Maker Pro
Maker Pro

DIY high precision linear pot. Cheap too if possible!

Hi all,

I'm a hobbyist taking his first (adult) steps into electronics. Apologies in advance for the incorrect terms I'm about to unleash on you :)

I need;
  • ~250mm linear/slide pot
  • high precision
  • cheap-ish (under USD$15)
  • durable (1 mil "slides")
  • free flowing, ie little (edit: constant is more important) resistance felt by user
I've looked at the market, and so far haven't found much. I found the Magnetopot by Spectra Symbol Corp, but it's a bit on the expensive side and (if I read things correctly), not very precise.

I've thought about making one with a "string" and a high precision rotary pot, but the pot alone is ~$80.

Standard slide pots are 100mm max, not the 250mm I need.

I can move a bit on the cheap requirement, but I can't alter my other requirements by much.

Any advice on a possible solution? Thanks in advance.
 
Last edited:
Um... not precisely!

I want accuracy (the slide is where it thinks it is) and also precision w.r.t decimal places. I've not thought about how many decimal places I need. It won't be absurd. At a guess, and with no knowledge of what's possible, I'd say down to 0.5mm. If the precision is worse than 1mm then I might not be able to do what I'm hoping to do. Thanks,
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
So, what you mean is that if you pace the slider in exactly the same position, you want the resistance to vary by less than 0.4% (or to state it the opposite way, if the slider is moved so that the same resistance is measured, the position of the slider should not vary more than +/- 1mm from the "average" position).

As for accuracy, this means that if you line up several pots next to each other, they will as a group behave in this manner, i.e. for the same slider positions on each, the resistance between them will be within +/-0.4%, and if all are set to the same resistance the difference in position between the extreme positions will not exceed 2mm.

I would thing that precision will be easier to achieve than accuracy. However if you purchase many pots from a single batch, you should be able to select those which are the most similar.

Now, to do that at low cost...
 

hevans1944

Hop - AC8NS
Hmmm. You did say DIY? If you can make one with your specifications for $15 you should become the richest Aussie in the world, practically overnight. I know I would buy a half-dozen of them. Seriously, for accuracy and precision over a ten-inch range of motion you need good bearings and a linear resistance medium with low temperature coefficient of resistance. This usually means small-diameter, stress-relieved, resistance wire wound on a non-conducting, temperature-stable mandrel. Inexpensive linear pots with good specs and a 10-inch range sell for about $500 and are widely available. Or, depending on your application (is a potentiometer really what you need?) a linear variable differential transformer (LVDT) might be more practical. The Magnetopot by Spectra Symbol Corp. is interesting, but I don't think it has the precision or accuracy you are looking for.

Please enlighten us as to exactly what you are trying to do... how many "linear, 10-inch, slide-adjustable doo-hickeys" do you need? Are they mounted side-by-side? If so, what is the minimum spacing between slides? Is this some sort of audio mixing board? What signals do the "slide pots" attenuate? Is your device to used as a linear position transducer? More details, please!

For a really inexpensive DIY device, I would investigate an optical, continuously variable-density, film strip ten inches long. The optical signal conditioning requirements are not trivial, but the cost can be very low on a per-device basis.
 
I would thing that precision will be easier to achieve than accuracy. However if you purchase many pots from a single batch, you should be able to select those which are the most similar.

Hmmm. Based on those definitions I'm not sure what's most important.

I'm testing a group of people's ability to slide the device in time with an external object. Several people will be using a device at the same time. If the devices aren't accurate, then it might make the testing irrelevant. I might be able to, though, "zero" the inputs and just look at the rate of change of each device.

I'll think about this a bit more. I may be able to favour precision over accuracy.
 
Hmmm. You did say DIY? If you can make one with your specifications for $15 you should become the richest Aussie in the world, practically overnight.

Bother. Point taken. As I said, I'm new to this field. I've seen dodgy slide pots for sale on Alibaba for $0.10 each, so I was hoping a better quality one would be available for 100x that.

Please enlighten us as to exactly what you are trying to do... how many "linear, 10-inch, slide-adjustable doo-hickeys" do you need? Are they mounted side-by-side? If so, what is the minimum spacing between slides? Is this some sort of audio mixing board? What signals do the "slide pots" attenuate? Is your device to used as a linear position transducer? More details, please!

Please forgive my ambiguity. At this stage it's a just a hobby, but I'd like to dream that one day I can find a market for this product.

For a really inexpensive DIY device, I would investigate an optical, continuously variable-density, film strip ten inches long. The optical signal conditioning requirements are not trivial, but the cost can be very low on a per-device basis.

For prototyping I'm happy to drop the "cheap" requirement. Once I've got the prototype and the software working, I can then look at cheaper options.

So, if price is no longer an issue, what components should I now consider? (I'll look into the film you've suggested, thanks)
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Now that you have described what you want to use it for...

If you have a look inside many printers you'll notice that they have a plastic strip which moves through a sensor. Looking very close at the plastic strip you will note that it has many many fine lines on it. The sensor detects these transitions from clear to opaque and these are counted, giving a measure of the position of the head.

This is a digital device rather than analogue and in printers they are very precise and accurate. A similar scheme may be useful for you. In your case, you probably only require one or two lines per mm and you can probably get away with something quite simple. The difficulty will be in retaining tension and knowing which way the slider is moving. (two sensors providing quadrature signals will solve the latter problem)
 

hevans1944

Hop - AC8NS
Hmmm. Based on those definitions I'm not sure what's most important.

I'm testing a group of people's ability to slide the device in time with an external object. Several people will be using a device at the same time. If the devices aren't accurate, then it might make the testing irrelevant. I might be able to, though, "zero" the inputs and just look at the rate of change of each device.

I'll think about this a bit more. I may be able to favour precision over accuracy.
So, it sounds like you are investigating a small population's ability to "track" a moving target using a slider. Perhaps a "bouncing ball" target on a video monitor? That could be an interesting experiment to measure the "frequency and phase response" of a human being in a closed-loop positioning system. For the slider itself, I think an LVDT would be most appropriate because they offer no-wear and very small inertia and friction. Many commercial versions exist at prices in the $500 to $1000 range for the length you require.

A linear potentiometer is a far-second choice, but the price is in the $200 to $300 range. The friction can be annoying for rapid movements by hand, and the resistive element will eventually wear out requiring replacement of the entire potentiometer.

Another approach might be a so-called "string potentiometer" which is a rotary potentiometer shaft connected to a pulley with a stainless steel "string" wound around the pulley. The string is internally spring-loaded to provide a retraction tension, and this could be a problem. However, these string pots are also available with quadrature-output incremental encoders instead of potentiometers. One example is the WS31/WS42 series from ASM Sensor. The problem with string pots is the spring tension required to keep the cable taut, and the possibility that rapid retraction could cause the cable to go slack, introducing an inaccurate reading. I have no idea what these cost, but the optical encoder type should be the least expensive. Visit the link and download the datasheet for the WS31/WS42 series if this sounds interesting.

Rotary optical shaft encoders are very inexpensive and are easily interfaced to digital microprocessors like the Arduino. All you have to do is figure out how to attach the slider to a rack-and-pinion mechanism to drive the shaft without backlash. Not a trivial problem, but potentially very inexpensive. There are plenty of ready-made anti-backlash rack-and-pinion gear trains available on line. Mount the rack to the bottom of the slider and mate it with the pinion attached to the incremental encoder. Add some electronics.

@(*steve*) suggested using a digital optical encoder ribbon and this may be the easiest DIY approach, if you can find a junk printer to salvage for a prototype. The optical position sensors (two in quadrature) will be the only part of the print head you will need. The optical ribbon must be carefully removed without damaging it and re-purposed as part of your slider mechanism. The quadrature outputs are derived by precise positioning of the two optical sensors with respect to the ribbon. Try not to disturb this positioning by carefully removing the optical sensors as a mated pair if you want to use the original optical ribbon.

In lieu of a transparent optical ribbon, I would investigate etching a pattern of parallel lines on a ten-inch long circuit board and use a pair of reflective IR LED and photo-transistor sensors to detect the line transitions. You could have the boards commercially manufactured at low cost, but you might have to experiment with the optical sensors (and perhaps some lenses) to obtain the resolution you need. One of the sensors needs to be adjustable in position with respect to the other one to obtain the quadrature output pulses needed for up/down counting. You could mount the PCB on the bottom of the slider though, eliminating the need for flexible wiring to the sensors. Or, in lieu of etched lines on a PCB, you could even produce a printed paper target of alternating black and white lines to attach to the bottom of the slider. Please send 0.1% of any profits you earn from this idea to me.:D

You can photographically produce your own optical encoder ribbon at whatever position resolution you require. The slider would be mechanically attached to the sensor optics with ultra-flexible ribbon cable, perhaps salvaged from the print head. You would have to design and build the optical signal conditioning electronics: a quadrature up/down counter and a means to read it, perhaps converting counts to an analog signal with a digital-to-analog converter. Discrete-circuit quadrature up/down counters are tricky to design because you must allow for count hysteresis at the quadrature encoder step transitions to avoid a cumulative count error. A microprocessor might be able to handle this problem better in software. A "reset" or "home" position sensor (magnetic reed switch?) attached to the slider would be needed to initialize the up/down counter to a known initial slider position. Despite all that, I like the optical encoder approach because of it has the potential to meet your "under $15" goal, especially if mass produced by an Asian vendor.

My idea of using a variable-density strip is probably not within the capabilities of the home DIYer. It may not even be practical. It requires a dark-room setup and a means to variably expose a moving strip of film to create the variable-density strip of the length (250 mm) your require. There are also problems with read-out stability to solve since this is an analog approach. The only "advantage" I see is a continuously variable (as compared to a discrete incremental digital) output. On further thought, I do not recommend this approach.

Please keep us posted on your project. It sounds like fun!

73 de AC8NS
Hop
 
Top