Implementing powerline communication on a battery-fed 12V line

Question

I was wondering if anybody had any thoughts on whether it is possible to implement powerline communication on a lead-acid battery fed-12V power network.

Characteristics:

• voltage: 12 V
• Fed by a battery-bank of 3 X 140 Ah, AGM lead-acid, non-spiral
• typical currents: 2-150 A
• wires: stranded copper 1.5 mm² to 50 mm², PVC isolation
• wires for powerline communication: 30 mm²
• typical wire lengths: upto 30 m

My worry is two-fold:

1. will I be able to drive up the voltage in the system to allow communication - or is the battery going to "absorb" it as "charging" current?
2. how "noisy" is such a network typically in the frequency range of 133kHz? I.e. how strongly am I likely to have to drive it to enable communication? (The system comprises several 12V DC motors, an alternator, chargers, etc.)

For implementaion I am looking at Cypress CY8CPLC10 and similar.

Note:

• Neither the wiring
• nor the DC voltage for power transfer can be changed.

A different way to understand this question is: "What are the ac characteristics of lead-acid batteries in the 100kHz range?"

Update 1: answer to Russel's questions

• required datarate: 500+ baud (I was looking at HART)
• yes, cable length is <30 m. The cabling cannot be touched as all cabling was put in when the structure was built. It is too restricted and convoluted to be able to put in any additional cables. IR is not possible due to no line of sight. RF is an option I am looking at but I was worried about interference with radio equipment on site. Though I realise this is a very real possibility with powerline as well...
• re filters: due to the high currents I was hoping to avoid putting anything in series with the the power carrying line and rely on the digital filters of the modem ICs
• similarly for coupling: at 150A, transformer coupling does not really seem to be a light-weight option. Do you think capacitive coupling is possible here?
• re feasibility of brute force modulation: how do you estimate that a few watts will be sufficient? I am worried about the battery just smoothing any modulation I try to put into the network. In DC conditions a SIGNIFICANT load is required to depress battery voltage by 1V. How easy it is to achieve about 200mV at 130kHz is the big question. I was hoping somebody would tell me: "don't worry about the battery: it's response is much slower than 10µs." :-)

Answer 1

You are right, the real question is what does the impedance of your 12 V power line look like at high frequencies.

I don't know, but this is something you can measure yourself. Try to feed a signal onto the 12 V line thru a resistor (with cap to block DC but large enough to not add significant impedance) and measure the attenuation.

133 kHz sounds very low to me. I would try around 1-2 MHz for starters. With a good filter, you can get by with very little signal at the receivers. After all, think of how little the signal is in the antenna of a AM radio.

It would probably help a lot if you can add even a little inductance to each connection of the 12 V line. At 150 A, that would be big and expensive to actually buy inductors, but maybe just wrapping the feed cable around a few loops would help. Three turns around a broomstick might make a difference. The reason I said broomstick is because then it will essentially be a air core inductor, so you don't have to worry about saturation current. 1 µH at 2 MHz has a impedance magnitude of 13 Ω. 1 µH is going to be hard to achieve with a few loops of cable, but it is easy to mentally work from there. 100 nH, which might be possible, will be 1.3 Ω

Answer 2

There is already a solution for UART over DC powerline at speeds up to 115Kbps. See Yamar's SIG60. It is really simple to interface and operates very well on noisy power lines.

Answer 3

Too long for comment, so:

What data rate do you need?
Do you really mean 30 metres wire length?
That is so small that running a parallel network or using RF or IR would be trivial unless there was some very very special reason not to.

You should be able to operate data transfer at 133 kHz in a very noisy environment as long as you had robust error correction and data rate could be adjusted to accommodate.

You would usually provide filters for the DC and signal. At 150 A that's annoying but doable. Bearing in mind that 0 dBm is about 0.6V and coms can operate at 10's of Db below that, brute force modulation of 133 kHz onto unfiletred DC circuitry should be doable at acceptable power levels. I've not done any caculations but I'd expect "a few Watts" should be ample.