pic: All Done!



2605 Sehome Seamonsters
Here is a few photos of a Prototype for a CIM encoder.
We use AMT 102-V3514 Encoders, which are magnetic.
This is the first version, and will be put to the test immediately for CAN testing.

that looks sweet!
Did you do the milling yourself? I know that those specs for the hole and everything are very precise. but what if you need to replace the encoder or the cim, seems like it would take a lot of time, and taking of a cim during competition is annoying and risky.

That looks really nice. How thick is that assembly? What are the counter sunk holes for?

-Hugh

Brilliant idea and nicely implemented.

I expect many teams will use this as inspiration for a similar implementation of their own. Definitely report back with results!

-Brando

Any concern about heat or electrical noise from the motor impacting sensor readings?

The AMT-102’s are capacitive sensing not magnetic. On their web site they describe how it works. They are designed to be mounted directly on a motor. This series comes in quadrature, commutation and absolute versions. Never had a problem with noise however the cable can always pick up noise and shielded cable can control that. I like that they come as a kit with many shaft adapters.

Thanks! Its exciting to be able to have an encoder directly on a CIM, but without a shaft extension it might be tricky to put to use in this fashion. The encoder’s instruction recommends removing the actual encoder no more then three times, and it is also very robust. But you are right, It take extra time to replace the whole unit if one part does fail.

We machined the part on a Centroid Conversion CNC, and added the mounting holes later on a manual mill.

The Counter sunk holes are actually mistakes The zero was a little off when it was first machined. The counter sunk allen-head screws are securing it to the CIM, and the two tapped holes are the same bolt pattern as the CIM, 10-24 to be used similarly for mounting.

I still love the idea, even though it takes more time to replace

Any reason you tapped them 10-24 instead of 10-32? I’ve found it’s easier to standardize on only one type of size-10 bolt.

And look at the spec sheet.

http://www.cui.com/product/resource/amt10-v.pdf

It also looks like DigiKey is the place to purchase from.

Very nice looking unit, thanks guys

Very nice work!

How thick is the assembly?

-Hugh

The sensor itself is 9mm thick.

The total width of the component was .535in. This just barely encompassed the width of the encoder and the length of the race on the CIM motor its self. The design does limit some of the usable motor shaft length however we still have about 3/4in of shaft length to work with.
The counter sunk holes were a mistake I made in the initial prototype. They did not fit exactly the way I wanted so I redid the mounting holes. The new design does not include the extra holes and has a few other features to make mounting the encoder etc a little easier.
I did use 10-32 for the tapped hole not 10-24. The face of the component exactly matches the face of the CIM motors. The idea was for us to be able to input the component anywhere a CIM motor typically is used without having to change mounting holes etc.
I plan on machining a few more mounts tomorrow and we will post some more pictures of the new design.

As for performance, they work brilliantly!

Interestingly, these encoders have the option to select the output resolution. This becomes handy when you’re dealing with the FPGA’s encoder counting, seeing as it’s only able to count about 38,000 counts per second with a quadrature signal, so the factory setting of 2048 lines per rev was just a bit too high. Currently, i have them set to 256 counts/rev and it tracks pretty reasonably even with the motor turned to 11. (or 12.9 as it may be)

They’re pretty accurate by my tests so far. Assuming we’re not going too fast for the FPGA, it doesn’t seem to lose count at all. I wouldn’t be uncomfortable with popping these things in for position control if need be.

We currently don’t have any CAN system set up to test them with a jaguar in PID control, but doing it in the FPGA seems to be working pretty well. Ideally, we’re going to be using CAN Talons this season, although having the FPGA as a fallback should be nice.

This looks pretty similar
http://firstchoicebyandymark.com/fc15-039

Yeah those encoders are from the same company, and have the same excellent resolution. BUT the pins output PERPENDICULAR to the metal face of the encoder. If you look closely in the picture you can kinda see this. For the design discussed in this thread they wouldn’t work.

I know we’ll be getting as many as we can of these regardless, because they are still awesome encoders, even if we are limited to putting them on the end of a shaft.