1
\$\begingroup\$

Is there an industry rule of thumb or application note etc for when differential signalling becomes necessary for reliable communication?

I'm thinking mainly in terms of distance and signal speed (and voltage levels), assuming a "normal" level of EMI from the environment.

Edit: To clarify, when I say differential, I mean using two wires per signal, with each wire having the opposite signal level to the other (one wire goes high when the other goes low, and vice versa). This is for two or three level signals (i.e. signals that have only high/low or high/zero/low). Examples include USB, LVDS, CAN, DDR clocks, and most other high frequency/high noise immunity signals.

The emi environment is, as mentioned, just the emi environment that you would expect in a normal city environment, not directly next to any rail switching gear or electric motors.

\$\endgroup\$
5
  • \$\begingroup\$ You need to clarify what you mean by "differential". There is the USB type which has a center point well above zero volts, then there is LVDS which is very fast, up to GHZ speeds. Better cables and trace control and drivers are pushing speeds up to 10gbps. Sata is now up to 6gbps. The fastest of these have severe trace or cable length limits. \$\endgroup\$
    – user105652
    Commented Dec 10, 2018 at 4:14
  • \$\begingroup\$ Define the EMI environment to tolerate (healthy data eyes at the receiver, despite Electric and magnetic and common-mode trash). An electric train speed controller ---- 2,000 amps at 5,000 volts, switching in 1uS---- induces nearly 100 volts in what should be a GROUND PLANE. In that situation, the existing methods use FIBER OPTIC triggers to the IGBTs. Differentianl signaling is not robust enough. \$\endgroup\$
    – analogsystemsrf
    Commented Dec 10, 2018 at 4:29
  • \$\begingroup\$ I think you are getting confused between balanced signalling and transmitting a differential signal. Transmitting a differential signal reduces emissions and doesn't really contribute to making a reliable communication. It's balanced impedances that make a reliable communication and differential signalling doesn't really bring anything to the party other than lower external emissions. \$\endgroup\$
    – Andy aka
    Commented Dec 10, 2018 at 9:45
  • \$\begingroup\$ @Andyaka wikipedia says this is correct, but doesn't provide any sources, or really give much info about the benefits of balanced impedances for non-differential signals. Do you have a good source that can explain further? \$\endgroup\$
    – BeB00
    Commented Dec 10, 2018 at 9:55
  • \$\begingroup\$ @BeB00 Only my brain and other answers given on this site. \$\endgroup\$
    – Andy aka
    Commented Dec 10, 2018 at 9:58

1 Answer 1

Reset to default
2
\$\begingroup\$

In general, when using long cables, the grounds a both ends could be different. And with either long cables or high frequencies, the signals are more susceptible to noise.

In practice, the limits depend on the actual noise levels.

TI's application note Comparing Bus Solutions mentions limits, which are appropriate for the environments these interfaces were designed for:

Electrical Data Interfaces

The speed and distances listed are estimated maximums and simultaneous operation at both maximums may not be possible. As cable length increases, you must lower the signaling rate to keep the same bit-error rates. This, and the increased exposure to noise with distance, makes the analog characteristics of line drivers and receivers important and helps explain the number of different standards. It also complicates the selection process.

\$\endgroup\$

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