3
\$\begingroup\$

I'm working in an application where I have to use different connector type system as a Ethernet plug. I did test this already in a single cable inserting this connector in the middle acting as a breakout and it works.

For the final design one of the solutions is to populate a standard RJ45 on the PCB and then make an adapter from the Ethernet plug to our connector system. This should work fine but the problem is that it takes quite a lot of space into the PCB enclosure.

Another solution which would solve the space issue is to use a magnetics IC like this one and solder the Ethernet pairs directly into the PCB board but not sure if this would work fine, I have not seen any working examples/applications of this...

Does anyone know if this last solution (magnetics IC + soldering pairs) would work fine?

\$\endgroup\$
2
  • \$\begingroup\$ A schematic or block diagram would be appropriate, there is a circuit tool to help if you edit your question \$\endgroup\$ – Voltage Spike Aug 31 '20 at 15:08
  • \$\begingroup\$ @VoltageSpike thanks for your comment, basically what I'm asking is if it would work replacing the RJ45 MagJack + Ethernet plug by the IC mag-s558-5999-m8 (instead the RJ45) and solder pairs directly to the PCB (instead Ethernet plug) \$\endgroup\$ – joe Aug 31 '20 at 15:54
3
\$\begingroup\$

The idea behind the magnetics is to isolate the grounding of the differential pair (for esd and stopping DC currents), the second is to create impedance with common mode chokes to stop common mode noise.

Locating the magnetics away from the phy could create problems, but is doable if you do these things:

  1. Ethernet differential lines need a characteristic impedance of 100Ω, make sure the lines cables and connectors leading up to the magnetics are routed as differential lines and have the proper impedance. So if you are soldering pairs, and you can maintain a 100Ω characteristic impedance then then go for it.

  2. Locating the magnetics away from the phy could introduce noise. Why? because you no longer get the benefit of the common mode choke in the magnetics. The longer you make the cable between the phy and magnetics, the larger loop area and more chance that the ground of the phy and the ground of the magnetics will be a different potential. Make sure that the ground of the phy side of the magentics and the phy are the same.

If the environment is noisy this could be a problem, the best way to check would be to build the design then check the eye diagram with a differential probe.

\$\endgroup\$
2
\$\begingroup\$

Ethernet "magnetics" are traditionally a board mount dual transformer component.

The idea of putting them in the RJ45, eg, a "Magjack" was a later innovation.

One could certainly retreat to the original norm with a separate transformer component, and if the non-standard connector or wiring interface meets the impedance match well enough for the desired speed mode, that should work.

It's worth noting that the term "integrated circuit" traditionally applies to multiple active components on a semiconductor substrate. It is not traditionally applied to mechanically assembled coils, resistor arrays and other non-semiconductor products even if those "components" or "modules" contain "circuitry" which is "integrated" and might even have resembled to some earlier generation sort of IC package (eg, a DIP with only sparse pin positions used).

\$\endgroup\$
0
\$\begingroup\$

Well you can solder the Ethernet cable directly to the PCB. This is often done with messurement adapters, when you only want to qualify the cable - not the connectors. Done right it's even better. Connectors are not ideal!

  • strictly ensure imperdance of the PCB traces
  • keep pairs twisted as long as you can
  • keep shield as long as possible
  • do not "roll" shield tail to something which equals a round wire (this will be a kind of inductance in series with the shield - the least you want is blocking HF here)
  • attach shield propperly

PS: why not using a indusrial M12(preferable) or M8 Connector? RJ45 is not very reliable, except if no movement, no vibration, etc.. is involved

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.