Mounting encoder wheels

[growler]

Does anyone have any experience with making/mounting small encoder wheels?

I've made a couple little boards based around these, http://www.sparkfun.com/products/9542, with the intent of making quadrature encoders for a pair of gearmotors that are 1 cm thick. The motors are being held flat against the chassis of the robot and the wheels need to be mounted to a 1mm diameter shaft coming out the back end of the motor. As a result, I only have 5 mm of clearance so the wheels themselves are pretty small.

I've been having trouble getting the encoder wheels mounted on the shaft in even a semi robust way. I've tried blue tack but I'm not happy with it. Does anyone have any advice?

[waltr]

I'd put the LED and sensor flat on a pc board (0.013 thick PCB if space is very tight). Then mount the board perpendicular to the shaft so the sensor 'looks' at the inside wheel face. Put the encoder disk on the inside of the wheel.

Like these:
http://www.societyofrobots.com/sensors_encoder.shtml
http://www.robotshop.com/productinfo.aspx?pc=RB-Sbo-31&lang=en-US
http://www.acroname.com/robotics/parts/R238-WW01-KIT.html
http://www.seattlerobotics.org/encoder/200010/dead_reckoning_article.html
http://code.google.com/p/wheel-encoder-generator/

[growler]

Thank you for your help.

I have already mounted the sensors onto a pcb, that part of things works fine.

I also have the actual wheel graphic taken care of, my problem is the physical mounting of the disc. I'm not going to attach it to the wheel itself for two reasons. The rear shaft of the motor will give me higher resolution on the output because I can get motor rotation pre-gearbox. The other reason is that I've already mounted the things.

I am trying to attach the printed wheels to metal shafts that are 1mm in diameter. The discs will then spin in front of the sensors, which are mounted perpendicular to the shaft.

I think that I will try layering some paper discs with krazy glue to get a better thickness/stiffness then put a hole in it and use some more krazy glue to hold it onto the shaft. Wish me luck in not gluing the rotors in place .

[Soeren]

Hi,

I also have the actual wheel graphic taken care of, my problem is the physical mounting of the disc.

What diameter is the disc?
And how long is the 1mm axle?

I think that I will try layering some paper discs with krazy glue to get a better thickness/stiffness then put a hole in it and use some more krazy glue to hold it onto the shaft. Wish me luck in not gluing the rotors in place .

Hold the glue... An aluminum (or plastic) disc, as thick as the axle will hold with a 0.98mm hole (drilled in a stand) would be a better option and can be heated for mounting (large heating expansion coefficient).
A clothes button may do as well (just drill the center hole a wee bit tight)

If you cannot make the hole to size, a brass or copper tube with an inner diameter of 1mm can be used as an in-between, as it can be crimped a teeny bit before mounting for a close fit.
A thin brass/copper disc can be soldered to this and trued before mounting.
The easy way is to use a lathe of course (if possible).

Your sensor has its peak sensitivity at around 0.7mm, but you should be able to get good result with a gap of up to 2mm, so make sure there's no wobble in the disc, or the output will wobble as well.


My eyes just fell on a box of spur gears made of nylon with a 1mm hole and one side smooth as a babys behind and they could be turned down if needed by spinning the motor and gently pressing a piece of  sandpaper against the wheel. I assume almost every hobby/craft shop has got something similar.

[growler]

Hello,

What diameter is the disc?
And how long is the 1mm axle?

The discs I'm currently working with are slightly less than 1 cm in diameter.
The axle is ~4.5 mm.

Thanks for all the suggestions. I have pretty limited resources as far as tools, no access to a lathe at the moment. I'll have to give those some thought and find out which of my friends has a drill press but your final suggestion about the spur gear sounds the most promising.

Appreciate the input on the sensor, this is the first time I've used these but they are working as I hoped. I was planning on running the encoder lines through a comparator against a reference voltage to clean up the signal in the final design but I'm reading logic level changes already without that.

[Soeren]

Hi,

The discs I'm currently working with are slightly less than 1 cm in diameter.
The axle is ~4.5 mm.

Not room for that many segments then (if the QRE1113 must be able to differetiate them securely), but the axle length should be sufficient for a very good grip

I know you told us, that you have the graphics under control, but since I made a sheet, just to better visualize it, I put it up for whoever might need 10mm discs (without quadrature)

(Click pic to get the .pdf - LASER is much better than inkJet - print on sticky labels for easy application).

[...] suggestion about the spur gear sounds the most promising.

Any plastic or aluminum disc of 3mm to 4mm thickness should do with this small diameter and one of your local machine shops, auto repair shop or similar will have a drill press and perhaps a kind soul willing to drill a few holes for you.
Have an eye out for useable parts when shopping - you might find just the right piece among toys, kitchen utensils or in your local hardware store (complete with helpful shop clerks, if you go there in the quiet hours)

Appreciate the input on the sensor, this is the first time I've used these but they are working as I hoped. I was planning on running the encoder lines through a comparator against a reference voltage to clean up the signal in the final design but I'm reading logic level changes already without that.

Test it in different types of light. Fluorescence lightning can interfere (as well as direct sunshine). Shade the sensors if light gives marginal readings.

If you see problems that isn't caused by having the segments of the disc too close for a well defined on/off signal, you can let the reference voltage always be half the amplitude of the resulting signal. A little hysteresis will help as well.

[growler]

Soeren,

You are a gold mine of helpful suggestions.

My plan is to have 12 (6 white, 6 black) segments on the wheel. I was able to find several resources with a quick google search for generating wheel graphics so that part was pretty simple. My hope is that I can get quadrature encoding done by having the sensor placement on the board such that they will be reading half out of phase.

I agree about the printer, I had to get a lazer to make the boards anyway. I had not thought of using labels but that will make that step trivial.

Any plastic or aluminum disc of 3mm to 4mm thickness should do with this small diameter and one of your local machine shops, auto repair shop or similar will have a drill press and perhaps a kind soul willing to drill a few holes for you.
Have an eye out for useable parts when shopping - you might find just the right piece among toys, kitchen utensils or in your local hardware store (complete with helpful shop clerks, if you go there in the quiet hours)

I should have time to do some shopping this weekend. I thought about it, and there have to be factories churning out parts everyday that will do exactly this that are used in things I can cheaply buy. I'll just have to keep my eyes open. Its' not like I don't have plenty of other aspects of this thing to work on.

Test it in different types of light. Fluorescence lightning can interfere (as well as direct sunshine). Shade the sensors if light gives marginal readings.

I was actually just going to test in one kind of lighting then get the whole assembly boxed up. I figure that's one of the advantages of having the encoders on the non-ouput side.

If you see problems that isn't caused by having the segments of the disc too close for a well defined on/off signal, you can let the reference voltage always be half the amplitude of the resulting signal. A little hysteresis will help as well.

If I understand you correctly here, that's what the current test set up looks like, my logic is running at 5V and I'm using 2.5 for the reference.

I've got a quad op-amp sitting in front of me and I'm trying to wrap my head around schmitt triggers.

Thanks again for your help.

[Soeren]

Hi,

I was actually just going to test in one kind of lighting then get the whole assembly boxed up. I figure that's one of the advantages of having the encoders on the non-ouput side.

Aha, they're going to be inside a closed box, didn't know - that will keep things in check.

If I understand you correctly here, that's what the current test set up looks like, my logic is running at 5V and I'm using 2.5 for the reference.

I was thinking more along the lines of a signal being say 1V to 3V (or whatever), but in a closed box, it will be set and forget, as you won't have the values wandering up and down with the ambient light, so don't worry.

I've got a quad op-amp sitting in front of me and I'm trying to wrap my head around schmitt triggers.

It's not hard. Just make a comparator circuit with a predefined shift point and a certain input impedance and inject a little bit of the output into the non-inve4rting input and you've got hysteresis.

Imagine a circuit with an op-amp having an input impedance of say 10k to each input. The resistor to the inverting input comes from an ideal (i.e. 0 Ohm impedance) 2.5V source and you mount a 1MOhm resistor from output to the non-inverting pin (the positive feedback). Assuming that the source driving the non-inverting input (through a 10k resistor) is 0 Ohms as well and that the op-amp has a true rail-to-rail output (just for making it simpler to calculate, at what input potentials (voltages) will the output shift up and down?

When you understand this circuit, it's so much easier to understand how impedances and not-quite-rail will influence this circuit.

If you don't quite get how it all goes together, just experimenting with something that looks the part, perhaps assisted by a trimmer resistor - this should still get you a usable circuit 


BTW. Did you consider using a 40106 (6 inverters with Schmitt triggers).

[growler]

Søren

A clothes button may do as well (just drill the center hole a wee bit tight)

I ended up doing this and it works great, thanks for the advice.

The encoders are not working reliably yet but I'm pretty sure it is because I don't have the enclosure over them yet.

[Soeren]

Hi,

I ended up doing this and it works great, thanks for the advice.

Glad it worked and hope nobody saw you nick their buttons off their coats

The encoders are not working reliably yet but I'm pretty sure it is because I don't have the enclosure over them yet.

As long as there's not any direct light on them they should work. Don't close up the thing without getting them to work first - more screw driving then

If you measure the voltage at the point between the resistor and photo transistor without power to the OLED part of of the sensor, what voltage do you get?
With the LED on, what voltage do you get when over white and black segments respectively?
Perhaps use less segments.

Get it to work optimally reliable before installing it, or you may have something that only works at certain temperatures and/or at low speeds.

[growler]

Perhaps use less segments.

This will be my next step. I think the way I have the sensors laid out on the encoder boards is non-optimal, but from what I've been able to observe everything should work with wider black stripes.

Anyway, changing the pattern on the wheel is currently the easiest change to make since I have the buttons securely mounted to the motor shafts and its easy to print another label. Much easier than making new boards.

Edit: Success, everything seems to be working, the speeds reported by the encoders both correspond with what I can observe with a stopwatch and match each other to within about 1% when running at 100% duty cycle which is one of the reasons I started down this road in the first place.

Thanks for the help.

[Soeren]

Hi,

Edit: Success, everything seems to be working, the speeds reported by the encoders both correspond with what I can observe with a stopwatch and match each other to within about 1% when running at 100% duty cycle which is one of the reasons I started down this road in the first place.

Great  -  Sounds like your motors are good in that respect 
And thanks for the feedback!