differential output signals shifted

[EDA_hg81]

I have used Spartan 3 FT256 generated 13 pairs differential output signals.

They are laid out on PCB with the same separate distance and length in each pair is same.

Right now the problem is that the positive and negative polarity output of one pair signals close to the middle of the group is not 180 degree shifted. Looks like they are 90 degree shifted.

What are the possible reasons for this?

Thanks.

 

[A.Anand Srinivasan]

please post which pair and the waveform i.e. the phase shift.... also tell about the external circuit connected if any....

 

[EDA_hg81]

Please wait one more day.

i have to capture those data.

Thanks.

 

[echo47]

Also check your PAR pad summary or use FPGA Editor to be sure the pair is connected to a normal differential driver, instead of somehow connecting to independent drivers.

 

[EDA_hg81]

Those four wave forms captured can explain what happened.

" Arrangement" shows 13 pairs differential traces.

The 6th pair from the top is the shitty pair.

“normal pair” shows the positive and negative output of 12 normal differential buffers.

“shitty pair” shows the positive and negative output of that abnormal differential buffer.

“positive polarity phase relation” shows the positive output phase relation between abnormal pair and normal pair.

“negative polarity phase relation” shows the negative output phase relation between abnormal pair and normal pair.

I also have checked FPGA editor, that shitty pair is assigned with the right differential buffer.

I am strongly confused what is the possible problem?

I am also going to contact with Xilinx about this issue.

Thanks.

"Arrangement"
https://obrazki.elektroda.pl/25_1189179177.jpg
"normal pair"
https://obrazki.elektroda.pl/83_1189178399.jpg
"shitty pair"
https://obrazki.elektroda.pl/100_1189178465.jpg
"positive relation"
https://obrazki.elektroda.pl/47_1189178518.jpg
"negative relation"
https://obrazki.elektroda.pl/83_1189178589.jpg

 

[Iouri]

Looks like immidance miss match

What do you have plug in your connector?

Also don't like idea of how you crossing layers on PCB. it should be Top straight top 45 degree via inner 45 inner straight

Also if you can do X-Ray on your BGA it could help, might be bad soldering joints

 

[EDA_hg81]

A display is connected.

But image on display lost some data.

I think this crappy differential output is responsible for what happened.

The most painful thing is if the impedance is not matching why the rest of them are ok?

Thanks.

 

[Iouri]

Do you have hyperlinx from MG?

 

[EDA_hg81]

Yes, I have used Hyperlinx.

only one board has this problem. The rest of board are fine.

May be this FPGA has some problem.