3
\$\begingroup\$

I would like to connect a small TFT display that uses LVDS, 4 channels for data and 1 channel for clock. Data rate is maximum 200 Mbps per lane. Very strict EMC requirements are applicable, namely automotive.

What would be the maximum length of a typical LVDS cable that would still work in this environment? Length could be up to 1 meter. I would use twisted pairs in any case. Shielding would preferably be avoided but could be applied if needed. Common mode chokes can be implemented on the PCB; clamps on the cable should be avoided if possible.

I am wondering from what length onwards I should use something like an FPD Link III SerDes (e.g. TI DS90UB927)? Or would that be overkill? BOM cost is important.

Theory like this suggests that I am still well within the regime for LVDS, but it does not quantify the EMC aspects, which are crucial in this application.

\$\endgroup\$
  • \$\begingroup\$ Could you use twisted pairs? Common-mode chokes? \$\endgroup\$ – CL. Jan 24 '19 at 14:39
  • \$\begingroup\$ @CL. I would use twisted pairs in any case; common-mode chokes could be added to the board (preferably not on the cable). Updated my question to reflect this. Do you think that with twisted pairs it could work at a length of 1 meter? \$\endgroup\$ – Martijn de Milliano Jan 25 '19 at 8:33
  • 1
    \$\begingroup\$ The additional cables cost more, and are harder to install. But I don't see much of a reason that the single SerDes output should be much less sensitive than the original LVDSes. Ultimately, you would have to measure yourself. \$\endgroup\$ – CL. Jan 25 '19 at 8:49
  • \$\begingroup\$ @Martin - 200 Mbps at 1 meter for LVDS should not be a problem. You should use twisted shielded pairs so as to provide the minimum loop area for any common mode currents that may flow. Make sure the shields are properly terminated at both ends of the cable. Improper shield connections (such as pig tails) are frequently a cause of failing EMI/EMC certifications. \$\endgroup\$ – SteveSh Jan 17 at 13:00
0
\$\begingroup\$

The standard deciding factor of whether a given transmission distance is practical, is how much jitter is observed by receiving nodes. This is application dependent; some applications require 5% or less jitter, whereas others tolerate up to 20%: https://www.analog.com/media/en/technical-documentation/application-notes/an-1177.pdf

There are a number of studies that look at how the signal integrity is affected by the transmission rate and cable length. However, normally you did not spec the maximum cable length for your application: http://www.ti.com/lit/ug/snla187/snla187.pdf http://www.ti.com/lit/an/slla127/slla127.pdf http://www.ti.com/lit/an/slyt163/slyt163.pdf

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ OP did say that the maximum cable length was 1 meter. \$\endgroup\$ – SteveSh Jan 17 at 13:01

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.