I am designing medical data acquisition equipment that requires moderately fast data transfer. As part of the design we are using a custom 100BASE-T PoE architecture (and cabling) to the data acquisition headstage, together with an intermediate "dumb" device to (1) provide medical-level isolation, (2) inject PoE power, (3) connect ethernet to the computer, (4) connect additional non-isolated equipment (via two extra ethernet pairs from the headstage, that's why we are limiting to 100BASE-T).

The only medical-grade isolated magnetics I can find are the HXU6200NL by Pulse, which will serve to connect to the computer and will reside inside the dumb device isolating the headstage itself.

The question is: Would it be ok to use two sets of ethernet magnetics back-to-back inside the dumb device?

That is, to use a second set of PoE magnetics to inject power and connect to the headstage with no intermediate circuitry (besides, perhaps, some common mode terminations for both cables).


  • There will be a standard set of magnetics in the headstage bringing the total count in the link to 4 (the absolute minimum to provide the needed isolation would be 3)

  • The total cabling length (headstage-box-computer) would be less than 10m, very likely less than 5m.

  • One reason to want to use a second set of magnetics is to provide "standard" common-mode termination (i.e., Bob Smith termination) to each cable segment.

  • Another reason is that ethernet magnetics are not symmetrical, as common-mode chokes and autotransformers are commonly present.

  • \$\begingroup\$ Welcome to EE.SE. You want to have the magnetics in series to double the breakdown voltage of normally 500VAC? The signal source needs to be close by even for 100BASE-T. \$\endgroup\$
    – user105652
    Oct 24, 2018 at 22:49
  • \$\begingroup\$ Rather than get a pre-certified magnetics, consider testing off-the-shelf products. Medical requirements might be easily met by many designs. \$\endgroup\$
    – Whit3rd
    Oct 24, 2018 at 22:58
  • \$\begingroup\$ The intermediate magnetics are to (1) inject power and (2) provide medical isolation. At least one of the sets of magnetics need to be "medically qualified" so as to provide the needed breakdown voltage and air gap requirements. Standard Ethernet magnetics are very far from meeting these. \$\endgroup\$ Oct 25, 2018 at 13:46
  • \$\begingroup\$ What standard are you trying to meet. Can't you just isolate the patient side and let all the communications happen on a non-isolated portion? \$\endgroup\$ Oct 25, 2018 at 15:33
  • \$\begingroup\$ Medical standard IEC60601 parts 1-11. Device in contact with the patient and possible use inside the operating room. And no. The amount of data being transferred from the patient requires 25Mb/s at a minimum. That is why we are using 100BASE-T. You somehow have to transfer this data between the isolated and non-isolated portions. \$\endgroup\$ Oct 25, 2018 at 18:12

2 Answers 2


You should be able to get away with two sets of magnetics, that is effectively what happens in POE midspan injectors (one transformer at each end plus one for the midspan injector).

  • \$\begingroup\$ Midspan injection magnetics seem to normally be just 1:1 transformers with access to the center tap, that is relatively simple devices. The more common ethernet magnetics incorporate at the very least a common-mode choke and sometimes an additional autotransformer. Although I think it should still work, that's quite a few extra components in a link. \$\endgroup\$ Oct 26, 2018 at 23:31
  • 1
    \$\begingroup\$ It will all come down to loss and capacitance parasitics. I have used standard POE magnetics for midspans in a couple of systems and experienced no problems with 50m cables (3W POE loads) for 10/100 links. I suspect that the losses associated with the common mode choke pair are small compared to the transformer. \$\endgroup\$ Oct 27, 2018 at 1:29

You do not need to invent bicycle - I am sure there're hub or ethernet switch devices available on the market certified for medical applications able to provide PoE to its connected devices. If you have custom connector in your design, then just provide appropriate cable with RJ-45 at one end and custom connector at another end.

I do not think connecting two magnetics back-to-back will work properly, or will work at all. Did not see anyone trying it, maybe because they simply do not see the rationale for it. Magnetics itself in your device will serve as decoupler, so you must ensure that connectors you use on your development are certified for medical application.

  • 1
    \$\begingroup\$ Being a passive ethernet pass-through is only one of the functions of this "box", for way too many reasons (including some listed in the question) I can't simply use a commercial PoE injector. \$\endgroup\$ Oct 25, 2018 at 13:40
  • \$\begingroup\$ Then the right question IMHO would be "How do I make passive ethernet switch with PoE". \$\endgroup\$
    – Anonymous
    Oct 25, 2018 at 14:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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