Detecting position -- Hall effect sensor?

Question

I have an extruder that is producing a molten plastic filament that cools after a brief while (2-3 seconds). The plastic stream needs to be a consistent diameter, and one method of doing this is to have a pinchwheel "pulling" the filament faster or slower based on the filament diameter.

My thought was to have a fixed-position toothed wheel pulling, and a spring-loaded idler lightly holding tension against the toothed wheel. The idler would have a long lever protruding from it (so a small change in diameter will create a large swing on the lever), with a magnet mounted on the end of the lever. A Hall effect sensor would be used to detect that the lever is in the correct location, and a microcontroller can be used to adjust puller RPM to keep the lever in the correct location.

The pinchwheel & idler configuration is a known working setup, my unknown is the interaction between the magnet and Hall effect sensor. Ideally the Hall effect sensor would act like a variable resistor or voltage divider, output voltage dependent on the strength of the magnetic field, but I'm concerned it may work on flux instead.

If I were to get one of the TO92-shaped 3-wire sensors (a passive one, I'll be looking for what's available locally rather than a specific model), what kind of behavior am I to expect here?

Alternative ideas welcome. Diameter of the filament should remain constant to 1.70mm +- 0.1mm, to give an idea of the resolution required. I'd love to get down to 0.05mm which I hope this solution can do.

Answer 1

I've used the Allegro 3503 ratiometric linear sensor. It's discontinued, but I suspect its similar to the 1301/1302 sensors. It is very linear when positioned between two opposing magnets. This particular config (push-push mode) is shown on page 27 of the guide that you can download at http://www.allegromicro.com/en/Design-Center/Technical-Documents/Hall-Effect-Sensor-IC-Publications/Hall-Effect-IC-Application-Guide.aspx, but there are some other interesting modes. Give that ref a good read. You are talking mm here, and not cm. I think if you have a small 1 mm range, you can probably get 0.5 mm resolution, but I can't say I've pushed it that far. Strong neodymium magnets would certainly help.

They're even easier to use than you describe. The output voltage simply changes as a function of flux density.

As a design alternative, you might consider an optical position sensitive detector from Hamamatsu. You'd just need a 1-D sensor. They're good to about 50 microns over two millimeters.