We need to check the thickness of a fiber board.
The thickness shall be within + / - 2 tenth of a mm. We could maybe go to 3 tenth of a mm. We have been talking about different ways of doing this - but not found the right way yet.

What do you think of Ultrasound sensor. Ultrasound manufacturers do claim to be able to measure +/- 3 tenths of a mm.

 

Is the board moving when you want to measure it, or stationary?

Can you touch the board, either with a probe extended periodically or with wheels that are in continuous contact?

Is the material generally smooth, without miscellaneous projections that you don't want to include in the thickness measurement? If smooth, how flat is each side and how parallel are the sides?

Is there dirt or other contamination that would accumulate on a mechanical measurement mechanism and have to be cleaned periodically?

Do you have at least periodic access to the location in order to "measure" pieces of pre-known thickness for calibration purposes?

 

> Is the board moving when you want to measure it, or stationary?
STATIONARY.

> Can you touch the board, either with a probe extended periodically or with wheels that are in continuous contact?
YES.

> Is the material generally smooth, without miscellaneous projections that you don't want to include in the
> thickness measurement? If smooth, how flat is each side and how parallel are the sides?
A LITTLE ROUGH ( AND SOFT) SURFACE, THE UPPER AND LOWER SIDES ARE PARALLEL.

> Is there dirt or other contamination that would accumulate on a mechanical measurement mechanism
> and have to be cleaned periodically?
NOT REALLY, A LITTLE DUST IN THE AIR.

> Do you have at least periodic access to the location in order to "measure"
> pieces of pre-known thickness for calibration purposes?
YES.

 

So how about a mechanism that releases spring loaded probes simultaneously against the upper and lower sides of the board, with the board thickness measured by a linear potentiometer attached to arms on the opposite side of a pivot from the contact points (to amplify the measurement)?

The probes can be of such a cross-sectional area and the spring force low enough that you won't leave a divot in the material.

BTW, what is the nominal thickness of the material, or are there different varieties?

 

There are also more precise probes that can measure that kind of variation directly, but they can be expensive.

 

I'd go with two LVDTs attached to a linear guide with rollers on them, a wheel wiper to keep dust off of the wheels, and a compressed air screen and maybe mechanical brush/wiper in front of the wheels to keep large contaminants out. If your thickness variation tends to be slow you can perform averaging on the signal to smooth out any dust that gets into the system.

Keyence has some prepackaged spring loaded LVDTs with linear ball bearings inside hooked up to a small display controller unit. I've used these where I needed to do contact measurement and wanted to have a "Good/Bad" signal coming out. The amplifier had built in digital thresholds. I'm not sure if it can combine two LVDTs though, so it might not be appropriate. If you have analog inputs on your controller you can use those instead and just do the combining/averaging/thresholding yourself in code.

The reason I'd go with a contact measurement is that ultrasonic or laser sensors will depend on the surface properties because it is a reflective process. If you have the room to experiment, get your sensor rep in (Keyence, Omron, Banner, Baumer, etc.) and usually they have demo units you can try on the spot with your materials. If you want to be at 0.2mm and the sensor is rated for 0.3mm don't go there because you will have other errors on top of the rated sensor error which is often for idealized targets.

KEJR

 

There is a lot of experience in the pulp & paper industry with measuring sheet thickness. I'm surprised their is no off-the-shelf measurement sensor for what you want to do. Is there some reason you don't want to use a commercial solution like this ?

http://www.acuitylaser.com/AR700/common-applications-plywood-thickness.html

Rob
www[.]lymac.co.nz

 

Ken,

I was thinking you could do it with one sensor if instead of measuring between each surface and some point fixed to the real world, you took the measurement between points attached to the suspension of the upper and lower roller. It's more work mechanically, but it would takes less sensors and would yield a single thickness signal.

 

Yes, I agree If the material and spring force of measurement wheel is stiff enough to force the material against the fixed wheel. Since the dancer wheel is going to need some sort of force anyway with a contact sensor then you could design it to apply enough force to the workpiece.

Good point, thanks.

Ken

 

> Yes, I agree If the material and spring force of measurement wheel is stiff
> enough to force the material against the fixed wheel.

Who said anything about a fixed wheel? I was going to float both upper and lower, within hard stop limits, to allow for the web not to be flat, then make a measurement between the two floating wheels. Depending on the stiffness of the material, you may also want to counterweight the entire assembly to avoid flexing the sheet.

If you have a fixed wheel, that's just a special case of the surface-to-real-world measurement.

 

Steve,

I'm with you now. I think I had the case of scanning your message too quickly. Sorry about that. Sounds like a good approach especially if the material does tends to warp. I'm afraid that without looking at the entire application we can only throw out possible solutions.

Ken

 

No problem, Ken!

Another trick, if you were measuring sheets whose nominal thickness varied by an amount large compared to the desired measurement of thickness error, would be to use calibrated setup spacers the same thickness as the nominal stock thickness, between the suspension (connected to the sensor) and one of the wheels.

That way you don't waste any sensor or signal resolution on the wheel travel just getting to the different board thickness. So say the product is particle board, in thickness from 3/8" to 3/4". Then the sensor is exposed to the same (small) amount of motion when measuring error on 3/8" stock with a 3/8" spacer as it is when measuring error on a 3/4" board with a 3/4" spacer. Otherwise the sensor has to encode the range we're interested in (probably +/- 0.010" or 0.020") in a signal that represents a total motion of 0.400" or so, or even more depending on the range of materials to be measured.

You could even equip the spacers with binary coded plugs so the controller could offset the measurement for the installed spacer/ensure that the correct one was installed.