Substrate losses question?

About losses - the tan(Delta) loss value for substrates, like 0.0008 for PTFE and 0.014 for FR4

Some softwares include models for how this might vary with frequency, called dispersion loss. There are names like "Debye" and "Lorenz". The present default option is "none".

Suppose we already know a couple of measured loss values at two different frequencies, say maybe at 2GHz, and another at 10GHz, hoping to maybe use with it.

Can anyone offer any hints about how to use any of these?

DEbye and LOrenz are theoretical models showing the loss vers. frequency. Practical applications are to use lower-loss substrates at higher frequency applications. Also use short line sections wherever possible.
Dispersion is a grave problem in bad substrates and design. It causes loss by unwanted radiation modes, and also unexpected feedback in amplifiers. Always select best substrates and save on line length.

My thanks jiripolivka.

I have so far done what most of us do. We choose the lowest loss substrate we can, trading off the loss factor against cost, ease of use, etc. For example, I choose some that have some ceramic loaded in with PTFE because vias can be made without a difficult sodium etch needed to adhere the metal. The dependence of frequency is a refinement we usually do not go to because Er and tan(delta) are normally fairly constant over a usual band of interest.

This time however, I was trying to model radome-type surfaces, not used in a circuit-building context.
I (think) I found the way. A description of how to use a couple of known values to establish at least the simplest order Debye rule for the material. I will hold off posting anything about it yet until I am reasonably sure I have got it figured out.

Meantime, I had hoped someone had gone further than entering only one Er, and tan(delta) when supplying material properties. Mostly, one is able to avoid even that, by just choosing a nice substrate from the default list.

Well, you are pretty advanced as I can see. With radomes the problem is easier in that the power density is much smaller than in transmission lines, so material characteristics become less important. I saw fabric radomes soaked in bakelite used a X-band and behaving quite well. Teflon fabrics are used for antenna radomes up to E-bad (80GHz).

Yes exactly.For instance FR4 patch works well at 5GHz antenna and probably up 10GHz. Arlon quite well up to 30GHz and above.