Tachometers

I've received the new pulleys and I'm in the process of swapping out the hardware. I'll get more into the details in my next project log, but to make everything line up correctly, I'm having to adjust the frame and had to pull the engine off to do it. More details to follow.

I've been thinking about the latest testing where some of the tests ended and the rotors were no longer aligned with each other. This is due to the belts skipping teeth. The new belts and pulleys should help prevent this from happening, but it'd be good to have a way of monitoring the speed/position of the engine and each of the rotors. Tachometers measure rotational speed and rotary encoders measure the rotational position, but I think I can use a a tachometer placed on the support arm at each propeller to measure the speed and then compare the signals to tell if they are in sync or not. This information will also be useful to have available to the Pixhawk to provide some closed loop control of the gas engine.

From doing some research it doesn't seem straight forward (or maybe possible) to have the Pixhawk directly measure the rotational speeds. Long term, the ideal setup would be a Gas Engine Control Unit (ECU) that would have at least (5) tach sensor inputs on it, relays for the starter and ignition, a simple processor for computing the actual speeds and an I2C interface to talk to the Pixhawk. (The overall concept of having a breakout board that interfaces with the Pixhawk over I2C is similar to the PX4FLOW board)

As far as the actual sensors go, typically hall effect or optical/IR sensors are used. I'm leaning towards using IR sensors mounted above the rotors on the support arms, so they will only be a couple inches away. If I did a hall effect sensor I'd have to add a magnet to the rotors. There's some good instructions on instructables by [electro18] using IR sensors. I'll probably start with these sensors , but use the Raspberry Pi I have on hand and work forward from there. I likely won't get started on this until I get Goliath back together and running again.

Since electronics is not what I usually do, if anyone has any thoughts on this I'd love to hear from you.

Discussions

Sean wrote 12/21/2014 at 06:11

There you go, even better than magnetic bolts. I knew they used a toothed wheel but I couldn't figure out in my head if that would work if the hall effect sensor was 90 degrees to the teeth and the magnet was on the opposite side of the wheel than the sensor. With a magnet embedded in the sensor, thats even better.

Are you sure? yes | no

Sean wrote 12/19/2014 at 17:35

The only comment I'd make is that typically hall effect sensors are used to measure rotation on safety-critical applications like antilock brakes. If you use IR sensors on your rotors, then you have to accept the fact of dirt/smoke/glare giving false counts and missed counts. Its your call if it is critical to your control system and to safety to have completely trustworthy feedback on rotation. Since you're using rotational speed to measure rotational position by dead reckoning and using it as part of a closed-loop control system, my personal opinion is that you'd want more reliability than IR can offer. Maybe you can use the hall effect sensor near the rotor mounting bolts. I think they sometimes use bolts with magnets in the head (no idea how much they cost), and if you replaced all of the mounting bolts together then everything would remain balanced.

Are you sure? yes | no

Peter McCloud wrote 12/19/2014 at 21:00

I really like the idea of having the magnets at the bolts. It'd be easy to add some ring magnets over the ends of the bolt. I had figured hall effects were more reliable, but I didn't know they were used for anti-lock brakes. Since I haven't ordered anything yet, I'll switch to that instead. Thanks for the input!

Are you sure? yes | no

PointyOintment wrote 12/20/2014 at 07:51

Wheel speed sensors (https://en.wikipedia.org/wiki/Wheel_speed_sensor ) usually use a Hall effect sensor with the magnet integrated into it, and a ferromagnetic (but not magnetized) toothed wheel. The wheel's teeth vary the magnetic field from the stationary magnet as they go by, and the Hall effect sensor senses that. That should be pretty easy to implement using the bolts instead of the toothed wheel.

Are you sure? yes | no

Peter McCloud wrote 12/21/2014 at 17:20

Thanks! This is certainly an elegant solution. I did a search on Mouser and found one (aka geartooth sensors)
http://www.mouser.com/ProductDetail/Honeywell/1GT101DC/?qs=sGAEpiMZZMs29kr3d%252bndI4OPdCdzrfug2RHPCTozkik%3d
They are $40 a piece vs. $2 for a simple hall effect sensor and $1.50 for a ring magnet. I'll probably set up both options and see how they work.

Are you sure? yes | no

Sean wrote 12/22/2014 at 18:30

Yes, thats perfect. Found a cheaper sensor from Hamlin:
http://www.mouser.com/ProductDetail/Hamlin-Littelfuse/55505-00-02-A/?qs=sGAEpiMZZMvWgbUE6GM3OZ9N4ycotAUnlfBwPBv3bJcUUB2Xowt2fg%3d%3d
I think its definitely worth your while to go with the more expensive sensor. Looks like these have some active circuitry in them to self-calibrate and clean up the hall effect sensor output. That will make your life so much easier when you go to plug it into your controller board. Stay in software land as much as you can.

Are you sure? yes | no

Peter McCloud wrote 12/23/2014 at 01:23

Awesome. When I looked over the weekend, the minimum order for the Hamlin sensors was 323. Looks like they are in stock now. I'll definitely get those.

Are you sure? yes | no

Roller Chain vs. Belt Drive and New Belts

Belt Upgrade: Frame Reconfigured

Discussions

Become a Hackaday.io Member