blog; Motor Controller Options for 2015

[Kevin Sevcik]

Originally Posted by Mark McLeod

Here is a height comparison shot.

Wait. You mean to tell me these new speed controllers AREN'T two-dimensional? I take back everything I said about how ridiculously small these are compared to all previous FRC controllers.

Seriously though, these things are even smaller than Spikes. I can see why you would only bother with 4 relay ports on the roboRIO if you knew these were in the pipeline. Why bother with a 20A max relay with a fuse you have to replace when you can have a 40A speed controller in the same footprint?

[AndreaV]

Originally Posted by Kevin Sevcik

Why bother with a 20A max relay with a fuse you have to replace when you can have a 40A speed controller in the same footprint?

There are 24.04 reasons why, but none of them are very good.

[Andrew Schreiber]

Originally Posted by Kevin Sevcik

Wait. You mean to tell me these new speed controllers AREN'T two-dimensional? I take back everything I said about how ridiculously small these are compared to all previous FRC controllers.

Seriously though, these things are even smaller than Spikes. I can see why you would only bother with 4 relay ports on the roboRIO if you knew these were in the pipeline. Why bother with a 20A max relay with a fuse you have to replace when you can have a 40A speed controller in the same footprint?

Cost and Complexity for the end user? A relay is a lot easier to use than an ESC if all you need is forward/stop/reverse.

[Jared]

Originally Posted by Kevin Sevcik

Seriously though, these things are even smaller than Spikes. I can see why you would only bother with 4 relay ports on the roboRIO if you knew these were in the pipeline. Why bother with a 20A max relay with a fuse you have to replace when you can have a 40A speed controller in the same footprint?

Driving solenoid valves, driving solenoid actuators, powering lights, saving money, and using parts you already have.

I still don't like the 4 relay ports on the roboRIO because we've used more than four relays multiple times. It seems kind of silly to me that we downgraded from a maximum of 16 relays to 4 on the new controller, considering the driving circuitry for the relays is probably not too complicated.

[FrankJ]

In my opinion, the spike is only useful for the compressor where you are allowed a self resetting breaker & very small motors like a window motors. With the mid size & larger motors, the least stall will blow the fuse.

[Joe Ross]

Originally Posted by FrankJ

In my opinion, the spike is only useful for the compressor where you are allowed a self resetting breaker & very small motors like a window motors. With the mid size & larger motors, the least stall will blow the fuse.

With the 2015 control system, the Pneumatics Control Module (PCM) drives the compressor directly. That accounts for 90% of the uses of spikes we've had over the years.

[FrankJ]

Originally Posted by Joe Ross

With the 2015 control system, the Pneumatics Control Module (PCM) drives the compressor directly. That accounts for 90% of the uses of spikes we've had over the years.

Yay.

[Kevin Sevcik]

Originally Posted by Andrew Schreiber

Cost and Complexity for the end user? A relay is a lot easier to use than an ESC if all you need is forward/stop/reverse.

I'll grant you cost, with the Victor SR being $25 more than the Spike. I'm pretty doubtful on complexity of the robot itself. (A Victor SR is obviously internally more complex than a Spike.) The Spike is somewhat different from all the speed controllers mechanically, programmatically, and electrically (with that 20A fuse to think about). So I disagree with your position that a robot with two rather different kinds of motor controllers on it is simpler than a robot with only one type of motor controller. Especially since I've seen our novice programmers get somewhat confused by the difference between relays and speed controllers in C++.

Originally Posted by Jared

Driving solenoid valves, driving solenoid actuators, powering lights, saving money, and using parts you already have.

I still don't like the 4 relay ports on the roboRIO because we've used more than four relays multiple times. It seems kind of silly to me that we downgraded from a maximum of 16 relays to 4 on the new controller, considering the driving circuitry for the relays is probably not too complicated.

See below. With the Pneumatics Control Module, all your solenoid valves and compressors are taken care of. That leaves solenoid actuators and lights. And I wouldn't be surprised if the GDC treated the PCM like the solenoid breakout and let you power solenoid actuators from it. So lights. Which you could drive from a Victor SR if you wanted, actually.

Originally Posted by Joe Ross

With the 2015 control system, the Pneumatics Control Module (PCM) drives the compressor directly. That accounts for 90% of the uses of spikes we've had over the years.

I'll ditto this. We've only used Spikes for compressors or really small loads that I just didn't want to find room for a Jaguar for. Now it's a Spike that I'd have to find room for.

[asid61]

Originally Posted by Ether

Using common sense as a guiding principle, would it be permissible to apply a thin film of non-curing thermal grease to the mounting surface of the new Victor SP (or Talon SRX) before bolting it to the robot?

To me not allowing that seems mildly ridiculous. It's the equivalent of putting a label on a CIM and having it count as a modification.
Grease would be a safety feature IMO. Nobody wants the robot to go up in flames.

[cgmv123]

Originally Posted by asid61

It's the equivalent of putting a label on a CIM and having it count as a modification.

Labeling is a listed exemption in the relevant 2014 robot rules regarding modifying electronics; adding thermal paste isn't.

[topgun]

Is there a drawing for the new Talon CAN controller available? What kind of connector will there be for CAN? Will we have to buy a special crimping tool?

I like the zip tie options for fastening, and the small form factor.

In 2013 we hit a limit with 8 CAN controllers. Anything after 8 was sporadic, so we had to back off down to 8. Will the new Talon have a similar limit, or is that more a function of the controller or bus?

I sense another FIRST Choice-esqu meltdown coming as 3000 teams attempt to order 30000 controllers on the first day of availability. Every year we are in the 8 - 10 range of controllers on our robot, in addition to several for testing.

[Kevin Sevcik]

Originally Posted by topgun

Is there a drawing for the new Talon CAN controller available? What kind of connector will there be for CAN? Will we have to buy a special crimping tool?

I like the zip tie options for fastening, and the small form factor.

In 2013 we hit a limit with 8 CAN controllers. Anything after 8 was sporadic, so we had to back off down to 8. Will the new Talon have a similar limit, or is that more a function of the controller or bus?

I sense another FIRST Choice-esqu meltdown coming as 3000 teams attempt to order 30000 controllers on the first day of availability. Every year we are in the 8 - 10 range of controllers on our robot, in addition to several for testing.

Here's your drawing.
CAN wires are permanent pigtails just like the power and motor wires. There's two twisted pairs. Based on the roboRIO, the new CAN bus is two wire, not four, so each pair is a leg of the bus. I'm guessing wago lever nuts or something will be the preferred solution for tying all those pigtails together.

Also, we had no problems with 12 jags on the old CAN bus, so something must have been odd about your setup.

[dmaki]

Originally Posted by ebmonon36

I am curious to see what kind of connectors are on the CAN leads.

I talked to CTRE at CMP and he said he was considering using a 2 row 20 pin 2mm pitch connector for CAN, limit switch, and encoder inputs in order to fit on the small body of the motor controller. The same type of connector commonly used for JTAG debugging on ARM microcontrollers. Based on the drawing, 2mm pitch looks about right, but went down to 10 pins for limit switches and encoders and moved CAN to dedicated pigtails.

Also, I've just mounted 2 prototype Victor SP on a testbot. The wires are very supple and flexible. "Noodley", good stuff.

[FrankJ]

Since I don't speak for the Cross the Road people, this is based on what I know about CAN. CAN is a two wire bus. It is intended to be daisy chained. That would be the reason for two pairs. One in & one out. The legs of the pairs would be connected together in the Talon. I know some people use the star configuration for CAN & get away with it. It is not to the standard & it in not a best practice.

[Oblarg]

Originally Posted by FrankJ

Since I don't speak for the Cross the Road people, this is based on what I know about CAN. CAN is a two wire bus. It is intended to be daisy chained. That would be the reason for two pairs. One in & one out. The legs of the pairs would be connected together in the Talon. I know some people use the star configuration for CAN & get away with it. It is not to the standard & it in not a best practice.

My only problem with the daisy-chain configuration is that it greatly exacerbates the failure mode. It's rather annoying to lose half of your motor controllers instead of just one of them.

It would be nice if there were a supported alternative.