Comms over DC power cable

[dougy83]

Hi all,

Does anyone know of a low part count and cheap solution for bidirectional (half-duplex) communications on the 48VDC supply line? Baud rate of 19200bps would be nice.

I did a quick test with an LM567 @300kHz, but it is slow and unreliable. Also, it requires a very high SNR of -6dB.

Device is powered on a pair of wires, without being locally grounded, and can be connected in parallel with others, with differing cables lengths. All devices must be able to communicate with a master device, which is presumably using the same comms hardware as the satellite devices. Cables could be 100m long, and could have a fair bit of noise on them. They can also have up to 4A of current flowing through them, which feeds into an SMPS, so large valued decoupling inductors would not be suitable. Cable is similar to 2-core mains power cable.

Any ideas of simple to use devices?

 

[canadaelk]

Search "power line communication". E

 

[aliarifat794]

You can check STMicroelectronics ST7540 Power Line Modem. The ST7540 is a Half Duplex synchronous/asynchronous FSK Modem designed for power line communication network applications. It operates from a single supply voltage and integrates a line driver and two linear regulators for 5V and 3.3V. The device operation is controlled by means of an internal register, programmable through the synchronous serial interface. Additional functions as watchdog, clock output, output voltage and current control, preamble detection, time-out and band in use are included. Realized in Multipower BCD5 technology that allows to integrate DMOS, Bipolar and CMOS structures in the same chip.

https://www.st.com/en/interfaces-and-transceivers/st7540.html

 

[ronsimpson]

Years ago I did it on the power line at 120khz. (120 volts) Very slow communication.
I know companies that do power line communication in the 7 to 20mhz range.
I have done 48V, at 100khz without a receiver IC.
I use inductors to increase the power line impedance at the communication frequency. If the power wires look like 100 ohms at 100khz then it does not take much energy to stand 2 volts on top of the 48 volts.

 

[dougy83]

Thanks aliarifat794. I've seen that IC, but it's quite expensive, and not too easy to source.

Thanks Ron. I was expecting more noise on my supply cable, but the devices are actually quite clean in that regard, so it might be fine without a specialised IC. What scheme were you using that slowed it down? Was it only transmitting at the zero-crossings (like X10)? From the signal I can see here, it looks like some filtering and envelope detection might be all that's needed (though I'm sure there'll be more noise when the devices are in their final places).

 

[ronsimpson]

If you are on a 60hz power line you need zero crossing. That slows thing down. On a DC power line, there is no zero crossing.
The carrier frequency probably should be in one of the dead spots in the RF frequency band. 72khz, 120khz, 455khz 10.7mhz. (memory is bad)
You might look at the X10 ICs again.
ST micro Here is something to read. Data sheet for power line communication. 60hz, 70khz and 120khz Might give you some ideas.
IEEE has approved a low-frequency (≤ 500 kHz) standard called IEEE 1901.2
Within homes, the IEEE 1901 standards specify how, globally, existing AC wires should be employed for data purposes.
You should not use AM radio or FM radio frequencies. Stay away from 2ghz. Look at FCC band allocations.

Last time I used 48V power & com, on the receiving side we had a LC filter tuned at 120khz.
You might think about the switching power supplies in the project. I used 1 mhz PWMs to keep away from the 120khz. Do not make power noise near the band you are using.
The old X10 is AM and FM is better with noise but harder to make the receiver.

Baud rate of 19200bps would be nice.

Bits are coming at 20khz. Each bit could have 5 cycles of 100khz or 10 cycles of 200khz.

 

[Tony Stewart]

Telephone Modems work full duplex on DC circuits using DSP's and hybrid transformers with high-compression quadrature phase modulation. It depends on how much DC power you need, affecting the network impedance and voltage. Since FCC limits restricted 64 kbps to 56k, this would no longer apply. This could also be scaled up to any bit rate with the carrier f but much work.

 

[ronsimpson]

Years ago, there was many radios that used 120khz IF parts. Now days those parts are very hard to find. If you choose 455khz there are many ceramic resonators available and many band pass filters.

I transmitted with volts on the power line but as I rethink this, it is possible to transmit with mVs and use a radio IC to pick up low level signals. See below.

TA7642a Here is a link to a very small AM radio receiver. It works with very small signal levels and boost that up to 15mV output. This IC would work well at 455khz. Just a thought.
RonS.

 

[dougy83]

Thanks again. I have simulated the simple differential AM transmission and reception, using a couple of LC filters, amplifier, a peak detector and bit-slicer comparator. Seems to work alright across a wide enough range of input level and temperature.

A 455kHz resonator should be able to be dropped in place of the LC bandpass filters when they arrive to see how well they perform. Cheers