Re: Jaguar VS Victor
 

From the discussion in this and other threads, it seems the consenses is that the Jags are beneficial only because of the CAN bus functionality and the more linear power curve. What I am interested in knowing is how many teams actually found these benefits indispensable and under what circumstances? We have been using Victors all these years and don't see the need to go with the Jags. I know CAN bus offers a lot of new nice functionalities but how many teams are actually using them in software and if they absolutely cannot do without these functionalities? Every year we had a discussion on switching to Jags but when doing cost and benefit analysis, Jags always lose out for us.

Re: Jaguar VS Victor
 

We used CAN for Breakaway to control a kicker. To get it in position, we used current mode control to move backwards until it felt a force from a hardstop. To fire, we used an intentionally under-damped PID looped around position on an encoder. This made it easy to reliably control the rest positions and contact speed.

I do have to admit that we had a slight advantage - 3 of the mentors were involved in the design of some aspect of the CAN system.

Re: Jaguar VS Victor
 

So if I understand you correctly, using CAN to control the kicker basically saved you a limit switch. And what does an under-damped PID loop has to do with CAN? I don't quite see it probably because I don't know your kicker's design.

Re: Jaguar VS Victor
 

It was a rotary kicker direct driven by a motor: the boot would rush 120ish degrees to kick the ball, return to 90ish degrees to get back within the envelope, and then slowly return to 0ish degrees to reset.

The kick (0 -> 120 ->90) was handled in a single CAN command of "Travel to 90 with constants P,I,D". Since those constants were under-damped, it would reliably "over-travel" to 120. It was surprisingly easy to tune.

The return was just a "apply amperage -A". Yes, we could have done this with a limit-switch, but it was just as easy to do it this way. One benefit was that it made it really easy to hold it in the ready position against gravity. Not enough reason to move to CAN, but since it was already there...

I've used this as an example with a few students, and I think it is a very interesting teachable moment. Plot the desired response on a torque/speed chart, and then compare it to voltage, current and speed modes. For kicker return, current mode is pretty darn close to optimal.

I really don't think that CAN should be used for every motor on every robot. It isn't a blanket replacement for PWM (though it can be used as one), and it isn't a silver bullet - it enables access to new control modes that can provide significant benefit when applied correctly in the appropriate situations.

I'm glad your team has the discussion, and I'm glad they are coming to a conclusion that works for them.

Re: Jaguar VS Victor
 

If jags weren't that fragile (i.e. get fried easily compared to victors and get cut-outs during competition), it could have easily won the cost benefit analysis. So our conclusion is that unless our robot design calls for a certain feature that jags can easily provide through CAN and cost too much to implement using Victors, we would avoid them. Hopefully, the next generation of jags will improve the reliability enough that will swing our opinion the other way.

Re: Jaguar VS Victor
 

The one problem we have had with both would be pwm cables falling out. We plan on trying the CAN system to clean up the wiring and add functionality. I didn't know that jaguars burned out so easily though. We have had one very old one go out, but that is all.

Re: Jaguar VS Victor
 

Personally, I find it really surprising to see everyone slamming Jags on reliability. Least year, we used 7 of them on the robot. The year before, we used 4. Year before that, we used 2. We've only had 1 issue with them between 7 different official competitions (and who knows how many off season events) competition, and that was due to the limit switch input on one getting screwed up (it worked if we jiggled our inputs for the limit switch, but replacing the Jag fixed it entirely). That Jag still works with jumpers. We've never burned out a Jag or a Victor in 6 years of competing.

The problem is, you seldom hear about teams using Jags successfully. It's the teams that have problems with them that come and post, so it looks like there's a much larger problem than there actually is.

Re: Jaguar VS Victor
 

One special case I can think of: When operating at stall, a Jaguar tends to heat up the motor a bit less than a Vic, for the same torque output.

Re: Jaguar VS Victor
 

This year, we are trying Jaguars and CAN bus because we are quickly running out of digital I/O channels with all the encoders (encoders can directly connect to the Jags). It sure does clean up our wiring. See my pictures on this thread (http://chiefdelphi.com/forums/showth...t=99554&page=2).
However, while making the Jags to work with the code, one of the brand new Jag is already malfunctioning. I don't know what happened. Here is the sequence of events:
- We have 4 Black Jaguars driving a mecanum wheel set.
- Flashed the latest firmware to all the Jags and the 2CAN.
- Successfully assigning IDs and running each individual motors by using the bdc-comm tool.
- We have four encoders (one for each motor), but for some reasons all the encoders are not working (could be because of the new way we connect them). We are still investigating it.
- While testing different modes of the Jags, one of them died. The status LED goes dark even though I have power to it. After replacing the dead jag and some tweaking around with the code, the robot now runs teleop fine.

Since this is a brand new Jag, is it under warranty? How would I send it back for repair?

Re: Jaguar VS Victor
 

There is a link to RMA information at http://www.ti.com/jaguar

Re: Jaguar VS Victor
 

I just want to throw this into the mix:

Pro Jag (black)

The 2Can controller is awesome. Nothing beats remotely a remote user panel to debug/modify. And it saves wiring space by having the the 2can-crio connection over Ethernet.

Re: Jaguar VS Victor
 

If your using PWM, Victors are your best bet. I've used both for years, and based on what you've said, Victors seem to be right for you. Jaguars only advantage is that they can run Can. Can will give you better feedback (Voltage/current/other), but are bigger, and less durable. Based on you saying "School TOY", I'm assuming this robot is in for a long life full of those autonomous mishaps that send your robot full speed into a wall. Victors will be able to take that abuse, while jags will not (Trust me, after testing, we used victors in our 120lbs battlebot). Hope this helps

Re: Jaguar VS Victor
 

I'd be all over the 2CAN interface, but I just can't justify the $200 item (PER ROBOT if you keep them around) when there is a $2 option that just happens to run 4 times slower. I haven't run into a situation where we've been limited by the bandwidth (yet).

While CAN is a nice interface that simplifies wiring, that's only one small benefit for the Jags. The BIGGEST improvement is minor loop control capabilities. Imagine driving to speed during the competition and then switching over to position mode to give you the ability to accurately traverse a specific distance.

While the Jags are fairly rugged, we had some issues last year with a couple of them. I'm of the opinion however, that the mechanical guys killed them with aluminum filings... If you're careful though, they work just fine.

- Bryce

Re: Jaguar VS Victor
 

My opinions on the topic:

-The serial bridge is ~1/8th the speed of the 2can, yes (115,200baud serial vs 1M?) - That's fairly significant if you update the Jaguars fast enough

-We have yet to use a CAN jaguar, or have any features we cannot do with purely a Victor and some software on our end. We like it this way for some things, as we can modify the control loop to fit our robot (a PID loop isn't great for everything), or add multiple control loops and add the output (e.g. control to distance + control to heading over the two sides).

-We will never use a Jaguar in drivetrains because of their high-current shutdown. We had a lot of issues on our Lunacy bot post-season, because after 5 minutes of driving (with super low cof wheels, remember) the shooter could randomly (predictably after a few minutes) shut down due to the extended current draw of forward-reverse slams.

-I've heard the black jags are far more reliable than the grey ones. We are willing to use a black jag if it gives us an advantage, but we haven't found an advantage yet. They are functionally the same over PWM, but if you only use PWM control, the Jaguars are physically bigger and can't handle the extended high-current draw that the Victors can.

-In my time on 33, I've only seen 1 Victor die, ever. We have some of the really old Victor 883's which we still use for practice and mock up robots, and they still work after many years of use. The Jaguars don't yet have that kind of track record, especially after the really bad grey jags.

Re: Jaguar VS Victor
 

Yep, you'd be right. However, we've run at least 6 Jaguars in CAN mode on a robot for the last couple of years now, and the serial to CAN interface has not been considered slow. How often do you plan on updating a Jaguar anyway? To give you an idea, the Jaguar messages that are sent are only 32-bits long. The therefore theoretical maximum bandwidth of messages is 3600 commands PER SECOND. In PWM mode, the Jags can be updated every 5ms or so and the Victors every 10ms. So that's only 200 and 100 commands per second, respectively. Assuming 10 Jaguars and a 60% efficiency, you still get about 200 updates per second per Jaguar. And with fewer, the update rate is higher. My point was that I can't justify the $200 cost for update speeds that I already get.

Sometimes it's just nice to tell a Jaguar to go to a position using a Set() command rather than having to write or use a PID in your code for it. :)

This year, rate limiting (built-in) will solve most of those issues. However, we have run Jaguars on the drive train for the last couple years and we could not make them reset when using a joystick to thrash the motors back and forth. We only run into those issues when the command for full forward is immediately followed by full reverse. And even THEN we only had that issue because we were doing that with five motors simultaneously. (It was a swerve drive train without completely rotating modules). Even THEN the issue was a voltage dip because of the HUGE current spike of 5 CIM motors reversing direction immediately. Rate limiting solved that issue nicely. It's typically considered unnecessary to slam them back and forth since even a quick rate limit will be imperceptible to humans.

I think that probably has to do with people removing the screws on the Grey Jags even though they aren't meant to be removed - and thus leaving metal shards in the case.

Now that, I'd agree with. Jags just need some babying and there are a few fundamental rules that have to be followed to have success. They have their problems, but they're pretty great at the same time.

- Bryce