I have stated working from home like a large portion of the population and I needed to put in another wireless access point on the other side of my house for better WiFi access to my new office work area and also has the benefit to extend the network out to the IoT switch on the grey water. I ran an Ethernet cable from the modem in the lounge to the eaves on the corner of the house, but didn’t want to run mains power to run the new access point.
100Mb use 2 pair for data and the simple PoE injectors use the 2 spare pairs for power. Whereas 1Gb use all 4 pairs for data meaning it is a bit more difficult. You do this by using a twisted data pair as a……description here
I pulled apart an old Telstra adsl modem for the Ethernet magnetics as they were centre tapped.
The issue with these transformers is that the centre tap is commoned between adjacent windings, so the pairs of the same voltage need to be on the same common.
Each of the 2 DIP modules had 4 channels so I could have an injector and an extractor. As the distance was not so far (25m) I figured I could get away with these old isolation transformers only rated for 100Mb and the attenuation wouldn’t matter. They had a pitch of 2mm and the veroboard I had in hand was 2.54, so I got the side of a milk carton to mount it on to give the future PoE injector some stability. I also added a RJ45 socket. The extractor I just soldered a halved patch cable directly to the pins for a quick test.
My first test of plugging these back-to-back into my network and did not work. I thought my theory might be flawed, but it turns out it was a faulty Ethernet patch cable I had cut up, so I used the other half then it worked perfectly.
I planned to use an old wireless router as the access point so I pulled it apart and checked out the magnetics. It had a center tap to extract the power and the tracks were easily accessible. This meant I didn’t have to have an external extractor.
I chose port 1 and not the routed WAN port as the modem is locked and I can’t access the settings to bridge it. I removed the 150R resistors coupling the centre taps to ground by a capacitor so they could be used for power. As a side note, having this value resistor also stops active injectors from suppling power as thy are looking for a higher value in the range of 15k.
I removed the resistors and soldered wires from the transformer side of the resistor pad to the dc input. Unfortunately the switched negative pin on the jack was grounded with the negative supply contact so if I ever power this from the jack it will also power up port 1. I could have put the positive supply on the other pole of the switch so it selected between PoE or jack power input, but I thought of that after soldering it together and thought it would be better to have the switch do as it said.
After mucking around with it a bit I pulled 2 of the resistor pads the wires were soldered to off the PCB, so I had to solder directly to the pins of the transformer, which was a much more solid connection.
The reason I am using all 4 pairs of wires for power is to reduce volt drop through the cable and also these transformers are not rated for a PoE load, so by using all 4 transformer windings I can halve the load on them to make this more reliable.
I want to increase the supply voltage to reduce the current further so I checked the input circuit of the router. It had a TPS54320 4.5- to 17-V Input, 3-A Synchronous Step Down SWIFT Converter (20v absolute max) along with a MP111 Dying Gasp Storage and Release Control IC 18v Max input (22v absolute). This meant I could have up to 17V input on the router without stressing the components. Capacitors were all rated 25V
I added a step up converter and a dc jack to plug a 12V power adaptor into the injector. I set the output voltage to 17.5V to live dangerously by 1/2 a volt.
I plugged it in and it worked. It also could power my PTZ Ethernet camera off its USB port