The Future of Onboard Realtime Communications

I don’t, The springs are only rated for a set number of insertions.

1 Like

I kinda liked them. They weren’t optimal, but any connector is better than the stick-in-a-bare-wire-end crap we have to deal with on the RIO and PDB.

If you’re sticking in a bare wire instead of a ferrule, you’re doing it wrong.

Granted, ferrules are better than bare wires, but only marginally. Ferrules at least partially solve the problem of not being able to inspect a connection after making it. However, trying to correctly connect ferrule’d wires to Weidmueller terminals blind, working by feel, is still a nightmare. Plugging in a D-sub blind is trivial.

In addition, on actual connectors, you can have corrosion-resistant plating on the contacts, and proper stress relief for the wires.


For those of us without much experience in wiring, what is the different between strain relieving a bare wire versus strain relieving a connectorized wire that makes the latter “proper” strain relief?

My naive assumption on strain relief is that the point is to reduce or eliminate stress on the connection point, and I’m not sure how that characteristic is lost without the use of a connector.


We like there to be something that is clamping the entire wire/cable, and keep it in a fixed position relative to the thing that it’s connected to. This way there is no movement of the part that is making the electrical connection.

A wire stuck into a spring clamp can move around, and the electrical resistance of the connection will change as this happens.

1 Like

I suppose I should say, connectorized wires can have built-in strain relief. Of course, you can (and should) strain relieve wires going into a Weidmueller by zip-tying them to structure or something. However, connectors can have backshells with integrated strain relief clamps:


You’re defending connectors and choose a picture involving the creative use of a zip tie… At least it was a Thomas and Betts tie.


So I don’t really have an idea of what I’m looking at other than that it’s cool.

Is what is happening there that some of these connectors have a separate piece that clamps the wire, and then can be secured to the connector housing to provide “integrated” strain relief? I can see something that looks like this on a couple of those pictures (specifically 3 and 4) but the first two just looks to me like some cool connectors with fancy heat shrink over them.

What are these connectors? Do you have any other information on them?

1 Like

Meh, it’s a testbed. If we didn’t have to demate and remate those connectors several times every day, we’d do something more secure.

You’re looking at a mix of MIL-DTL-26482, MIL-DTL-38999, D-sub, and micro-D. Here’s a 61-socket 24682, a 15-pin D-sub, and a 15-socket micro-D (banana for scale):

(Technically, the D-sub and micro-D are connector savers, not connectors that you would use to terminate a cable. Cut me some slack, it’s what I had in my office.)

Here are some 38999 plugs:

Yes. So for example, to install a 26482 plug, you first crimp pins onto the wires, and insert them into the housing (that ring terminal is to provide an electrical connection to the connector shell):

The backshell then slides over the cable and screws onto the connector housing:

In this case, I needed a few wraps of Gore tape to bulk up the cable so the clamps could grip it:

Finally, the clamps get screwed on to secure the cable (this is also where the ring terminal gets attached to the connector body):

Heat shrink can be part of strain relief, but it’s not the only part. What you’re looking at in the first picture is some small 38999 connectors:

Notice that shiny horizontal band just above the connector body? What’s happening here is that the backshell extends upwards, to where you see the heatshrink start to narrow. The cable’s ESD shield (braided metal) goes down over the backshell, and is clamped to it with what’s basically a specialized steel zip tie. Thus, if there’s any tension on the cable, it gets transmitted through the shield to the connector body, rather than being applied to the wires in the contacts. Also, the point at which the wires meet the contacts is buried down in the connector, and protected by the backshell.

Same deal on the micro-Ds. The actual connector is pretty narrow:

Most of what you see is the backshell, which protects the crimp joints and provides a robust place to secure the cable to:

So you first wire up the connector, then slide on the backshell, then secure the cable shield to the backshell with a clamp band.

Partway done (bottom) and complete (top).


A properly sized and crimped ferrule will push into a Weidmuller terminal without having to depress the button. I find that a lot easier to do when working by feel than trying to screw in d-sub screws.

I have enough experience with circular connectors and other MIL-std connectors at work to never want to use them in FRC. I’d very much like to spare my students the pain of having to either use the pin insertion/extraction tools or the solder cups, not to mention the all too frequently instances of an inexperienced user cross-threading backshells.


Wise man once said that Crimpin ain’t eazy.


Sooooooo difficult.

But it’s necessary.

As nice as all the military-grade circular connectors are, at ~$200/connection they are a hair out of the price range I would like.


I will concede that point.

(Consider, however, that you can get bayonet-locking waterproof connectors, known at JPL as “Bendix connectors”, for relatively cheap: They’re not flight quality, but they’re not bad.)

And here I am having a heart attack looking at the budget with the price of powerpoles :joy:

1 Like

I’m a fan of the DT-series from Deutsch as the next step down - tooling is still expensive, but the actual connectors are fairly economical.

However, still too big and overkill for FRC IMO.

Connector-wise, I’d tend to say what you’re really looking is:

  1. Small/lightweight
  2. Vibration durability
  3. Dust (ie metal shavings)
  4. Foolproof assembly

All that being said. Absolutely loving the pictures and technical details. CD needs more things like this!


First of all, I have a really hard time answering this question without considering the feasibility (And I think the results also show that). With that in mind, I would like to describe both.

I personally would very much like an Ethernet-based system, 100base-t would be acceptable for basically any mechatronics components but I think having 1000base-t in key places would be very handy (co-processor, cameras, screens). HEBI Robotics motors are a great example of this. I would definitely be open to a variation of ethernet (EtherCAT) especially something with power but that is secondary. The only thing is that, the variation still needs to be able to use standard ethernet transceivers and such to make it cheap and easy to build out.
If a system like this were to be created I think that using M8/M12 connector would be a no-brainer (@jnicho15). M12 Connectors are highly adaptable and ubiquitous in industrial environments. This would make getting cables a little harder but there are plenty of suppliers and there is no reason the cable has to be that expensive. Also, you can all ways convert them to 8P8C (RJ45). Why I don’t see more use of the M12 Connectors in FRC is something I do not understand (vs DB9). Space cable is cool but even more of a stretch.

As for the most practical solution, it would definitely be multiple CAN busses. I would like to see some of those CAN busses be CAN-FD. That way you can support legacy and updated components. The only problem with CAN is power delivery. So getting a good connector that can do both would be critical. I think of the M12 connectors but that is not practical for small sensor boards. A more practical solution would need to be found for the CAN connectors that could accommodate the market and have greater reliability and flexibility. The locking PWM connectors are what come to mind first (Locking Molex and PWM connectors). Maybe one with more pins but that can still connect to smaller pin count connectors. CAN is a relatively inexpensive and quite robust protocol. There needs to be a CAN switch so that people can do “Star Topology” without getting the spec thrown at them (Rightfully so).