Sensor to read the size of a laser shadow? Moderators: adafruit_support_bill, adafruit

[ianjohnson]

I am working on an extruder to make filament for 3D printing. It is important that the diameter of the filament remains consistent within .05mm. I can check this by cutting off sections to measure with a caliper, or try to use the caliper as the filament feeds to the spooler, but I'm looking for a better way.

I thought of running the filament under a line laser, close enough that the shadow cast by it is considerably larger than the width of the filament so that any variance in its diameter would be exaggerated. However a reasonable throw distance with sharp edges to the shadow works out to only about 1-2mm of change in shadow length to indicate the changes in filament diameter that I'm watching for. That's pretty tough to eyeball on a ruler, and throwing the shadow further to increase its size begins to create indistinct edges which reduces precision.

Is there any kind of sensor that can be used to track the distance between the ends of the laser line at each edge of the shadow, that would be DIYable? One challenge is creating a path for the filament that holds it absolutely still in relation to the laser as it feeds past. It's a little hard to track the width of the shadow by eye when both ends are moving back and forth. Something that can read the length of the shadow without one end being fixed in place would help.

[adafruit_support_bill]

I worked on a similar issue: Measuring the diameter of a red-hot steel rod as it was coming out of a rolling mill. We used a collimated beam (but a line laser should work as well) and a linear CCD sensor. You can find these sensors in scanners. I have seen them for sale from surplus sites as well.

[franklin97355]

This http://www.parallax.com/StoreSearchResults/tabid/768/List/0/SortField/4/ProductID/566/Default.aspx?txtSearch=line+sensor might be of interest.

[zener]

That is basically what I was going to recommend, Taos. Here is the sensor that is in that, in case you don't want the lens:

http://www.mouser.com/search/ProductDet ... rm=TSL1401

It looks like you give it a clock, then at some point give it a start pulse "SI" Start Index? Then it clocks out 128 analog values I think, that you read and decide what is going on.

[ianjohnson]

That looks like what I need. This is my first electronics/Arduino project so there is plenty to learn. I'm thinking the strategy is to read values from each pixel. If it is bright/saturated, go to the next. If it is darker, start counting, and stop counting when another bright pixel is encountered. I'll need some calibration step where it can present a # of dark pixels and have the user enter a known measurement that they relate to. Then it would take that relationship between pixels and mm to output a measurement.

I'm thinking of adding a second laser/sensor combo at 90 degrees to compensate for any movement of the filament in relation to the lasers. If the filament moves up or down in relation to the laser, the diameter would appear to change. A change in diameter would cause the first and last dark pixels would move in opposite directions (scaling). If the filament moves, the first and last dark pixels would move in the same direction (translating).

If the bottom sensor shows the shadow scaling but the sideways sensor shows its shadow translating, then the bottom sensor reading can disregarded. Likewise the bottom sensor can detect sideways translation and disregard the sideways measurement.

[zener]

You might look at some of the commercial sensors made for that purpose and see how they do it. I am guessing Keyence might have something. I have seen them before but I don't remember what brands they were. You want to search on "laser micrometer"

[flipperr]

I put together a prototype sensor for a similar purpose, using the Taos sensor and LED illumination. The plan is to use it to adjust the rate of extrusion on a 3D printer to compensate for variation in filament width. I found that it was possible to get good results with guiding the filament over the sensor. Still early stages, though...

http://www.thingiverse.com/thing:89044

[tldr]

this is probably daft, but could you interpolate the size of the filament just from the amount of light hitting a sensor behind it?

[adafruit_support_bill]

could you interpolate the size of the filament just from the amount of light hitting a sensor behind it?

Possibly with a controlled light source. You would probably have to calibrate it for each color though...