REV Sparks or Vex Victor 888?

[KrazyCarl92]

Originally Posted by Alan Anderson

Using CAN is no more difficult in software than using PWM. Wiring it is actually simpler, if you can plan the location of the speed controllers in advance. Using the advanced CAN-only features of a Talon SRX will of course add complexity, but you shouldn't avoid CAN just because of things you don't think you will be using.

Price is a good reason to stick with a PWM-only controller, though.

I'll bite and share my experience:

I've been a part of a team that has used CAN in 4 seasons. In 3 of those seasons the CAN out right failed at some point and we switched to PWM on at least some of the motors (granted those were all with Jaguars...I have my own opinions on those too).

In 2012, I personally spent over 70 hours during a week troubleshooting with a practice bot to get CAN working in preparation for the next tournament, only to have the CAN fail on the competition bot shooter motors at competition and ultimately switch to PWM.

In 2015, Team 20 used CAN with Talons relatively successfully. However, there were still quirks with the code which definitely burned lots of time during build season. I wasn't personally coding, but I recall us taking over a week to iron out the code for switching between current setting and position setting modes of CAN motor controller operation. There was also one practice match where a motor inexplicably drove the opposite direction it should have throughout a match. We were unable to reproduce the issue and it never happened again, but it was very puzzling. Not sure of the cause. Taking advantage of any of the elegance or benefits of CAN does involve added complexity. And it comes with added risk; if one motor controller fails the whole robot is down (the poor durability of Jaguars exacerbated this issue, but Talons are far more durable, haven't had one fail yet).

And having not used CAN in 2013 and 2014, we were able to get much improved robot performance which I believe is in part attributable to not fussing around with CAN. Every hour we spent other years trying to get CAN to work as intended was another hour we couldn't spend practicing. PWM was set it and forget it...it always just worked. I have yet to be a part of CAN going as smoothly, and I have yet to see a robot do something on CAN that wow'ed me and made me say "darn, there's no way a robot using PWM could achieve that same performance". And that isn't for lack of smart, dedicated people trying.

You are correct that if you just used the set voltage mode of motor controller operation, then CAN is no more complicated to code than PWM. But if that is your planned mode of operation, then why not save yourself $30 per motor controller, or get 9 (7 for vets) motor controllers for free?

Perhaps this is a knee-jerk reaction or I am holding a grudge over past experiences. My point is that for the team I am working with now, using CAN is not the low-hanging fruit for proper allocation of our resources (monetary, time, man-power). I am basing that assertion on my and my past team's experience working with CAN.


Back to the topic of the thread, the real point of my post was to illustrate that you could get 4 Victor 888's for free through FIRST Choice if you are not using the 200 credits (50 per) on something else. Or the SPARK can be purchased for $45 a piece if you would rather spend your FIRST Choice credits elsewhere. The SPARKS are slightly more linear, but this study shows the Victor 888's aren't all that far off from linearity:
http://www.fightingpi.org/Resources/...12_Day_9.shtml