EMI EMC Problem in DDR2 RAM

[dharithothi]

My board is having problem in EMI testing in R.E. Problem is found in 133MHz Harmonics of DDR2 RAM clock frequency. I am getting strong radiating frequency at 533MHz and 666MHz which is 4th and 5th harmonic of DDR2 clock frequency. As known DDR2 has differential clock. Although I have tried to make RC filter and RLC filter. there was no any effect of them. I also made low drive option for DDR2.

My PCB is of 6 layer and processor and RAM is having proper ground reference.

Please help for this issue. :-?:-?:-?

 

[FvM]

DDR2 clock is only one of many possible EMI sources. I would also suspect RAM and processor power supply nets, RAM control, address and data lines.

Is it a discrete RAM or RAM module design? Are you sure to have suffcient power supply bypassing?

As a first step, I would try to localize the strongest interference sources with inductive or capacitive probes and SA. There may be a problem that the ground plane is excited and radiating as a whole, but you still be able to find "hot spots" that are the primary interfernce source.

 

[dharithothi]

Hello FvM,

Thanks for replay.

My RAM and processor is on same board only. there is no separate module for RAM. Both are in same board and in one layer only. I am using processor of Atmel 9G45 series. I have put enough bulk caps as per atmel guideline suggest. In my design I have put 47Ohm series registers to connect DDR2 ram with processor. I have checked with near field probe and main source is near processor to series registers.

 

[malli_1729]

how the clock,DQS, DQ lines were routed....Any Plane splits etc.....Layout review

 

[marce]

A picture of the layout between the RAM and the FPGA would be a great help here...
How many routing layers for the DDR interface and where are the return paths. How many memory chips etc etc.