pic: 221 Robotic Systems Swerve Demo Bot Steering Feedback

[ajlapp]

[cdm-description=photo]35885[/cdm-description]

[jspatz1]

Could you give more info on the "continuous-rotation pot?" Does it have a dead zone?

[Brandon Holley]

I'm assuming the actual tensioner pulleys are like a black delrin? If so, have you had any issues with wear and tear on them in your findings? I only ask because I used an almost identical steering gearbox setup as yours (kind of eerily similar actually) and had to replace the pulleys every so often.

-Brando

[Aren Siekmeier]

Our team does something similar (http://www.chiefdelphi.com/forums/showthread.php?p=962804#post962804) to tension our drive chains with black delrin, and we haven't experienced much wear. Our only problems have been the bolts loosening and the cams rotating away from the chain (loctite would probably fix this) or the mounting bracket being insufficiently supported. You also have to make sure the cam has enough play to take out more slack when the chain stretches (we had to drill a second hole in one of ours at competition for this very reason).

Btw, I like how you cut a slot for the chain in the tensioner material to keep it in line. We'll have to consider something like that for future drive trains. How did you machine it with the guide grooves?

[ajlapp]

Could you give more info on the "continuous-rotation pot?" Does it have a dead zone?

From Digi-Key....made by Bourns. 6639S-1-103-ND

My controls engineer may have to clarify, but I don't recall any dead band. They simply rollover at the zero point.

I'm assuming the actual tensioner pulleys are like a black delrin? If so, have you had any issues with wear and tear on them in your findings?

This unit has low hours, but I'm planning to see some wear over time. The black delrin rotates quite smoothly on the aluminum cam. This rolling action helps minimize wear.

How did you machine it with the guide grooves?

These pulleys were made on a simple lathe with a parting tool.

[Joe Ross]

From Digi-Key....made by Bourns. 6639S-1-103-ND

My controls engineer may have to clarify, but I don't recall any dead band. They simply rollover at the zero point.

They do have a deadband at the rollover point, the rollover isn't a perfect "sawtooth". We've used Vishay 357s which after the rollover remain zero for 10-20 degrees. These appear to have similar specs.

If you can live with the deadband, we very much like them. Continuous pots have been much more robust then any encoder we've used.

[Ether]

Continuous pots have been much more robust then any encoder we've used.

What failure modes have you encountered with encoders?

[Marc P.]

I spy an arduino with an ethernet shield. Is it controlling the bot on its own, or is it working with another microcontroller or IFI/cRio/other control system?

[jspatz1]

They do have a deadband at the rollover point, the rollover isn't a perfect "sawtooth". We've used Vishay 357s which after the rollover remain zero for 10-20 degrees. These appear to have similar specs.

If you can live with the deadband, we very much like them. Continuous pots have been much more robust then any encoder we've used.

So could you explain how you deal with the dead band in your logic?

[Nikhil Bajaj]

As an alternate sensor, one might consider the US Digital MA3, which has a much smaller dead band (it is pretty minuscule.) They are a little more expensive, though, at around $40+. However, you can get a ball-bearing version that can sustain higher rotational speeds.

http://www.usdigital.com/products/encoders/absolute/rotary/shaft/ma3/

This is what we like to use for certain things on 461. I found out about them when asking some folks on 469 about their swerve drive a few years back.

I guess my method of dealing with dead bands is avoiding them entirely if possible!

[Gdeaver]

I noticed that the sprocket on the steering drive is not the same diameter as the sprockets on the pivot modules. That combined with the dead band of the potentiometer would lead to some tough programing problems. I don't see any indexing on the steering system. So I want to know how are you guy's handling these problems? We used the window motors also. We have one for each module. Our steering sprockets are 1:1, so the value from the sensor always matches wheel orientation though 360 degrees. As to absolute rotary position sensors, I have not found any company that makes a classical potentiometer that has 360 degree mechanical and electrical ratings. The US Digital MA3's mentioned would be good for this application. We used the Cherry AN8 series. The Cherry AN8's are IP67 and automotive qualified. By being Automotive qualified you gain reverse polarity protection and higher ESD protection. Some teams have had problems with MA3's and ESD. They also are 0 to 5 volt and the out put is sloppy at the rails. The AN8's are 0.5 to 4.5 volts and have a sharp saw tooth slope.

[AdamHeard]

I noticed that the sprocket on the steering drive is not the same diameter as the sprockets on the pivot modules. That combined with the dead band of the potentiometer would lead to some tough programing problems. I don't see any indexing on the steering system. So I want to know how are you guy's handling these problems? We used the window motors also. We have one for each module. Our steering sprockets are 1:1, so the value from the sensor always matches wheel orientation though 360 degrees. As to absolute rotary position sensors, I have not found any company that makes a classical potentiometer that has 360 degree mechanical and electrical ratings. The US Digital MA3's mentioned would be good for this application. We used the Cherry AN8 series. The Cherry AN8's are IP67 and automotive qualified. By being Automotive qualified you gain reverse polarity protection and higher ESD protection. Some teams have had problems with MA3's and ESD. They also are 0 to 5 volt and the out put is sloppy at the rails. The AN8's are 0.5 to 4.5 volts and have a sharp saw tooth slope.

The sloppiness of the MA3's have been annoying, and the fear of ESD has bothered me (but never affected us). What do the AN8 costs?

[Gdeaver]

The AN8's run about 40$ depending on the supplier. The AN8's come with a magnet assembly but it would not work in our use. We mounted a 15 tooth #35 sprocket on to the window motor and then mounted a 1/2" x 1/8" diametrically magnetized neodymium magnet on top of the sprocket. The Cherry sensor was then mounted 2.5mm away from the magnet. Look at the picture in this link.
http://wiki.team1640.com/images/thumb/6/67/DB6_steering_motor_with_encoder_100124_csm_2.jpg/400px-DB6_steering_motor_with_encoder_100124_csm_2.jpg
You'll also need the matching Delphi connector to make a cable for it. There are getting to be more and more sensors of this type in the market. Many are much more expensive.

[Joe Ross]

The sloppiness of the MA3's have been annoying, and the fear of ESD has bothered me (but never affected us). What do the AN8 costs?

Looks to be about the same as the MA3. http://www.mouser.com/Sensors/Motion-Position-Sensors/_/N-6g7q6?Keyword=aN8&FS=True

So could you explain how you deal with the dead band in your logic?

We don't use crab drive. We've used continuous rotation pots instead of encoders for odometry (http://www.chiefdelphi.com/media/papers/1743).

I'd also be interested in hearing how 221 Robotic Systems dealt with the deadband in this application.

[ajlapp]

So I want to know how are you guy's handling these problems?

The short answer is that we don't currently deal with the dead band. :eek:

Our first focus was developing the three driving modes; tank, crab and monster truck steer.

Our wheel rotation is currently limited to 75 degrees in either direction from neutral....this is plenty good for monster truck steer.

[BEEKMAN]

Was this the one on demo at IRI? If so, have you been able to control the jaguars with the Arduino yet?

[ajlapp]

Was this the one on demo at IRI? If so, have you been able to control the jaguars with the Arduino yet?

This is the unit from IRI.

We already control the Jaguars using the Arduino via pwm. We would like to add CAN communications using a 2CAN device but we haven't worked on this yet.

Its on the short list. :D