# Single capacitor for Powerline communication - AC coupling

First of all: I am not talking about traditional powerline communication (ethernet protocol etc) but about a very simple and primitive communication, like serial data transmission (https://en.wikipedia.org/wiki/Serial_communication)

I have the idea to use a simple high pass filter to "filter out" the AC voltage and just have the raw serial signal left. This idea sounds so simple to me that I can't immagine why such devices don't exist. At least I didn't find anything.

simulate this circuit – Schematic created using CircuitLab

On the left side you can see the transmitter which "connects" to the power line using a high pass filter. I did some calculations with those values and the mains AC signal should be filtered out very good. (Should be only around 100mV left so this should not fry the electronics).

On the right side there is the same setup with the same highpass filter.

I assume the communication to be a serial communication with a baud rate (frequency) of 115,200 Hz.

So with the circuit above, the AC power signal (50Hz) should be filtered out and the data signal (115 kHz) should be able to pass through. This sounds a bit too easy and to simple to me so I am sure I am missing something here. Please let me know what? Or am I missing nothing - should this work??

• You might have a look at the "X10" powerline communication system. When this first came out the modules were somewhat overpriced and it became popular to roll-your-own and interface them to computer of the day. That included things like the Altair 8080, Commodore KIM-1, and the Rockwell AIM-65. It takes a bit more than just a capacitor but not much more. There are plenty of "HOWTO" articles on making your own X10 interfaces. – jwh20 Mar 19 at 16:20
• A 1.4 pF capacitor has an impedance at 115200 Hz of 986 kohm so it's not really up to the job of putting any reasonable signal level on top of an AC line voltage that might have an impedance of a few tens of ohms at 100 kHz. – Andy aka Mar 19 at 16:47

Just to be exact some series impedance should be added to your AC voltage source to make possible the receiver see something else than the mains AC voltage.

It's also practical to have the series capacitor in both lines because in normal AC outlet it's not at all sure which of the 2 wires plugged in gets connected to the hot wire and which to the neutral.

I have built a 50kHz transistor oscillator which was able to output say 2 watts. It was amplitude modulated by swinging its operating voltage (=10VDC) with an audio signal which was got from an amp which had loudspeaker output transformer. The operating DC to the oscillator was fed through the secondary of that transformer.

The signal was well detectable and loud with a 50kHz crystal receiver in another building. The test was made in rural area. The proven working distance was about 50 meters. Some electric motors caused noise to the signal when they were ON, but I'm sure low speed data transmission would have been well possible even by pulse by pulse detection with no clever signal processing.

The LC resonant circuit in the crystal receiver was far more frequency selective filtering than our RC highpass filter idea could be.

BTW. The transmitter was bad. The harmonic components it produced made my signal well audible with an AM receiver at numerous frequencies.

Why they do not exist:

The allowed signal power levels are low. Attenuation in power lines and noise caused by electric devices restrict the communication distance. Wireless radio is more cost effective. I guess the nearest distribution transformer would stop the signal - it's designed for 50 or 60 Hz.

Your model schematic shows an ideal voltage source in the center, which will force the center of your schematic to always have exactly the voltage produced by source. In other words, the transmitter will not be able to change the voltage on the line in order to send any information.

You need a much better model of the power line.

What error rate would you like when a switch off creates a logic 1 instead of 0 which is because the noise spectrum goes thru your broad near baseband data with a sync pulse to restore DC.

This is why multiple consecutive narrow-band carriers (with a small guard band) are used to filter out noise using and a DSP to do the parallel math functions to extract the data. BTW this is how 56k modems and cable modems work too.

SNR improvement factors might also use more BW per bit or move it higher away from the glitch risetime with spread spectrum modulation.

The grid distribution impedance in the home also is taken into consideration with 1uH/m or so on cable so resonant filters must factor this when coupling the carrier to the line. Spectral testing of the grid with adaptive equalization filters improve the BER along with FEC.