LVDS Communication problem - skewing maybe?

[sebmaster]

Hello,

I wasn't quite sure where to put this so apologies if it shouldnt be in this forum.

I am trying to drive an LCD with a DS90C363 FPD-Link LVDS transmitter.

The LCD acted as if it were not receiving data, so to debug I connected a DS90CF364 - the receiver for the 363 - to the output of my controller to check the signals.

The 363/364 are designed to take 21 bits of TTL data and a clock and transmit them down four LVDS lines before driving TTL outputs with them again. 21 inputs and outputs that should map exactly.

When I put a scope on the 364 what I see is that the ouputs are being driven but with the wrong signals, for example the biggest problem is the DE signal, going to TX20 on the 363 turning up at RX17, 18 & 16 on the 364, although not looking as clean as the waveform before transmission.

I can't think why this would be, the cable between the two is only a few inches long and for the most part the LVDS cables are shielded twisted-pair.

Is there anyone who knows LVDS a little better and has some ideas?

Here is the datasheet for the chips: **broken link removed**

(I am not using any terminating resistors or the like - the connection between the chips is direct. They are powered from the same supply.)

 

[FvM]

I am not using any terminating resistors or the like

By specification, LVDS requires receiver parallel termination, independant of distance. Otherwise, no regular operation
can be expected. If you have doubts on this, check the receiver signal, if it's a halfway regular LVDS level. Most likely,
you'll observe total overload.

 

[sebmaster]

Hi FvM,

Thanks again; I was under the impression that since the load impedance is necessary for the operation of LVDS that it would be part of the receiver IC.

Once I put 100ohm resistors accross the LVDS pairs of the reciever the waveforms cleaned up and I started getting the outputs at the right pins.

My concern now is that I don't know why my LCD is not working, since the datasheet states that that has the same resistance accross all LVDS pairs. I will try and put resistors accross the outputs of the transmitter and see what affect that has.

EDIT: LCD is working well now, there was a problem with VCC oscillating quite badly due to what I can only assume was a faulty connection on the controller board. I bypassed the track showing a resistance and it started working.
I can show solid colour perfectly, when I try and draw a pattern it seems as if the DE signal is not synced perfectly but I suspect this could be to do with how I am driving the RGB outputs in the VHDL (assigning integer values rather than switching like the actual controller would).