# Potentiometers

V

#### vishal

I am a student working on a project which requires position sensing with range of 8" and resolution of atleast 0.001" (1 mil). Since my budget is limited I can't spend more than \$200 at the same time can't sacrifice the specs required.

I tried many companies, first of all their spec doesnt mention resolution, all it says is infinite resolution for conducting plastic type potentiometer, which is ok but the confusion arises when the same resolution is available at \$150 and also at \$400+.

Regards,
Vishal

A

#### Anil HP

Hi friend,
Are you particular about using potentiameter for position control. why dont you try using an encoder. We can stop a stepper motor to any perticular pulse of the encoder feedback by using some PLC (Without using any other hardware) for eg, Unitron PLC. if you need more information mail to me [email protected].
Anil

T

#### T. Connolly

Any potentiometer type position transducer manufacturer will tell you their pot has infinite resolution, which is true (even for \$1.00 Radio Shack pots), but it is not the entire picture. There are at least four other things you need to consider.

1) Potentiometer linearity, or how well the resistance changes in a linear manner as the pot is changed. This will have a huge impact on your accuracy. You can overcome this by dividing the pot into sub ranges and appling a calibration multiplier to the result for that range. For high accuracy you will have to dervie this empirically.
2) Your resolution on a pot is determined by your AD converter resolution, not the pot. If you have an 8 bit converter then your resolution is 8"/(2^8) or "0.031 inches, if you have a 12 bit A/D its 0.0019". So you are going to need at least a 13 bit converter (unipolar) and you'll need to make sure your pot voltage is full scale, ie, if your analog input is 10 volts, the the pot output at 8" will have to be the full 10 volts, unipolar. If its bipolar -10,+10V your resolution is half. Anything less than full scale starts cutting into your resolution.
3) Accuracy of A/D converter and conditioners. Typical accuracy is 0.05%, so out of 8" full scale you get 0.004" accuracy.
4)Stability of the power supply to the pot. Obviously if it is drifting around your going to get error in your result. Use a good quality supply.

While it is do-able, .001" accuracy out of 8" is going to be hard to come by with a pot. Can you use an encoder? They can be had for real cheap. You can get a light duty one from "http://www.automationdirect.com":http://www.automationdirect.com for around \$75.00. Bourns also makes a small inexpensive plastic encoder. You migh also try Dynapar.

I often do joint projects with the local university, so I understand that its sometimes tough to do automation projects on a student budget but there are ways for the resourceful.

V

#### vishal

Dear Sir,
I would consider myself very fortunate to get really great advice from you and dont have words to thank. Practical world is so different than the one we design on paper, I guess thats what makes it becomes more exciting and challenging. It becomes a pleasure when one gets help from an experienced person.

I would ask for another favor, after a lot of search I could decide on potentiometers made by

"www.midoriamerica.com":http://www.midoriamerica.com ,

LP-Fxx series, it costs around \$165 and good linearity etc. can you please go to this website and give your opinion.

Regarding the encoders you had mentioned, though its a great option but I cant use it becuase even if I put a rack/pinion arrangement (convert the linear motion to rotary), the tolerance/backlash in the teeth arrangement will defeat the purpose of accuracy.

I hope you agree with my understanding of the difficulty with encoders.

Regarding the accuracy of A/D convertor I was thinking of using a 16 bit convertor instead of
13 bit thus giving a resolution of .00012 thus I wont be affected by the error in the A/D by disregarding the last three LSBs, I am still working on this problem.

Regards,
Vishal

M

#### Michael Griffin

Resolution isn't the same thing as accuracy. It is possible to have high resolution with low accuracy. Check the linearity specifications for these pots. If you are looking for high accuracy, I don't think you are going to find a 200mm linear pot to be good enough.

--

************************
Michael Griffin
************************

T

#### T. Connolly

I have actually used something similar to the midori sensor. It worked OK, but I wasn't after the kind of resolution you want. I also usually use 16 bit A/D converters. The big issue will still be your signal conditioning in the A/D, that error won't be mitigated by increased resolution. High quality A/Ds are available though, the 0.05% I mentioned is typical, but not the limit. Have you also considered LVDTs? Of course the best accuracy still comes from an encoder. Sony, Heidenhan, Dynapar, and others also make linear encoders, but those are probably out of reach for a student budget, Dynapar has one thats not too expensive but the interfacing will probably put you over the top, I havent priced it though so you might look it up "http://www.dynapar-encoders.com":http://www.dynapar-encoders.com You might be able to adapt some ditigal calipers that have an RS232 output for just a little more. Usually the rotary encoders like the ones I suggested are attached to the motor shaft or the ball screw, but that depends on how your system is moved. Preloading the ball screw nuts or gears can remove most of the backlash, but again your working on a student budget and that stuff can get expensive real fast. I am working on a large welding machine right now that has well over \$200K just in the motion system alone.

C

#### Curt Wuollet

I doubt that many potentiometers will be repeatable to 1 part in 8000. That might be why they don't want to talk about resolution. The element is stepless but contact resistance and mechanical tracking will give larger errors than what you are looking for I think. Of course, I
haven't tested pots for many years and an 8" long slider might come closer than a rotary. At least for a while. Perhaps you could look into the
inexpensive rotory encoders that have largly replaced pots in consumer gear. It should also be possible to make low cost linear encoders with
IC mask type technology but I suppose there isn't much demand.

Regards

cww

--
Free Tools!
Machine Automation Tools (LinuxPLC) Free, Truly Open & Publicly Owned Industrial Automation Software For Linux. mat.sourceforge.net.
Day Job: Heartland Engineering, Automation & ATE for Automotive Rebuilders.
Consultancy: Wide Open Technologies: Moving Business & Automation to Linux.

C

#### Curt Wuollet

Using a band or cable rather than a rack and pinion to convert linear to rotary might just get you in the ball park. You would want several turns over the 8" length. Eight inches to .001" is hard to do cheaply but not impossible.

Regards

cww