Cheap high resolution rotation sensors, do they exist?

[Marco]

Do I have a blind spot or are there no devices to make a cheap endless rotation dial with high resolution? The Hall based devices and the high precision optical encoders are ridiculously expensive.

I've only seen some suggestions for low cost hacks, like ganging up two infinite turn hollow shaft potentiometers or connecting a dial to a XY joystick. Are there any such cheap potentiometers to gang up? I see the Piher ST15, but no one is selling it, and the N15 which isn't particularly cheap.

[Benta]

Define cheap.

[Marco]

Define cheap.

Couple bucks.

[ataradov]

Things like this http://www.ebay.com/itm/Rotary-Encoder-400P-R-6mm-Incremental-Optical-Shaft-Working-Measurement-5-24V-ED-/131969802647?hash=item1eba036197:g:bAsAAOSw4GVYQDcC are pretty cheap for what they are. And they are actually pretty well-made.

I doubt you will find anything for much less than $10.

It all also depends on what's your input speed. 1 RPM is very different from 1000 RPM.

[Marco]

As I said, just for a dial (ie. a jog/shuttle dial). Was a bit surprised that there is simply nothing out there except low resolution rotary encoders and the stuff meant for machine automation.

You'd think there'd be something like a ready made ganged up potentiometer for that.

[artag]

These are good enough for most purposes :

http://www.ebay.com/sch/i.html?_from=R40&_trksid=p2047675.m570.l1313.TR1.TRC0.A0.H0.Xalps+encoder.TRS0&_nkw=alps+encoder&_sacat=0

But the cheapest rotary encoder with fairly high resolution used recently is the rubber-ball mouse, universally made using a plastic wheel and a pair of IR diodes. Since most cheap jog controls are injection mouldings, it probably makes no sense to have an encoder component. Just a slotted wheel and IR sensors on the pcb.

[Ian.M]

If a relative encoder like a slotted wheel doesn't do it for you, (and its resolution and precision is only limited by its diameter and how accurately you can mount the prefocussed LEDs and phototransistors to your PCB) you could mount a reasonably strong magnet cross-wise on your shaft and use two linear output hall sensors at rightangles to each other.to get the absolute sine and cosine components of its angular position.

[Marco]

I was rather looking for something low effort Guess a rotary encoder will have to do.

[Ian.M]

Its the consumer market race to the bottom -  cheap custom rotary encoders based on the quadrature encoder disk principle + touch sensing have driven all the pre-packaged solutions either out of the market or to a higher price point.

The hall sensor approach is fairly low effort - just mount the sensors in the correct places - easy if there is a PCB behind the front panel and sample them with two ADC channels then use the C library atan2() function to directly calculate the angular position.

[artag]

The Austria Microsystems magnetic position sensors will do that hall effect function for you.

http://ams.com/eng/Products/Magnetic-Position-Sensors

[james_s]

Those $10 ebay optical encoders look pretty good to me, I wasn't aware anything like that existed in that price range, that sort of thing used to be quite expensive. At that price you could buy one to play with.

[ataradov]

that sort of thing used to be quite expensive.

Yep. And I have a few, they are very well made, run perfectly true and smooth. They are just a shaft on two bearings (one outside and one on the inside) with an optical wheel attached to it. Plus some optical sensors and a small PCB with logic.

[joeqsmith]

Pull a mouse out from the trash bin and steel the parts for the thumb wheel encoder from it.  Instant free rotary encoder. 

[james_s]

That does work, although removed from the mouse there is no mechanical framework to hold it all together. It would be easier to buy one of those $10 optical encoders, unless you have a lathe and want to fabricate something.

[Cerebus]

Here's a thought.

I know high resolution linear encoders have been made using the moiré effect. They have two transparent bars that slide over each other, one has dark lines on it, one has slightly slanted lines* and then the whole thing is sensed with an LED one side and a photodiode the other. Observed at one point, the moire effect causes the movement to be 'amplified' so that you see many more flashes of light/darkness than you would if you simple had vertical lines of the same spacing. It also effectively multiplies the precision of your line markings and sensor aperture.

Now imagine this wrapped around into a disk instead of set out linearly. Then add a second in quadrature and a couple of opto-interrupters. I'd imagine that you could achieve something on the close order of 1000 counts per revolution with lines that could be printed quite easily onto a piece of acetate with a laser printer or inkjet printer. For the basic mechanics it could be built on top of an existing cheap rotary encoder.

*There's a variant that uses parallel lines for both parts, with a vernier like difference in the line spacings.

[bson]

Not that hard to find, though I don't know what constitutes "high resolution" in your mind.  Bourns: 30 detents (15ppr), 6mm shaft, press function.  I use these and find they're quite good.  They're mechanical.  Not superlative, but certainly better than the $2 price suggests!
http://www.mouser.com/Search/ProductDetail.aspx?R=PEC11L-4220F-S0015virtualkey65210000virtualkey652-PEC11L4220FS0015

[Marco]

The Austria Microsystems magnetic position sensors will do that hall effect function for you.

I hadn't seen the AS560x yet, that's an interesting price point. Especially the AS5601 with button press detection.

[mark03]

The annoying thing about optical encoders (besides the fact that Ebay isn't a long-term sourcing strategy), is that someone, somewhere has apparently decreed that all of them shall require high voltage (5V) to operate.  This tends to rule them out in portable equipment, for me anyway.

I'm just waiting for an enterprising Chinese factory to resurrect the cheapo old-mouse-style plastic encoders for the hobbyist market, revamped for the 21st century:  IR LED without internal series resistor so you can run it at 1.8V or 3.3V if you like, which also sets the digital output level.

[Gyro]

Have you got any old USB optical mice lying around? A lot of the three button ones use an optical image sensor and separate microcontroller to handle the buttons and USB interface. The interface between the sensor and the micro?... Quadrature, both X and Y axes  .

All you need is to strip it down and use a disc with a bit of surface texture (eg, randomly sanded aluminium) and you have a very high resolution encoder!

P.S. That's assuming you want one-off of course!

[james_s]

The annoying thing about optical encoders (besides the fact that Ebay isn't a long-term sourcing strategy), is that someone, somewhere has apparently decreed that all of them shall require high voltage (5V) to operate.  This tends to rule them out in portable equipment, for me anyway.

I'm just waiting for an enterprising Chinese factory to resurrect the cheapo old-mouse-style plastic encoders for the hobbyist market, revamped for the 21st century:  IR LED without internal series resistor so you can run it at 1.8V or 3.3V if you like, which also sets the digital output level.

There is nothing in the encoders besides a pair of LEDs and phototransistors, they will probably work fine on 3.3V. If you need a large quantity for production you can probably buy them direct from the manufacture. Depending on quantity they can likely customize them for whatever voltage you want, it's only a matter of adjusting resistor values.

Mechanical encoders seem to have taken over the low end of the market, I never cared much for them though, I've always had difficulty obtaining the right balance of performance and glitch resistance, especially as they get worn.

[CM800]

The annoying thing about optical encoders (besides the fact that Ebay isn't a long-term sourcing strategy), is that someone, somewhere has apparently decreed that all of them shall require high voltage (5V) to operate.  This tends to rule them out in portable equipment, for me anyway.

I'm just waiting for an enterprising Chinese factory to resurrect the cheapo old-mouse-style plastic encoders for the hobbyist market, revamped for the 21st century:  IR LED without internal series resistor so you can run it at 1.8V or 3.3V if you like, which also sets the digital output level.

The reason that'd be the case for optical encoders, is that most applications for them are 5V, they go on the end of a cable to a servo drive or the like to do 3.3V max would limit their consumer base.


1.8 V? .-.

[yngndrw]

The annoying thing about optical encoders (besides the fact that Ebay isn't a long-term sourcing strategy), is that someone, somewhere has apparently decreed that all of them shall require high voltage (5V) to operate.  This tends to rule them out in portable equipment, for me anyway.

I'm just waiting for an enterprising Chinese factory to resurrect the cheapo old-mouse-style plastic encoders for the hobbyist market, revamped for the 21st century:  IR LED without internal series resistor so you can run it at 1.8V or 3.3V if you like, which also sets the digital output level.

There is nothing in the encoders besides a pair of LEDs and phototransistors, they will probably work fine on 3.3V. If you need a large quantity for production you can probably buy them direct from the manufacture. Depending on quantity they can likely customize them for whatever voltage you want, it's only a matter of adjusting resistor values.

I have one of those optical Ebay-special encoders, it's really smooth and light to turn and the cable is shielded. I've also used it at 3.3V and it works perfectly as you suggested, which is really handy when using them with FPGAs and modern microcontrollers.

The reason that'd be the case for optical encoders, is that most applications for them are 5V, they go on the end of a cable to a servo drive or the like to do 3.3V max would limit their consumer base.

1.8 V? .-.

Despite the spider in your avatar (), I tried mine on 1.8V but sadly it didn't work. However as I said above, it works fine at 3.3V. They are open-collector outputs with no in-built pull-up resistors.

It should also be noted that there are no detents and they are rated for a maximum of 5,000RPM. This was the exact one that I purchased: http://www.ebay.co.uk/itm/131515172625

[voltsandjolts]

This was the exact one that I purchased: http://www.ebay.co.uk/itm/131515172625

How about a mini-teardown?

[yngndrw]

How about a mini-teardown?

So needy. You'll have to live with my phone camera.

Three small Phillips screws hold the outer case on, the cable just pushes through a rubber grommet.

The cable has a cable tie to prevent it from pulling through the case, hot-melt glue is used to protect the cable joints and the board is screwed onto the face place with some brass spacers. The shaft looks to have two separate ball bearings but the (Plastic ?) disk has a slight wobble so the encoder may not be perfectly linear.


The board is single-sided, apart from the optical sensor which is mounted on the back. There's a 7805 regulator (Glued down) on the board and the soldering looks to be of decent quality. There looks to be a diode at the top by the shield wire, but it's covered in glue.




Just had a quick buzz around with my multimeter and have worked out the main part of the circuit.

The shield is connected to the metal case, the ground is not. The input power wire gets regulated through the 7805.

The output of the optical sensor looks to be open collector - It has a 4.7k pull-up resistor to +5v and then a 1k resistor to the base of the transistor. The transistor pulls the output down to ground through a 20 Ohm resistor. There's an up-populated resistor pad for each output for an optional pull-up resistor - This would pull the output of the transistor up to the input rail, not the regulated 5v rail.

[LaserSteve]

Not cheap, but I've had good luck with US Digital for small projects over the years. They will sell just the disk..

http://www.usdigital.com/products

Steve