Strange transmission curves in HFSS for a particular substance at a very specific thickness

[Zachak]

Ok, so I have simulated some transmission curves for different powders modeled as dielectric clads. The screenshot taken from excel shows all of the ones that are nice and smooth (at 50um here, but remain nice and smooth regardless of thickness within my limits). The problem comes when I try to run Dry Milk only at 50um-- all of the other thicknesses of dry milk produce nice smooth curves (which you can see is the S21 graph from HFSS with the messed up, spiked curve). Again, this is only a problem for Dry Milk at the thickness of 50um... If I make it 50.5um instead of 50um, I get a nice smooth curve like the rest of them, regardless of their thicknesses and dielectric properties. What could be going on here???

 

[pgib8]

I may be completely off and I don't even know this program but maybe all the other powders are [mostly] made up of one thing but powdered milk is probably made up of different elements mixed together. What if you modeled each element separately, I don't know if that would give any clues.

 

[Zachak]

I may be completely off and I don't even know this program but maybe all the other powders are [mostly] made up of one thing but powdered milk is probably made up of different elements mixed together. What if you modeled each element separately, I don't know if that would give any clues.

Basically, HFSS doesn't know what the material is made up of. The dielectric is defined by the real and imaginary component of the permittivity in the 100GHz range, where the dielectric loss tangent is epsilon_Re/epsilon_Im

 

[pgib8]

In that case maybe it is a bug. I don't know if this is possible but what if you made materials that slightly deviated from what you have to see how it changes. Maybe some multiplier overflows to negative or an internal scaling issue. I've never used this kind of software and sorry I can't be of any help, hopefully someone else will come through for you.