Re: FRC Blogged: - Motor Controllers
 

This is a good example of an anecdote that proves my point -- it doesn't even provide evidence that the Jags are inferior to other speed controllers.

If your practice bot weighed as much as your competition bot, and was geared the same way as your competition bot (a la Poofs @ 17-18ft/s), then no wonder you kept tripping it. To me this is either a fundamental misunderstanding of how the Jags worked, or a misunderstanding of how much current your drivetrain actually pulled in 2011. If your requirements were to pull more than 40A for more than the trip time, you shouldn't have ordered the Jags in the first place.

While there isn't enough detail to comment on your 2011 season, it's reasonable to deduce that your drive train in 2011 was geared at least as fast as 2010 due to an open field. If that's the case, then see my comments to Adam. Your last sentence makes me :rolleyes:



I don't foresee the removal of the current protection as a good thing. It, by itself, at least provided a way for my team to realize the 'oops, I forgot to account for that' moments without burning out a motor or a $100 speed controller.

Re: FRC Blogged: - Motor Controllers
 

We fully understand how much current we were drawing; our drivetrains are completely analyzed before we run them, and their top speed, and time to distance match our analysis pretty closely.

For reference, 254 was geared 1-2 fps faster iirc.

This current was not enough to ever even trip the 40 amp breakers; not once on the comp bot or practice bot during season (when it had victors). We've never blown a Victor from overcurrent, we've also never blown a CIM.

We still tripped the Jags routinely when we switched in slower geared gearboxes at a top speed of about 13 fps.

Overcurrent protection would be an awesome feature, if it was tuned to trip at the proper point; it's simply too conservative now.

Re: FRC Blogged: - Motor Controllers
 

I guess we just have different experiences then.

After some tweaking to our code to prevent immediate full-forward to full-reverse changes, we've never tripped our Jags on the field with 145lbs @ < 12ft/s (and the code complexity is in-line with what the Tom posted here, so I bet the driver->bot response time is on-par). From 2006-2009 we consistently had 2-3 Victors fail over the course of 3 competitions each year whereas we only had 1 Jag in 2011 fail (0 in '10/'12).

Granted, we never put those code changes on the Victor-based robots, so maybe that's a differentiator.

Re: FRC Blogged: - Motor Controllers
 

The 40A breakers will let 60A through for up to 47 seconds (or as little as 3.9) [1]. You always design the fuse or breaker to blow before anything else so this should be the limiting factor of the circuit, not the speed controller

We had the same problem and used the same solution, although we implemented it using the limited-acceleration feature built into the Jag (I don't remember it's proper name). We have never blown a victor due to load.

[1] http://www.snapaction.net/pdf/MX5%20Spec%20Sheet.pdf

Re: FRC Blogged: - Motor Controllers
 

I believe Al Skierkiewicz has stated on multiple occasions that the 40a breaker is there to protect the wiring, not the device.

Re: FRC Blogged: - Motor Controllers
 

There are many ways to kill a device other than high load, but the breakers should protect from high load. Not just because it's proper design, but because the FETs can handle it (at least the ones in the Victor, I haven't checked on the FETs in the Jag).

Re: FRC Blogged: - Motor Controllers
 

Man, 47 is everywhere on this website!

Re: FRC Blogged: - Motor Controllers
 

230 is testing the Talon as well. Unofficially, of course.

Re: FRC Blogged: - Motor Controllers
 

Jason,

Good design works from the end device back towards the supply.

1> You start with the expected average current draw od the device, in this case a Jaguar -- 40Amps (note that the Jag can pull much higher loads but that the 'expected load is 40 Amps).

2> You then set the gage of the wire to handle the expected load (10 AWG can handle a ~55Amp average)

3> you then set the breaker to protect that wiring (40 Amp breaker)

I believe the FETs on the Jag are designed for 60 Amps. They can handle a lot more, but that will cause them to heat up. Heat is the primary failure mode with power FETs.

The overcurrent on the Jags was too conservative (~40Amps) since the expected load could jump to ~132 Amps with a stalled CIM. For short periods of time that high current is fine since the Jag (FETs), wire, and breaker can also handle that for the time it takes for the breaker to trip. The real issue comes whenb your pulling a constant 60-80 Amps and it becomes which tripps first, the FETs (dying) or the breaker.

Re: FRC Blogged: - Motor Controllers
 

Correct... Did I say anything to contradict that?

If you're not sure which is going to blow first, the FETs or the breaker, that is an indication of a poorly designed (not necessarily improperly designed) system. There are three solution: decrease the load to what the system can handle (but who's going to give up CIMs), increase the capacity of the FETs (so that the breaker blows before any expensive equipment blows), or decrease the capacity of the breaker (for the same reason).

Re: FRC Blogged: - Motor Controllers
 

This blog really made my day.

I've believed in the Jaguar for the past two years now. It still has its flaws, but like any product, the improvements are an iterative process.

The one thing that was not touched upon in the blog was price. I've got my fingers crossed that the Jaguar will be offered at a similar cost to what FRC teams had in the past.

I'm really excited to hear about both the conformal coating, and the firmware tweaks. Both are weaknesses in the product that we've fought with.

My dream is that one day we see a speed controller that gives us the best of all our current options.

-Conformal coat
-Sealed unit with heatsink
-Linear output
-15kHz response
-Integrated velocity and position PID (with a derivative term that actually works...)
-Velocity PID with feed forward support
-Direct limit switch, potentiometer and encoder inputs
-CAN communications
-status LED that blinks progressively faster as speed increases
-small footprint

The best part is, if this happens, and it's in a package reliable enough to put into an industrial machine, you have a product viable enough to displace industrial motor controllers that cost 4+ times as much.

Re: FRC Blogged: - Motor Controllers
 

The Jaguar is listed at $89 on vexrobotics.com

Re: FRC Blogged: - Motor Controllers
 

As promised, I am rejoicing.

FIRST and IFI both deserve great big shiny gold stars for playing well with others. :D

Re: FRC Blogged: - Motor Controllers
 

Well said, Richard. I totally agree.

Andy B.

Re: FRC Blogged: - Motor Controllers
 

Ether is correct, the breakers have always been sized to protect the wiring. This is an attempt to keep fires to a minimum in the event of catastrophic failure of wiring or devices. When examining the wire tables, the max current specifications for chassis wiring are used. As stated above, the max continuous current for #10 wire 55 amps. So if a chassis part were to fall across the input to a speed controller, the breaker would trip.
Since all of our breakers are capable of passing 600% over current for a few seconds, they are not appropriate devices to use for protecting electronics. In the case of the 40 amp breaker, it will pass sufficient current to damage FETs in any controller under the right conditions. All competitors should remember that all motors draw stall current when current is first applied. A quick check of legal 2012 motors will show that many are rated for more than 40 amps stall.
Jason, there is a fourth way that most teams use without realizing it. That is to add resistance to limit current. Often this comes in the form of extraneous wire length. I have used a term for many years, "wire foot" to demonstrate this concept. At 100 amps, one foot of #10 wire will introduce .001 ohms of resistance and drop 0.1 volts. Other stranded wire sizes per foot are roughly:
#6=0.5 WF
#12=2 WF
#14=4 WF