hi everyone,

i have application where I need to detect vertical position of the object.

the arrangement is rectangular U-shaped frame that is about 10" wide (~250mm) and about 4" high (~100mm) and the top side is open.

the object is rod that can be lowered into this "wide-U shaped frame" and it is about 1" diameter (~25mm). it is free floating and can be lifted out of the U-frame. As mentioned I only care about vertical position (not horizontal, not diameter etc.). I cannot use camera to look at it from "front" so I was considering light-grid type of arrangement (could be discrete sensors or "light curtain" like Balluff BLG-1 for example). Accuracy does not have to be great 3-5mm should be fine.

Any ideas?

Regards...

It sounds like you found the solution with the BLG-1. Is this not an option?

Can you touch the part with a probe arm connected to an encoder or something like that?

Can you sweep a laser distance sensor slowly over the area on a slide and watch an analog input for a max value?

what i gather from Balluff literature, BLG is used to check size of the object (diameter).
If the object size is larger than sensing range of the BLG so that BLG only looks on part
of the object (one edge), then BLG output can be used to tell position of the object
(it is actually measuring "diameter" from the edge of the sensing field to edge of the object
like in example shown on top). This is not what I need since in my case object is smaller than
sensing field (see sketch) and the situation shown on top is just one case. I don't think the
BLG output will change between other two scenarios (middle and bottom).
I could easily do this in program if I could read individual beams of the BLG.

I cannot touch the part with probe or anything. I cannot use camera because view shown below is
actually cross-section (object is very long). Looking from side (either from left or right side in image below)
with camera will not work because the depth (change in focal length).

Looking at you pictures gave me an idea.

What about a Banner Diffuse Retroflective that has an analog output something like http://www.bannerengineering.com/en-US/pro...E2%80%94Analog/ mounted on an LVDT arm.

You start at the bottom or top and move up or down until you get the strongest return signal and then read the LDVT.

Hope this makes sense.

Might even use a pair of S18U's, one to sense the rod and one to sense the firs sensor's location.

unfortunately cannot have moving parts since object is moving so I need to get the position information fast (at least every 5ms).

http://www.joescan.com/index.html

[attachmentid=7447] [attachmentid=7448][attachmentid=7449]

These scanners are used for lumber scanning.

I haven't used these in years, but I found them to be very good.

They can be programmed to send you the position of the lowest or highest beam blocked, or the number of objects in the beam, or the number of beams blocked. Just about anything you can think of. You can get the data serially, or from an analog board.

Right out of the box, the analog version we used were always exactly calibrated, and we had no failures to the best of my knowledge.

STI Valuscan Measuring Light Curtain

Paul

EDIT: Resolution may be too coarse...best resolution using double scan mode is 1/4"...

the Sti item looks very close to what i was expecting but it's too large.
i was not aware of JoeScan, picture of a guy on a log made me laugh, hehe...
unit looks very good but it's too large (4x4x33"),

available width is about 1.5" and I may get as much as 2" but height also has to be within those 4" (may go to 5", not sure yet) and can't have anything exposed (cable has to run through frame).
i need something that is like stick or pancake (1-2" wide, not 33") or mini light curtain. if this goes beyond quoting stage, i will just make my own PanicScan curtains but it's probably not worth for 2-3 units to start with.

Banner makes a "mini light curtain" that I recall does something similar to what you are describing.

In addition it may be contrasty enough (especially with the right lighting) to outright use an el cheapo vision system. Again, Banner has some pretty good stuff here.

Otherwise if you can get at it from another angle then I'd definitely consider laser distance scanners if you can pick off a reference point, if you could look at it from the top or bottom. In this case, IFM Efector and PepperL Fuchs would be the ones I'd go for.

thanks paulengr,

this is closest match so far (still on the chubby side but i guess this is as close as it gets without making something custom).