Society of Robots - Robot Forum

Electronics => Electronics => Topic started by: krich on February 25, 2009, 01:26:58 PM

Title: ADNS Optical Mouse Sensor as Wheel Encoder
Post by: krich on February 25, 2009, 01:26:58 PM
Has anyone evaluated whether a Avago ADNS optical mouse sensor could be effectively used as a wheel encoder?  This has been something I've been contemplating for a while, but it's been on the back burner in favor of other projects.  I've decided to put this project on the front burner because of some issues I'm having with the reliability of my current wheel encoders.

I'm wondering if anyone has tried and failed or tried with some success.  I'm looking for an iterative dialog. 

In general, I'm considering mounting the sensor and associated lens pointing at the inside hub of the wheel where I've secured a scrap piece of paper with print on it.  The print will give the sensor an appropriate basis with which to calculate it's optical flow algorithms.  Solid colored paper simply won't do.

I've tentatively decided on the ADNS-2610 because I can get it in quantity 1 and it comes in a 8-pin DIP package and I'm concerned about the footprint (the lenses are bulky).  Another alternative is the ADNS-2051 because it has quadrature outputs and slightly better specs, but it's a larger package and I'm not sure I can source it in singles.

Looking at the datasheets for the ADNS-2610, I see that it has the following specs:

Accurate motion detection of up to 12 inches per second (move the sensor closer to the axle to detect higher speeds at the expense of per rotation resolution.)

400 counts per inch of resolution  (that's a lot of resolution.  I wonder if my poor atMega168 can handle that effectively)

In a previous project, I was able to interface the chip via SPI and do things like read deltaX/Y and take snapshots.  I'm pretty sure I can do this project, but I'd like to draw upon other's experiences.

Any opinions would be greatly appreciated.  Thanks.
Title: Re: ADNS Optical Mouse Sensor as Wheel Encoder
Post by: Soeren on February 25, 2009, 04:00:34 PM
Hi,

[...] I've decided to put this project on the front burner because of some issues I'm having with the reliability of my current wheel encoders.
[...] 400 counts per inch of resolution  (that's a lot of resolution.  I wonder if my poor atMega168 can handle that effectively)
[... ]Any opinions would be greatly appreciated.  Thanks.

Here's an opinion:
Describe the issues with the "old" wheel encoder and get it sorted out. 400CPI is not much in a mouse, but overdoing precision in a wheel encoder is just excess luggage, and it is not hard to get good readings with a much simpler discrete build.