Could someone tell me which model your team has used, and how well they worked?
I want to know because I have heard they are much better than the gyro for measuring angle and movement, ect.
We’ve been using the $3 10K linear taper pots from Radio Shack, simply because I’m too forgetful to order good ones, and they have them in stock at RS.
They work fine for measuring angular movement, up to about 300 degrees or however far it is they turn.
We use multi-turn potentiometers (10-15 turns if I remember correctly). these are good for some applications, such a lead screws and joints that move beyond the limits of a normal potentiometer (or even joints that approach the limit since the extra range provides a buffer that protects the pot from damage).
Some teams use fancy ones called continuous potentiometers which can turn infinitely. The trouble with these is that they typically have a dead zone of 5-10 degrees where they won’t provide any feedback, but if you can account for this they are excellent for things that don’t have a mechanical rotation limit.
All potentiometers have the disadvantage of not working to find the heading of the entire robot since you must physically attach the shaft or wiper actuator to the rotating part and the body to the stationary part (or vise versa). Also, only linear taper pots should be used since logarithmic tapers are harder to use due to the fact that the logarithmic or piece-wise linear function must be determined before you can know which resistance values correspond to which positions.
We use Bourns 3590, 3543 and 3545 series potentiometers and they have always worked very well.
We’re using some of those for our control board, and boy do they have a terrible deadband near the extremes.
Our team did not use potentiometers this year mostly due to bad experiences last year. I was not on the team at the time, so I do not know the details, but potentiometers can cause trouble in some designs, but can also be a effective position sensor if you know both their capabilities and limitations.
Potentiometers are (mostly) mechanical devices. They produce an absolute position signal rather than incremental, but often at a lower resolution than many encoders. They also produce an analog output, and can be noisy particularly with lower cost potentiometers so using them sometimes requires filtering either in hardware or software which can make them too slow for some applications. And the dead band region is an issue for continuous motion unless you find a way to mount two of them together with enough overlap in the signals so that your software can account for this.
As with any sensor, there are advantages and disadvantages and experience with them will help you pick the best places to use them and also to work around the disadvantages. Like I said, our team had a bad experience with them and want to avoid them. My response was not to push for them to be used during this build season. But I will want the electrical and software team to get some experience with them in the off season and gain some understanding of them so any future decision on using them is based on real data.
Good luck with your application, and take some time to think specifically about what you really need to measure. How accurate does the measurement need to be, how quickly does the data need to be updated, can you tolerate some noise or dead spots? This is some real systems engineering. I believe that selecting the correct sensors for various parts of a system is often not given enough consideration. But, how well a system works when complete is often highly dependent on these choices.
What was the bad experience?
Like I said, I was not with the team last year. But from what I can tell they were having noise problems due possibly to low quality potentiometers or perhaps electrical noise on the low level analog signal being run through long unshielded cables. The team did not have access to the test equipment needed to fully understand the problem so they came away with a lack of trust for the components involved. I am hoping if we try again in the off season there will be time to give this technique more of a chance to work.
As a side note. The system that was having trouble was using a CAN interface to Jaguars that was also running a PID loop with potentiometer feedback. Because of the problems they encountered some members of the team do not want to use potentiometers, Jaguars, PID, or CAN this year. But many other teams have had great luck with CAN for example.
I believe that many of the problems encountered last year with CAN, PID, potentiometers, etc. may have been the result of poor construction skills coupled with inadequate debugging skills. But those issues are now getting attention. This summer I hope to show the team that with adequate attention to the required practices they can use CAN, and the Jaguars built in PID control features without issues, and with great advantage.
A gyro is a great device to sense rotation of the entire robot and we’re using two of them for that (pitch and yaw). We’re using a potentiometer for feedback of our shooting turret position and they are well suited for such a task. A few guidelines for choosing the right one are:
Use a linear taper pot unless you want deal with the logarithmic output of an audio taper pot (you don’t).
Choose a pot with enough mechanical rotations for the job to avoid hitting the internal stops. Breaking the stops will often take out the wiper too. The 10-turn pot on our turret sees about 6 rotations when the turret moves stop to stop.
Choose a relatively low resistance pot to avoid interference issues from nearby PWM and other noisy signals. I would go no higher than 10k if I couldn’t get a 1k unit. A 1k pot will only draw 0.005 amps from the analog bumper. You won’t even notice that.
Choose a pot rated for enough mechanical cycles to last as long as you need. They aren’t all the same and some have startlingly low rotational lives.
Radio Shack pots are junk. If you believe nothing else here, believe this.
We used one last year to position our arm. We planned to use one this year on our turret, but never got it mounted. Ended up not needing it with the camera.
We got ours from McMaster-Carr. They have a good selection of multi-turn pots in various grades. They are hard to find in the catalog–you have to look up “variable switches” to find them!
You give them the part number and specify the resistance. I’d stick to 10K or so for good results without pulling too much current.
Dr. Bob
Chairman’s Award is not about building the robot. Every team builds a robot.
Is 5 ma too much ?
Adding a 0.01 to 0.1 uF capacitor at the input of the Analog module should filter out most of the noise that gets picked up. It will slow down the response but probably not enough to matter in applications such as these.
Definitely use a 1k pot vs a 10k. It reduces the effect of external noise sources by 10 times relative to the effect you would get with a 10k.
It may also help if you used wires that are twisted together and shielded. This is commonly sold as “control cable”. The shield drain wire can be connected to the “GND” or “-” terminal at the Analogue module.
The signal out of the pot may appear “noisy” if it has not been used for a long time (several months) and the wiper contact has become oxidized. Exercising 10 times ought to clean it up. If it is still noisy as you move the wiper, you need to get a better quality one, like the ones from Bourns.
The “blue” analog breakout has an RC filter already, 100 ohms and 1 uf (1600hz).
The 5V regulator on the analog board can source at least 2 amperes.
That’s the “old” red analog breakout, with a switching power supply and no filters. The new blue analog breakout has a linear regulator and an rc filter, however it can only source 0.25 amps. In either case, a 1k pot wouldn’t cause a problem.
I would recommend a us digital absolute magnetic encoder instead.