Part Detection Engineering Problem (a challenge actually...)

[talosinnov]

I'm reaching out to the eng-tips community to share a problem I'm working on to get some help. (see attachment)

The problem that I'm working on involves measuring the relative height on the left and right bottom edges of a tile as it moves past a detector arrangement. I need to know the height difference to a resolution of at least 0.0254mm (.001 inches). I have further explained the problem in the attached PDF.

I am assuming that some of the optics engineers on this site might have come across something like this in the past and would have a couple obvious-to-them suggestions that would point me in the right direction as I try to design the solution.

Cost & dimensional constraints have made the solution to this problem beyond obvious to my Mechanical Engineering background. Simply casting a shadow onto an off-the-shelf Si photodiode array from companies like Hamamatsu is a far too expensive solution ($200 per sensor). This also requires a lot of space to get the resolution I need.

http://www.hamamatsu.com/eu/en/product/category/3100/4001/4204/index.html

I believe I need some optics to expand/condense the lightsource (signal) to get the resolution I need in the space I have. An EE friend has suggested a linear array of surface mount photodiodes would be the best sensor arrangement for cost and resolution... now I need to get the signal to them in the space I have.

Many thanks in advance for your suggestions!

 

[IRstuff]

I can't really understand your drawings and terminology:
> When you say height, are you referring to thickness? Seems like you are, but just checking
> Your little insert didn't make sense to me, re. "bad," "good," the blue chevron thingies don't compute yet.
> Why are there two lasers?
> Have you considered more brute force approaches, like a Dektak profilometer?
> I don't think your price numbers are achieveable, $5 price equates to $1.25 cost, and the sensors need to come with a processor; the cheapest Arduino style processor is more than a $1
> Why does it need to be that cheap? It looks to be some sort of manufacturing tool, but I can't see my way to 50,000 units/yr.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[MikeHalloran]

The blue chevron thingies seem to be partial cross sections of an edge of the test article, suggesting that it comprises two truncated rectangular pyramids bonded 'top to top'. The up and down motion during translation suggests that both outer faces are being ground, and the subject sensor array is intended to monitor that the stock removal from both outer faces is equal-ish. The drawn angles on the chevrons suggest that the edge to be detected is sharp. Proposing optical measurements suggests that it is also delicate. ... and hopefully opaque to the wavelength of the laser.

I'm having a hard time with objections to the cost of what seems to be production tooling. Maybe talosinnov is new to industry?

I'm also thinking that diffusing the laser beam is unnecessary of both lasers and both sensor arrays are reciprocated in the up/down direction in sync with the test articles, which may require using the arrays as tracking sensors while doing differential edge detection to fulfill the primary edge match detection mission.

It's an interesting problem; probably less so if I ever really understand it.


Mike Halloran
Pembroke Pines, FL, USA

 

[IRstuff]

Maybe a real picture or real drawing might shed some light. I also have difficulty with the concept of "measuring the relative height" given that any nonuniformity on the conveyor or the bottom of the tile could result in a relative height difference.

In any case, vertical measurements would seem to dictate sensor like rangefinders, which are in the realm of several thousand dollars. Measuring from the side, using cameras would be in the realm of several hundred dollars.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[talosinnov]

Thanks so much for the responses. I have attached answers to hopefully all of the questions you posed above and also added some 3D models to clear up the confusion from the original posting.

The crux of the problem is that if I simply cast a shadow with the edge of the part onto a surface mount sensor array - the sensor array will need to be a foot away (or greater) from the part to have the sensitivity I need. So I am sharing the problem with you to see if you can think of something more crafty or can recommend how I can shrink this distance in order to get the sensitivity I need.

One way to shrink the distance might be to reflect the signal (laser chalk line) back and down (I have more freedom in the enclosure vertically) from each chalk line to get the length I need but to collapse the overall width of this "module" in my product. This should all make more sense when you see the attachment.

Thanks again...I look forward to hearing your suggestions.

 

[IRstuff]

OK, much clearer, but:
> Is it a single measurement per tile, or a continuous measurement?
> Are the tiles reflective at all?

I would suggest looking at some sort of capacitive measurement,

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[IRstuff]

> How do you control the tilt of the tile?

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[talosinnov]

Tiles are not reflective.
Each tile might have 3 measurements along the length of the tile ... Continuous is not critical.
Capacitive suggests some sort if contact...even if just to set a spacing. I would prefer no contact.
The tile is firmly clamped and this controls the tilt.

 

[MikeHalloran]

Replace the sensor arrays shown with somewhat convex cylindrical first surface mirrors, or somewhat tilted planar first surface mirrors, and move the sensors 'down' below the mirrors.


Mike Halloran
Pembroke Pines, FL, USA

 

[IRstuff]

http://www.lionprecision.com/capacitive-sensors/

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529