2
\$\begingroup\$

I'm trying to measure the power consumption of a PoE device using Mode A.

I assume I would need 2x current readings -- one for pin 1 (I1), one for pin 2 (I2), and then 1 voltage reading of from either pin 1/2 to either pin 3/6 (V)? Then my consumption would be $$P = (I_1+I_2)\times V$$

I'm currently splicing the wire in the middle and sticking 2x current reading devices in between to measure the current as well as the voltage as seen here: circuit diagram

I'm using 2x TI INA169's for my current readings and a simple voltage divider for my voltage reading. This is then being processed by an Arduino ADC for now but will move over to something more sophisticated in the future. I'm confident the INA169's work as intended as they work with some other (simpler) DC circuits I've tested. I just want to make sure I have the right idea for measuring PoE current before I start redesigning/simulating my INA169 circuits again.

\$\endgroup\$
2
  • \$\begingroup\$ Did you ever get this working? I'm hoping this can work, but currently unclear on whether connecting a typical multimeter in current mode attached on lines 1 and 2 would completely destroy the signaling or not. I will conduct a test soon and hopefully it will work, or degrade to 10 or 100 Mbps and still work well enough for me to make power measurements. \$\endgroup\$
    – Steven Lu
    Mar 3, 2022 at 23:48
  • \$\begingroup\$ I think since it is impossible not to introduce a huge loop in the twisted pair, a more reasonable approach is to put suitably valued, say, 1Ohm resistors where you've got the ammeters there, minimizing disruption to the twists, and measure the voltage drop for deriving current. \$\endgroup\$
    – Steven Lu
    Mar 4, 2022 at 0:11

1 Answer 1

1
\$\begingroup\$

This technique works.

As I mentioned in my comments, I think that it is best to try to reduce the introduction of loops to the twisted pair wires.

enter image description here

Here I am using a 1 Ohm and 1.2 Ohm resistor spliced into pins 1 and 2 of my PoE cable, which is a Cat 5 cable. I will use a constant current power source later on to accurately measure their actual resistances, but keep in mind these type of resistors are subject to variance under temperature change.

The voltage across the resistors combined with the resistance value will yield current measurements, which can be summed up and multiplied by the line voltage to determine total system power consumption.

I did notice that after setting this up the signaling rate has dropped from 1Gbit down to 100Mbit (probably a higher quality cable could maintain 1Gbit, but I wanted to sacrifice a crappy cable this time), but power is delivered to the device just fine and the network connection remains functional so signals are traveling through fine.

I will make a measurement of the overall PoE voltage later, but I don't need a super precise measurement of power, I can cross reference my results against the (very slow-refreshing but seemingly precise?) value reported from my Ubiquiti PoE switch.

In the specs of this switch it mentions that pins 1 and 2 are "power +" and this is consistent with the "Mode A" power delivery. As far as I understand it, since Gigabit is supported this means that all 4 twisted pairs are being used for signaling. Given that I'm getting current through both even when the signaling drops to 100Mbit, and it is in the right ballpark:

(63mV/1Ohm + 70mV/1.2Ohm)*48V = 5.8W

The switch reports 6.5W.

\$\endgroup\$
1
  • \$\begingroup\$ Appreciate the detailed and thorough reply, this seems to be exactly what I needed. \$\endgroup\$
    – fullyjosh
    Mar 15, 2022 at 20:53

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.