pic: cRIO CAN Jag

[Daniel_LaFleur]

Originally Posted by squirrel

Unfortunately the RJ series connectors are inherently flaky....

Please cite your source ...


... and compare the 'flaky'ness to the 3 pin non locking PWM cables we've used in the past.

[MrForbes]

My source is 35 years of playing with electronic stuff.

The PWM cables are not wired in series.....the CAN bus is a daisy chain. Any one poor connection can kill the rest of the chain.

[Jon Jack]

The saying 'You Get What You Pay For' comes into play with RJ connectors. If you want reliability you're going to pay for it.

[Gdeaver]

I hope some time and money is being invested in to the short commings of the jag. Next year we may be back on carpet and It's my opinion that the are several points of failure that could cause some serious problems when first teams start to abuse them. I would rather have a solid Hobby PWM jag that is robust than a weak can jag.

[Jon236]

From my rudimentary knowledge of CAN bus topology, I believe the Jags can be arranged in a ring, thus providing some connection redundancy. The Luminary folks at Championship thought this could be done as well. Does this make anyone sleep better?

[MrForbes]

Yes, that would be good....

The thing about connectors is: Good connectors have both a positive mechanical connection for the housing (the housing halves cannot be pulled apart without first releasing something), and a positive pressure contact connection (the pin is encased in the socket). The PWM cables lack the housing lock, and the RJ series has only a single side contact.

Originally Posted by Nick Lawrence

Well, ideally your electronics would be in a low impact area on your robot, right?

-Nick

Wait, our robots have low impact areas? How did I miss this?

With all seriousness, daisy chaining these makes me want to mount my electronics on shock mounts.... Just in case, y'know?

[Jon236]

Originally Posted by CraigHickman

Wait, our robots have low impact areas? How did I miss this?

With all seriousness, daisy chaining these makes me want to mount my electronics on shock mounts.... Just in case, y'know?

Well, in his presentation in Atlanta, Al Skierkiewicz mentioned that WildStang uses zip ties to secure their controllers to allow quick swapping. The controllers would not be as rigidly secured compared to using 4-40's and thus be able to absorb shocks.

[daltore]

If you think the RJ-45 connectors are going to be a problem you can still use the 3-pin molex if you want to...

This year, we made a nice little electronics box near the end of build season because our electronics boards had begun to look like we'd planted a wire bush and it grew. We ended up with our electronics vertically mounted on 1/8" Lexan with another piece of 1/8" Lexan on the outside to shield them. The only connection problem we had was the ethernet cable coming out because we hadn't checked to make sure it clicked before we went on the field (we'd been doing some tethered debugging). All of our Jaguars were on 1/4" mounts with two Jaguars per mount, so they were able to flop around a bit. The whole thing provided enough rigidity for nothing to break and enough flexibility that no shocks were conveyed directly into any components. Same type of system should work well for RJ-45, because the two Jaguars per mount would be easy to attach to, and then you have the two cables going to and coming from that mount that you can tie down in cable ties.

I'm not really worried about these connectors. We used them this year to carry the signal from our encoders to our digital breakout board and we used the little phone line couplings to change over between wires so we could replace them in segments. Since we were using swerve drive, after a couple of trips around a rogue wheel, the hot glue we used to mount the coupling popped off, and one of the encoders lost the connector on its board (not the RJ-45). Never had any problems at all with the RJ-45 connectors. And considering the build quality of these Jaguars (never replaced one once this year), I think the connectors should be of a high enough quality that they'll serve you well as long as you take care of the cable properly and keep the Jaguars protected.

[Nick Lawrence]

Originally Posted by Jon236

Well, in his presentation in Atlanta, Al Skierkiewicz mentioned that WildStang uses zip ties to secure their controllers to allow quick swapping. The controllers would not be as rigidly secured compared to using 4-40's and thus be able to absorb shocks.

We find using Velcro is perfect for mounting electronics. This year, our electronics panel was all carbon fiber made in our shop too, so it made mounting even easier.

Gee, low impact areas on FRC robots? That's an oxymoron.

-Nick

[DonRotolo]

Originally Posted by squirrel

The only thing that worries me is all those RJ45 connectors in series

Point of order: The 4-pole RJ series is an RJ-10*. Doesn't affect any other statements though.

Originally Posted by squirrel

the CAN bus is a daisy chain. Any one poor connection can kill the rest of the chain.

Nope, the CAN Bus is a parallel protocol, without repeaters or gateways (generally). I would fully expect an unbroken signal path from input to output; the CAN Bus controller would hang off the side of those conductors, not in series with them.

CAN doesn't work well in series or daisy-chain, and that's not the way it is supposed to be used.

It works more like a Radio channel, where everyone hears everyone else, and indeed everyone must hear everyone else (including yourself) in order for the bitwise arbitration to work. (That is the method used to ensure that messages don't collide on a CAN Bus).

A 1 Mb/s CAN bus is a Class A bus, and definitely requires terminating resistors. Their optimal value depends partly on the chips being used, and mostly on the expected medium (type of wires).

There are many CAN Bus resources on the web, however for FIRST I don't think a detailed understanding of the physical layer will be necessary to use it at the application level.

*Also RJ-9 and RJ22 - I cannot tell from LM's data sheets which specific implementation of the 4p4c standard they have. I suspect they did it to take advantage of inexpensive (unreliable?) telephone handset cables.

[MrForbes]

Thanks for the explanation, Don!

I guess I wasn't quite clear expressing what my concern was, and as usual got the terminology wrong because I'm not an expert on the subject. What I meant was that the wiring method had a bunch of RJ connectors in series. The signal has to go thru all the connectors in the chain before it gets to the end. If any fail, the signal won't get through to those past it. Is my understanding correct?

[skimoose]

Considering that in most cases you'll be plugging in your CAN bus only once when you wire your robot, I don't see a concern with the RJ series connectors. Your land line telephones are also connected with RJ series connectors and they also take a fair amount of abuse, their connections have a fairly low failure rate in that environment as well as in network systems.

As a model railroader, we use a similar connection for our data bus standards for digital command control with very few problems. It should work well.

All I can say to FIRST is... "When CAN we play?"

[Stuart]

Originally Posted by skimoose

All I can say to FIRST is... "When CAN we play?"

if Im reading the tea leaves correctly . . next year, I will admit Im much better at reading coffee grounds.


I just hope that they fixed the issue where if the jags sensed a 6v drop ( for any length of time) it would power down to protect itself( I can replace a bad jag, I cant replace a bad match).

[Joe Ross]

Originally Posted by Gdeaver

I hope some time and money is being invested in to the short commings of the jag. Next year we may be back on carpet and It's my opinion that the are several points of failure that could cause some serious problems when first teams start to abuse them. I would rather have a solid Hobby PWM jag that is robust than a weak can jag.

That sounds like a false dichotomy to me. Every indication is that the CAN bus was fully functional this year. Whether or not NI implements CAN hardware, NI or WPI write drivers for CAN, or the GDC allows CAN should be completely independent of the time Luminary Micro spends to improve the design. I haven't seen anything to indicate that Luminary Micro isn't taking the problems seriously.