small, accurate rotary encoders?

Critical question, is whether the unit can establish a 'zero' point, using some other method (moving the joint to a 'limit' perhaps during boot up)?. This makes the difference between having to use an 'absolute' encoder, or a 'relative' encoder. Now, if you use a 'relative' encoder, 1 degree, only needs a 90ppr encoder (remember if you decode the four possible quadrature states, you get 4* the quoted number of 'steps' round a circle). The smallest 90ppr encoder I know of, at a reasonable price, is the Nemicon OMS-T encoder, which is just 1.2cm across, but is 1.25cm long. I don't know of anything quite this small, for an absolute encoder (look at the CUI MAS-10, which gives 256 locations, and is only 13mm*17.5mm, but the

Best Wishes

I had to dredge the toilet bend of my brain to remember this company, but

martin

Yabbut, where is that BBC stuff that let's you build your own?

DNA

A pot would be much better in that application. It's self zeroing, it's able to tell you position immediately from power-on, and it has a much higher resolution (limited by your ADC). Open up a hobby servo and take a look at their pot-driven control loop.

How are you with making little tiny parts? Digikey seems to have quite a selection of 1-turn trimpots, which might work in your app. Obviously, you wouldn't get a full range, but a cermet pot is "infinitely" variable - you might have to watch for repeatability, but as others have mentioned, a zero-set at bootup should take care of that; and with sensors, your controller could compensate. :-) Just go to

My idea is you could find a way to mount a little pot on the axis of a joint, and just measure the angle by the resistance.

Good Luck! Rich

Maybe also look at their E4P model:

Maybe wrap a monofilament around the joint, with spring tension, tied to a sliding potentiometer? Even connecting a rotary encoder in addition to providing mechanical rotation is a tricky job.

I've has soem good success with the Austrian Microsystems electronic encoder IC's.

AM5040 Comes to mind - 10bit (0.35degress resolution) SPI connection to micro.

Enjoy

Joe

The HMC 1512 from Honeywell might be worth a look, it senses magnetic field direction down to 0.05=B0 and they're pretty small! you'd need a little magnet too but there's plenty of places that sell them now :)

The usual solution for non-industrial designs is to use a pot. I've seen some really small ones among the kazillion of them out there.

Luhan

much better in that application. It's self zeroing,

Hi Ben - my worry with using pots is that they will wear out with repeated usage. Maybe there are some pots designed to be constantly turned?

-Mike

I can handle little parts. Problem is that cermets, in my experience at least, are designed to be used to adjust a setting, and then left and forgotten. I think they would break down with constant usage.

-Mike

much better in that application. It's self zeroing,

Well, like he said, open up a hobby servo, and get another one of whatever they're using.

Good Luck! Rich

an absolute/Gray code encoder is what you need how ever, the space and resolution you're looking for may put the kibosh on that!

in your case, use some precision pots into ADC of a PIC or something like an Atmega that has lots of ADC's on them and perform a commutations link to your master CPU.