Best way to modulate serial data over DC power line

[Gandalf_Sr]

I want to have 2 PICs communicating with each other over a DC power line that's running at 12V and maybe carrying 10A. It will be driving motors so there will me some noise. Max distance is probably around 50m. The attachment shows the basic approach but I'm not sure how best to approach putting the data onto the power line.

What I've read so far suggests that a frequency of 1 MHz is fairly easy to filter on/off the power line but the data rate of I2C, SPI isn't that high, it would vary, and I can't just push serial data onto the power line, or can I?

If I need some sort of modulation, how about PSK or something?

Anybody like to chip in with suggestions?

Other Info
The data will have to be bi-directional, there will only ever be 1 master and 1 slave - one on each end, there will only be 1 signal wire - I'm thinking of sending the data and power over coax with the +ve and signal on the inner conductor.
Amount of data is fairly low so I am prepared to add a lot of parity, multiple transmission and acknowledgment etc. to get reliable comms.

 

[betwixt]

Carrying 10A over co-ax sounds a bit risky, the core wire is well insulated and can't dissipate heat easily, you might get a melt-down unless it's really heavy grade cable.

A better solution if it can be done is to use the co-ax to carry the signal and use a wire to carry the power, you can use the co-ax screen as the return for both.

I have something similar planned here but nothing actually built yet. I need to control 15 room lights so each can be switched on, off or dimmed, from a central box. My house has LED lights which run off a 12V supply. My idea is to interrupt the supply momentarily with serial data, probably at 1200 bauds. Each receiver will extract its supply from the incoming 12V but have a local reservoir to keep the power up while the data is being sent and in turn will PWM control the LED current. Each receiver will respond to an individual address sent in the data.

The only problems I forsee in adapting this idea to your system would be the difficulty in working bi-directionally and that 10A is a lot of current to pulse quickly.

Brian.

 

[Gandalf_Sr]

The coax I'm thinking of using is easily able to carry 10A; at 12 V DC that's only 120W. It has the advantage that the DC power noise and the signal on the coax center will be screened by the coax shield.

Using one of the protocols supported by the PIC family will allow bi-directional data to be passed.

If your house has a constant load from the LED lights then data over the power simultaneously will probably be fairly straightforward.

Original question still stands, suggestions for how to modulate data onto DC line still welcomed.

Ted

 

[FvM]

You already considered modulation. That's clearly the regular way for any kind of powerline communication. You have two options:
- use one of the existing powerline modem chips respectively chip sets, available from various vendors
- design a basic bidirectional modem on your own, e.g. using FSK

 

[betwixt]

You haven't told us the data rate you want to use. If it is low, you could probably use audio tones with either DTMF or your own 'home brew' protocol. They are fairly easy to produce and detect in software and you could couple to the power line with a small signal isolating transformer and series capacitor across the supply line. Alternatively, you could emulate an IR protocol such as RC5 and pulse it's carrier across the lines.

If you needs something faster, I would be inclined to use a modulated RF carrier but at a higher frequency than 1MHz, possibly at 433MHz where inexpensive transceiver modules are plentiful. The isolating filter would be much smaller and all the data handshaking is taken care of for you. Basically, you have a 'data in' and a 'data out' pin at each end.

Brian.

 

[FvM]

If you take a look at existing powerline communication standards, e.g. X10, LonWorks, EIB/KNX, BACnet, you'll find a rich choice of ready to use solutions, including single-chip modems, media couplers and complete modules.

Designing a new method from the scratch or applying a possibly inappropriate technique from another field (some suggestion of this kind have been already made, in my opinion) brings up a certain chance of re-inventing the wheel or (depending on your skills) even re-inventing the square wheel.