Proposed Hardware Changes #68
Replies: 12 comments 65 replies
-
|
I'd step back a bit and ask, why are the CAN and "Smart I/O" connectors different from each other? It would be nice to have just one way to do low-voltage signal connectors: if we have to get good at tiny crimps, fine, we'll use tiny crimps for everything. Regarding the power connector, I agree that a Microfit, or similar ground-protected connector, would be better than the Weidmuller. |
Beta Was this translation helpful? Give feedback.
-
|
The percentage of teams that will have nothing plugged into the SmartIO will be very high, many teams will never need to crimp those connectors. |
Beta Was this translation helpful? Give feedback.
-
|
In general, I believe FRC has rarely required teams to crimp critical connections with specialty connectors. On the RIO, power is a screw terminal and CAN is weidmuller. PWM motor controllers come with pre-crimped connectors. Really old PWM motor controllers could use pre-crimped servo wire on both sides. I think maybe very long ago the controller power took spade or ring terminals, but those are much easier to crimp and the crimps and crimpers are much more widely available (e.g. from a local hardware store). The vast majority of teams probably don't crimp a single connector on their robot, and it would be nice to keep it that way. |
Beta Was this translation helpful? Give feedback.
-
|
I'm a fan of connectorizing to latching+polarized connectors. Bare wire connections to spring loaded connectors are subject to multiple failure modes that connectors are not (incorrect strip length, incorrect insertion depth, swapped wires, failure to tug-test). They're also not actually tool-less, as you have to strip the wire, and often a tool such as a small screwdriver is necessary to depress the spring. With a standardized latching+polarized connector and pinout across the ecosystem (which would naturally be encouraged by the SystemCore having such a connector) would mean far fewer wiring errors such as having the polarity swapped (and sometimes swapped twice). Standardized connectors also bring benefits with respect to ease and speed of installation and reconfiguration. Even in early 80s computers, almost everything was connectorized, and connectors still are the universal approach in consumer products for these reasons. In FRC, since there are a large number of CAN products on the market currently, there will of course be a transition period when some "dongle" cables are necessary (or working with bare wire/crimping), but I would see vendors rapidly converging on whatever connector (and pinout) SystemCore uses for new product releases, and that would be an overall benefit for ease of use for all teams. Notably, using a connector does not mean teams have to crimp their own connectors. If you really still want bare wire to work with, pigtail assemblies exist (and a pack of these could even be included in the box). For teams that really wanted to avoid crimping, COTS pre-fabricated extension cables with connectors on both ends in various lengths can be produced and sold cheaply by vendors (see COTS servo extension cables). In the pre-CAN days of FRC when PWM servo cables were used for all motors, entry-level teams seldom crimped their own cables--they would get PWM servo extension packs to wire everything up, and never touch bare wire. |
Beta Was this translation helpful? Give feedback.
-
|
The other nice thing is Wago 222's exist and are easily and cheaply available from vendors if teams want to go to bare wire, especially if those same vendors start selling Molex SL to bare wire pigtails (And maybe even ship bundles together). So for teams that want to crimp their own or buy premade cables, they can, but for teams that want comptiblity with all the old bare wire stuff, thats also easy. |
Beta Was this translation helpful? Give feedback.
-
|
We've had enough issues with welidmullers that I'm interested to see locking connectors have their time to shine. https://refiresolutions.com/refire-quick-connect-for-brushless-motor/ - ReFire has been pumping out 2-pin locking connectors for CAN and I do like it. One potential option: Use a 3-pin connector for CAN, with the middle pin grounded for shield. It would provide some backward compatibility with Rev's existing cables. |
Beta Was this translation helpful? Give feedback.
-
|
Crazy thought: Why not both? This way, teams can chose one or the other dependent on their own risk profile? Also, Microfit+ is a wonderful connector and I love it, but there doesn't seem to be a low-cost, compatible crimper available for teams to use. Is there something FIRST can do to try and reduce the cost of the connector + crimper? Mouser Link |
Beta Was this translation helpful? Give feedback.
-
Thankfully, this path isn't on the table. |
Beta Was this translation helpful? Give feedback.
-
|
There's a table? |
Beta Was this translation helpful? Give feedback.
-
I'm curious why it's not. If we're willing to discuss multiple built-in options, why not allow adapters from "what the hw ships with" to "what your team prefers". Not saying I like the idea as-is, but it's weird to me to shut down that discussion. |
Beta Was this translation helpful? Give feedback.
-
Swyft basically attempted to do this with their early breakout boards for the roboRIO and CTRE modules. They sent me a box of them for review and I didn't hate the concept but we also never ran them in competition. But if vendors wanted to go that path, having Weidmuller connectors on SystemCore starts sounding far more sensible. |
Beta Was this translation helpful? Give feedback.
-
|
Specific to the "let teams choose" - If systemcore itself had Molex connectors on top, it would be feasible for teams/vendors to supply a PCB that snaps into all the molex connectors together (maybe only locks into the end ones?), and breaks out the connections to weldimullers. The reverse is much less feasible, as removing a weildimuller-to-molex board would require pushing down on all weildimuler releases (10x) together. |
Beta Was this translation helpful? Give feedback.
-
|
Am not testing SystemCore, but I have run a team long enough. We've already had to chop up countless cables to interface with things. Some are replaceable (SPARK MAX, Thrifty Nova, most newer CTRE), some aren't (CANcoder, brushed VEX/CTRE controllers, looks from pics like Pigeon 2 and SPARK FLEX). Our bus ended up a mix of bare wire, WAGO 221/222, and REV's OEM 3-pin connectors. Even if you got every vendor in the game to play ball, any new connector is going to feel like xkcd_standards.png whether it's better or not. That's not to say I have a burning love of Weidmuller connectors, but I know the coping mechanisms (ferrules, hot glue, wire stripper, WAGO 222) are cheap and abundant. With the several modules that seem to remain part of the new system, bare wire ends aren't going anywhere and I hope the cable selection accounts for that (presumably, at least friendliness to WAGO 222s and/or heat shrink solder connectors). |
Beta Was this translation helpful? Give feedback.
-
|
If the connector is going to change, it would be very helpful for teams in Turkey if the new connector is available at Digikey . Since they have a distributor in Turkey, it's much easier for teams to buy from them. Getting connectors from other suppliers can be quite difficult here. |
Beta Was this translation helpful? Give feedback.
-
|
Could you say more about what's special about the crimper? The crimp area of the terminals looks absolutely generic. What goes wrong if you use a low-cost all-purpose tool like the Molex 63811-1000? |
Beta Was this translation helpful? Give feedback.
-
The expensive one does things like ensure the crimp is in the right position, and ensures the wire is at the right position in the crimp. It makes it trivial to get a correct crimp. |
Beta Was this translation helpful? Give feedback.
-
|
oh, yeah, but it's certainly possible (with a small amount of extra attention) to get the same quality of crimp from the simple tool, isn't it? |
Beta Was this translation helpful? Give feedback.
-
|
The insulation and wire crimp regions require different crimp forces as well. The "real" crimper uses a stepped die. Yes, it is possible, but it takes experience and attention - both things that students don't often have because they're still learning. |
Beta Was this translation helpful? Give feedback.
-
|
According to the reviews, these are apparently good for crimping Molex SL: https://www.amazon.com/IWISS-IWS-3220M-Connector-0-03-0-52mm%C2%B2-Ratcheting/dp/B078WPT5M1 |
Beta Was this translation helpful? Give feedback.
-
|
Another thought: What fails first on these cables? The side with the locking tab (board side), or the wired connector? What happens when the locking tab on the board side breaks? We've seen a number of RIOs getting dragged across the carpet by just the Ethernet and power cables. |
Beta Was this translation helpful? Give feedback.
-
|
? the terminals just come out, use a new housing. |
Beta Was this translation helpful? Give feedback.
-
Do they? Are you sure that the terminals don't break in the housing? Again, how do these connectors fail? |
Beta Was this translation helpful? Give feedback.
-
|
oh, sorry, i think i lost track of which connector we're talking about. are you talking about the Microfit or the SL? i've never extracted pins from microfit connectors but the internet says you use the pin extractor tool, or something like it. https://www.reddit.com/r/VORONDesign/comments/yvjq0w/is_there_a_trick_to_removing_pins_from_a_molex/ |
Beta Was this translation helpful? Give feedback.
-
The extractor tool is actually a set of small, thin blades that sneak between the connector. It's really finicky to use (like the photo shows) |
Beta Was this translation helpful? Give feedback.
-
Unfortunately the housing on the device is the first thing to break in my tests. I really hoped they'd engineered these so the connector would break first. Locking connectors are still nice because they won't vibrate out so we won't have to use hot glue on everything but if your robot rips a cable out, it will break the connector on SystemCore. In my tests, the port is still use-able because the pins are still there but you'll have to start hot-gluing again. Also if teams don't want to risk breaking their connectors, they could use non-locking 0.100" spacing connectors like people use on the Rio now. |
Beta Was this translation helpful? Give feedback.
-
|
Another thing worth highlighting is that the Microfit+ premade cables aren't sold in pairs. This means you'll need to manage bundling them yourself. 
|
Beta Was this translation helpful? Give feedback.
-
|
is this not what we're talking about? |
Beta Was this translation helpful? Give feedback.
-
Connector Suggestion/Recommendation:Why not use something that's more widely accepted and common?
|
Beta Was this translation helpful? Give feedback.
-
|
i thought that was a good idea too but the minifit is much bigger than the microfit. |
Beta Was this translation helpful? Give feedback.
-
Sure, but there's plenty of PCB real estate and (in my opinion) the nearly-universal availability of the connector makes it significantly more compelling than the Microfit+ option that's only manufactured by Molex. |
Beta Was this translation helpful? Give feedback.
-
|
I think you might be overestimating the importance of availability, and also overestimating the availability itself. if you don't have this thing lying around (almost nobody does), you have to go buy one from some specialty shop. digikey, mouser, etc, have all this stuff, and long pigtails could be included in the KOP. we didn't really like the RoboRIO power connector, but it wasn't actually a problem in any sort of objective sense. you buy the weird-thing from the weird-thing-sellers. |
Beta Was this translation helpful? Give feedback.
-
The RoboRIO connector didn't require any special crimp tool because it was (essentially) a screw terminal. Also, I can guarantee that you own at least one of these connectors. They're exactly the same thing used for +12V rails in desktop PCs. This is exactly what makes them such a good choice for this application since it enables FRC to leverage the broad availability developed by that market. |
Beta Was this translation helpful? Give feedback.
-
|
I like connectorization, and I really like eliminating all moving parts (weidmuller buttons) from Systemcore. Still, I agree that crimping is basically a non-starter for low-resource teams (and a bad experience for everyone else). With weidmullers, all you need is a wire stripper and Wago inline connectors. Also, teams can "convert" a weidmuller port into a Molex SL port with a single pigtail. With that being said, the "connectorization" and "crimping" discussions are separable if we look at the kit of parts. How do opinions change if the KOP is loaded with an excess of pigtails, extension cables, twisted pair wire, and wago 221s every year? What if we can guarantee a surplus these items at every event and within every team workspace? For a low-resource team, short CAN connections would be made with extension cables, and long CAN connections would be made with two pigtails and a cut-to-length twisted pair section in-between. |
Beta Was this translation helpful? Give feedback.
-
I've found three flavors of Wago 221 issue:
|
Beta Was this translation helpful? Give feedback.
-
|
I understand concerns about the weidmullers breaking or wearing out, but I've heard both sides of the story (i.e. teams who never had any issues with them, and teams who have had many issues with them). If the concern is robustness, I hope any decision made is based on additional data as opposed to just anecdotes (either positive and negative). Additionally, the mating cycle rating for the Microfit+ (couldn't find the datasheet for the SL) looks worse than the weidmuller (although it's unclear whether one side of the connector cycles more quickly than the other). 100% connectorization for FTC is something that I hadn't thought about. If that's a hard requirement, it obviously changes a lot of the tradeoffs. Including reasonably-lenghted pigtails for power seems quite feasible, as a power run from the PDH would sensibly be no longer than 2 ft. CAN gets a bit trickier, since a team could want to run a dedicated can bus for a single faraway mechanism without any intermediate connectors, but that might be uncommon enough to not significantly matter. It would still be a shame to lose the flexibility that the weidmuller provides, but I understand there are tradeoffs. |
Beta Was this translation helpful? Give feedback.
-
I'll respectfully disagree. Given adequate tools, we've been extremely successful teaching students (even inexperienced ones) proper technique to prepare, crimp, inspect, and assemble a variety of crimps, including MolexSL, PowerPole, ferrules, and main power. We get excellent results, and it's just not a big deal. For those who don't want to / don't have the right tools for MolexSL crimping, pigtails to bare wires are a cheap alternative that can be readily supplied.
True enough, and that's what we'd do, but I see an opportunity to move the ecosystem towards a better standard here. Consider: if the MolexSL becomes the standard connector for CAN - which having it as the native interface on SystemCore would go a long way towards - it encourages other vendors to adopt it, displacing the other "standard", the terrible non-locking Dupont connector (which has no place on a robot IMHO). We're already inching in that direction, what with Kraken powerpole adapters / Refire both using it. |
Beta Was this translation helpful? Give feedback.
-
In a world where it is expected for teams to reuse control system parts, I have to ask: what about the installed base? Easy enough for SPARK MAX (make a v3 cable) or components that are themselves Weidmuller-ed (ship Molex to bare wire), but most vendors have at least a couple current products that aren't that simple. (The interim fix might be as simple as heat shrink solder and a bit of wire stripping, but it merits contemplation and verification.) |
Beta Was this translation helpful? Give feedback.
-
A reasonable question to consider, although choosing to continue an inferior standard just because others are already using that standard means we never make progress... Regardless, the nice thing about CAN being a series of connections is, they don't all have to be the same to function. If you've got a pile of Dupont-connectorized CAN wires and you're not able/willing to swap them out (for separable wires) or re-connectorize them with MolexSL, you can absolutely keep using them for whatever portion of your CAN bus you like. |
Beta Was this translation helpful? Give feedback.
-
|
There's another path to discuss here - assuming there are no other changes to the current design, how do you feel about about dropping the weidmuller power input? What if FRC teams used the bridge port / 4 pin Microfit+ for power input? We can assume that every KOP, every year, would include an excess of prebuilt cables designed for the PDP-Systemcore connection |
Beta Was this translation helpful? Give feedback.
-
TBH I don't love it. Weidmullers (used correctly) are pretty decent, and don't require introducing yet another different connector (there are no extant uses of Microfit+ in FRC that I am aware of, at least not in the mainstream). From what I've seen, Weidmullers fail when a too-big wire/ferrule gets jammed into them, or something breaks off inside them. Both are pretty easy to avoid. We've been using the AndyMark green push-in adapter for the Rio for years, which is basically a Weidmuller. It's vastly superior to the stock screw terminals!
Lots of us opt out of the KOP, so unless it's in with the game piece kit, this would mean either:
The problem with prebuilt cables is they're never the right length. Either they're too long and you cut them (likely meaning they can't be reused, since inevitably you'll need a slightly longer one) or they're too short and you have to extend them (potential point of failure if not done carefully). |
Beta Was this translation helpful? Give feedback.
-
Well, you don't have to cut them. You can just bundle up the extra wire. Run lengths really don't matter for CAN or power input to SystemCore. |
Beta Was this translation helpful? Give feedback.
-
This point cannot be understated. I encourage the team reviewing this feedback to please keep this in mind. |
Beta Was this translation helpful? Give feedback.
-
You can opt out of the drive base separately from the black tote, and you still get game-specific items and access to FIRST Choice even if you opt out of both. If for example the pigtails are bundled with the black tote, it's still completely your option whether you choose to get them for free through that, or opt out and buy (or maybe get through FIRST Choice) a crimper and terminals (or maybe get just those pigtails through FIRST Choice). I'm not sure why it's surprising that if you opt out of getting something in the KOP that you might have to buy other things to fill the gap. |
Beta Was this translation helpful? Give feedback.
-
Not sure if this comment was to me or @juchong, but regardless - my primary point was that if we have Weidmuller power terminals, nothing extra is needed. Teams have wires and ferrules already, thus we don't have to carry anything additional with us or buy anything we don't have today. My only point about the KOP was that the stated premise of "an excess of prebuilt cables" negating the disadvantage of Microfit+ (or any other "new" connector) isn't generally true. Yes, obviously opting out of the KOP is our choice and has tradeoffs, but that decision isn't going to be driven by this single cable - historically the KOP had declined precipitously in usefulness so we look at its overall value. I'd love to see the KOP contain Actually Useful Things again instead of throttle motors and spade connectors, but that's a topic for an entirely different thread. 😃 |
Beta Was this translation helpful? Give feedback.
-
|
Taking a bit of a left turn, a few concerns from looking at the current design:
|
Beta Was this translation helpful? Give feedback.
-
Counterpoint: users with colorblindness may not love a single RGB LED. This could be engineered around with funky blink codes, but I don't think the current setup is out of line. |
Beta Was this translation helpful? Give feedback.
-
The RSL is designed to be a super reliable basic signal that the robot is enabled. RGB would just add complication, as now you would need to either run 4 wires to it, or make it addressable, and then still you'd need 3 wires. The 2 wire setup is super nice and easy for conveying what its supposed to convey.
From the experiences we've had with micro SD in FRC, micro SD cards are not the greatest in our environment. If we want an alternative to USB, I'd probably prefer some vendor either make or import an A+E NVME SSD, so teams can expand their storage with the PCIe slot. |
Beta Was this translation helpful? Give feedback.
-
To be clear, I'm not a systemcore tester. However, I do have some experience with PCB design and product dev, and I would say that if you're doing RGB LEDs, at least have two. PC motherboard codes come to mind. I remember seeing a video of a motherboard beep code, and it was very difficult to tell "long long long short" and "long long short short" apart, and I think you'd run into the same issue here. IMO this issue would be remedied if you just had 2 RGB LEDs, and didn't mess with the timing at all, as you could encode binary (long/short vs LED1/LED2) in this way as well, which is what I'm choosing to build into anything I make which needs debugging. |
Beta Was this translation helpful? Give feedback.
-
I think this would be a better fit for a DS device than it would SC. Since a DS device is going to be required regardless to drive the robot, it could host a web browser client to the on-robot development environment. |
Beta Was this translation helpful? Give feedback.
-
|
FWIW, converting a "blinking" signal into RGB is fairly trivial. It's 2025 and RGB is commonly used for safety-critical systems all the time. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
I thought it would be worth having some discussion about https://github.com/wpilibsuite/SystemCoreTesting/blob/main/hardware-changes.md.
Personally I'm not a fan of this. It's easy to make crimping mistakes with small AWG connectors especially those with housings. I think this change would result in less reliable connections for two of the most critical connectors on the robot for many teams.
Beta Was this translation helpful? Give feedback.
All reactions