Microstrip Coupled line filter design using Rogers Ro4003
I have designed a coupled-line filter at X-band using Ro-4003 with 8mil of thickness.
Rogers Software (MWI-2016) suggests dielectric constant of 3.81 for this substrate, this thickness, and at X band. this number has significant difference with popular dielectric constant for RF design using RO4003 (I mean 3.55).
What is your suggection? should I use 3.55 in my design (and expect shift in response!) OR use 3.81 and be optimistic!
I'd suggest using 3.55 until proven otherwise. When working with a board material that I'm not familiar with, I also make a 50 ohm line to verify my trace width. Sometimes this 50 ohm line can be chopped from the input or output trace and connectorizing this. 4003 is quite common.
Recently I asked PCB fabrication house about RO4003 and was surprised by this fact. They have recommended Er values for design (without stating frequency, only thickness) , while still providing standard Er value of RO4003. Why still providing standard value? Is it frequency and thickness dependent value?
Er is related (to some extend) on a number of factors. I would expect that John Coonrod @ RogersCorp could do an all day discussion on this topic.
h**ps://www.rogerscorp.com/documents/726/acm/RO4000-Laminates---Data-sheet.pdf
The above data sheet shows Er (Dk) vs Frequency. Design engineers have to start with some number to select substrate and thickness... and keep track of assumptions. I do mostly octave and multi-octave designs.
I haven't tried the MWI-2016 version yet, but using MWI-2014 at 10GHz, the software provided a frequency related Er value of approx 3.7 for RO4003 of 12mil thickness and the filter centering was pretty good.
Thank you.
This answer is what exactly I looked for (experience with dielectric constant, suggested by MWI).
Another question is about non-isotropic dielectric constant of RO4003 (it shows slightly different dielectric constant in X, Y, and Z directions). Therefore, its dielectric constant may also be depend on structure that is printed on substrate.
Is it correct?
It it possible to define a non-isotropic substrate (as mentioned above) in HFSS (or CST)?
I've only had trial exposure to HFSS and CST. I did use the Er value from MWI for my full-wave EM simulation in AWR's Microwave Office.