AWR frequency shift

[Mahruz]

Hi All,

I designed a 1.1GHz microstrip low pass filter using AWR.

After running axiem simulation, the results looks good. During the simulation, I used the following information which was provided from the PCB manufacturer.
Dielectric -3.5 (Rogers 4003)
Thickness -0.813mm
Top Copper- 1oz
Bottom Copper- 1 oz (this is polygon pour )
There is NO soldermask on the top copper trace.
There is soldermask on the bottom layer.

After manufacturing the design, the results was slightly shifted by nearly 200 MHz.


Is there any reason for this frequency shifts after manfucaturing despite the EM simulation shows excellent reason?

Thanks

 

[vfone]

Dielectric constant variations could shift the frequency of a microstrip filter.

Here is written that RO4003 have Dk of 3.38 +/- 0.05, and not 3.5 which you used in the simulation.

RO4003C™ Laminates

RO4003C™ materials are proprietary woven glass reinforced hydrocarbon/ceramics with the electrical performance of PTFE/woven glass and the manufacturability of epoxy/ glass.

www.rogerscorp.com

 

[Mahruz]

Hi Vfone,
The reason why I used that 3.5 for the di-electric is because the PCB manufacturer said to use that value.
My question is that, is there anything else that has to be taken care of to ensure that is no shift in the frequency post manufacturing?

thanks

 

[BigBoss]

The manufacturer might have been used a "replica substrate". Because in order to reduce the prices, they generally use replica substrates that are equivalent to original Rogers or any others.
If the specifications are tight, I recommend you to use original Rogers substrates.
In additional to, the sensitivity of a planar structure against manufacturing tolerances depends also on structure itself and manufacturing process. ( very thin lines, very close design, grounding etc.)
Do an Pareto analysis and find which part of your circuit is sensitive against tolerances.