1
\$\begingroup\$

I have a system driving LVDS over ~ 140 ft of CAT6 cable. I am using the SN65LVDS series devices from TI to accomplish this. I'm aware that there will be low-frequency distortion in the signal at the receiving end due to the attenuation of the cable itself.

I realize there are good ways to equalize the LVDS signals at the receiving end, but I am after a really simple solution. Can I put caps in series with the LVDS high/low lines to EQ the signal a bit passively? How would I size them, a decade below signaling frequency? If there is a 100 Ohm termination resistor between the two lines (H/L), do I use 100 Ohms as the R in the RC calc?

I found this app note that describes what I am talking about a bit, but it doesn't quite answer my question.

\$\endgroup\$
5
  • \$\begingroup\$ Are you using a signal encoding that ensures no DC content? \$\endgroup\$
    – The Photon
    May 11, 2017 at 15:45
  • \$\begingroup\$ Probably not. I don't know what the drive end to me looks like, I just receive the signals and convert them to LVDS for the long distance drive. \$\endgroup\$
    – jareddbh
    May 11, 2017 at 15:49
  • \$\begingroup\$ By "low frequency distortion in the signal at the receiving end due to the attenuation of the cable itself," are you referring to attenuation by the voltage divider formed by the cable impedance and the terminating impedance at the receive end? Consider that series decoupling caps will only serve to add effective cable impedance at low frequency, and thereby only increase this effect... \$\endgroup\$
    – user49628
    May 11, 2017 at 16:08
  • \$\begingroup\$ I meant harmonic distortion caused by cable parasitics. I read somewhere that the signals tend to have low frequency noise after traveling through a long cable (parasitic caps, etc). So a way to 'boost' the high end is to put series caps. My signal will be switching on the order of 10-20 MHz. My thought was to use series caps with an RC set at around 1 MHz. I could be totally off here though. \$\endgroup\$
    – jareddbh
    May 11, 2017 at 16:31
  • \$\begingroup\$ Boost is the wrong term, attenuate the low frequency distortion is a better way to say it. \$\endgroup\$
    – jareddbh
    May 11, 2017 at 16:53

2 Answers 2

2
\$\begingroup\$

I think you are getting somewhat off track.

In comments you say you don't know the signal encoding and it may have some DC (or very low frequency content). In that case, a simple series capacitor termination will block some of your desired signal, probably causing worse distortion than what you are encountering from the cable loss.

How would I size them, a decade below signaling frequency?

If you did this, you would be basically trying to make the filter not affect any of your signal frequencies. It wouldn't have any equalizing effect on the frequencies you are actually using.

Rather than just a series capacitor, maybe try a simple de-emphasis filter:

schematic

simulate this circuit – Schematic created using CircuitLab

The resistors provide some attenuation (de-emphasis) for low-frequency signal content, while the capacitors let high-frequency content (whatever has made it through the transmission line) pass through.

Appropriate values for the filter capacitor and resistor depend on the characteristics of the transmission line between the transmitter and receiver as much as they do on the characteristics of the receiver or the signal. You'll have to experiment (or simulate) to find the best values for your situation.

\$\endgroup\$
1
\$\begingroup\$

Assuming you only need some first and 3rd harmonic, consider this for 20MB/second LVDS

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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