From Scratch - Designing a tool to calculate "dip"

Hey everyone,

So glad I came across this forum. I hope that you can help me with this project and I can get it built. I'm working on a tool to help me at work be more accurate and more efficient as I do my quality inspections.

Essentially what I need to do is design a tool which will help me to accurately and quickly determine two measurements: dip and warp. In order to do this I intend to build a frame which I can put the varying products in, and have a vertical row of measuring sensors (still need some advice on what the best sensors to use might be... the surfaces will not be reflective). I figure with 10-20 data points (adjustable based upon the height of the product) I can plot a curve and determine the depth of the concave surface. I need the resolution to be accurate within .001inches. What will make this difficult on the math end of things is that the surface being measured will not always be true to the y-axis, and so the top and bottom points will need to serve as the reference against which all the calculations are done. Here's a rough sketch...View attachment 60719

Thoughts on type of sensors necessary, price point of those sensors? What type of "brain" I will need so that it can record and process these measurements 100x/second as I rotate the item, store the greatest depth in the surface, and display the max on a small digital screen?

I'm sure I've left a lot out, and I know I've been slightly ambiguous, but ask away, or direct message me and I can try to give more detail.

You have described dip pretty well. How about warp?

How non-reflective is the surface? If it is completely non-reflective, that could be a problem.

You mention a level of precision of 0.001", which is consistent with being from Western Mass. Welcome to ETO. What is the total dip?

Have you read about using knife edges to cast shadows and measure dips?

John

Sounds like you might be able to use a LVDT (linear variable voltage transducer) I used to work on some Lasers from MA. (Coherent General, Sturbridge) that used them to detect the edge of the sheetmetal that was being cut.
https://thesensorconnection.com/cat...pring-loaded?gclid=CPmB-Y21_a0CFU2b7QodoRXKtA
Here is just a sample found on a quick search. Pick your tolerance and travel specs.

It sounds to me, and I may be missing the objective, like you want to plot a concave surface. There are a number of ways using different sensors to do this ranging from somewhat expensive to I can't believe it cost that much.

Likely the least expensive sensors that will return a good accuracy and uncertainty would be Linear Potentiometers from a company like BEI. The BEI 9600 series being low end. Since you don't mention the depth this is pretty generic. This is an example of the BEI 9600 series I am speaking of. They are inexpensive at about < $25 USD per copy.

Next there are higher end versions like the BEI 600 series and here is an example of them. At roughly between $250 to $450 USD per copy you can see where ten of them can get a tad costly.

BEI is one example, you can also look at ETI Systems for like items as well as Honeywell MLT Series Linear Position Transducers.

That would be an inexpensive starting point. Depending on how accurate you want to get running surface profiles you could move to LASER Interferometer sensors. I use the Micro Epsilon line with a focus on these units but be advised these units run about $3,500 USD per copy so if you want ten it won't be cheap.

Charting the data becomes easy once you have a sensor head configured and designed. There are also likely off the shelf solutions out there unless your need is very unique. Anyway, this may give you a few ideas. Hopefully some other members have some thoughts that will meet the uncertainties you need.

Ron