pic: Custom encoder mount for Mini CIM/CIM



This 2729’s first try at a custom encoder mount for 8mm shaft motors (it works with both the mini CIM and CIM). We CNC milled it out of 2.5" diameter delrin stock ‘pucks’, cut about 5/8" thick. The milled channel perfectly holds a CUI AMT10 incremental quadrature shaft encoder, while still provided a solid mounting surface for the motor.

The AMT10 has a configurable resolution ranging from 48 up to 2048 at a max of 7500 RPMs which would work great for either motor, and is about $24 from Digikey.


http://www.digikey.com/product-detail/en/AMT102-V/102-1307-ND/827015

We discovered the AndyMark CIMcoder after designing this around the AMT10, but honestly the AMT10 provides configurable (and higher) resolution at less cost than the CIMcoder. [cost of CNC mill not included :slight_smile: ]
http://files.andymark.com/PDFs/CIMcoder_Spec_Sheet_1-27-16.pdf

Best of luck to everyone this year!

I like it! Although is there any reason you’re direct mounting it instead of on the output shaft of a gearbox or other?

Does that still leave you much space for pinion gear mounting inside a gear box? Very similar to this one.

Actually, according to CUI’s datasheet, the maximum RPM is 15,000RPM if you go down to 512PPR

Great design BTW! How much of the CIM shaft is left sticking out?

We wanted an option for cases where either a gear box isn’t used (like our shooter this year) or another type of encoder mounting just isn’t a convenient option.

I’m not sure, it might not work for some pinion to gearbox applications, but it works well for an 8mm inside diameter keyed shaft (or equivalent shaft adapter).

I hadn’t seen the one in the link before! Here I thought we had a novel idea! :slight_smile:

Right, it can go up to 15,000 RPM at a lower encoder resolution. I only quoted the range that a mini CIM or CIM could reach, but good catch.

I’ll measure the final shaft length tomorrow and let you know.

What exactly is CUI’s definition of “PPR”? “Pulses Per Revolution” is ambiguous.

Do they define it as

  • rising edges per rev on one channel?**

  • rising and falling edges per rev on one channel?**

  • rising and falling edges per rev on both channels?

Couldn’t find the term defined unambigously anywhere on the website or in the product datasheet.

Note: US Digital rates their encoders according to CPR, which they define as “Cycles Per Revolution”, which is unambiguous. For example, a US Digital 360 CPR encoder would mean 360 cycles per rev, which is the same as:

360 rising edges per rev per channel
720 rising and falling edges per rev per channel
1440 rising and falling edges per rev both channels

Good question sir, I’m not sure yet. I plan to test that tomorrow and I will let you know.

if you want to save cnc time and have a 3d printer.

Assuming1 for the moment that 2048 means “rising edges per rev per channel”, let’s do the math if you set this up for quadrature at 7500 RPM with a 20ms control loop period:

CPR: 2048$

decoding:4$

secondsPerControlLoop: 20/1000$

FPGAsamples: 127$

edgesPerControlLoop: RPM * (1/60) * (decoding*CPR) * secondsPerControlLoop = 20,480

secondsPerEdge2: float((1/RPM) * 60 * 1/(decoding*CPR)) = 0.9766e-6

secondsPerSamplesize: secondsPerEdge * FPGAsamples = 0.1240234375e-3

As can be seen, that is way overkill for operation at that speed.

1 I am inferring that because there are a couple of odd-number PPRs in the list, namely 125 and 385. You can’t have an odd number of “rising and falling edges per rev per channel” or “rising and falling edges per rev both channels”

2 This number is even smaller if quadrature tolerance is included. But the 40MHz sampling rate of the roboRIO theoretically should handle it.

There appears to be 13/16" of the shaft remaining.

I didn’t get a chance to test the encoder signal today.

We ran the encoder class on the encoder with the resolution set to 256. We got roughly 25,000 from the encoder class. 25,000ticks x (60sec/min) / (256ticks/rev) = about 5,859 RPMs. Since this was a full speed minicim I’d say the data was valid and I’d say the encoder’s resolution is based on rising (or falling) edge and not both. Otherwise the RPM data would not be right.

The beauty of this encoder is that we can set the resolution lower if there was a sampling rate issue, simply with dip switches.

You left out one piece of information crucial to your argument: decoding mode.

The mode doesn’t change the value, just the resolution. One cycle of the quadrature signals is one count.