Electrical Board or Distributed System?

Team 5404 had a single electrical board with almost all of our electrical components attached to it this year. We ran power wires to the motors and actuators distributed throughout the robot and signal wires to our various sensors.

We are considering switching to a more distributed system, with a primary electrical board with the PDB, roboRio, VRM, and PCM attached. All of the other electrical components (motor controllers, spikes, etc.) would be distributed throughout the robot to be close to the actuators they are powering.

We think that a distributed system would be a lot more modular, though possibly more difficult to make, as the electrical board couldn’t be fully made off the robot and then attached.

Which method does your team use? Why do you use it?
Thank you for any guidance you can offer.

We have used a centralized control board each year so far. Every year we re-consider whether the board will be centralized or distributed. The greater ability to program and trouble shoot a removable board has so far trumped the modularity aspect. After our first two years, not only do we keep the board centralized, we have it oriented so that we can see all (or at least most) of it at one time.

We did move our pneumatics control module to the back with the compressor and other pneumatics components.

We also considered moving individual Talon SRXs to be near their motors and encoders to shorten the encoder cable lengths.

  • That is, easier to see all the lights at once, and easier to trace the shorter signal runs.

Our team has always used panels. One thing we used for many years was something we referred to as a “smart door”. The panels were hinged and able to fold out of the robot to allow it to be easier to work on.
This past year we used an electrical box that 1405 came up with. RoboRio, PDP, VRM, Talon SRXs, Gyro, Radio, main breaker, and RSL were all inside. The idea behind this was to have everything well protected and compact and allow for easy swap outs in case of electrical failures.

There’s a good argument to keeping sensor wires as short as possible. We use the TALON SRX almost exclusively, and try to locate those near the motor/encoder to reduce noise pickup. The CAN bus is quite reliable, and not as subject to noise pickup.

We’ve done both ways. Having things spread out always seemed to be more prone to damaging wires in robot mechanisms, but that can be addressed. Having things centralized is good for plan-in-advance neatness, but it can end up being harder to work on and might make it more difficult to see all the important status lights.

I prefer to have the electrical system as a single module, with sufficient space allocated for it from the beginning of the robot design. That space is in three dimensions, and needs to be accessible and kept from getting buried under other mechanisms.

Having everything in one place is really cool, but the ability to put things in other parts of the robot shouldn’t be discounted. I’ve done both and as long as you can get to the connections / devices for repair & replacement it’s all good.

I think having an electrical person on the mechanical design team is your best bet. I always harken back to my first car, an AMC Rambler. It was a clear case of the mechanical team and the electrical team never met. Since on that car to replace the starter you needed to pull the engine. :frowning:

For our team, we’ve found a nice balance between the two.

Firstly, we try to have a traditional spider layout w/ banked motor controllers for the drive train eg https://media.team254.com/2013/01/image.png

The key thing is that the path from battery to all the drive motors is as short as possible. This means we usually end up with a layout similar to the above, except with only four-to-six motor controllers in the banks.

The rest of the electronics is more loose. We quasi centralise all the “intelligence”, namely the roboRIO, radio, camera-processor etc. mainly for convenience. However when using the SRXs, we prefer to place them as close as possible to the mechanisms, keeping sensor cables short.

This allows us build mechanisms complete with most of their electronics, so that we only have to connect/wire power and CAN to the rest of the robot. It also has the secondary advantage of not needing to deal with all the SRX CAN wires when they are right next to each other.

This looks lovely! Can I ask what the base (the diamond cutout area) is called? My electrical team is looking into adopting that for our next robot, but we can’t seem to figure out what it’s called! Thanks

Bellypan.

Most robots have them, this one’s just lightened by using a diamond pattern.

Newer team here but we made a panel based on a team from FL design. You print the corner pieces and then you an customize sizes dimensions to whatever you need based on your overall design. All wires use color coded Anderson connectors. Worked great for us!
Google Photos

The finest wiring job of any team I’ve ever been on: 4901, 2014. Every other robot was a poorer imitation of that basic arrangement: centralized power distribution panel, run motor controllers close to that so you can troubleshoot them all at one time.

The electrical board can be assembled outside of the robot chassis and then attached to the assembled chassis very quickly if it is designed that way. It can be assembled and pre-tested while the construction of the chassis proceeds in parallel. Some teams might find this approach to be advantageous over the approach of having to dedicate a block of time for installing and testing the electronics in the robot after the chassis is (mostly) assembled.

We used a sheet of corrugated plastic plastic to hold all the electrical components. It was attached to the kit chassis using small zip ties through holes punched in the corrugated plastic. We were able to remove it and install it in an old kit chassis to create a practice robot very quickly. It was brought in as part of our withholding allowance and was quickly re-installed in the competition chassis along with the new versions of the scoring mechanism.

I’m a big fan of the centralized electronics board. I like having everything neatly laid out on the belly pan similar to how timytamy’s picture has it. And as far as modularity is concerned, if you connect all components to each other using Anderson Power Poles for the larger wire, 2 pin JST connectors for CAN and smaller wires, and use 3M Dual Luck to secure components down then you have your modularity in the bag. I believe that this is the best price but you have to buy a whole lot from here. This is probably more practical for most people. But I like the first link where you have to buy 150 feet. $2.46 a foot for this is a really good price :smiley:

As this thread has been necro’d, and we did it a bit differently this year, time for an update!

For 2017, to improve gear cycle time and simplify climbing, we opted for a small and light robot (starting configuration fit in BOTH sizing boxes, though we had lateral extensions during the match; under 75 lb at bag time, WITH bumpers and batteries, though we added 3 or 4 more pounds before competition). We also used SRXs for our four drive motors to better support autonomous operations. We used the “back” 5/8 of an AM-14U2 chassis (4 wheels), but we wanted to leave most of the belly pan open for a floor gear pickup (that we never executed). As such, we had three separate control boards!

  • On the port (left) side, the PDP and motor controllers for the manipulators
  • On the starboard (right) side, the RoboRIO, radio, VRM, and relays for the decorative LEDs we never installed.
  • Under the robot, particularly under the CIMs, a small board with the four SRXs for the drive motors (we called this board the “hindbrain”).

Hiding the four SRXs where we could barely see them was an ongoing problem; while it made great sense at the time and used less wire, I’m going to be advocating for all the control system devices (esp. those with diagnositc LEDs) being clearly visible from here on.

i will have to vote against distributed system:

  • your robot will look very bad
  • it is easier to route a 24 AWG sensor wire across the robot than to route a 12 AWG power wire across the robot. (even tho you will have to do it anyway)
  • problem solving is so easy when everything is in the same place.
  • easier to use CAN with a single board (if you do)
  • inspectors like it all together

Ahh, love these necro’d threads!

I’ve seen both ways done very well, looking very aesthetically pleasing. That being said, I do believe it is easier to have a decent looking electronics board with centralized electronics.

I agree, that is significantly easier, but as you stated, you’re going to have to route 12 AWG power wires as well. The choice you have to make is longer sensor wires or longer CAN bus wires)

This is one of my defining reasons. As long as everything is VERY well labeled and the wires are orderly this system is immensely easier to problem solve and fix; most everything is centrally located.

Another HUGE pull towards a centralized electronics system. You can CAN bus incredibly easily with very little wire.

In my experience inspectors have to deal with all kinds of scary stuff that as long as everything is reasonably organized they don’t mind too much one way or another. Though if yours looks good it may be more of a benefit to the judges than the inspectors…

It doesn’t really matter what you do if you follow a couple of Al’s rules when you assemble.

  1. High current wires should be short. These include all wiring connected to the battery, PDP input and main breaker.
  2. High current loads (drive motors) should be short and as nearly the same length as possible. Long runs to one side of the robot while the other side is short will cause your robot to turn all by itself. High current in long wires drops voltage.
  3. The PDP is best centered in the robot. You can then connect speed controller wiring directly to the PDP and then the motor wires directly to the controllers. (Think star with the PDP as the center.)
  4. Following Al’s corollary to Murphy’s Law (Anything that can go wrong will go wrong, ON EINSTEIN!) make electrical components easy to replace. I prefer to use ‘non-hardware’ methods of attaching components. If you are in a hurry and drop a screw, you will spend a great deal of time looking for the lost hardware.
  5. Components with lights on them need to be visible from outside of the robot without trying to look through a lot of mechanical gack to see if an LED is on.
  6. Low current wiring like power to RoboRio, PCM, and VRM can be longer to allow those devices to be mounted as needed. Can bus, serov and PWM wiring can be as long as needed. Do not run signal wiring near high noise sources like motors and compressor. Use wire ties liberally to keep everything in place.
  7. Use care when stripping and terminating wiring. The PDP needs wires to be stripped 1/2" to 5/8" to be firmly held in the compression terminals. Perform a “tug” test on all terminations to be sure they won’t pull out.
  8. Many teams use some kind of label on the wire. We use a Scotch wire marking tape you can get from Digikey. It has ten colors in a plastic holder. The tape is 1/4" wide. It is $47 but worth every penny when trying to search electrical problems. STD-C-ND on Digikey website.

With your permission AL, I would like to print these rules out and post them around our electrical area. I think this post has even helped me figure out why our bots in previous years didn’t go in a straight line (though it still could be something completely unrelated).

These are all great points, and as you have always done, you have added more and more value and help to this community than I could have imagined. Thank you and your team very much.

Whether to do distributed electronics on multiple boards is exactly the problem of “partitioning” in VLSI chip design (whether to integrate everything in a single chip or come up with a multi-chip set design). To make this decision, one also needs to consider the following:

  • Minimize number of interconnects between boards: if you move a component to a different board, how many interconnects will you create? Are the interconnect lengths critical?
  • Keep it modular: one may argue separating subsystems into separate boards made it more modular. But remember one thing, if the module has a lot of interconnects to other boards, it’s not modular when you need to disconnect many different wires in order to take the “module” out for repair. This could be alleviated by grouping interconnects into block connectors. Generally, two connectors, one for power block and one for signals. On the other hand, connectors create points of failure. So if there is really no reason to put components off to a separate board, don’t do it.
  • Spatial consideration: The only valid reason for separating components to a separate board is because they don’t all fit on one board. In that case separate them wisely with the above rules.

In general, we found we often ended up with two boards, one for all pneumatics and one for the rest. The pneumatic board connects to the main board with two connectors: power for the compressor and pneumatic valve signals/pressure switch signals in a large keyed connector with locking mechanism. The two boards can be taken out with minimum electrical disconnects and mechanical screws from the robot.

Phil,
Anytime, I hope it helps.
People have been bugging me for a few years to put down in a book some of the things that I have seen on robots over the years. I really need to do that. There are so many to chose from.

To revisit the above post, please add:

  1. Securely mount the robot battery. It should not move. A team learned that the hard way this past season when their battery moved and the wiring bumped the main breaker “off” button.
  2. Mount the radio high on the robot if you can and don’t surround it with metal objects or robot chassis. The internal antennas work best when the radio is horizontal.