Hollow-shaft Hexagonal 8mm potentiometer

[tru3533]

Hi,
I have a 8mm hexagonal steel rod, where I neeed to mount a potentiometer or similar, to find out what turning position the rod is at.
Rod rotates similar to one turn potentiometers. 340 degrees.
Also have limited space available on the rod 2-3mm hight.

So I'm looking for a hollow-shaft Hexagonal 8mm potentiometer.
Any suggestions where to find?

[TimFox]

Have you tried cascading two shaft couplers?
An appropriate round bore with two set screws will fit the hex side, and reducing round adapters can get you to a standard pot shaft (1/4", 6 mm, etc.) diameter.

[langwadt]

not cheap, https://dk.rs-online.com/web/p/potentiometre/8427169

[Ian.M]

You wont fit a 8 mm across flats hex rod through an 8 mm bore.   Depending on how sharp the corners are, it could be up to fractionally under 9.24 mm across them, so Langwadt's find wont help you.  A 10mm bore one with a precision hexagonal  center hole bushing might work, but you'd probably need extra height for the bushing as one that fitted entirely in the bore would only have 0.38 mm wall thickness at the points so would be rather fragile.

[TimFox]

One can use a precision reamer to fit a measured rounded-point to rounded-point value for the diameter in a shaft coupler.
For precision purposes, our lab machinist always bored out shaft couplers to couple motors to lead screws and similar shafts.

[tru3533]

Sorry It looks like I need a better explanation.
The rod is coming out of a gear assembly, total length is 9mm.
The rod is then going into a valve housing. The hexagonal hole in the valve housing is 13mm deep.
Gear assembly and valve housing is mounted together and can be splitted. 

I was thinking if I add spacers between them, I can get 3mm access to the rod and can
place a potentiometer there.   

[langwadt]

a thin plastic gear on the shaft and on a regular pot?

[tru3533]

Absolutely a solution langwadt 
If I can't find a pot with the right thickness

[Infraviolet]

As it is powering a potentiometer it isn't going to be under all that much torque load, a 3d printed adapter between it and any potentiometer could probably work. Depending on the precision you need you could print a herringbone gear with a hex hollow shaft, squeeze this on to the rod, then have another herringbone gear mate with it in a 1:1 ratio and have that second gear use a potentiometer as its axis. Given the low forces a potentiometer will be requiring a carefully sized friction fit with the aid of some cyanoacrylate superglue would likely work, for large torques properly mating a 3d printed part to a shaft (hex, D, round, splined...) becomes a whole art of its own. You could also print the gears to inclde places to put nuts and run screw through them to clamp the shaft the way grub screws would, although if you've space enough to allow screw heads then using proper screws is often better than grub screws as the bigger head lets you tighten them harder than the little recessed hex in a grub screw would. The mating gears would lead to some backlash (amount of wobble shaft can make before potentiometer responds), you'll have to check your application can tolerate this.

[tru3533]

Thank you so much Infrared for the in depth explanation 

[PCB.Wiz]

Hi,
I have a 8mm hexagonal steel rod, where I neeed to mount a potentiometer or similar, to find out what turning position the rod is at.
Rod rotates similar to one turn potentiometers. 340 degrees.

What precision do you need ?

a thin plastic gear on the shaft and on a regular pot?

3D printing was mentioned too.
The op could look up single track Gray codes, or even standard Gray code optical disk + readers ?

[f4eru]

one option is to add a couple of gears.

[langwadt]

As it is powering a potentiometer it isn't going to be under all that much torque load, a 3d printed adapter between it and any potentiometer could probably work. Depending on the precision you need you could print a herringbone gear with a hex hollow shaft, squeeze this on to the rod, then have another herringbone gear mate with it in a 1:1 ratio and have that second gear use a potentiometer as its axis. Given the low forces a potentiometer will be requiring a carefully sized friction fit with the aid of some cyanoacrylate superglue would likely work, for large torques properly mating a 3d printed part to a shaft (hex, D, round, splined...) becomes a whole art of its own. You could also print the gears to inclde places to put nuts and run screw through them to clamp the shaft the way grub screws would, although if you've space enough to allow screw heads then using proper screws is often better than grub screws as the bigger head lets you tighten them harder than the little recessed hex in a grub screw would. The mating gears would lead to some backlash (amount of wobble shaft can make before potentiometer responds), you'll have to check your application can tolerate this.

since it doesn't do a full rotation, pulleys and wire could also work

[tru3533]

PCB.Wiz
Single track Gray codes is totally new for me. Very interesting, Thanks!
The valve opening position feedback is not high precision 0-100% in 5% steps

Makes me wonder if it's possible to use a Hall-Effect Sensor?

[PCB.Wiz]

Single track Gray codes is totally new for me. Very interesting, Thanks!
The valve opening position feedback is not high precision 0-100% in 5% steps

So that's just 20 steps ?

I found some higher precision examples
This one is an encoder disk for 9 sensors, equal spaced 360 steps, so each sensor need to be 40°, That's impressive for 1° absolute resolve.
https://www.thingiverse.com/thing:4590077

but I personally prefer the other solutions with sensors more packed, as they can fit into a ~ half circle (slightly less than 180°, as N+1 fits into 180° )
That allows more practical construction, and also support linear track sense.

Here is a nice built solution of 60 step, Six sensors, on a 30 degree pitch (150° total )
https://www.instructables.com/Absolute-Position-Encoder-With-Single-Track-Gray-C/
They used reflective optos there, which are cheap and easy to source and mount.
The track code as ASCII string is '000000000000111000000110000001111111111111000111111001111110'

Makes me wonder if it's possible to use a Hall-Effect Sensor?

There are magnetic angular sensors, but they usually rely on the magnet being above the centre of the sensor.
You could get that with gears.

There are also commercial linear track magnetic absolute sensors, quite clever, but not low cost.

There are magnetic sensors like TLV493D-A1B6, that give 3 axis raw readings over i2c, so maybe that's enough to sense say 8? tiny magnets of varying orientations spaced around a circular disk on your axle ? 
It would not be very linear, but you just need it to be monotonic.

[PCB.Wiz]

So I'm looking for a hollow-shaft Hexagonal 8mm potentiometer.

To reply to your original question, Panasonic have what they call Center Space Rotary Potentiometers, 300  degrees and 20mm or 25mm hole diameters,
They are too tall for your physical situation, but may interest others looking at the heading.

https://na.industrial.panasonic.com/products/electromechanical/encoder-potentiometers/lineup/rotary-potentiometers

[Ian.M]

Sorry It looks like I need a better explanation.
The rod is coming out of a gear assembly, total length is 9mm.
The rod is then going into a valve housing. The hexagonal hole in the valve housing is 13mm deep.
Gear assembly and valve housing is mounted together and can be splitted. 

I was thinking if I add spacers between them, I can get 3mm access to the rod and can
place a potentiometer there.   

The thing that would concern me is the hex shaft's engagement with the valve if you significantly increase the spacing.  Separating them by 3mm will only leave 6mm of shaft in the valve, and with the engagement significantly less than the shaft diameter, there is a high risk of rounding off the corners or wallowing out the hex hole in the valve if the valve ever gets stiff to operate.

A solution with minimal extra separation is therefore desirable.  Personally I'd be looking at the possibility of laser cutting a spur gear 1 mm or so thick, and if backlash is a concern,  stacking on top of it a spring loaded annular spur gear ring to eliminate backlash.  The center of the annular gear would clear the boss on the valve housing so no extra height would be required for it.  The  sensor would have a plain spur gear of double the thickness.  Use a metal plate as the spacer and you'd have something to mount the sensor to.

If any fluids containing water, or corrosive gasses are involved, it would strongly favour  magnetic or inductive sensors, with optical as a less desirable third choice, as the humidity from any weeps from the valve seals or plumbing connections would be likely to rapidly degrade any moving contact sensors.

[tru3533]

Ian.M
Yes, you have a good point, and it seems to be the best solution to use spur gears reducing the opening to 1mm.
I think 26mm outer diameter on the gears will be OK.

Using the https://evolventdesign.com/pages/spur-gear-generator I get the included calculation, not shure how to optimize it to my use.
Next question will be where to print it and material to use

[Ian.M]

I think it would be simplest to have the same tooth count (and thus diameter) for both the shaft and sensor gears.

While you *can* run plastic on plastic gear tooth contact, you need a high grade surface finish and accurate tooth profile for that to be reliable and remain low backlash long term, so personally I'd prefer to find a commercially available brass metric spur gear with bore and hub to fit your proposed sensor, and print/lasercut a plastic gear to match as plastic on brass is more forgiving.  If you've only got a FDM printer you aren't going to be able to print fine pitch gears with an accurate enough tooth profile and good enough surface finish, so unless you farm it it out to a specialist 3D printing shop, finer than module 2 probably isn't a great idea unless you have a SLA (resin) printer or other technology printer capable of much better surface finish and dimensional accuracy than FDM.  Even with FDM, you'll struggle producing high enough quality module 2 gears unless its well tuned with spot-on slop and shrinkage compensation.

IMHO laser cutting with kerf compensation is your best shot at making a hassle-free standard pressure angle and module gear to fit your shaft.  You'll probably want it to be a close sliding fit (or only slightly tighter) on the shaft and maybe run it between a pair of 0.2mm PTFE washers so it doesn't wear on the gearbox or valve bosses.

Choice of plastic is going to depend on the fabrication technology.  ABS would be the minimum spec for FDM printed gears unless you can guarantee it will never be above room temperature.   Delrin or an Acetal copolymer would probably be the best choice for laser cutting, though other plastics that you know cut well in the laser cutter you are using may have good enough engineering properties to be suitable.  If possible avoid Acrylic - while it cuts beautifully and can make excellent gears, it is far too prone to stress cracking and embrittlement if exposed to a wide list of common solvents, cleaning agents and lubricants.  I cant comment on choice of SLA resins, though due to the work involved cleaning and recalibrating the printer to change resins, you are generally stuck with whatever its set up for!

[mariush]

You could make a custom round circuit board with a hexagonal center and just make tiny holes or slits going from center towards the edge, and at various degrees.

Have on one side of the disc a row of leds with very narrow angle and very small produce light that goes through those small holes, and on the other side have some sensors to pick up the light when present.

You could have for example 6-8 concentric tracks ... at 0 degrees only light from leds 1  and 4 would shine through,  at 5 degrees you'd have light from leds 1 ,4 but also 2 and 8 , and 10 degrees some other mix of leds ... this way you'd also know the direction

See for example : https://www.researchgate.net/figure/A-5-bit-rotation-encoder-using-a-Gray-code-pattern-and-an-optical-sensor_fig2_26498639  - you could stretch it to more than 5 bits if you need more precision.

[Ian.M]

That's a good idea, and as thin as my suggestion.
However I think that retroreflective optosensors would be a better choice, so no need for components on or contacts to the rotating PCB.   Bare ENIG pads and matte black solder mask should give plenty of contrast for the sensors.

[PCB.Wiz]

That's a good idea, and as thin as my suggestion.
However I think that retroreflective optosensors would be a better choice, so no need for components on or contacts to the rotating PCB.   Bare ENIG pads and matte black solder mask should give plenty of contrast for the sensors.

For standard Gray code designs (as opposed to the single track Gray codes I linked above) I did find these sensor arrays
https://www.pololu.com/category/123/pololu-qtr-reflectance-sensors
Those are standard, jellybean, high volume reflective sensors, so are not super small but they give a great mock-it-up ability.
You can get smaller reflective sensors, but I've not see pre-done boards using them.

Have we seen a diameter-max for the disk that can fit here yet ?
The boards above in 4mmx5 and 4mmx6 come in 21mm and 25mm

[PCB.Wiz]

You could make a custom round circuit board with a hexagonal center and just make tiny holes or slits going from center towards the edge, and at various degrees.

Have on one side of the disc a row of leds with very narrow angle and very small produce light that goes through those small holes, and on the other side have some sensors to pick up the light when present.

Yes, if you were doing a thru-a-disk design, the smallest radial length using practical sensors could be to use reflective sensors like ITR1502SR40A, but not actually used in reflective mode.

Instead, they would be placed in rotated pairs above and below the disk so each pair reads through two tracks.
Those sensors have 1.6mm pitch lens, so a staggered layout on 3.2mm package centres of 3 packages each side, would sense 6 tracks over 8.0mm or 10.4mm total package radial length.
Such an arrangement is more than twice as radially-dense as the standard pololu boards, if space is tight.

[Infraviolet]

The thing to really note here is that while you may only have 3mm of space on the hex rod itself for your system to couple to, as soon as you get to a larger radius you've got a lot more vertical space to work in on either side of that 3mm thick region. A gear, for example could be printed to be thin enough to fit the 3mm hex where it mounts to the hex, but then when reaching a wider radius where the teeth are it can get thicker. Similarly for an optical transmission encoder the disc might only be 3mm thick (or 3mm thick at the centre and thicker at the radius where your code tracks are) you don't need to fit the circuitry either side of the disc in to the same 3mm thick zone.

You could also  consider TLE49 or AH3772-SA-7 style magnetic latching hall switches, and a 3d printed disc in to which you could embed 2mm thick magnets. I've done it for single track incremental quadrature encoding but there's no reason you couldn't do an absolute position gray code with more sensors at diferent radii and patterns of magnets at different radii too(because of how magnets affect the hall latches you might not be able to have a series of magnets together all keep the latch in the same state, so single track might not work here).

P.S. what exactly is the gear assembly, could it be modified or otherwise replaced by something with a longer output hex shaft if you needed mroe length to work in whilst still having deep engagement of the hex shaft in to the valve socket.

[PCB.Wiz]

If any fluids containing water, or corrosive gasses are involved, it would strongly favour  magnetic or inductive sensors, with optical as a less desirable third choice, as the humidity from any weeps from the valve seals or plumbing connections would be likely to rapidly degrade any moving contact sensors.

Good point, if the environment excludes optical, there are also capacitive contactless sensors
Attached is an example Capacitive PCB patterning that claims high fractional degree precision.
For the OP's use, a single modulated track + coupler (tagged rough in the image) may be 'good enough' with the 4 quadrant read offs.