0
\$\begingroup\$

I am working on a project in which we have an an i.MX53 to driving an LCD through the LVDS connector and we fail the required EMC criteria. There are multiple peaks at 40 MHz intervals (starting at around 500 MHz) which I believe corresponds with the LVDS clock rate of the MX53:

measurement

The yellow is with the standard cable and the blue/cyan is the measurements with a twisted pair LVDS cable. I know we are in the red with that broadband noise between 200-300 MHz but I'm fairly confident we will have it fixed with the new expansion board that's in the works. Also I'm pretty sure it's the LVDS cable causing the trouble because those spikes go away when I unplug the cable.

My question is: how can I reduce the harmonics from the LVDS cable? I've tried ferrites (although I got no specs for the ferrites so it was a toss up) and twisting the pairs with very little effect.

Also is it possible that the i.MX53 sends out way too strong signals? Based on the schematics, they didn't overdo filtering.

i.MX53 LVDS out

I would be happy to hear any suggestions. Here's what we've tried so far:

  • twisting the differential pairs
  • ferrites
  • shielding the cable (although that one is anecdotal, I didn't work at this company than)

And here's what I'm considering:

  • Spread Spectrum Clocking (not sure if possible on the LVDS)
  • 47 Ohm resistors in series (an act of desperation)
  • shielding (give it another go)

Thank you,

David

\$\endgroup\$
7
  • \$\begingroup\$ A link to the driver's data sheet would be good. What loading did you put on the end of the cables at the LCD? \$\endgroup\$
    – Andy aka
    Feb 12, 2018 at 11:46
  • \$\begingroup\$ LCD datasheet: glynshop.com/erp/owweb/Daten/DSS/Promate/Products/… \$\endgroup\$
    – David Baka
    Feb 12, 2018 at 12:15
  • \$\begingroup\$ David, did you find a solution for this? I'am asking because I have the same problem. Regards, Greg \$\endgroup\$
    – Grzegorz Pawłowicz
    Dec 23, 2020 at 21:54
  • 1
    \$\begingroup\$ The solution was a hardware revision change by the vendor. Turns out they have been shipping the revised display for a while but our engineering sample was one of the old ones. Once I swapped it with a new display we were compliant. EDIT: Merry christmas :) \$\endgroup\$
    – David Baka
    Dec 25, 2020 at 10:31
  • \$\begingroup\$ Thx for yout reply, in our case source of problem could be harness, I will try several option (types of harnesses)and I wll know (I believe). I have also second type LCD so I will do some research. Curious, PCB alone is very quiet, problem exist with harness plugged one the source end (no LCD on the other end), plugging LCD change nothing. I don't know how important is good harness (especially tight twisting of twisted pairs)? Thx, David, Merry Xmass :) \$\endgroup\$
    – Grzegorz Pawłowicz
    Dec 25, 2020 at 22:58

1 Answer 1

Reset to default
1
\$\begingroup\$

Are you working with an OTS board or do you have control over the i.MX53 Board? If you can change the board, you could have a look at the following:

  • Correct (differential) trace impedance
  • Good layout: ground or power layers next to every signal layer
  • very good decoupling of i.MX and fast switching components
  • do use twisted LVDS pairs
  • use a suitable connector for high speed signals
  • correct pinout on the connectors

In all cases, even if you can't influence any of the above: look into drive speed and drive strength of the i.MX ports. It might or might not be available, but reduce as far as possible.

\$\endgroup\$
2
  • \$\begingroup\$ We're using the off-the-shelf i.MX53 Quick Start board and really wouldn't want to replace it due to legacy issues. I'll look into reducing the drive strength, it seems like it could be done and it might solve our problem, thanks. \$\endgroup\$
    – David Baka
    Feb 12, 2018 at 12:11
  • \$\begingroup\$ Experiment with soldering tiny SMT bypass caps from VDD pins to the GND plane. Just for fun. Just to learn. Some folks came to me with a EMI-failing LIN bus driver, failing at 400MHz. Turned out the Cbypass + PCB inductance from VDD to GNDplane was resonating at 400MHz; I had them alter the choice of Cby value, to alter the Fresonance. And to call the certifying society about the PCB resonance. \$\endgroup\$
    – analogsystemsrf
    Feb 12, 2018 at 17:48

Your Answer

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

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