0
\$\begingroup\$

I am primarily software. I do debug and test a lot of hardware, so I know enough to be dangerous.

Twenty years ago we built Panasonic-compatible analog-signalling PCBs for controlling Panasonic's line of "hybrid" telephones.

What I remember was that it provided power and signalling over a single pair, much line a telephone line itself if you consider the audio to be the signalling.

I would like to reproduce this configuration with the following specifications: the signalling will be a continuous 50% duty-cycle clock with a possible range of 5 to 20 MHz, the DC supply would be in the range of 9V to 24V, with a maximum of 600ma at 5V needing to be produced at the far end (using a SimpleSwitcher). This would be using a twisted-pair ethernet-style cable.

I was thinking there are probably 10 different ways to accomplish this, and numerous what-ifs are swirling around in my head. But since I've little experience in the analog possibilities, I thought I might ask some experts.

Any suggestions would be appreciated. My main goals are: cheap, simple, hand-solderable, and minimum PCB real estate.

Thanks in advance.

\$\endgroup\$
8
  • 1
    \$\begingroup\$ Why reinvent the wheel? Can't you just use 1wire? \$\endgroup\$
    – Hearth
    May 7 '20 at 14:43
  • \$\begingroup\$ Not sure how well the Dallas 1-wire spec would perform at 50 feet. Do you know? \$\endgroup\$
    – GaryLa
    May 7 '20 at 14:47
  • 1
    \$\begingroup\$ It's called Phantom power and, for your application needs filters at both ends. \$\endgroup\$
    – Andy aka
    May 7 '20 at 14:48
  • \$\begingroup\$ I'm not sure how well it'd work either, but it's reportedly a longer-range protocol than I²C. You could also consider a 4-20mA current loop, but that wouldn't be able to provide your 600mA current requirement. \$\endgroup\$
    – Hearth
    May 7 '20 at 14:49
  • 1
    \$\begingroup\$ Look into Power Over Ethernet (PoE). It has been done 10 different ways and if you are using twisted pair anyway, the mass produced nature of PoE devices and Ethernet cables will likely make this cheaper than anything else. \$\endgroup\$
    – Alphy13
    May 7 '20 at 16:08
0
\$\begingroup\$

Phantom power and data: -

schematic

simulate this circuit – Schematic created using CircuitLab

There is some detail missing such as line terminations in case the signalling frequency causes line-reflections.

EDIT showing a balanced drive

enter image description here

\$\endgroup\$
8
  • \$\begingroup\$ This is the sort of starting point I was hoping for. I thought it might be fairly simple. I guess I can determine termination resistance requirements by looking at the signal after 20 feet or so of cable. I guess one of the negative consequences of this is the signal cannot be differential. I thought the original telephone concept had that benefit. Was this due to the use of a special transformer? \$\endgroup\$
    – GaryLa
    May 7 '20 at 15:41
  • \$\begingroup\$ @GaryLa You can certainly make this method differential (just duplicate the filters on the negative rail too). The transformers are for isolation more than anything else in telephone and ethernet systems. \$\endgroup\$
    – Hearth
    May 7 '20 at 15:49
  • \$\begingroup\$ Would a "Mode A" Power-Over-Ethernet modular jack provide a similar effect? \$\endgroup\$
    – GaryLa
    May 7 '20 at 15:49
  • \$\begingroup\$ There is nothing much in the above circuit that needs to be added other than inductors to make it differential. \$\endgroup\$
    – Andy aka
    May 7 '20 at 15:49
  • \$\begingroup\$ Ethernet - @GaryLa this question is showing signs of moving the goal posts. \$\endgroup\$
    – Andy aka
    May 7 '20 at 15:50
0
\$\begingroup\$

You could always do something like the following block diagram:

schematic

simulate this circuit – Schematic created using CircuitLab

I'll leave the details on working it out to you, but simply putting filters where appropriate and adding your clock to the power line should do you fine.

If your circuitry on the receiving end is sufficiently tolerant of input voltage fluctuations, you could even leave off the low-pass filter for the power entirely, though it would increase noise in the circuit and I don't recommend it.


As the asker has said they've already worked out this much, I'll give a bit more information.

On the receiving side, you should use an LC input low-pass filter before feeding the voltage into your regulator. I say LC because that gives you lower losses; RC will work as well but provide a harder current limit and will waste some power. An RC filter would be my choice for the high-pass filter, however, since you don't need significant power going through that line.

On the sending end, adding the clock on top of the supply voltage can be done with a simple LC filter like this:

schematic

simulate this circuit

There are, needless to say, better ways of doing this, but I think you'd have a hard time finding a much cheaper or easier way of doing it short of using a resistor instead of an inductor. Again, the choice to use an inductor is to limit power loss. Putting resistors in series with your power source is usually not the best idea.

\$\endgroup\$
4
  • \$\begingroup\$ I certainly appreciate your effort, but I already have this much figured out. :) But the details are exactly what I was hoping to get some suggestions on. \$\endgroup\$
    – GaryLa
    May 7 '20 at 15:10
  • \$\begingroup\$ @GaryLa I didn't realize you'd already gotten that far. I've added more to the answer, which I think should be enough to make this system work, short of any problems with the clock signal being able to drive the line (you may need an amplifier on it). \$\endgroup\$
    – Hearth
    May 7 '20 at 15:20
  • \$\begingroup\$ Is it possible it's this simple? That would be great. \$\endgroup\$
    – GaryLa
    May 7 '20 at 15:42
  • \$\begingroup\$ @GaryLa It is indeed this simple! This isn't the best way to do it, I'm sure, but it's a way to do it that will work. Data integrity will probably be minimally affected, your transmission line itself will be more of a problem than the filters. Of course you'll have to spec your inductors for the full DC current through them that the supply will be providing, and you'll want to avoid resonances, but those are just details. \$\endgroup\$
    – Hearth
    May 7 '20 at 15:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.