Displacement sensor

[eugeni]

Hi,

I do have a project. I need to measure a gap of 100 microns in real time. One surface is metalic and moving the other is plastic. I tried a proximity sensor (inductive) with a analogue output (it is not a displacement sensor), i wrote a program to make it linear and is working fine as long the temperature is constant. It seams that i need a real displacement sensor, but I have some :-( budget limitations. There is a displacement sensor 0 to 5 mm range and of 5 to 8 mm diam. not expensive?. I need only the sensor (not a system). Thank you for any answer and/or input you have.

Eugen

 

[nbucska]

Capacitive sensor ? <nbucska@pcperipherals.com>

 

[buzzp]

You should try banner engineering, automation direct, gem sensors. Did the proximity switch not meet accuracy specs within your temp range? Seems they should perform well, is it current output or voltage? Maybe an ultrasonic or laser sensor would work. Ultrasonics are not too terribly expensive and used for distance. We use some in hydraulic cylinders (6&quot;stroke) and for detecting an objects distance from another (accuracy depends on range-span, we get 15 mils accuracy with distance between objects of 7-12&quot;, $300US). I am not for sure they make them for these small ranges but I believe they do. When I checked lasers they were quit a bit more expensive but accuracy was awesome.

 

[eugeni]

Dear Sirs,

Thank you for the answers.

1.I forgot to tell you that the gap i have to measure is filled with some water based substance. The compositin of this gel like substance changes in time. I presume that capacitive sensors are not apprpriate for this application.
2.From the same reason i can not use the ultrasonic sensors.
The suond speed is diferent. I think the precision of the ultrasonic sensors is limited by the wave lenght.I need a range of 5 mm (20 mils)with an accuracy of max 5 microns (0.2 mils).


Thank you
eugen

 

[buzzp]

I think the ultrasonic would work by changing the oscillation freq of the sensor. But it would not fair well with the substance between them. If you could lower the freq then it may (like transmitting through water at lower freqs for further distances) but I believe this would have a huge impact on the accuracy at the levels your talking about.
What if you could use a sliding pot and attach this between the two pieces? Is this possible? You could put voltage across the pot and measure the voltage across the wiper to determine the position.
Cylinders have this option but I don't think they make one this small.
Something like the above two is all I can think of to do what you need.

 

[melone]

Can you characterize the temperature factor in your current setup and modify the results based on temp? Then all you would have to do is put a thermocouple on your sensor and modify your results based on the temperture seen at the sensor. You might get lucky, the relationship might be linear.

 

[nbucska]

Could you send a sketch and spec's ? (range, accuracy,
composition of gel, etc )? <nbucska@pcperipherals.com>

 

[eugeni]

Dear sirs,

1. I am not lucky. The relationship is not linear. The temperature is rotating the characteristic somewhere in the midle of it. I could use this point for the calibration but this implies a very fine tunning. I could compensate using some math, but there is no guarantee that the next sensor will behave similar.
2. The gel is waterbased and not transparent. The gap is 2 mils. The gap error should be 0.1 max 0.2 mils. The gap is between a stainless steel plate (the moving one) and a plasic surface (the one with the sensor) Using a uP with ADC I can do all the readings within 0.1 mils. The things becomes ugly when temp. changes (and it is changing).I am using a proximity sensor and not a displacement sensor (expensive). The proximity sensor in some degree has a analogue output (I am using the slope).
3. The sensor has to be non - contact.
4. For the ultrasonic sensor (if i remeber corectly) there is a relation between the minimum lenght and the wave lenght (1/4?).At 1000 ft/sec (in the air) the frequency should be too high in order to have the accuracy of 0.1 mils

I think the use of inductive displacement sensor is a must(at least is designed for this and is compensated). I see no other option (for now).

Dear sirs , thank you very much for your inputs.

Eugen Iacob

 

[nbucska]

I don't understand your setup. If one part moves in the liquid, there must be a way to measure the distance outside of the liquid, too.

<nbucska@pcperipherals.com>

 

[djs]

Hello;
have you tried a proximity sensor with an analog output. For instance Allen-Bradley has such a sensor:

http://www.ab.com/catalogs/c114/2indprox/206986.pdf

page 17
These units come in different sizes, and have linear output with respect to displacement.

 

[melone]

Build an enclosure around your sensor to keep it at a constant temp.

 

[nbucska]

Can U FAX me a sketch ? (303)-439-8651 ATT:Nick <nbucska@pcperipherals.com>

 

[IRstuff]

Are you allowed to mount things on either or both surfaces?

> camera? place camera to side with macro or microscope lens and simply look at displacement in image directly.

> light position displacement sensor? bounce light from source onto moving part and locate reflected spot on CCD. Lateral motion of spot on CCD proportional to tangent of angle and separation

> variation of above. semi-calibrated clip-on light source mounted on moving part. Use CCD on the side and monitor light spot position on CCD.

> Hall sensor? not sure it can be run as straight displacement sensor. Put magnet on moving part and sensor on fixed part, but needs calibration and gain characterization.

> Hall or eddy current/magnetic? with something like an inside-out LVDT

> kludge. mount magnet on moving part. Hall sensor on linear stage with encoder. move stage to maximize hall output and read encoder position.

TTFN

 

[melone]

nbucska, are asking me for a sketch? If so, I will try to work on it over the weekend, but the basic principle is questioning the assumptions. If the ambient temperature modifies the measurement, remove the ambient from the equation. By making a small temp controlled enclosure around the sensor, he might be able to use the existing sensor (and save all of the development / calibration / etc. time). It already sounded like eugeni had a sensor that was working just fine in his application except for the fluctuations caused by the ambient temp.

I understand that not only will he need to build an enclosure, but he will have to include either a heater, fan, and/or cooling unit to this box. Perhaps, if the temp variations are small enough, he might be able to get away with using a fan, thermocouple, and a micro with A/D inputs. Then he can turn on the fan when it gets too hot, and shut it off when the enclosure ambient is acceptable.

Obviously, this is a lot of work, but he seems like he is willing to consider non-traditional solutions. It would have been easier to just find a sensor that can handle the temp with the kind of accuracy that he wants, but it looks like that isn't going to happen.

Sometimes, some of the most absurd ideas cause a light to go off in your head. I am not saying that this is the best solution, but it is definitely a different one. I just hate getting locked into thinking that there is only 1 solution to a problem.

Let me know what you think....

 

[eugeni]

Dear Sirs,

It seams that the path gets clearer (I' ve seen the light... ). I will have a thermistor on the sensor, a initial calibration (done only once - 3 degree and no local maxims or minims.) It is like &quot;factory setup&quot;. And a displacement calibration, done periodically. Actually it is two variables calibration (a surface in three dimension). Did first test and is running nice. I did the math and the errors are acceptable. Now, the real life can be… different.... Tests...
One reason to choose the proximity sensor is that I am limited in space. The Displacement sensors are more expensive and more of it, they are > 1/4 (8 mm) in diameter.
I want to thank everybody it was very helpful and insightful.

Once again, thank you
Eugen

PS sorry, English is not my first language so inherent errors and misspells occurred (for sure) Appreciate the understanding

 

[nbucska]

Hi Eugeni:

Can U fax me a sketch ? <nbucska@pcperipherals.com>