RF signalling over powerlines

[richdrich]

I've got an application where I want to have one or more receivers underwater, where signals are unlikely to be received, but there's a power connection.

I was thinking of using the Moteino in a configuration with, instead of an aerial, the transceivers connected through an LC blocking circuit to the (low voltage DC) power lines. Has anyone tried this?

[Felix]

richdrich,
Can you give more detail?
How deep underwater?
You mention there's a power connection - you mean the low voltage DC lines underwater? Any more details about that?
If I understand correctly, you plan to use one of the power lines as transmission (with LC in between)? But then don't you break that DC line?

[richdrich]

It'll be a metre or two underwater.

So in ascii art:


+Vpwr->--/\/\/\---------------------------------/\/\/\---+Vcc
                   |                      |
     AntA---||------                       --------||--AntB



Where -/\/\/\-- is an inductor (choke) and --||-- is a capacitor

The RF passes through the capacitors and is blocked by the inductors
The DC power conversely passes through the inductors and is blocked by the capacitors
The power line will carry DC with a small superimposed AC signal.

[Felix]

Ok makes total sense now. What I don't know for sure is how the RF is impacted by the offset of the DC voltage on those lines. You also mentioned there's a superimposed AC, but not sure if you meant that's the actual RF or a separate additional AC wave. If there's additional AC my guess is the RF will not see a "constant" impedance, but i might be totally wrong about that.
It's an interesting problem of signal integrity.
I think the easy way to test this is to apply the RF with and without the other DC/AC signals on the power lines. If there's a significant loss in reception with the DC/AC present (or no reception at all) then you know you can't share the lines.
Also worth mentioning that the line better be close to 50ohms (ie coaxial). Which means you have to ground the Moteinos to the jacket of the coaxial so you could get that impedance.

[TomWS]

I have to agree with Felix re 50 ohm impedance (at the antenna connector, not the power connector) with one additional caveat, the power line needs to be shielded with the rf signal completely contained within the shield.  Otherwise the water surrounding the cable will produce a heavy load on the RF fields within the cable. 

If all you're powering is a moteino and a few low power sensors (and cable length is not too outlandish), this might actually work.   RG178 is reasonably flexible and would be able to support Moteino level power requirements.  If this is a long term thing (underwater for days or weeks),  however, then you'll need underground/underwater RF cable which will be a LOT stiffer.

You'll probably have to play around with the AGC controls to prevent overpowering and distorting the signal. 

Tom

[richdrich]

not sure if you meant that's the actual RF or a separate additional AC wave

It's the actual RF.

the power line needs to be shielded with the rf signal completely contained within the shield

Ah, yes. I hadn't thought of that.

[TomWS]

[Not sure if this should go here or in projects]

I've got an application where I want to have one or more receivers underwater, where signals are unlikely to be received, but there's a power connection.

I was thinking of using the Moteino in a configuration with, instead of an aerial, the transceivers connected through an LC blocking circuit to the (low voltage DC) power lines. Has anyone tried this?

There are a couple of key factors that will influence your implementation:
1. What is the instantaneous data rate required?
2. How much data do you need to transfer over time? (minute, hour, day)
3. How much power do you anticipate requiring on the circuit that is underwater?  (peak and average)
4. Must you only have 2 wires?  Three or four is not acceptable?

Tom

[richdrich]

There are a couple of key factors that will influence your implementation:
1. What is the instantaneous data rate required?

Low: Say 100bps

2. How much data do you need to transfer over time? (minute, hour, day)

Not sure: maybe a 10 byte command every minute

3. How much power do you anticipate requiring on the circuit that is underwater?  (peak and average)

Of the order of maybe 20W average, 100W peak. It's an LED lighting array. And either 5 or 12/24 volt

4. Must you only have 2 wires?  Three or four is not acceptable?

The idea of using 2 wire is simply to reduce complexity and cost. 3 wire is feasible, and I'd just use RS232. I did look at baseband modulation of the power line (low baud rate, signalling by polarity swap), but that would involve big mosfets in the driver.

i did wonder about one wire with a ground return through the water. It'd need a very good ground, though (I'm inspired by underwater HVDC cables, like this one https://en.wikipedia.org/wiki/HVDC_Inter-Island - 2500 amps through the water and apparently no smell of chlorine!).

[TomWS]

Of the order of maybe 20W average, 100W peak. It's an LED lighting array. And either 5 or 12/24 volt

This might be a challenge if you want to do RF over power line.  You'll need a choke that can handle 8-9Amps and still operate as an effective choke at 400+Mhz...  I'd say trying to do that AND match for 50ohms is going to require some engineering.  The good news is that the radios have a pretty large dynamic range and might be forgiving enough to handle this application.

The idea of using 2 wire is simply to reduce complexity and cost. 3 wire is feasible, and I'd just use RS232.

This seems the least complex...  One Wire Protocol is another option, if you need bidirectional signalling.  There might be some One Wire GPIO expanders.  I looked for a one wire expander and didn't find anything that seemed more useful than RS232.  If you need bidirectional over a single wire, it would be relatively simple to set up a half duplex protocol that would work.

I did look at baseband modulation of the power line (low baud rate, signalling by polarity swap), but that would involve big mosfets in the driver.

Yeah, that seems like a lot of extra HW to do what you want to do.  However, if you could turn OFF the load while signalling, then a relatively small MOSFET could be used at both ends to isolate the load from the signal wire and you wouldn't need humongous drivers or a complex network for the signalling.  A small battery or supercap could keep your Mote alive during signalling.  In this case you wouldn't even need a radio and you could get away with 2 wires.  It sounds like you could easily complete the transfer in just a few milliseconds so you may not even notice that the LEDs are powered off... 

Tom

[richdrich]

My thought was that the TX would be a Mosfet bridge and the RX a bridge rectifier with a capacitor to remove the switching spikes, so that even during signalling, there's always power available. Voltage drop on the bridge would be an issue unless I did distributed regulation - e.g. have a regulator at the receiver end.

I might do some tests with an underwater wire though just to see...