search

victoriarobotics / victoria_mechanical

Place to house all of the mechanical diagrams, BOMs, etc.
MIT License
0 stars 0 forks source link

Design bumper #4

Closed griswaldbrooks closed 7 years ago

griswaldbrooks commented 7 years ago

So, I did a quick test with the neato hall effect sensor to see what kind of responses there were. There were several testing conditions. The primary one was voltage vs distance, but I also tested three different magnet sizes that I have, and the presence of a metal slug that is usually behind the sensor that channels the flux through the sensor.

Here is a chart of the response: (https://github.com/victoriarobotics/victoria_mechanical/files/784392/hall_effect_data1.xlsx)

Obviously the largest disc magnet with the slug gives the best response but if that interferes with the compass I believe the small disc with slug would suffice. I feel like the ball magnet is going to be trouble. In general, we should use the slug.

Here is the data: hall_effect_response_chart1 [hall_effect_data1.xlsx]

Picture of the sensor without slug: 20170217_134656

Picture of the sensor with slug: 20170217_134704

Picture of the magnets tested and the testing blocks I used. 20170217_134914

The discs were of 1 inch and 3/8 inch diameter and the ball was 3/16 inch diameter.

griswaldbrooks commented 7 years ago

Forgot to mention, the sensor was powered with 5V and the baselines were: No magnet, No slug: 2.494V No magnet, Slug: 2.579V

griswaldbrooks commented 7 years ago

I was told that the new version of the sensor (I didn't realize I was using the old version) uses a more sensitive hall effect. I'll have to retest.

markwomack commented 7 years ago

Looks like Small Disc w/slug, allowing for an inch or more of movement by the bumper arm? Seems like that will be sensitive enough to know we ran into something, but not too noisy if we bump around. Assuming there is a spring on the bumper arm...

griswaldbrooks commented 7 years ago

Right, well, we'd be comparing Small Disc with Slug at 1 inch: 2.768 V to No Magnet with Slug: 2.579 So, it'll probably be fine, depending on testing with vibrations. I think the mechanism is sound, and we'll get things as close as we can, but we can also swap sensors, adjust their position, and swap magnets if we have a problem without fundamentally overhauling the mechanism.

markwomack commented 7 years ago

I was thinking it would be more like 3.8v (when very close, .25) compared to 2.75v (when 1 inch removed). Maybe I don't understand that chart.

griswaldbrooks commented 7 years ago

Consider bumper extended vs bumper compressed. bumper_extended1 bumper_compressed1

The sensor is inside the robot, facing orthogonal to the motion of the magnet. Therefore it only sees the magnet or not. The distance is meant for how close we can get the sensor and magnet together in one of these states and still allow the robot to be opened up. I guess the magnet can be very close to the shell, since it only moves relative to it, but the sensor can only be so close.

griswaldbrooks commented 7 years ago

An overhead view would look something like this. bumper_compressed3

The distance here is what I was testing. bumper_extended3

griswaldbrooks commented 7 years ago

Updated test with the other sensor.

hall_effect_response_chart2 hall_effect_data1.xlsx

griswaldbrooks commented 7 years ago

So as a quick test, I "mounted" the hall effect sensor in the robot 20170217_200519 20170217_200504

It's inset by the amount it is because the upper shell is inset 20170217_200536

So, using a spacer about 0.2 inches thick

20170217_200543

I placed the small disc magnet over the "new" sensor until I got the max response. 20170217_200555

Then I moved it a bit to the left. The results are:

Displacement (in) Voltage (V)
0 1.83
0.25 1.5
0.5 1.09

Given these results, I'm guessing we can build whatever is practical and calibrate it.

griswaldbrooks commented 7 years ago

Bumper is installed, no hall sensors installed

https://youtu.be/ipIWtrX_vfM

griswaldbrooks commented 7 years ago

Hall sensors installed https://youtu.be/vJRPw-XxklI

griswaldbrooks commented 7 years ago

Pictures of the sensor on the inside. Still need to crimp new connectors and get the design files up and then this can be closed.

20170226_232825 1 20170226_232821 1 20170226_232815 1

griswaldbrooks commented 7 years ago

Note, right now when the bumper is extended (not bumping), the magnet and the hall sensor are aligned and receiving full signal. The advantage of doing it this way versus the other way is that if the cover (top shell) is removed, the robot will think it's bumper is compressed (bumping) and will not run.

griswaldbrooks commented 7 years ago

As the integration of the sensors, electrically, is going to be addressed in https://github.com/victoriarobotics/victoria_electrical/issues/9 I'm going to close this one.