Can Victor and Jaguar motors work together on a single gearbox?

[Ether]

Originally Posted by Mr. Lim

I'd like to hear an explanation for this one.

Do you still have last year's bot?

Could you bring it back to life and run a few quick tests?

1) wheels up (bot on blocks): test voltage* at Jag's and 888's motor outputs at several throttle settings

2) bot pushing against a wall: carefully repeat above test with low throttle settings (to avoid overheating)


* measure both true RMS and arithmetic average. true RMS requires a voltmeter that specifically has that capability.

[Ether]

Originally Posted by Ether

Are you also certain that:

all the controllers were properly calibrated

and

your code was using the appropriate input pulse width range for the Jags and the 888s?

Originally Posted by Ether

Do you still have last year's bot?

Could you bring it back to life and run a few quick tests?

Pending answers to the open questions and results from any testing Mr Lim (or anyone else) is able/willing to do, my working hypothesis is that the answer lies somewhere in this code:

/**
 * Common initialization code called by all constructors.
 */
void Jaguar::InitJaguar()
{
	/*
	 * Input profile defined by Luminary Micro.
	 * 
	 * Full reverse ranges from 0.671325ms to 0.6972211ms
	 * Proportional reverse ranges from 0.6972211ms to 1.4482078ms
	 * Neutral ranges from 1.4482078ms to 1.5517922ms
	 * Proportional forward ranges from 1.5517922ms to 2.3027789ms
	 * Full forward ranges from 2.3027789ms to 2.328675ms
	 * TODO: compute the appropriate values based on digital loop timing
	 */
	SetBounds(251, 135, 128, 120, 4);
	SetPeriodMultiplier(kPeriodMultiplier_1X);
	SetRaw(m_centerPwm);

	nUsageReporting::report(nUsageReporting::kResourceType_Jaguar, GetChannel(), GetModuleNumber() - 1);
}





/**
 * Common initialization code called by all constructors.
 * 
 * Note that the Victor uses the following bounds for PWM values.  These values were determined
 * empirically through experimentation during the 2008 beta testing of the new control system.
 * Testing during the beta period revealed a significant amount of variation between Victors.
 * The values below are chosen to ensure that teams using the default values should be able to
 * get "full power" with the maximum and minimum values.  For better performance, teams may wish
 * to measure these values on their own Victors and set the bounds to the particular values
 * measured for the actual Victors they were be using. 
 *   - 210 = full "forward"
 *   - 138 = the "high end" of the deadband range
 *   - 132 = center of the deadband range (off)
 *   - 126 = the "low end" of the deadband range
 *   - 56 = full "reverse"
 */
void Victor::InitVictor()
{
	// TODO: compute the appropriate values based on digital loop timing
	SetBounds(210, 138, 132, 126, 56);
	SetPeriodMultiplier(kPeriodMultiplier_2X);
	SetRaw(m_centerPwm);

	nUsageReporting::report(nUsageReporting::kResourceType_Victor, GetChannel(), GetModuleNumber() - 1);
}


/**
 * Set the bounds on the PWM values.
 * This sets the bounds on the PWM values for a particular each type of controller. The values
 * determine the upper and lower speeds as well as the deadband bracket.
 * @param max The Minimum pwm value
 * @param deadbandMax The high end of the deadband range
 * @param center The center speed (off)
 * @param deadbandMin The low end of the deadband range
 * @param min The minimum pwm value
 */
void PWM::SetBounds(INT32 max, INT32 deadbandMax, INT32 center, INT32 deadbandMin, INT32 min)
{
	if (StatusIsFatal()) return;
	m_maxPwm = max;
	m_deadbandMaxPwm = deadbandMax;
	m_centerPwm = center;
	m_deadbandMinPwm = deadbandMin;
	m_minPwm = min;
}

[Mr. Lim]

They were Jags and Victor 888s.

The Jags were not calibrated, as we found they came out of the box with acceptable centring, and consistent extremes.

The Victors we did calibrate. We noticed that uncalibrated, the problem was even worse. Calibrating them helped slightly, but did not solve the issue entirely.

We program in Java, and instantiated the proper object classes for everything, so the pulse widths should have been proper for each.

Unfortunately, we don't have the time to take out last year's robot right now. It was involved in the 2015 Alpha Testing, and has been significantly re-wired and re-programmed as a result.

[Tem1514 Mentor]

The quick answer is one motor is being run as a generator due to the difference is applied voltage/current/duration (it is a chopper drive circuit).

An application of this is the tread mill drive clean test for cars. aka dyno

[Ether]

Originally Posted by Tem1514 Mentor

The quick answer is one motor is being run as a generator

That quick answer is not true.

The difference in voltage would have to be so large that the higher-voltage motor was driving the system at a speed exceeding the lower-voltage motor's free speed*. Not likely.

What actually happens is that the higher-voltage motor carries more of the load. It supplies more torque and power. So it tends to get hotter.

The lower-voltage motor is still generating torque and supplying power to the load as long as the speed is lower that that motor's free speed at that voltage.

* at its lower voltage

[Richard Wallace]

Originally Posted by Ether

... What actually happens is that the higher-voltage motor carries more of the load. It supplies more torque and power. So it tends to get hotter.

Ether's right, as usual.

The 2015 PDB with its datalogging feature will make it easier for more teams to see this effect directly.

[dsirovica]

To conclude:
Different controllers is not recommended, but if needed can be used. The only negative result may be that one motor works harder than the other and will get hotter.

On a related topic: what do people think are the issues with using different motors on the same gearbox (with same controllers). Eg. A CIM and a mini-CIM? My gut level tells me that is perfectly fine, but a soon-to-be Mech Engineer told me that there will be a "torque fight" and that we should adjust each motors' power via a lookup table that we should develop by actual measurements of each of the motors.

[Ether]

Originally Posted by dsirovica

On a related topic: what do people think are the issues with using different motors on the same gearbox (with same controllers). Eg. A CIM and a mini-CIM? My gut level tells me that is perfectly fine, but a soon-to-be Mech Engineer told me that there will be a "torque fight" and that we should adjust each motors' power via a lookup table that we should develop by actual measurements of each of the motors.

There's no "torque fight" until you reach the free speed of the motor with the lower free speed (the CIM in this case). Above that speed, the CIM acts as a generator (causing negative torque). However, you'll likely not reach that speed in a typical FRC application (e.g. drivetrain).

You can't adjust the motor's power without knowing the motor's speed. But if you want to assume a given operating speed, you can adjust the motor commands to match power or current.

[Ether]

Originally Posted by dsirovica

To conclude:
Different controllers is not recommended

I would make a distinction between

a) I recommend that you not do that

and

b) I don't recommend that you do that (but I don't say not to either). In other words, the jury is still out.

I'm in the (b) camp.

The only negative result may be that one motor works harder than the other and will get hotter.

We don't know for sure yet that's the only negative result, but it seems to be the case. In any event, two things appear to be true:

1) one motor working harder than the other happens all the time even with the same model motors and motor controllers, albeit on a smaller scale, due to manufacturing tolerances and wear.

2) it can be mitigated in code by running the joystick outputs through a simple LUT or polynomial.

[Mr. Lim]

Okay, I'll chime in with my conclusion:

Yes, a Victor and a Jaguar can work together on a single gearbox.

BUT,

It won't work well unless you code something to compensate for the unbalanced load sharing between the two different types.

[dsirovica]

Thanks Ether for your thorough post (as always!)

I realize this is not quite the correct thread for different motors with same controllers on one gearbox. But to close on that one. Based on your posted Torque/RPM charts, in most properly sized applications one is not likely to run anywhere near the free running speed of the slower motor.

On the correct thread, I do think one needs to be careful when mixing motor controllers. The easiest way would be to check the current into each motor with a DC Amp Clampmeter and make sure they are reasonably close. The power dissipation in the motor is roughly I^2R*, so any difference in I(A) will cause an exponential difference in Power per motor. To fix it you can use a LUP (as mentioned before) but given all the hassle - I would just avoid it - its bad practice anyway.

*R is a fusion of restive R and back-emf, but since they should be nearly the same for same motors its a fair assumption (I see a sledge hammer coming down on me from Ether on this one? )

[Ether]

Originally Posted by dsirovica

...The power dissipation in the motor is roughly I^2R*...

*R is a fusion of restive R and back-emf, but since they should be nearly the same for same motors its a fair assumption (I see a sledge hammer coming down on me from Ether on this one? )

OK, here comes the hammer:

The "R" in I²R is resistance, and has nothing to do with back-emf.

(The "I" is RMS current)

[dsirovica]

True (hm - that is a redundant statement for characterizing Ether's posts),

what I meant to say was this:

P(motor heating)=I^2R
But I is not purely Vbatt/Rmotor, it is:
I= (Vbatt-Vback-emf)/Rmotor

Therefore each motor will have its own R and its own Back-emf characteristics. If they are the same model and same vintage these characteristics are likely very close to each other.

[Ether]

Originally Posted by dsirovica

I meant to say was this:

P(motor heating)=I^2R
But I is not purely Vbatt/Rmotor, it is:
I= (Vbatt-Vback-emf)/Rmotor
.

That's closer. "Vbatt" should be Vmco (motor controller output). But depending on the motor's inductance and the motor controller's switching frequency, there may be ripple current. That causes extra motor heating and power loss. (That's why I said "I is RMS current").

[AdamWiwc]

Thank you everyone for your help and support. We have the situation all sorted out now and all of us at Team 5185 appreciate it immensely.

Team 5185
Dial Up Grizzlies