I want to make a 5K Pot, what is the best material?

[kolleamm]

TL;DR Summary

What kind of material should I use that can have a 5K resistance for a 3-4 inch length?

I need to make a custom potentiometer that will be like a ring that goes onto a shaft. The shaft is about 1 inch in diameter. I couldn't find anything like that online so I have no choice but to make one.

My idea is to print a plastic ring that I can glue a circular resistive material into. Since the shaft is about 1 inch in diameter the length of the wire will have to be greater than the circumference (3.5 to 4 inches about).

So to summarize things, what kind of material should I use that can have a 5K resistance for a 3-4 inch length?

Thanks in advance

 

[Baluncore]

Material selection decides durability. A resistive material will wear unless the wiper runs on a commutator. I would have thought this was an application for an optical or magnetic pickup.
You will need to specify;
1. What angle does the shaft rotate through, 270° or 360° ?
2. What is the speed of rotation?
3. What number of rotations before failure?
4. What maximum current will flow through the wiper?
5. What maximum voltage across the 5k resistive element?
6. What angular resolution do you require?

 

[kolleamm]

Thanks for your answer Baluncore

1. What angle does the shaft rotate through, 270° or 360° ?
270
2. What is the speed of rotation?
3-10RPM
3. What number of rotations before failure?
roughly 10,000
4. What maximum current will flow through the wiper?
100 mA, at 5V
5. What maximum voltage across the 5k resistive element?
5v
6. What angular resolution do you require?
5-10degrees

After doing some research I have found that there is a conductive filament which can be printed with a 3D printer. This would allow me to control the volume of the material and therefore the amount of resistance. If anyone is interested I found it here.

https://www.proto-pasta.com/products/conductive-pla?variant=27767315720
But if you have any better ideas please do share.
Thanks

 

[sophiecentaur]

An optical encoder would probably do the job and it's a lot easier to print a pattern in ink than to lay down a linear resistance track. Electronics can be as complex and cheap as you like and could deliver a varying voltage, proportional to angle, indistinguishable from an 'ideal' Pot.

 

[Baluncore]

I don't think a metal is a good solution for 5k when you consider conductive polymers. Google 'conductive rubber' and you will get sheet products through to silicon rubber that conducts after it cures.

 

[Rive]

10000 rotations is a killer for any cheap potentiometer, especially DIY ones, whatever material. Maybe some wire based ones would be able to deliver, but the mentioned resistance would require thin wires -> not really tough stuff. @sophiecentaur is right about encoders, especially since he mentioned optical encoders right off the bat.

 

[jrmichler]

Try to avoid building something with a sliding electrical contact. Any variation in contact resistance is a direct position error.

Hollow shaft optical encoders are available. Here is one manufacturer: https://www.dynapar.com/products_and_solutions/rotary-encoders/hollow-shaft/. Dynapar is a well known manufacturer of encoders, another is Heidenhain. There are others.

An alternative to optical encoders are resolvers. Search term: hollow shaft resolver. My experience with resolvers is that they are more robust, less precise, and lower cost than optical encoders.

Another type of encoder is magnetic encoders. Search term: hollow shaft magnetic encoder. Here is a good place to start looking: https://www.dynapar.com/technology/encoder_basics/magnetic_encoder/. I have no experience with magnetic encoders, but they have been around for a long time.

Expanding on @sophiecentaur post #4, you could apply a tapered black or white track on a white or black background, then a light source and photocell, and shield the entire thing from ambient light. It should work well enough to meet a specification of 10 degree position resolution.

Edit: It is well worth your time to look up the encoder catalogs from companies such as Heidenhain, Dynapar, Renishaw, and others. In addition to a complete list of encoders available, they have a lot of practical information. Trying to buy and use technical products without this understanding leads to a seemingly endless supply of annoying little problems that just barely prevent the system from working. My authority for saying this is: BTDT.

 

[Baluncore]

Magnetic encoders to sense rotation are available.
AMS AS5048A/B
Avago AEAT-6600-T16
Those need access to a magnet in the end of the shaft, or to an idler gear running off the shaft.

 

[kolleamm]

Optical encoders do seem pretty tempting, but would they give me a relative postion or an absolute one? For example when I first start my servo motor would it know it's position or just how much it moved since it was powered on?

 

[Baluncore]

Optical encoders do seem pretty tempting, but would they give me a relative postion or an absolute one?

There are two sorts of rotary encoder.
1. Incremental encoders; that produce sin and cosine waves, and maybe a zero reference mark.
2. Absolute encoders; that produce a gray code output that is always correct when turned on.
https://en.wikipedia.org/wiki/Gray_codeGray code can be converted to binary.

You will need two limit switches for your servo motor. You could use one of those as a reference.
If the servo is a simple DC motor, each limit switch can have one parallel power diode, so if the servo exceeds the limited range the forward power will be cut, but it can still reverse away from the switch and out of trouble.

 

[kolleamm]

There are two sorts of rotary encoder.
1. Incremental encoders; that produce sin and cosine waves, and maybe a zero reference mark.
2. Absolute encoders; that produce a gray code output that is always correct when turned on.
https://en.wikipedia.org/wiki/Gray_codeGray code can be converted to binary.

You will need two limit switches for your servo motor. You could use one of those as a reference.
If the servo is a simple DC motor, each limit switch can have one parallel power diode, so if the servo exceeds the limited range the forward power will be cut, but it can still reverse away from the switch and out of trouble.

Interesting, I'm thinking an absolute encoder would be great if it has that zero reference point.
The limit switch idea would work, but it would require on startup for my robot to zero out all it's positions and then move back to the right place. It could work though.

 

[Asymptotic]

Interesting, I'm thinking an absolute encoder would be great if it has that zero reference point.
The limit switch idea would work, but it would require on startup for my robot to zero out all it's positions and then move back to the right place. It could work though.

That is the way many motion axis using incremental encoders are set up. Two physical limit switches (to sense overtravel in either direction), use one of them as a "hard" home, then add an offset to define an operational home used during normal cycling.

Digikey has a decent gloss of how incremental vs. absolute encoders work.
https://www.digikey.com/en/articles/techzone/2018/dec/how-to-use-rotary-encoders-convert-mechanical-rotation-digital-signals

If DIY is preferable, and although I've not seen it done before, imagine instead of a code disc (borrowing the 4 bit binary example from their tutorial) it was a "code sleeve" that slid onto the shaft. This sleeve would consist of n number of bands each lined up with a phototransistor. I'm visualizing it similar to an old-fashion slot machine with each bit position being a separate reel.

A quick look at 3D printers suggests they ought to be able to lay down clean lines at least as small as 0.75 mm. For a sleeve OD of 1.25 inches this is about how wide the lines are at the 7th bit position.

At some point the lines will be so close together that a particular phototransistor design won't be able to sense the light/dark difference (and I don't know what that point may be; perhaps the 7th bit isn't practical), but all you need is 6 bits to get an angular resolution of 5.6°.

Keep in mind the above is a simple binary code disc. Grey code is preferable, and although I haven't seen one in awhile, another option is a BCD (Binary Coded Decimal) encoder.

 

[Rive]

At some point the lines will be so close together that a particular phototransistor design won't be able to sense the light/dark difference...

Quite an overkill, but if this design will be part of a computer controlled something (and more accuracy is needed) then sacrificing an USB port for a camera (and some programming to recognize the actual visible pattern) might be able to cut it.

 

[Baluncore]

Interesting, I'm thinking an absolute encoder would be great if it has that zero reference point.

The absolute encoder always wakes up knowing it's absolute correct position.
An incremental encoder has to search for the special zero reference mark, like when using a limit switch, only that zero reference can be locked to anywhere in the allowed range.

 

[anorlunda]

Wow, lots of excellent engineering advice in this thread.