0
\$\begingroup\$

We have a device with two RJ45 ports. One of these ports is for receiving 24vdc on pins 7(-) and 8(+) to power the device. The other RJ45 port is an Ethernet connection. The ports are of course labelled to advise against such a thing, but is there guaranteed damage if someone accidentally plugs the RJ45 power cord into the Ethernet connection. Or, would it depend entirely on how robust the implementation of the NIC card is?

This strange situation is due to this being an engineered system of several different products from different companies.

\$\endgroup\$
  • 1
    \$\begingroup\$ This is unanswerable without the Ethernet pin assignment in use, typically only gigabit or atypical schemes use those pins. It would also be good to know the current limit of the supply. \$\endgroup\$ – Chris Stratton Mar 7 at 17:51
1
\$\begingroup\$

Pretty sure that 24V connected to pins expecting 3V or 5V is going to damage something. Starting with the magnetics in the ethernet connector.

\$\endgroup\$
  • \$\begingroup\$ Unless it's gigabit there may well be no connection to those pins \$\endgroup\$ – Chris Stratton Mar 7 at 17:50
  • 1
    \$\begingroup\$ What makes you think the NIC is expecting 3V or 5V? \$\endgroup\$ – marcelm Mar 7 at 20:10
  • \$\begingroup\$ I read the question a POE type cable might be plugged into an RJ45 connector that may not be set up to handle 24V on other pins. I could have have mis-read that. \$\endgroup\$ – CrossRoads Mar 7 at 20:45
1
\$\begingroup\$

If it is Gigabit or at least uses a connector including magnetics for Gigabit (which may be the cheapest option), pin 7/8 is a bad idea, because these form a pair.

It is completely normal to have 48–56V between different pairs for Power over Ethernet, but pairs are 1-2, 3-6, 4-5 and 7-8, with PoE using either 3-6/4-5 or 1-2/7-8. The resistance between wires in a pair is usually rather low.

Your options to make this safe:

  1. a soft-start power supply that tests the resistance between the outputs, and turns off if it is too low.

  2. a custom board with two RJ45 plugs in the right distance as edge connectors. This way, you get a double connector that cannot be inserted the wrong way.

2a. a custom board with a PoE extractor and 24V regulator. Same, and eliminates the need to ever connect a power supply there.

\$\endgroup\$
1
\$\begingroup\$

Yes. Using pins 7 & 8 in this way (pair 4 of an ethernet connector) would almost surely guarantee damage to the Ethernet magnetics and associated traces. Unless the device only works at 10/100 BaseT speeds (which does not use, need, or require this pair) there is very little a designer can do to avoid such damage and still remain within Ethernet specifications.

In the unlikely case that the designer chose to add TVS diodes on the field side of the magnetics, which is generally not recommended as this counterintuitively would tend to reduce ESD reliability and degrades signal integrity, you would destroy these diodes together with the magnetics.

Of course that depends on how much current the power supply can provide, but for the sturdier PoE magnetics, a maximum of 350mA for common-mode current in a pair is the maximum specification but the differential mode is normally specified at less than 10mA. Given the <2Ω winding DC impedance, you could be talking as much as 12A going through the magnetics.

Using whole pairs instead (e.g., 7&8 for Vdd- and 4&5 for Vdd+) is closer to the normal way for PoE devices to transfer power (PoE 10/100 mode B) and, if the device is new enough, these common modes are isolated by capacitors in the RJ-45 magnetic jack or in components associate with discrete magnetics.

But the damage will mostly be relegated just to the RJ-45 jack, or to the jack, magnetics, and associated PCB traces--if using discrete magnetics. Any transients that make it to the other side of the magnetics are unlikely to transfer enough power to destroy the ESD-protection diodes in the PHY ICs.

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

Ethernet magnetics look like this:

enter image description here
Source: https://www.analog.com/media/en/technical-documentation/application-notes/EE-269.pdf

And like this for regular POE, the difference is the rectifier:

enter image description here
Source: POE external Magnetics Alternative to magnetic RJ45 jack ieee803af

The difference is in how the powered device is connected to pins 7\8 and 4\5. (Pins 1\2 and 3\6 have transformers on the end so if you send in the same voltage on both pins (like the 1/2 pair), nothing will happen because transformers block DC.)

Some devices have a DC to DC converter that aren't true POE devices, the ones that are POE, have a rectifier to avoid problems with polarity.

The other RJ45 port is an Ethernet connection. The ports are of course labelled to advise against such a thing, but is there guaranteed damage if someone accidentally plugs the RJ45 power cord into the Ethernet connection. Or, would it depend entirely on how robust the implementation of the NIC card is?

So it depends on what the implementation of the powered device is, if you send it the wrong voltage it could burn out the components if the polarity was wrong for your device. If it has a rectifier, then the rectifier should take care of the polarity.

\$\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.