receiving two same signals when using SN65LVDS95/96

Hello, thanks for your question.  The LVDS95 is a simple 21:3 serializer, and the LVDS96 is a simple 3:21 de-serializer.  I think you will find that the data inputted to the LVDS95 matches the LVDS96 output.  The serdes treats all 21 bits equally, regardless of whether they're used for data, Vsync, or Hsync.

Thanks,
RE

Hi,

  Thanks for you reply.

  I have known it. The D0-D20 pin of LVDS95 should be equal to D0-D20 of LVDS96, but our PCB has a problem, that is D15 pin of LVDS95 is equal to D15 pin of LVDS96, and D16 pin of LVDS95 is NOT equal to D16 pin of LVDS96. We have made several PCBs, which all have same problems.

  We think that maybe we don't use the serdes correctly, but how can we find where the problem is. Is there any testing method to help us to find it? Or Do you have

any advice?Thanks a lot, RE!

Hi,

  We find that the Y1P pin of sn64lvds95 is always lower than Y1M pin, and Y2P pin of lvds95 is always lower than Y2M pin. Do the differential signal have problems?

  Thanks a lot.

Hi crystal28,

1. Do the inputs to the LVDS95 swing from 0V-3.3V?
2. Is your frequency between 20-68MHz?
3. Is the CLKOUT from the device correct?

For debugging basic functionality, it can be useful to apply static inputs to the LVDS95 inputs, and measure the serialized output with a scope.  For instance, apply 3.3V on all 21-bit inputs and verify the output is always differential-high (Y1P=1.4V, Y1M=1.1V).

Thanks,
RE

Hi,

  The inputs to the LVDS95 and CLKOUT are all correct.

  Our frequency is about 27MHz. When we decrease it to about 20MHz, the VSync and HRef both work fine.

  Can you give some advice on increasing working frequence? Maybe our hardware design or layout has some problem.

  In the attachment, there is a 100ohm resistor between A0P and A0M, and we don't use it. Is it necessary?

  Or is there some other advice? Thanks RE.

Best Regards,

Thanks for the schematic.  That's very interesting that your problem goes away when you use 20MHz.  Sorry, but I'm not sure what you mean that you'd like advice on increasing the working frequency.  The frequency is probably based on the resolution, refresh rate, and blanking that's being output from your CMOS sensor; if you change one of those, it will change the frequency.  But it would be best to understand the reason why 27MHz causes a problem.

You should not put a 100ohm resistor between the differential pins, for both an LVDS transmitter and LVDS receiver.  The LVDS96 receiver has an integrated 100ohm termination.

What is the distance between the LVDS95 and LVDS96?  Is it all across FR4, or are you using a cable?  Can you see a difference in the LVDS waveform at the LVDS95 compared to near the LVDS96?  It would be useful to measure waveforms at every spot in the path.  If the LVDS path is a long trace, there may be loss that's degrading the signal.  Or there could be an EMI issue.  Debug can be simplified if you are able to apply static voltages to the LVDS95 input.

I hope that helps.  Let me know what you find.
RE

Hi,

  Thanks for your reply.

  I'm sorry english is not my mother language. Increasing working frequency means the LVDS95/96 can work under high frequency, such as 40 or 50 MHz .

  The distance between the LVDS95 and LVDS96 is about 12cm. We use copper wires to connect them. We will check the difference at LVDS95 compared to near  LVDS96.

  It has some difficulty to apply 3.3v to the inputs of LVDS95 on our PCB, but it's easy to apply 0v to LVDS95. Can you tell me the voltages of Y0P and Y0M when applying 0v to inputs of LVDS95?

  Thanks a lot!

Best Regards,

So, does your problem go away at 20MHz, or at 40-50Mhz?

12cm is very short at this frequency, so that's not much of a concern.

If you apply 0V to the LVDS95, the LVDS output should nominally be Y0M=1.425V, Y0P=1.075V.  The Vcommon-mode and Vod can each vary by 200mV, so you might measure a little differently.

Thanks,
RE

Hi,

  Our PCB can work only under 20MHz , and it seems it's enough for us now.

  we'll raise questions If we have other problems.

  Thanks RE!