CIM motor testing

[efoote868]

I'm looking to create a test to calculate (and hopefully rank) the performance of the many CIM motors in our robot closet.

The approach would be to have a CIM hooked to a winch, and hoist a known load up a known distance, and measure the time it takes.

I've assigned several students the task to determine what these values should be, based on the datasheet curves and gearboxes we already have laying around. I'm going to try and target the test taking about 3-5 seconds, while using the FRC robot control system and components.

I'm looking for any advice others might have, specifically how to control confounding variables that might skew the test results.

One item I've already identified is battery discharge over the course of testing; I'm looking at purchasing a power supply to run the motor (in lieu of a battery).
Was wondering if powering the setup from this would work (having trouble finding anything above the 1000W 12V range):
http://www.ebay.com/itm/RC-POWER-SUP...-/301361360478

Keeping in mind that I'm not looking to stall the CIM for any amount of time.
Also I'm wondering what sort of variations we'll see.

Once again any advice is appreciated, I'd love to hear everyone's thoughts.

[Ether]

Originally Posted by efoote868

any advice is appreciated, I'd love to hear everyone's thoughts.

Assuming you're mostly interested in relative ranking of the motors, you could build a simple test rig using an encoder to measure speed, an ammeter to measure current, and a CIM to generate a load.

With the load switch open, close the power switch and record speed and current. Close the load switch and record speed and current. Open the power switch and install the next CIM to be tested. Repeat.

You could also include a double-pole-double-throw switch to quickly reverse the power leads, and test in both directions.

[GeeTwo]

Originally Posted by Ether

Assuming you're mostly interested in relative ranking of the motors, you could build a simple test rig using an encoder to measure speed, an ammeter to measure current, and a CIM to generate a load.

An even easier-to-build equivalent test rig would be a TB-mini or other 2-CIM gearbox with a built in encoder hub. Just wire the load CIM in reverse, adjust the encoder counts for the gear ratio, and otherwise it's the same. Let the gearbox output shaft spin freely.

[efoote868]

I must admit, I didn't think of that.

I like the simplicity of the design, but the other part of this is I'm trying to keep a group of high schoolers busy with the project
I might do that as a secondary test to see if the results align.


Doing a little more research on power supplies, seems I might be able to buy a used server power supply with a single 12v rail and use that.
HP Integrity RX6600 Server Power Supply Model RH1448Y PN: 0957-2198
Looks like it's rated at ~130A, and can be purchased for around $80.

[Richard Wallace]

Originally Posted by efoote868

I must admit, I didn't think of that.

I like the simplicity of the design, but the other part of this is I'm trying to keep a group of high schoolers busy with the project
I might do that as a secondary test to see if the results align.

The test Ether and Gus recommend will give you results that are much less sensitive on variation due to internal motor heating, compared to the winch lift test. Lifting will tempt you to use heavier loads so that elapsed time is longer, and therefore easier to compare. However, longer test duration will heat the motors, causing maximum available power to decrease significantly.

See my results from a few years ago, and VexPro's more recent three-minute power test results, to understand how internal motor heating reduces maximum power.

[efoote868]

Originally Posted by Richard Wallace

Lifting will tempt you to use heavier loads so that elapsed time is longer, and therefore easier to compare. However, longer test duration will heat the motors, causing maximum available power to decrease significantly.

If the goal of testing is to find the 4 best motors for a drive train, would that help the goal? As long as the motors are subject to the same condition (i.e. motor in test starts room temperature, is subject to test exactly 10 times, and then wait X hours before testing with motor again).

Also thanks for the links, they're very good resources. As an aside, it looks like the VEX folks had motors catching fire in under a minute.

[Richard Wallace]

Originally Posted by efoote868

If the goal of testing is to find the 4 best motors for a drive train, would that help the goal? As long as the motors are subject to the same condition (i.e. motor in test starts room temperature, is subject to test exactly 10 times, and then wait X hours before testing with motor again).
...

Yes, if you carefully pre-condition the motors to be tested. Checking case temperature is not sufficient to tell you if the internal temperature is stable. I think Aren has some results indicating time to reach stable internal temperature when the motor is being cooled by a fan.

[James Juncker]

If you are using some system I would recommend the use of a micro switch at the end of a lift should you go with that method to measure the time from start to stop. Similarly I would recommend to lift weight as the motors won't be used at their no load ability on a robot. I also think that some consideration ought to be given to building a gearbox to change motors in and out of and measuring their individual power consumed and difference in speed.

[Ether]

Originally Posted by James Juncker

Similarly I would recommend to lift weight...

The proposed test rig with a "load CIM" puts a substantial load on the CIM being tested when the load switch is closed.

Originally Posted by James Juncker

...the motors won't be used at their no load ability on a robot.

No load (or low load) speed (load switch open) can be an indicator of motor health, and since it's so simple to test it's worth testing.

[efoote868]

Originally Posted by James Juncker

If you are using some system I would recommend the use of a micro switch at the end of a lift should you go with that method to measure the time from start to stop. Similarly I would recommend to lift weight as the motors won't be used at their no load ability on a robot. I also think that some consideration ought to be given to building a gearbox to change motors in and out of and measuring their individual power consumed and difference in speed.

Those are all very good suggestions, thank you.

Part of the testing will be to get the students to program the FRC controller to measure the time it takes for the test to elapse, and whether it's a micro switch, encoder count or some sort of photo switch (or combination thereof) is still yet to be determined.

I'll be interested in seeing the difference in power characteristics. I'm not sure what we'll have available in terms of data logging, and I'm very curious to see how well the motors fit the datasheet.


EDIT:

Originally Posted by Ether

The proposed test rig with a "load CIM" puts a substantial load on the CIM being tested when the load switch is closed.

Would you think that the energy the "load CIM" dissipates would change its resistance characteristics? Are there any steps we can take to mitigate the effect?

[Ether]

Originally Posted by efoote868

Would you think that the energy the "load CIM" dissipates would change its resistance characteristics?

If the load CIM gets hot, it will change its load vs speed characteristics.

Are there any steps we can take to mitigate the effect?

Richard addressed that concern in his earlier post. With the proposed load CIM test rig, it takes only a second or two to load the test CIM and record the loaded speed and current. The short test times mitigate the effect: close the load switch, record the speed and current, open the load switch.

Safety warning: If you decide to use such a setup, make sure to use a heavy-duty load switch and that is properly insulated/shielded (including cables). I'm not sure how much instantaneous voltage will be generated when the load switch is opened. Maybe someone (some adult) could test this. If the voltages are high enough to be a concern, a condenser (old-school term) -- like the ones that were used to keep the automobile distributor points from pitting back in the pre-electronic-ignition days -- could be placed across the load switch.

[efoote868]

Originally Posted by Ether

Safety warning: If you decide to use such a setup, make sure to use a heavy-duty load switch and that is properly insulated/shielded (including cables). I'm not sure how much instantaneous voltage will be generated when the load switch is opened. Maybe someone (some adult) could test this. If the voltages are high enough to be a concern, a condenser (old-school term) -- like the ones that were used to keep the automobile distributor points from pitting back in the pre-electronic-ignition days -- could be placed across the load switch.

I think I'll use a speed controller instead of the power switch (bonus if it has current monitoring built in), and I don't have any intention of opening the load switch with the cim being tested under power. But if I absolutely have to do that, I'll probably invest in a contactor or snubber diodes to do it safely.