Need some feedback on techniques to make an old analog bus work with bidirectional data(half duplex is fine) at speeds upto 256kbps

The cable itself is a shielded twisted pair with a nominal impedance of about 50 Ohms. It's wired to various points in large buildings in anyway the installer felt was optimal (this includes daisy chaining, branching off into a star at a daisy-chained node etc). I believe it was used to pass audio and/or very low bit rate security data up to 500meters in length

Some of things I considered include:
1. A variant of 10Base2 at a lower data rate. The manchester encoding and csma/c(d/a) shouldn't be too hard to implement but the uncontrolled bus cabling worries me. At 256kbps(kHz) and some controlled slew rate the reflections and signal integrity shouldn't be too bad?
2. Restricting bandwidth to 40kHz or even 20kHz and increase bits/symbol by using QAM. This is not my preference because of the complexity of implementation unless if I can find a single chip SoC. Maybe there are bits and pieces of the old school v92 soft modem libraries I can re-purpose
3. Adapt Power line Communication techniques like G3-PLC

There is not much I can do to change the cabling as it's an existing installation. There won't be anything else on it, if it helps with reflections I can terminate it at each and every node. These cables are by and large interference free because of the shielding and differential signaling

Any thoughts/stories that you can share are most welcome. Thank you.

EDIT: The solution is not specific to one site, the wiring is different in each site and there about two dozen sites all up.

EDIT: There can be between 12 to 15 nodes on each bus

  • 1
    \$\begingroup\$ I think you need more information. Can you hook a 50ohm signal generator and drive e.g. a 250kHz square wave to see what kind of reflections you get? This cable might be lossy enough that the grody topology doesn’t really hurt. \$\endgroup\$ – Bob Jacobsen Mar 24 '18 at 5:03
  • \$\begingroup\$ I will be on site in 2 weeks so will be able to do more tests then but for now all I have is that the resistance is about 18 ohms for every 100 meter stretch of the cable. Reported by a field technician. \$\endgroup\$ – user183368 Mar 24 '18 at 5:33
  • 1
    \$\begingroup\$ 50 ohms is a very low characteristic impedance for twisted pair, closer to 100 ohms is more likely... \$\endgroup\$ – Henry Crun Mar 24 '18 at 6:25
  • \$\begingroup\$ OFDM is the magic bullet for squeezing high data rates through low symbol rates on a multipathy (stub reflections) medium. Whether or not you can leverage ADSL, or need to roll your own, is another matter. \$\endgroup\$ – Neil_UK Mar 24 '18 at 6:28
  • 1
    \$\begingroup\$ Sorry, I was being sarcastic about client's thinking \$\endgroup\$ – user183368 Mar 24 '18 at 17:53

WIth powerline modems it is possible to get 500Mbps, 300 meters in point to point but this requires equalization training for each point to a hub.

In P2P this would have to be trained for every combination of P2P and saved in the baseband modem which has to be done in parallel. So if anyone can be talking then you must have a low speed negotiating phase when idle.

300KHz with Er=4 has a full wave of 500m so using 10% of wavelength for transmission line low reflection bandwidth leaves 30kHz of low reflection bandwidth at 110 Ohms =/-20% est. at HF limits which could support 250kbd easily and possibly 500k with EQ using ODFM.

But PSK,bi-ph has the best resilience (Steepest curve) for BER at low SNR. Compression factor depends on content if already compressed or not. FEC is desireable.

Consider R&D budget cost vs COTS modems, I would be looking for HD-PLC with Ethernet to a hub and loop back to get any to any using point to point PLC ( power line carrier) technology but without the AC power using choke grounds and RF caps to suppress AM ingress.

In other words a bus/tree with MIMO Hub loopback to get P2P at max rate.

For low bit rates (TBD) simpler solutions exist...but assumption was up to 256kbd.

| improve this answer | |
  • \$\begingroup\$ Thanks for a very practical and considered response. I won't accept it as an answer just yet. Out of curiosity, what simpler solutions are out there for say sub-100kbps data rates? \$\endgroup\$ – user183368 Mar 24 '18 at 17:39
  • \$\begingroup\$ I am accepting this as the answer since this put me in the right direction for further research. Thanks Tony. \$\endgroup\$ – user183368 Mar 26 '18 at 5:00
  • \$\begingroup\$ upload.wikimedia.org/wikipedia/commons/0/07/… \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Mar 26 '18 at 6:23

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.