I put the router in “Client Bridge Mode” and wirelessly connected it to an access point.
The following table shows the voltages measured when powering the router through PoE.

The following table shows the voltages measured when powering the router through the normal power jack.

*The iperf test creates network traffic between the PC behind the bridge and a PC on the other side of the access point.

Parts list

Radioshack Size M Panel Mount Coaxial DC Power Jack - $1.99
Leviton Quickport 2-Port Surface Mount Housing - $1.79 at Home Deport
2 X Leviton Quickport Snap-In Cat5e Jack - $4.95 each at Home Depot
Scrap Cat5e cable

Many of the discussions online about homebrew PoE adapters suggest starting off with a new AC adapter that is a higher voltage than the original AC adapter that comes with your network device. This is to compensate for voltage drop across a long Cat5 run and still deliver the necessary voltage to the router. I wanted to see if I could avoid using a new power supply. They are quite expensive after all. As I learned from this experiment, the Linksys WRT54Gx series of routers is quite capable of handling a voltage range in the neighborhood of 5 to 20 volts. I planned on testing the voltage as I added each piece to the injector. I needed a baseline to start with. I tested this AC adapter that comes with my WRT54GL.

The output voltage of the power supply with no load on it is over 17 volts.

At the DC input on the powered router, the voltage shows up at right about 15 volts.

I built the PoE injector using the TIA/EIA 568A standard.

Just for fun, I made the data input jack blue for “cold” and the data/power output jack red for “hot”.

The DC input jack compatible with the WRT power supply connector is size 5.5mm outer diameter and 2.1mm inner diameter. There is a jack with a 2.5mm inner diameter, but the center pin is too big for the plug to fit.

I got it all hooked up and tested the voltage coming from the end of a 7 foot manufactured Cat5e cable.

While this voltage is reading slightly higher than the first measurement of the unloaded power supply, it is in the same ballpark. I found that there were slight variances each time I measured.

I identified the positive and negative contact in the router’s board. I then soldered wires connecting them to the unused pairs on Port 4 of the router’s built in switch. This will allow the hot Cat5 cable to send power to the electrical bus of the router.

I hooked up the hot Cat5 cable to port 4 and measured the voltage again. It is essentially identical to the voltage using the original power jack. I will measure this again when I can try it with some very long Cat5.

It works!

I hooked the router up to a network and it works perfectly! It turns out that the OEM power supply provides well more than 12 volts to the DC input bus on the WRT. The voltage regulators inside are both an AP1509-33 and AP1509-50. The operating voltage of this model of voltage regulator is 4.5 to 22 volts. Even with voltage drop on a very long Cat5 run, the voltage should still be well within the range that these voltage regulators should still be able to provide the 3.3 and 5 volts that the router uses internally.