4-cim vs. 6-mini-cim drivetrains

Over the summer we went from a 6 CIM drive, to a 4 CIM drive to a 6 mini CIM drive using VEXpro’s 3 CIM ballshifter.

6 CIM: Worked well when we didn’t trip the main breaker, Did get very warm but we were only playing defense both days (so hot our alliance members were fanning them with clipboards in between elim matches)

4 CIM: Worked very well, didn’t have any problems with overheating

6 MINI CIM: Worked amazingly, brought or adjusted top speed in high from 20fps to 22fps, Slightly less pushing power and slightly harder to turn. However half way through the semis they get almost too hot to touch. Also works okay enough with 5 turning and one ‘sleeping’.

We loved using the 6 mini cim drive this summer and will be using it again in the near future

canTalon.setCurrentLimit(amps);
canTalon.EnableCurrentLimit(true);

We used 40 amps and it works great. It impacts performance far, far less than voltage ramping and works much better, in our experience.

If you do this, then your CIM gearbox could easily have less torque than an otherwise roughly equivalent miniCIM gearbox.

And some of the issue is the heat caused by the bushing, which would be if anything more of an issue at a higher RPM?

When data shows you one thing, I think it’s more prudent to act on that than to act on “well this data that doesn’t exist might show a different thing maybe”.

The CIM “ordinary” CIM stats show a higher torque at basically every current value than the mini-CIM.

Of course, this may still change in use due to heat accumulation even with the current-limiting.

Current-limited CIMs may also not offer enough benefit over mini-CIMs to be worth the extra weight and space.

We intend to test this.

Than the mini-CIM at the same current, instantaneously. The whole point is analyzing the problem over time. I look forward to seeing more data.

The firmware updates to the VRM and PDP (either in 2016 or 2017) fixed most robot-wide brownout issues. 6-CIM drive trains no longer need current limiting in order to prevent a robot reboot. In this video, we were trying to characterize what would happen to the robot with no current limiting on a 6-CIM drive train in high gear. Hilarity ensued.

Caveat: we did have SRX voltage ramping set to 12V per ~0.1 seconds, which was approximately how long the theoretical acceleration curve kept the current draw above 240 amps. The intent was to prevent issues with swift reversals of the joystick inputs.

Anecdote

(Side note, although this was with 2016 firmware for the PDP)
PDP current reading seems to be sllllllllllllow.

All,

This whole thing started because I stated in a post last season that my teams haven’t used CIMs for a few seasons and some people asked why.

I don’t expect everyone to run out and buy miniCIMs (please don’t as we will not have enough if that happens), but I wanted to share things I noticed and then tested.

Most of the discoveries were made during the dyno up/down test. As an aside, everything on motors.vex.com has a description of how we tested it. You do not need to assume how we tested, because we tell you in the “how we test” section.

We tested about 20 of each CIM and miniCIM. We tested CIMs you get from AndyMark and from VEX (they are the same motor from the same factory). They were all within 3-4% of each other. The stated motor manufacturer variance for these is 10%, but I have yet to see that much variance.

What we found is the CIMs ALWAYS heated up more than the miniCIMS during those tests. Now, we didn’t do anything with current limiting as we were just trying to quantify the motor specs as motor manufacturers do not do the up/down tests so their published performance data is a little higher than reality.

All of you need to make your own decisions on what to use and I’m not really interested in arguing with anyone on this subject. This is a complicated issue and I have found that testing with your specific configuration is always best.

Yes, I am sure our motors would have heated up more if we had more wheel scrub. However, I found it interesting that the motors didn’t heat up and our darn large breaker got so hot it would trip after about 2 or 3 hours of continuous practice. The motors (including the 775s) were happy.

I like distributing the heat across 6 motors. I just don’t like the weight hit of doing it with CIMs. 6 minis worked great for us (so did 4WD with all omnis) and we will likely use them again this year on our drive.

We have found that if you design your robot to win a pushing match you probably won’t be a high scoring robot (there are exceptions, but not many) in the latest styles of FRC games. What we have found to be more important is to make yourself “immovable” when trying to score. We did this by going to the rear peg when under heavy defense this last year. Depending on the game, 4 omnis may not make sense no matter what motors are driving them.

Why do I like 4 omnis? Simple. The T-bone! After 2011 Einstein when 973 made 217 their plaything by t-boning us I said, “never again”. If we used 4 omnis that year, then the 973 strategy would have been useless. Since we were protected when scoring omnis would have made perfect sense.

Anyway, I digress.

Paul

Maybe you can clarify? Because if I’m understanding you correctly, your statement is the exact opposite of what is being shown in the video. Wouldn’t current limiting have helped the robot not brownout in the video?

Yeah, the point of current-limiting is to avoid brownouts, not necessarily reboots. We’ve found it absolutely crucial, especially with closed-loop control which tends to cause very large current spikes when setpoints change suddenly.

Brownouts are when one or more (but not all) subsystems do not get the amount of voltage needed to stay powered on. I’ve (inherently) extended that meaning to be ‘critical’ subsystems. In 2014, we would have seen this robot sit idle for 45 or more seconds as either the radio or RIO rebooted. In this video, we hear the results of the software cutting output to the compressor in order to prevent a RIO or radio reboot. We also see the effects of the motor outputs being temporarily cut as the software allows system voltage to recover.

I agree that current limiting would have prevented wasting energy (which wasn’t under consideration at that time). At that point we were focused on the opportunity to do a few hours of real drive practice before bag & tag.

I suppose my point of view is “if a robot browns out and no one notices, did it matter?”.

I think this is a terminology issue; the motor on-off behavior seen in that video is exactly why we implemented current-limiting, and is what we (and a few other teams) have called “brownouts.”

Seems like we need more specific terms for the specific things the RoboRIO does to deal with voltage drop.

I mean the math is pretty simple. 6 > 4.

2010 was definitely the exception. Play soccer like real soccer players in order to score goals.
And to the comment by Cory on pushing power independent of no. of motors. We quickly figured that out that same year, running just 1 CIM/1 FP-AM planetary gearbox) on each side of our drive.
Cory wont remember, but I asked him during build season about our holonomic drive motors stalling. He suggested an 8 wheel drive instead during week 4 of build season. Until this day, its our best season…ever.

If that’s your concern, limit it to 80A and you’ll get better torque than a miniCIM at 89, with less current draw. Personally, I’m thinking of a current limit in the 50-55A range, but planning to experiment. I used 42A in the case above to maximize the difference between the two motors when run with the miniCIM’s profile in the max power test.

If that bushing were such a problem, it should show up in the free run conditions. The CIM dissipates 32.4W running free at 5330 rpm, and the mini-CIM dissipates 35.8W at 5840. If we assume the friction is viscous (proportional to speed), they’re practically twins. Unless you’re putting significant lateral forces on the shaft for long periods of time, I’m dismissing this as a marketing rather than engineering consideration. So far, all but one of our CIM/mini-CIM motors have had spur pinions driving spur gears on well-constrained axles. That one exception directly drove a belt; that case would likely cause me a bit of concern today.

The peak power data shows what happens in a specific, somewhat pathological use case - you’re constantly pushing two quite different mechanisms to get everything you can get from each in the short (instantaneous) term and asking which one lasts longer. If you use a CIM and put in some controls so that the CIM does not draw greatly more current than the miniCIM, it will outperform the miniCIM up and down the power curve. Physics lets you fill the gaps in the measurements (subject to later measurements). Physics makes me pretty confident that the heat being generated is quite close to the difference between the electrical energy in and the mechanical energy out. I’m ready to make engineering decisions based on the theory of conservation of energy.

Paul,
Did you post how you did the “max power” test? I could not find a description, and can think of several ways to do the test, though it is likely that none are exactly what you did.

I didn't see a description of the "locked rotor test" either, but that one seems pretty obvious; lock the rotor to a torsiometer, regulate the voltage, measure.  The only variable is the torsion constant of the meter; I'm assuming that it was stiff enough that total rotation was less than a right angle.  If that wasn't it, please elaborate.

Gus,

Right here: http://motors.vex.com/#testing

We explain all three tests exactly how we do them.

Also, I have to make sure the viewers at home understand something about your claims about the spur gear setup you describe here:

If that bushing were such a problem, it should show up in the free run conditions. The CIM dissipates 32.4W running free at 5330 rpm, and the mini-CIM dissipates 35.8W at 5840. If we assume the friction is viscous (proportional to speed), they’re practically twins. Unless you’re putting significant lateral forces on the shaft for long periods of time, I’m dismissing this as a marketing rather than engineering consideration. So far, all but one of our CIM/mini-CIM motors have had spur pinions driving spur gears on well-constrained axles. That one exception directly drove a belt; that case would likely cause me a bit of concern today.

Spur pinions put a significant lateral load on the CIM motor shaft unless you support the other side of the shaft. Even then, it still sees load. Let’s neglect separation forces for now due to the 14.5 deg pressure angle and just deal with loading.

At max power, the CIM motor is spinning at 2670 RPM and has an output torque of 1.2Nm. If you are using the 12T pinion, then the pitch diameter is 0.6", or a radius of 0.3". So Rpinion = .00762m. At 1.2Nm, then the force at the pitch diameter is 157.48N, or 35 lbs. The pressure § on a bronze bushing is then this force divided by the projected area of the bushing, which is 8mm (.315") * .5" (length of bushing). So the pressure is 35/(.315*.5) = 55.56 psi

PV is the rating for bushings and the CIM bushing is oil impregnated bronze, which is rated at max PV = 50,000; max P at 2,000 psi; max V at 1,200 ft/min. The CIM surface speed 220.3 feet per minute at 2670 RPM so the PV is 12,240. Well within the rated range for the bushing, but definitely not 0.

More importantly, the coefficient of friction between a bronze bushing and steel is about .10 so the frictional force on the bushing is about 3.5 lbs, or 15.6N at a distance of 4mm that equates to .0623 N-m or about 5% of the actual output torque at max power.

Not extremely high, but definitely not 0. You’re dismissing the difference between a bushing and bearing at high RPM as marketing vs engineering considerations? I can tell you that is not true, because I made the decision to use the bearing as our testing showed it had much less friction than the bushing.

Maybe I misunderstood.

How much less friction is in the bushing vs. the bearing when running the CIM/miniCIM at max power on a small spur gear? The idea about running the same current on both CIMs and MiniCIMs is a valid one that I want to see the results of.

inb4 someone brings up 6-8x 775pros.

But hey, who likes saving weight anyways.

And if/when they burn out, that’s even more weight saved! Gotta consider the mass of the magic smoke :wink:

What about the mass gain from lifting a heavier robot?

Thanks!

I must have slipped a digit or something when I did the calculation a few years ago; I got something about a tenth of that. Yes, that’s significant.

It would be interesting to test the CIM with the max power test but with the target speed 4664 rpm, where the power output curve crosses 215W on the fast side.