Separate from that is the 317's initial voltage tolerance. This is the variation in the internal voltage reference over the operating temperature range and from one chip to another. The datasheet says it is +/-4% over temperature. It also says the output regulation error can be over +/-1% on top of that.

Thanks for sharing. If I use 0.1% shunt resistors, and hand select the LM317, I might give this a shot. I could use a larger voltage if that would help, say 7.5 or 9v, perhaps lithium chemistry...

If I buy a few dozen LM317 to hand sort, which manufactures should I choose, Texas Instruments , STMicroelectronics, onsemi, Analog Devices, or?

 

If you use a ratiometric measurement as suggested in post #12 by measuring the voltage and current at the same time, then the accuracy/stability of the current source is not critical.

Where do I measure the current/voltage, at the current supply probes?

The problem with this is now I have two measurements to take down, calc/convert. If I can get a single reading, millivolts, then eventually, an operator can grab a reading, compare it to a chart, and make the necessary machine adjustments without an engineer needing to always intercede.

Although, I am open to this approach if I find that the needed accuracy is unobtainable otherwise.

 

Where do I measure the current/voltage, at the current supply probes?

See below

If I can get a single reading, millivolts, then eventually, an operator can grab a reading, compare it to a chart, and make the necessary machine adjustments

You should be able to use a ratiometric voltmeter/ammeter such as this, to simultaneously measure both the voltage and current to give a single ratio measurement, which can then be referenced to a chart.

Your measurement accuracy is thus mainly determined by the meter accuracy, and independent of the current accuracy.
You could probably just use a power resistor in series with a voltage power supply, or a power supply set to the constant-current mode to generate the measurement current, if desired.

 

This thread is ridiculously lengthy for a premise that won't work in any case.

It doesn't matter how much "precision" You have,
a tiny-single-point-of-contact, on the surface of a sheet of Foil will
not reliably give any indication of the Foil's thickness.

You will need 2 points of contact, that preferably,
have a precision-fixed-distance between them of around
100X the average the average Foil-Thickness being measured.

Then getting a consistent Contact-Resistance with the Foil is the next challenge.

2 plastic Tubes, each containing a spring-loaded, gold-plated-ball, to contact the Foil,
could be used to insure reasonably repeatable Contact-Resistance.

Then, the next major barrier would be the surface that the Foil will be measured on.
The exact same surface will have to be used for each measurement,
with a clean, hard-Rubber surface, probably being the best bet for consistency.

The Gold-Plated-Balls should be cleaned before each use.
The Foil surface must be impeccably clean as well.
The Probes must be positioned equidistantly from all edges of the Foil.

It ain't about the accuracy of the Current-Source ..............
And, You would have far fewer problems using
a standard 9-Volt-Alkaline-Battery for a Power-Supply,
and a single Op-Amp set up as a Current-Regulator.

A ~$5.oo Digital-Panel-Meter,
will then give You reasonably repeatable numbers that are
directly correlated to the thickness of the Foil being tested.
A single-point probe will not be consistent, if it works at all.
.
.
.

 

It doesn't matter how much "precision" You have,
a tiny-single-point-of-contact, on the surface of a sheet of Foil will
not reliably give any indication of the Foil's thickness.

That is not my understanding of what is proposed.
The 4-probe measurement should work well.
Two to apply the current and two to measure the voltage drop across the foil.
That way contact resistance is not a significant measurement factor.

 

Trimming a resistance to an exact value by means of a higher value parallel resistance is a well known method that allows quite fine adjustment, and it can use resistors of a much lower power rating and tolerance, as well. for 5 ohms , using 5 resistors marked 27 ohms, all in parallel, plus a single 100 ohm adjustable in parallel would allow easier adjustment using cheaper parts.

The higher voltage current supply does not need to be regulated, either, although it probably should be full-wave rectified and have an adequate filter to avoid having much ripple and noise.

And the "four terminal arrangement" is commonly called a "Kelvin Connection", and it is the only reasonable scheme for accurate resistance measurement in an industrial production environment.

 

There are many significant reasons that Commercial Thickness Measuring Devices
do not use the above proposed method, and cost thousands of Dollars.

I previously outlined a possible DIY method that should prove to be fairly accurate, and repeatable,
but it's not gonna happen with a single-point, hand-held, "Pocket-Clip" style Probe.
.
.
.

 

Certainly there are other very good schemes for measuring thickness, almost all of them better than a resistance scheme, which, by the way, is affected by the width of the material between the connection points. (Sheet resistance effect)
But using the 3 terminal regulator to set resistance measuring current is a stable and reliable way to check low valued resistances, which several of my systems have been monitoring fuel injector harnesses for over 20 years at a major auto company engine plant. Of course, that is checking harnesses using the 4-point connections, not checking foil thickness using surface contacts.

 

There are many significant reasons that Commercial Thickness Measuring Devices
do not use the above proposed method, and cost thousands of Dollars.

I previously outlined a possible DIY method that should prove to be fairly accurate, and repeatable,
but it's not gonna happen with a single-point, hand-held, "Pocket-Clip" style Probe.
.
.
.

Sorry, but, "resistance per square" is commonly used for estimating thicknesses in foil-rolling and sheet-rolling plants. Some companies even use it as a specification for foils and thin sheet. Test methods require the supply current to go up significantly with thicker samples (10 to 100Amps) to get a stable voltage drop with the 4-point probe. Resistance testing shows voids and contaminants. The same resistance per square is also the common method for testing conductive inks and thickness of copper plating over some conductive inks.

 

I couldn't find any company selling "Current-Based" Testing-Equipment,
plenty of other very different, and even proprietary methods, but not Current.
It was an interesting Search.

~10 to ~100-Amps doesn't sound like measuring "Foil",
the "Foil" would just vaporize instantly.
.
.
.

 

I couldn't find any company selling "Current-Based" Testing-Equipment,
plenty of other very different, and even proprietary methods, but not Current.
It was an interesting Search.

~10 to ~100-Amps doesn't sound like measuring "Foil",
the "Foil" would just vaporize instantly.
.
.
.

That's why I said, "and thin sheet". It's really hard to discuss a topic with someone who only reads the first few words of each sentance.

 

There are many significant reasons that Commercial Thickness Measuring Devices
do not use the above proposed method, and cost thousands of Dollars.

I previously outlined a possible DIY method that should prove to be fairly accurate, and repeatable,
but it's not gonna happen with a single-point, hand-held, "Pocket-Clip" style Probe.
.
.
.

I plan on implementing a FOUR POINTS of contact scheme (Kelvin), that informal testing has shown the way forward in repeatability and the currents needed to achieve the necessary resolution across the full range of our products.

I'm conceptualizing a tool that is hand-held, a bit cumbersome, and battery operated. NOT a "pen-light" sized, single point of contact, shirt-pocketed tool.

We already have several other "COMMERCIAL THICKNESS MEASURING DEVICES" for testing, which cost thousands to implement, & offers a multitude of opportunities for generating inaccuracies. Useful as they are in concert to provide quality assurance standards to the clients, they live apart from the factory floor, and do not offer immediate results.

I need to provide machine operators an instantaneous go or no-go feedback.

 

Definitions .........
Thin-Sheet is not Foil, Foil is not thin-Sheet,
At some point one becomes the other.

I still haven't seen any references to an existing Commercial-Product using
Current as a measurement method, especially using a hand-held, single point of contact, device,
and I spent a good ~45-minutes looking, and I'm fairly competent at finding what I'm looking for.

Of course that doesn't prove that it doesn't exist, I just haven't seen it done yet.

If it's so pervasive, how about some links to some manufacturers
selling Commercial-Devices which operate on this principle ?

The above comment came in as I was writing this,
and sheds some light on the situation ............

"" ........... & offers a multitude of opportunities for generating inaccuracies. ""
This is exactly my point, repeatability.
.
.
.

 

Certainly there are other very good schemes for measuring thickness, almost all of them better than a resistance scheme, which, by the way, is affected by the width of the material between the connection points. (Sheet resistance effect)

In my investigations, the "sheet resistance effect" is negligible for my application. I was much concerned initially that this effect would play havoc with my attempts at implementation, particularly when nearing the edge of the sheet. Just as bypassing a 1 ohm resistor with a parallel 1K ohm resistor has negligible yet non-zero effects, so similarly with sheet resistance.

 

"resistance per square" is commonly used for estimating thicknesses in foil-rolling and sheet-rolling plants. Some companies even use it as a specification for foils and thin sheet. Test methods require the supply current to go up significantly with thicker samples (10 to 100Amps) to get a stable voltage drop with the 4-point probe.

We're using eddy-current for our inline production guidance, and let me say, the promises of the of this non-contact induction method are larger than the chest on Dolly Pardon, and similarly, have yet to reveal themselves.

 

Thin-Sheet is not Foil, Foil is not thin-Sheet,
At some point one becomes the other.

Right, and that's why different currents are needed for different thicknesses. If you think I meant 10-100 amps are used for all thicknesses, you made an incorrect assumption.

especially using a hand-held, single point of contact, device,

Could you discuss why you keep talking about hand-held, single point testing when nobody else on this thread is talking about that? I have no doubts that you'd have trouble finding anything about single point resistance testing.

 

If you use a ratiometric measurement as suggested in post #12 by measuring the voltage and current at the same time, then the accuracy/stability of the current source is not critical.

Missed that. Very nice.

ak

 

Something else I might have missed - how many of these are you building? Is this for in-house QC work, a commercial product for sale, or other?

ak

 

Something else I might have missed - how many of these are you building? Is this for in-house QC work, a commercial product for sale, or other?

ak

In-house production work only. (QC aint my dept.) I got the idea from a 4-pt. probe @100mA that I made years back to make my own power resistors.

 

Why do I have a feeling of DejaVu? Why does this seem recently familiar?

Oh wait, I think I know.

The Van der Pauw (VDP) method measures the average resistivity and Hall coefficient values on thin samples (approximately two-dimensional) by applying current and measuring the voltage along the sample perimeter . This allows irregular shapes as well as more traditional structures to be evaluated.

So are we done with uV? What you are looking to do is not easy. In addition to the probe spacing the amount of force the probes exhibit on the material is critical. Additionally for true results the polarity of the probes should be reversed during measurement without disturbing the probes position or pressure. The results of forward and reverse need to be within 3%. The systems which are designed to measure what you want to measure are not inexpensive home brew projects.

As a current source while the LM317 may be considered I see it as a poor choice. Your current source for running Kelvin four wire test is going to need to be rock stable. Lacking stability your voltage measurements will be all over the map. While I haven't a clue what resistance you expect to encounter on your film you need to consider current source compliance voltage. Compliance voltage is the maximum voltage a current source will go to in its attempt to source the programmed current. Has that been figured in? This is why anticipated resistance of the load needs to be known when designing a current source.

Yes, as mentioned measuring film thickness for QA acceptance is an expensive proposition as is measuring film resistance. You can't make a silk purse out of a sows ear. This is a link to some of the suggested systems available. Sorry but I do not see an easy hand held solution. However wish you the best. Again, looking at your anticipated resistance numbers don'y forget to consider compliance voltage on your current source.

Ron