Dirty environment Ethernet connector

Thread Starter


Joined Apr 24, 2016
Hello Everyone,

so this is my first post on AllAboutCircuits...
I'm a mechanical engineer by training so please be lenient :D

I want to design and build an ethernet connection between two points using an inductive or capacitive coupling device. The idea is to be able to connect and disconnect the link even in muddy, snowy, or rainy environments, such as are found on a railway coupler.

In ethernet circuit diagrams, all i find is a 1:1 coupling with no details added.

I want to build this on a breadboard.
I had the idea of buying a couple of ethernet magnetics, take them apart, take out the toroidal cores, "chop" these in half, re-wind them, then align the core-halves in a plane and set the whole thing in resin:

What do you think? Would this work? From what I've read here ( http://www.encyclopedia-magnetica.com/doku.php/pulse_transformer ), I need low leakage inductance and distributed capacitance as well as high open-circuit inductance. I'm not that sure what that means and particularly what that means about how to wind my pulse transformers...

If you can give me some additional theory (also more mathy stuff than in the textbook offered here) to read up on, please don't hesitate. I'm actually supposed to know all this stuff, but in class it was all quite theoretical...

Thanks in advance for any replies :)


Joined Nov 30, 2010
The first thing I'll tell you is that slicing a torroid destroys it.
Are you talking about pushing the ends back together and hoping the magnetic coupling resumes? That's possible, up to a point.

Somehow I don't think you have the problem defined very well.
You are asking for what you want instead of telling us what you need.
This is a normal thing around here. People get focused too tightly and tend to eliminate excellent solutions they just haven't thought of yet.

Can you put some numbers on low leakage and low distributed capacitance, or better yet, go back to basics. What is the overall goal? Is this just a data link that you think is not the normal way these cables are used?
Last edited:


Joined Apr 10, 2015
#12 is right about cutting the torroids. It is unclear whether your intent is actualy a non-contact, non-electrical, communications "coupling", or an extremely mechanically durable conductive electrical connection.

Ethernet is in fact a communication network standard, which we implement electrically.

Thread Starter


Joined Apr 24, 2016
Yeah, the idea is to push back the ends of the half-toroids and hope the magnetic coupling resumes. I hope a tenth of a mm of dirt between the faces will affect this coupling a lot less than dirt in an equivalent pin connector.

I have no idea what values the low leakage and distributed capacitance need to be for ethernet to work, I was hoping someone could tell me here.

But back to the goal:
The problem is transmitting ethernet from one railcar to the other, preferably at high bandwith.
These days, this is performed via electric pins in a box with a cover on it like this one:
Cover open, bottom mounted electric coupler: ( http://www.railway-technical.com/Scharf-Coupler.jpg )
Full coupling process with side-mounted electric connectors:

It is a major headache keeping all these beautiful pins clean & dry. To make matters worse, in the railroad environment, there is powdered rust everywhere, because train brakes are cast iron on steel... Another problem is you need a drive system which opens the cover and moves the electric boxes forward for electric coupling. You even have to make this happen after the actual mechanical coupling couples and before it uncouples, if you don't do this you end up damaging all these beautiful pins (yeah, the ethernet ones are even gilded :D )

So yeah, I was hoping to do away with all this flimsy crap and mount the two halves of the toroid securely in the striking surface of the coupler face or in an arrangement that would push any snow and dirt out of the gap between them during coupling.

Thread Starter


Joined Apr 24, 2016
I think a picture is best for explaining what I would like to achieve:
So yeah, the goal is basically an extremely rugged contactless inductive connector which does not mind some dirt in it.
Last edited:


Joined Nov 23, 2012
I think a picture is best for explaining what I would like to achieve:
View attachment 104943
So yeah, the goal is basically an extremely rugged contactless inductive connector which does not mind some dirt in it.
Awesome idea but, inductors cause lags in current so you end up slowing the high/low and low/high transitions of a digital signal. 100 MHz and, more recently, 1GHz is the desired signal speed. Your inductors will have to be low values to avoid issues. With these low values, less and less energy can be transmitted or higher and higher energy will be needed to get your signal through. Granted, you only need a very short distance but you have a very noisy environment (by Electromagnetic interference perspective). Motors from the train, steel wheels on iron tracks, sparks, hundreds of people on the train hoping to connect to Wifi, ...

This will be a big challenge and you will need some type of a team for RF communication and materials to select the right ferrite or un annealed iron powder for the toroid (assuming a toroid is the right shape). I like the idea but it is just not something that you build in an afternoon. It will be a challenge but, if effective, will improve connections and reduce the dangerous situations of people entering the track area to clean and check connections.


Joined Jun 8, 2013
Is there a reason the communications devices need to be in the coupler?
Can they be wireless communications devices? Bluetooth perhaps?
Do they need to be failsafe?

Thread Starter


Joined Apr 24, 2016
Wireless or Bluetooth is out because I need the connection to instantly fail when the coupler fails. Nowadays, we have air brakes which instantly lock if the brake pipe is ruptured.
Also, I want to be sure that a car connects only to the car physically connected to it, I don't want people to have to enter pairing codes to avoid connecting to the car on the adjacent track...

@ Gopher: with regard to inductive lag: AFAIK all Ethernet connections already have magnetics (pulse transformers) in them, so my question is whether it is possible to use this feature to perform the task at hand.
Have you considered optical coupling?

OK I know you are in a dirty wet environment but I suspect some form of mechanically cleaned plug / socket would be easier to handle than mucking about with big magnetic s on fast signals.
How about a rotary system with built in wipers of some sort, think flanges with holes that rotate. potentially mechanically when a shroud engages. IR windows would only be uncovered when the shroud was closed, pull the coupling apart and all signals break.
The shrouds could be magnetically attracted, with electromagnets to stablise the coupling without latches whilst still allowing relative movement during coupling and any ferrous particulate would be ejected when the couplings is not engaged.

Ok so you will not be making this with Lego and blue-tack but if you use standard none contact fast optical link components, possibly even off the shelf data links, your design process becomes essentially an engineering exersize.

What technology do the deep sea wet data couplings use, have you looked?

Sounds interesting in general, I doubt I can help but will be watching.

Thread Starter


Joined Apr 24, 2016
I have thought about optical links, but believe me, the rail industry is pretty much one of the worst places to try that out...
I guess it would work inside the sealed boxes I mentioned earlier, but this is precisely the thing I want to avoid...

I think what I am looking for is Near Field Magnetic induction communication. This article really hioghlights the benefits over the other principles auggested...

I guess the rail industry seems a nice market here because I need to transmit only for a few millimetres, a requirement where most applications would just have a wired link...
If anyone has tried near field induction communication on a breadboard I would be very happy to hear... :)
NFC is inherently slow, essentially for the same reason that cutting a torpid in half probably inst going to work, and if you factor in all those huge current spikes and EM from high voltage panto-graphs I suspect it would be a none starter...

That said, I am just a hobbyist and don't know enough to make a sensible prediction but If I were putting money on it I would go for anything that didn't rely on unshielded magnetic coupling, particularly if said coupling method was already at the practical limit before you factored in the worst imaginable environment... But that's just me.

Watching with interest