# Recommened chips for datacommunication over a low-voltage AC power line?

A client wants to send data (think 100 ... 10k baud) over a combined power / cable-TV coax cable, distances up to 1 km. The lower end of the spectrum (say up to a few MHz) is free, but the cable carries the 50 Hz 48 .. 60 Volt power. What would you recommend for this? I was thinking power-line modem chips, but those seem to be out of fashion (and hence difficult to get), and require quite a few external components. A DIY approach could be to couple a modulated carrier into the line and detect it with a PLL chip or even a PIC with DFT. But I hope chips exist for exactly this purpose?

-
 How about higher frequency ranges? It would be helpful to know more accurate figures on what frequency ranges are in use. – Majenko Jul 5 '11 at 17:58 From my sketchy notes: 5..65 MHz occupied, 85..1000 MHz occupied. It is a cable TV system. – Wouter van Ooijen Jul 5 '11 at 21:00

Have you considered looking at the other end of the spectrum? Low-frequency communications, as you have rightly noticed, are not as common as they used to be.

There are many chips around that handle all the modulation, demodulation, etc at >1GHz - take the MRF24J40MC from Microchip (maybe not the cheapest solution) for example - 2.405Ghz to 2.475GHz. Or the ADF7241 from Analog Devices - again in the 2.4GHz range. There are many many more.

Yes, I know these are aimed at wireless communications, but then, what is Cable TV if not wireless communication with no aerial? The cable is basically a closed loop aerial between two points. A modulated high frequency signal being sent between two points through a medium - just using copper instead of air.

-
Most cable TV hardware is specified to 1 or 2 GHz, so you'll have to make sure you don't have 1 GHz gear in the pipeline. The more you go past 2 GHz, the more attenuation you'll have. – Mike DeSimone Jul 6 '11 at 12:13
My question obviously has no single 'correct' answer, but this answer was the most usefull to me. – Wouter van Ooijen Aug 24 '11 at 18:01

Have you considered just connecting some v.90 modems? The 50 to 60 Hz is below the audio range those modems use, so can be filtered. The television VHF and UHF channels likewise are outside the modem frequency range. The modems are 600 ohm impedance and signal in the 300 Hz to 4000 Hz range.

Telephone wiring (600 ohm) to CATV wiring (75 ohm) will require a suitable coupling transformer as well as a power filter that blocks the power frequency.

Telephone ring detect and onhook/offhook features won't work, of course, you'll need to give the modems commands to force connection.

-

Maxim has a good range:

http://www.maxim-ic.com/products/powerline/

The MAX2990/2991 combo looks good. There's an evaluation kit available too.

-
Hmmm, the customer was selecting chips (for the other parts of the system) and mumbled something about availability. My immediate resonse was "stay away from maxim/dallas!". Quick www.findchips.com check: no stock at DigiKey (bad sign), low quantities at Mouser, total price > $20 @ 1. I hoped for something much cheaper (and smaller)! – Wouter van Ooijen Jul 5 '11 at 20:57 I think you'll have to do it manually. One approach is to use a zero-crossing detector and transmit a packet while the mains voltage is close enough to zero. Then you can use regular modulation chips and the only part you'd have to implement is the gating. – Optimal Cynic Jul 5 '11 at 21:05 Can use suggest some regular modem chips, circuits, app notes? Speed can be low, no isolation needed, price/size/availability is important. – Wouter van Ooijen Jul 5 '11 at 21:30 All the ones I'm familiar with seem to have vanished into the murky depths of RoHS. Sorry, you may have to implement this manually after all. One thing that I'd forgotten between first and second answer is that you're driving a thousand metre coax cable - that's going to be tough to do on the zero crossings. I wonder if you could do something clever with inductive coupling and simple baseband signalling? – Optimal Cynic Jul 5 '11 at 21:43 One more question - is this unidirectional or bidirectional? – Optimal Cynic Jul 5 '11 at 21:43 show 2 more comments Since you mentioned ASK in a comment, the NXP TDA5051 is an ASK modem designed for home power networks, and operates at either 600 or 1200 baud, which is in between your 100 .. 10K baud range. It is RoHS compliant, and is available at Digi-Key for under$3 in quantity. The interface circuitry required seems to be fairly straightforward.

-

You've got a megahertz of bandwidth available and you only need 10 kbits/s of data? You could just use a series capacitor or RLC circuit (basically, a bandpass filter) to couple transmitted data onto the line and a 200 Hz high-pass filter/1 MHz low-pass filter combo to filter the received signal, then do a software modem with a 100 ksample/s or so ADC and DAC. Higher-end (96 or 192 ksample/s) audio codecs might do the job nicely. You could do it with a dedicated low-power 32-bit chip or similar, which might even have good enough ADC and DAC on board. If it has PWM outputs instead of a DAC, then just use them to drive a push-pull stage to get a class-D transmitter.

Assuming you have the people needed to write a modem, of course. But, again, the excess of bandwidth means it doesn't have to be a particularly good one, just one that can deal with the cable irregularities. ^_-

-
 Yeah, there is plenty of bandwith. But I hoped for an (or a few) cheap chips (like an old-fashioned FSM or ASK modem, but in one chip, few or no externals, cheap, and maybe the modst difficult: available), not a 32-bitter that must be programmed for modem duties! Maybe I hoped for too much. – Wouter van Ooijen Jul 6 '11 at 6:49 It really depends on what resources you have available. If you have no software developers, this won't work for you. But usually debugging software is easier than hardware since you don't have to get the board revised. FSK and ASK algorithms aren't that hard to implement. – Mike DeSimone Jul 6 '11 at 12:12