Victors verses Jaguars?

[Al Skierkiewicz]

Originally Posted by Ether

I'm not sure anyone knows for certain why this happens, but one prevailing hypothesis is that the PWM frequency of the Vic is low enough to create torque ripple near enough to the resonant frequency of the pins to knock them loose and keep them from jamming.

I believe that the low mass of the locking pins cause them to oscillate into a lock mode when the motor is powered through the 15kHz switching of the Jag. Removing the locks fixes this problem in many applications. While locking is still an issue with Vic, the problem seems to occur far less frequently.
While the newer Jags have gone to lower "ON" resistance FETs, the manufacturer has lowered the number to two in each leg while the Vic remains at three. I think this gives a slight edge to the Vic in terms of dissipation at sustained high currents. No one has mentioned that the Vics are conformal coated. This coating prevents a lot of the contamination that is prevalent with the Jags.
Heating of the CIMs over a two hour practice/demo is not unusual. The CIM motors are sealed and intended for intermittent duty. Internal heat has very small paths to dissipate to the outside air.

[Hawiian Cadder]

personally, i would say avoid jags for any of the high current motors. the cims and rs-775's generate a lot of electrical noise, and this is what we believe to be the culprit behind our CAN problems. the Current control is very nice for some things. i think that a mix of jags and victors would probably be ideal. jags on the smaller motors where more finesse is necessary (550's, 395's) and victors or spikes everywhere else.


another reason to use victors on the high current motors is that they are typicaly in the drive-train, which means that they should have encoders on them already, and if that is the case then the sensing capability of the jag is unneeded. with voltage and speed, current and anything else you want can be calculated.

[Ether]

Originally Posted by Hawiian Cadder

the cims and rs-775's generate a lot of electrical noise, and this is what we believe to be the culprit behind our CAN problems.

Is this just speculation, or has your team run designed tests to verify this?

[Hawiian Cadder]

at CMP another team appeared to be having the same problem ( motor pulsing ) we had already solved the problem with the timeout, and that had helped tremendously, however we were still getting intermittent dropouts in the Can buss. when we talked to the other team (i dont remember name or number, maybe 399?) they said that they had been having the same problem and that attaching a filter capacitor on the leads of their motors had helped tremendously. another thing that lead me to this conclusion, is that we never had any problems with the can while we were testing our lift or manipulator, which were all rs550s or fisher prices. when we tested the drive however, the can network continued to drop out intermittently. the FP on the lift was drawing at least as much current (approx 30 amps continuous) so i was left to think that electrical noise was the culprit. we have not done any tests, but before next season we might try the filter capacitor trick and see if it solves the problem.

[AustinSchuh]

Originally Posted by Ether

Is this just speculation, or has your team run designed tests to verify this?

While it may be speculation, I've noticed what I believe to be a similar thing. I'd love to get a scope in the same spot as the robot though and see what's really happening though...

971 was testing our elevator, and when we would tell the control loop to move the elevator down a foot, (with a FP, and also with a FP and 775 both running off Black Jaguars), we could reliably get the robot to reset the Jaguars and loose communications. This happened more often when we had lots of D in the control loop. I haven't seen anything like that in the drivetrain with CIMS and Jaguars when doing control loops.

[Jon Stratis]

Originally Posted by Hawiian Cadder

personally, i would say avoid jags for any of the high current motors. the cims and rs-775's generate a lot of electrical noise, and this is what we believe to be the culprit behind our CAN problems. the Current control is very nice for some things. i think that a mix of jags and victors would probably be ideal. jags on the smaller motors where more finesse is necessary (550's, 395's) and victors or spikes everywhere else.


another reason to use victors on the high current motors is that they are typicaly in the drive-train, which means that they should have encoders on them already, and if that is the case then the sensing capability of the jag is unneeded. with voltage and speed, current and anything else you want can be calculated.

FWIW, we've used CAN now 2 years running (for drive train and some manipulators), and haven't had any real issues with it at competition. Sure, it can be a little finicky when getting it all set up during the build season, but once it's working, it just keeps on working. further, we've only lost 2 Jags so far... one was bricked by a bad firmware update, and the other due (we think) to metal shavings. At least, its replacement worked just fine over 2 competitions afterwards! We've seen 0 problems using jags with CIM's.

[Kevin Sevcik]

57 has used CAN for 2 years and we're heading into our 3rd this year. We've had occasional problems with the CAN network dropping out, but I'm pretty certain they were all self-inflicted. Things like the CAN-Serial adapter coming loose from the cRIO, or a finicky terminator that would occasionally short if you looked at it funny. I think CAN is plenty reliable if you're careful about your wiring. The major drawback is that a bad connection will take your whole network down, instead of a single motor.

As for reliability, we've had a handful of the Gray Jags go bad on us, but we haven't lost a black one yet. Again, I think being careful around your electronics is the main thing. The Jags are more susceptible to metal shavings than the Vics, so you need to cover up your electronics if you're producing shavings. Which you should be doing in any case, even if your Vics are less likely to fail because of it. There's plenty of other stuff in the control system that's vulnerable to metal shavings nowadays.

[RyanN]

Originally Posted by pfreivald

In my very limited industry experience, I've seen too much vapor-ware to trust 'functionality' that hasn't been proven...

You can substitute "very limited industry" with "decade-long FIRST" and make the same statement. Unfortunately, FIRST tries to vet too much unproven product. For a $6K price tag (at a very bare minimum), I expect products to work out of the box.

While this is true, you have to start somewhere. That somewhere, or sometime in this case, is now.

I'm not trying to start a fight or anything, but your statement represents something that a lot of engineers I work with do, and personally, I hate it.

We, as an engineering society, won't get anywhere using the same stuff as we always have. I'm an intern at NASA, and I hate to say I see this all too often. Some things are done the same way as they were done back in the 1960s, and there are much better, easier, and cheaper ways of doing the same thing.

It might work for a while, but I'd say replace it once support for that product is up. And sadly, it rarely happens. When that product fails, and there are no replacements, then quick engineering fixes must go into place to mitigate the problem.

In this rant, and in all my posts here, I never did say to use CAN. I said it gives a lot more functionality than PWM does. I recommended that teams start to look into it.

We tried to use CAN for a few days, and I can personally say that it sucked. It didn't work as intended. It wouldn't work for competition. It had random drop outs. One motor wouldn't work for some reason. A lot of issues. But hey... guess what, it was working. We didn't use it for competition. We used PWM for competition because we couldn't afford to lose a match because the CAN network went down. Now though, we're rebuilding the control system for our 2010 swerve drive robot, and we're going to implement CAN again. This will be a demo robot, and a loss of functionality won't cost us a match here.

CAN, or Controller Area Network, is not something new. It has been used in vehicles for over a decade. Every vehicle produced since 1996 (in the US anyway) has been required to include an OBD-II port. Part of the OBD-II's specification is CAN, along with a few other protocols, but as of 2008, the CAN protocol is required. Also, much of the vehicles equipment talks with CAN. With my car (2006 Mazda3), the stereo talks with the ECU to get the trip computer information over CAN. CAN is tried and trusted in the industry. So don't put the blame on CAN. (Source: http://en.wikipedia.org/wiki/On-board_diagnostics)

The issues that arrise are with the Jaguars and the cRIO images and our own software. There has been a steady release of updates, resolving most of the issues with CAN. Our team is taking the initiative to try to iron out some of the remaining bugs.

Out of the 40 or so Victor 884s we've had, we've probably blown 10 of them.

Now out of the 10 or so Jaguars we have, we've burnt out just 2 of them.

That's nearly the same failure rate. The difference though is that we know that most the Victors failed due to user error, whether it be metal shavings or pushing the robot too fast while off. The Jaguars failed because of a manufacturing defect (Gray).

Also to note about the Window motors & Jaguars...
Removing the locking pins seem to help most people.
http://wiki.team1640.com/index.php?t...r_Locking_Pins
http://www.chiefdelphi.com/forums/sh...ad.php?t=83973

As part of our 2010 robot, we have the Window motors for steering, and we're planning on trying to use the CAN network for better control using the voltage mode. I'll report if I have any problems later this summer if we have any. I will be removing the locking pins. The only difference I can see between the Jaguars and Victors is the frequency at which power is sent to the motors. The Victors 'refresh' the power slower than the Jaguars. This can change the way the thermal cutout works. Remember that window motors are designed to run at either -100%, 0%, or 100%, not anything in between. We're using these motors outside of their intended purpose.

[pfreivald]

Originally Posted by RyanN

While this is true, you have to start somewhere. That somewhere, or sometime in this case, is now.

I'm not trying to start a fight or anything, but your statement represents something that a lot of engineers I work with do, and personally, I hate it.

I hear what you're saying and largely agree -- I just don't think that the Kit of Parts is the place for vetting new technologies. It should be proven to work through independent testing on FRC bots (through beta test teams, for example) before it's rolled out as an option, IMO.

[RyanN]

Originally Posted by pfreivald

I hear what you're saying and largely agree -- I just don't think that the Kit of Parts is the place for vetting new technologies. It should be proven to work through independent testing on FRC bots (through beta test teams, for example) before it's rolled out as an option, IMO.

From what I understand, the use of C++ and Java (and now Python) are all beta products that NI is currently testing. I don't believe any of these languages are officially supported, or were officially supported when the Compact RIO was released.

I might be wrong, but one of the LabVIEW developers at work was talking about how FIRST teams are spoiled and get to try new products before they're even released.

But I also totally agree. CAN was hardly even half baked at the beginning of the 2011 FRC Season.

[Greg Needel]

Originally Posted by Webberbots

We ended up burning out one of our CIM motors.

Really? After 2 hours of driving your CIM motors will be hot, but I have NEVER seen a CIM "burn up" My guess is that your speed controller is shot and your motor is fine.

A simple test is take one of your other speed controllers on your robot and switch the leads to that drive motor and see what it does.


As for the Victor vs Jaguar debate, In my opinion if you are not using CAN there is no reason to use a Jaguar*.




*This opinion has nothing to do with my company affiliation. I never have and never will use a Jaguar when a Victor is an option due to proven reliability.

[qnetjoe]

Please let me chime on a few things here:

• Jag Over Current Protection
  • Gearing - A good number of these issues really center around the lack of proper gearing. There is no subsuite for good mechanical design. For example, one of the teams at a regional I supported was trying to direct drive a 6" wheel with the CIMple gear box. This is about half of the proper gear ratio.
  • Stupidity - I don't know for sure, but I suspect that the reason that it was implemented had more to do with the rash of team that did not fuse their victors during the post 2008 off season. I personally saw at least 4 or 5 teams that wired victors directly to the rockwell PD block without any fusing.
• Reliabilty - I have personally fried at least 10 or more victors and jags. In almost every case the problem was environmental and not a design flaw of either. Metal shaving and debris are not friends with any motor controller. One of the big problems with the Gray Jags is that the screws were not met to be taken out thus creating metal shavings. It would not surprise me if a third version of the Jag is released that it will be conformally coated.
• CAN - CAN is still a work in progress and will be for the near future. CAN provides a gateway into the power of the Jag. The Jag can do 40+ MIPS (million instructions per second). Moving your control out to the Jags can have a major advantage if you are will to put in the time to get it to peform consistently and reliaibly.
• Firmware Update? - Have you updated the firmware on your Jags? I know they changed some of these settings in one of the firmware updates. I updated the firmware of my gray jags to V92 vs the original firmware and the behavior is night and day.

The biggest piece of feedback that I can provide is that if something fails find out why. This is what engineering is about. Too many people blame the jags since they are just a symptom and don't find the root cause. The whole window motor is a great example. The Jag provides a more linear output (most of us could argue that this is better) and this creates an issue with the locking pin. The problem is not with the JAG but with the locking pin on the window motor (root cause); Yet people blame the Jag (symptom). You could replace the Jag with an industrial drive and have the same problem.

One of challenges of designing things for FIRST is that it can be used in so many ways; commonly in ways you never expect.

[Ether]

Originally Posted by Webberbots

Recently we had a 2 hour driving session with our robot--it had been a while since our last competition and our students wanted to use the robot. It turns out that 2 hours of continuos use puts alot of stress on CIM motors (lesson learned). We ended up burning out one of our CIM motors.

May I ask:

What kind of drivetrain, and how many CIMs? What was the total gear ratio from the CIM to the wheel, and what was the wheel diameter?

Did the practice entail a lot of starting, stopping, turning, or pushing?

[Alan Anderson]

I just remembered one small benefit of using Jaguars instead of Victors. It doesn't take a lot of practice and experience before you can reliably plug a PWM cable into a Jaguar.

[jamie_1930]

In my opinion the victors win hands down I've never seen one break and I've seen at least 4 jags break.