2
\$\begingroup\$

I can't use a magnetic jack for this POE PD (ieee803af), because we use another connector (M12).

Looking for components to replace the internal magnetics, I found that almost all components and schematics use this no nonsense layout: POE implementation

But the internal magnetics of magjacks look like this: Internal Magnetics RJ45 If looks more fancy: - first tap off the DC - then a choke - then transformer coupling

It is hard to find transformers that are rugged (temperature to -40degC) and implement all these characteristics.

Should I be worried about not using the first coil where the DC is tapped off separately? If not, why is it implemented like that in the internal magnetics jack?

The transformer I've chosen for now is HX2260FNL (datasheet HX2260FNL), so my design looks like this: (with external Schottky bridge rectifiers) (for future reference, so shoot it if it is not good!). enter image description here

EDIT after post asndre:

 H1102FNLT this H1102FNLT transformer has the common mode choke on the cable side in the same package as the first chosen transfo.

EDIT 2 after manufacturing:

After assembling the board: H1102FNLT does not work but HX2260FNL does! I am looking into it.

\$\endgroup\$
  • \$\begingroup\$ I don't follow your reasoning for not using a proper RJ45 jack with internal magnetics. \$\endgroup\$ – Andy aka Apr 23 '15 at 7:27
  • \$\begingroup\$ I can't, we want to use this in the field using a rugged cable with a 4pins-M12 connector. (clrwtr.com/PDF/N-TRON/N-Tron-108M12.pdf this is a potential candidate as switch) And standard cables. I did not find M12 connectors with internal magnetics. And it does not feel good to make a five centimeter cable interface just to use that RJ45 jack. Besides, I am really wondering how it functions! \$\endgroup\$ – lode Apr 23 '15 at 18:27
1
\$\begingroup\$

The first coil (from the jack point of view) in the magjack is an autotransformer needed to reject common mode noise. The coil is followed by (surprise!) a common mode choke needed to reject common mode noise too. Don't be afraid, this is a two step rejection.

What you need to be worried about in your design is that you must connect HX2260FNL by its common mode choke enabled side to the line, not to the PHY. And if you will not use Auto-Crossover (Auto-MDI/MDI-X) feature, try to find an asymmetric transformer.

This memo by Pulse Eng. (the vendor of your transformer) explains well the purpose and operating principles of both CMC and AT.

\$\endgroup\$
  • \$\begingroup\$ ow yes! I see. I found the H1102FNLT transformer, thank god in the same package. I wonder why the other type exists though. I don't get the asymmetric transformer remark. Would it be better ? \$\endgroup\$ – lode Oct 14 '15 at 19:41
  • \$\begingroup\$ @lode It will be better if you will not use Auto-MDI/MDI-X. In a Auto-MDI/MDI-X xformer both sections are equal ("symmetric"): an center-tapped transformer at the PHY side then a CMC (and then AT if any) at the RJ45 side, as you can see in HX2260FN. In a non-Auto-MDI/MDI-X xformer ("asymmetric") the transmit section has a center tapped transformer at the PHY side and then a CMC [or CMC+AT] at the jack side while the receive section is opposite: it has a center tapped transformer at the jack side and then CMC at the PHY side. \$\endgroup\$ – asndre Oct 16 '15 at 16:25

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.