[Rookie] How to set up electrical the best way?

[otherguy]

Originally Posted by slijin

Just out of curiosity, how exactly did you conduct this test?

When experimenting with CAN last year, we ran into a Jaguar that terminated the network when the chain reached it (but we've had far more than our share of bizarre Jaguar phenomena).

We pulled breakers out of the PD board. on jags through the line while maintaining comms with the Jags along the chain.

The schematic for the tan Jag is available online (to substantiate my claim): http://www.ti.com/lit/ug/spmu022c/spmu022c.pdf see page 42 upper right hand corner.


Could the problem you saw have to do with wiring?

[slijin]

Originally Posted by otherguy

We pulled breakers out of the PD board. on jags through the line while maintaining comms with the Jags along the chain.

The schematic for the tan Jag is available online (to substantiate my claim): http://www.ti.com/lit/ug/spmu022c/spmu022c.pdf see page 42 upper right hand corner.

Could the problem you saw have to do with wiring?

Oh, I see. I was wondering if you had taken an actual failed Jaguar (i.e. one that wasn't functional) to see if it would facilitate a network connection.

The problem we had wasn't due to wiring (confirmed by swapping wires between connections to confirm functionality), but more likely due to internal Jaguar complications (which seem to pervade all our attempts to coax them into being functional).

In any case, I would believe that the rationale behind a backbone would be that it would allow one's CAN to continue functioning even in the event of a bizarre Jaguar breakdown in which the hardwired connection is somehow shorted or cut.

[otherguy]

Originally Posted by slijin

The problem we had wasn't due to wiring (confirmed by swapping wires between connections to confirm functionality), but more likely due to internal Jaguar complications (which seem to pervade all our attempts to coax them into being functional).

In any case, I would believe that the rationale behind a backbone would be that it would allow one's CAN to continue functioning even in the event of a bizarre Jaguar breakdown in which the hardwired connection is somehow shorted or cut.

Our team ran the entire season last year using CAN without a single wiring or programming issue. We made the cables once and forgot about it. We went to two regionals and 3 off season events. The robot was driven all summer for driver training, brought to demos for fund raising, and is being used now to demo code and test functions for the 2012 games. And everything's still working, and everything's over CAN. I personally can't say the same for any of our robots PWM cables in the past.

To me, the backbone element gets rid of my favorite feature of CAN, clean simple wiring. And I don't think I quite understand the rational that it would help resolve shorts/cut of your communications cables. I would think this system is still susceptible to shorts, and if a wire is getting cut during a match, then me or my electrical team made a grave mistake in how we ran the wires...

That said, the wiring in that backbone box is really clean. If that's the route a team wants to go it looks like a nice reliable solution.

[Alan Anderson]

Originally Posted by slijin

When experimenting with CAN last year, we ran into a Jaguar that terminated the network when the chain reached it (but we've had far more than our share of bizarre Jaguar phenomena).

Based on a smattering of reports from a couple of teams, that Jaguar might just have had a connector with the pins not sprung out far enough to make good contact with the CAN cable.

[slijin]

Originally Posted by otherguy

I don't think I quite understand the rational that it would help resolve shorts/cut of your communications cables. I would think this system is still susceptible to shorts, and if a wire is getting cut during a match, then me or my electrical team made a grave mistake in how we ran the wires...

The hardwired connection I was referring to was the internal Jaguar connection; i.e. daisy-chaining is susceptible to internal Jaguar complications, whereas a backbone is less so (only the Jaguar on that branch will be affected). Sorry if that wasn't clear.

@Alan, thanks for that info - I hadn't been aware that that had been an issue. I'll keep that in mind when we play with Jaguars in the future.

[mikets]

Originally Posted by otherguy

Just so you know... If a jaguar in a daisy chain fails, down stream jags are NOT affected. The contacts on the PCB are wired through (in to out) without going through active components.

Yes, that's true. I realize my mistake after I post. That's why in my subsequent post I said the following instead.

So there is no issue of daisy chaining where one segment has problem and all down stream Jags are disabled.

If you make your own CAN bus cable, which I do, you may have cable problem or connector problems that would affect down stream Jaguars.

[EricVanWyk]

Originally Posted by mikets

Here are the pictures of the 12-port CAN hub (part 1 of 2).

Sorry that I'm late to the party, I found my way here by a link from a different thread.

I'm curious how the CAN hub is holding up during testing. It is an interesting tradeoff between cable reliability and signal reflection. How many places are you terminating? What are your cable lengths?

If I remember my CAN math correctly (and honestly it has been >2 years since I've done any form of line reflection analysis), at 1 megabit those taps should be limited to a third of a meter. But I think that was assuming an electrically significant distance between the taps.

Could you scope the bus and see what the reflections look like? If it works, this could be really handy.

PS: Maybe this needs to be spun out into a different thread.

[mikets]

I have spun off a different thread.
http://chiefdelphi.com/forums/showth...15#post1121115