gap inductor. One could also try an old audio tape or even floppy head.

[m3vuv]

I used to work in ndt on cross country gas pipelines,one of my tasks was to measure the coating thickness on the pipes,we had ultrasonic testers but mostly used a banana gaugesimple to use with nothing to go wrong.

[StuartA]

From what I can find out, the "banana" gauge was a device which used to be made by Elcometer, but their website makes no current mention of it. It worked using a magnet and a mechanism for detecting how much force was needed to pull the magnet off a painted surface. Your guess would probably be better than mine as to why it seems to have been replaced with electronic gauges.

Paint on modern cars seems to be typically around 100um (4 thou), consist of three layers and I'd guess that you really want a resolution of ~10um. This may be thinner than on infrastructure like pipelines and might be asking a lot of a banana gauge?

[techman-001]

From what I can find out, the "banana" gauge was a device which used to be made by Elcometer, but their website makes no current mention of it. It worked using a magnet and a mechanism for detecting how much force was needed to pull the magnet off a painted surface. Your guess would probably be better than mine as to why it seems to have been replaced with electronic gauges.

Paint on modern cars seems to be typically around 100um (4 thou), consist of three layers and I'd guess that you really want a resolution of ~10um. This may be thinner than on infrastructure like pipelines and might be asking a lot of a banana gauge?

I imagine that the 'banana guage' could work very well  providing it had the sensitivity needed, which means precision bearings in the force balance mechanism and so on. Cheap and nasty bearings would mean a lot of non repeatability.

Battery less for sure and probably simple but also fragile like all sensitive mechanical instruments I'll wager. Once dropped it was probably useless.

A perfect case for robust and cheaper electronic device replacement.

[MosherIV]

Hi. If you had good results with the inductive approach, why have you stopped?

I think that the inductive appraoch needs individual tunning/calibration for each inductor.
This means it is not cost effective to mass produce but perfectly ok for individuals to build at home.

Interresting to see so many approaches to the problem.

[techman-001]

Hi. If you had good results with the inductive approach, why have you stopped?

I think that the inductive appraoch needs individual tunning/calibration for each inductor.
This means it is not cost effective to mass produce but perfectly ok for individuals to build at home.

Interresting to see so many approaches to the problem.

I'm quite sure the 'Banana Guage" would also need individual calibration due to manufacturing tolerances including magnet strengths. In this regard it would be no different to a inductive meter needing calibration.

[Kleinstein]

The inductive and also some other meters would need individual calibration. This may be needed just once but possible even later. By providing a piece of coated steel, the calibration could be done by the user, possibly just before use to also compensate drift with aging. This would not add much costs.


I did a quick test with a tape head:  it did not work  , at least not really sensitive, but only some 10% change in inductance (tested at 1.6 kHz).

[StuartA]

Hi. If you had good results with the inductive approach, why have you stopped?

I've not stopped, but I now need to make or source some suitable calibration standards. I quickly found out that the clear plastic strip that I started with is not the same as paint loaded with pigments. There's a learning curve in making them!

In its day, the Banana gauge was probably the best thing around, and may well still be adequate for some work