Integrating Encoder Into Gearbox

[Owen Busler]

Good evening CD,

Next year my team is looking to create our own custom gearboxes. We are quite confident in our shop bot and we have made a few test gearboxes so far and everything is looking great. Now we come to the question of how to mount our encoders into our robot. We have about 10 of these encoders:

http://www.usdigital.com/products/en...otary/shaft/H1

and although they are bulky we really like how they perform so we would like to keep them. In the past we have had them on our drive axles with 3d printed mounts and it never came out nicely because the mounts didnt hold up. Next year we were looking into a more robust system. One plan of attack was to lathe a 1/4 inch hole into a shaft in the gearbox and hold the encoder in place with a set screw. Here are links to two different angles. My biggest concern with this design is losing a lot of integrity in that hex shaft. I was hoping to get thoughts on whether or not this would hold up?

https://drive.google.com/open?id=0B6...VFSaVFpTDRobk0

https://drive.google.com/open?id=0B6...WhZeGVUVFZESGM


Thanks so much!

[AustinSchuh]

You are being too concerned with space in your design. Poke the shaft 3/8 - 1/2" past the gearbox plate, and put the set screw on the other side. That's what we do.

When you *turn* the hole into the shaft, finish it with a reamer. You'll get a nicer fit. Probably a .2505 reamer.

For mounting, we've been bending a lexan z out of 1/32" lexan and using that to constrain the encoder torsionally. It seems to be working fine. There should be some pictures on our picasa site of how we've solved this exact problem before.

[asid61]

Originally Posted by AustinSchuh

You are being too concerned with space in your design. Poke the shaft 3/8 - 1/2" past the gearbox plate, and put the set screw on the other side. That's what we do.

When you *turn* the hole into the shaft, finish it with a reamer. You'll get a nicer fit. Probably a .2505 reamer.

For mounting, we've been bending a lexan z out of 1/32" lexan and using that to constrain the encoder torsionally. It seems to be working fine. There should be some pictures on our picasa site of how we've solved this exact problem before.

+1 on the Z-shaped lexan/polycarb mounting. We used 1/16" and it's fine as well. also +1 on sticking the shaft out of the gearbox farther, it helps with making mounting easier.
We have used set screws in the past and it's worked fine. We used the plastic/nylon tipped ones from McMaster to avoid killing the encoder shaft.
Some teams have used surgical tubing for a connection between shaft and encoder, but personally we've experienced lots of slippage both in absolute and incremental situations. That being said, it definitely helped avoid breakage due to the flexible coupling.

[AustinSchuh]

Originally Posted by asid61

Some teams have used surgical tubing for a connection between shaft and encoder, but personally we've experienced lots of slippage both in absolute and incremental situations. That being said, it definitely helped avoid breakage due to the flexible coupling.

We tried that for years, and eventually figured out that the rubber connection was actually acting as a filter, and was causing controls problems. We were kind of shocked.

You can also make a male end to your shaft rather than female, and use a shaft coupler. There are flexible shaft couplers. Unfortunately, that's a bit more pricy of a solution.

[Bruce Newendorp]

Aluminum flashing found in the roofing department at your local Home Depot or Lowes also works well to make the Z bracket.

[mman1506]

We just ream a .25 hole and use some retaining compound to permanently attach our encoders to the hex shaft. The wires were then zip tied down without any slop preventing the encoder from rotating. I'm fairly sure 254 does the same thing. Never had any issues with structural integrity.

[AustinSchuh]

Originally Posted by mman1506

We just ream a .25 hole and use some retaining compound to permanently attach our encoders to the hex shaft. The wires were then zip tied down without any slop preventing the encoder from rotating. I'm fairly sure 254 does the same thing. Never had any issues with structural integrity.

254 makes it a slight press fit, and secures with the wires.

I spent a long time tuning a PID loop for one of 254's bots, and was only able to fix the issue by securing the pot by a bracket instead of by the wires. Sure, it works most of the time, but I'm no longer willing to take most of the time as an answer.

[The Swaggy P]

The shaft should easily hold up to the stress, you should focus more on reducing the profile of the encoder outside the gearbox.

[InFlight]

Originally Posted by The Swaggy P

The shaft should easily hold up to the stress, you should focus more on reducing the profile of the encoder outside the gearbox.

The the encoder is adjacent two 2.5 Inch OD CIM Motors that protrude 4.5 inches out the rear of the gearbox. Minimizing the encoder profile outside the gearbox provides no advantage, and is counter to having a more reliable gearbox system.

The mechanical specs for this encoder are on the website. Please take note of the hole tolerances. CAD is one thing; student repeatable machining of center holes to a 0.0003 inch tolerance on center is quite another.

[asid61]

Originally Posted by InFlight

The the encoder is adjacent two 2.5 Inch OD CIM Motors that protrude 4.5 inches out the rear of the gearbox. Minimizing the encoder profile outside the gearbox provides no advantage, and is counter to having a more reliable gearbox system.

The mechanical specs for this encoder are on the website. Please take note of the hole tolerances. CAD is one thing; student repeatable machining of center holes to a 0.0003 inch tolerance on center is quite another.

According to the product page (link's broken) the maximum TIR allowed is 0.006", so there's a bit more breathing room than that. Still not much though. I think the 0.0003" is the encoder shaft relative to the encoder, otherwise anything less than an ABEC-5 bearing could be out of tolerances.

[scottandme]

Originally Posted by InFlight

The the encoder is adjacent two 2.5 Inch OD CIM Motors that protrude 4.5 inches out the rear of the gearbox. Minimizing the encoder profile outside the gearbox provides no advantage, and is counter to having a more reliable gearbox system.

The mechanical specs for this encoder are on the website. Please take note of the hole tolerances. CAD is one thing; student repeatable machining of center holes to a 0.0003 inch tolerance on center is quite another.

971 Encoder Mount Example Here

It's a pretty simple operation in the lathe - just drill undersize and then ream (as suggested by Austin). The doc you linked just establishes that the encoder shafts are made 2 to 6 tenths undersize, just means that you can probably get away with a 0.2500" reamer if that's all you have on hand.

You want a mount that prevents the encoder from rotating, and doesn't create any angular or parallel misalignment. The bent lexan design will allow the encoder to "float" a bit, which solves the parallel/angular problem.

[InFlight]

There seems to be an assumption that a basic lathe comes from the factory, and continuously maintains perfect alignment between the tail stock ram axis and the spindle axis; to allow perfectly on-center holes to be drilled.

The reality is it takes a somewhat experienced machinist with a test bar and a dial indicator to adjust the tail stock to be aligned perfectly laterally and vertically. The average FRC lathe is use is likely only accurate to 0.01 inch on-center.

Drilling center holes to the stated -0.0000/+0.0006 on center tolerance would be a non-trivial challenge for many FRC teams. To do it consistently you would need to use a collet holder in the spindle.

[scottandme]

Originally Posted by InFlight

There seems to be an assumption that a basic lathe comes from the factory, and continuously maintains perfect alignment between the tail stock ram axis and the spindle axis; to allow perfectly on-center holes to be drilled.

The reality is it takes a somewhat experienced machinist with a test bar and a dial indicator to adjust the tail stock to be aligned perfectly laterally and vertically. The average FRC lathe is use is likely only accurate to 0.01 inch on-center.

Drilling center holes to the stated -0.0000/+0.0006 on center tolerance would be a non-trivial challenge for many FRC teams. To do it consistently you would need to use a collet holder in the spindle.

You're confusing the shaft diameter tolerance with TIR. The H1 encoder has a maximum TIR of 0.006", and the shaft itself is between 2 and 6 tenths undersize (-0.0002 to -0.0006).

So the shaft diameter has a tight tolerance, but the encoder might have up to 6 thou of runout, an order of a magnitude larger than the shaft diameter tolerance.

In any case - using a reamer would give you a nice clearance fit, and then use a set screw/retaining compound/whatever to hold.

If your lathe is poorly setup, and doesn't drill holes on center:

1) Fix that
2) It doesn't matter - since the 1/32" piece of lexan holding the encoder will happily flex/float if the encoder has any axial/parallel misalignment. Same idea as using a helical beam coupler/flexible coupling that can deal with axial/parallel/radial misalignment.

[InFlight]

Originally Posted by scottandme

You're confusing the shaft diameter tolerance with TIR. The H1 encoder has a maximum TIR of 0.006", and the shaft itself is between 2 and 6 tenths undersize (-0.0002 to -0.0006).

So the shaft diameter has a tight tolerance, but the encoder might have up to 6 thou of runout, an order of a magnitude larger than the shaft diameter tolerance.

In any case - using a reamer would give you a nice clearance fit, and then use a set screw/retaining compound/whatever to hold.

If your lathe is poorly setup, and doesn't drill holes on center:

1) Fix that
2) It doesn't matter - since the 1/32" piece of lexan holding the encoder will happily flex/float if the encoder has any axial/parallel misalignment. Same idea as using a helical beam coupler/flexible coupling that can deal with axial/parallel/radial misalignment.

If you measured the current runout of your own lathe you might be surprised. Most will be over the 0.006 value which was my point.

If your drilled off center; your applying a bending load into the resolver shaft every rotation with direct mounting. At 100 rpm your going to be challenging a floating mount, and getting higher loads in the resolver bearings.

This is the reason many of us prefer alternate non-direct mounting. For close mounting a no contact magnetic rotation sensor would make far more sense.

[Owen Busler]

Thank you everyone for your replies! I really like the robustness of this design in addition to to how it doesn't require much sacrifice in terms of material in the load bearing part of the shaft. I used some 1 inch leg .75 inch tall 1/16" thick Z 6061 to support it and I think this will work very well thanks again!

https://drive.google.com/open?id=0B6...VpFZ05IY0lNLUE