Stripline feed slot antenna for water filled steel pipe

Hello to you all,

hope you have a nice summer everyone in the basement in-front of your computers... I know I have :D

I have been struggling with an antenna design all summer and quite frankly, I need some pointers and help. And hopefully some interesting discussions.

My design is like this (see PDF for figure):

I want to transfer microwaves into a steel pipe filled with water. My frequency band of interest is 2-3Ghz (fc = 2.5GHz, FBW = 40%).
The antenna type is going to be a slot antenna manufactured on a PCB. The feeding is from a stripline. This PCB is sandwiched in-between a steel block and the steel pipe. The steel block forms both mechanical support and EMC shielding to the outside. In the pipe it's supposed to be a corresponding slot as in the PCB trough the pipe. This slot will be filled with vulcanized rubber (εr approx. 2.5 with low loss). The thickness of the steel pipe is variable and so is the size of the slot in the pipe (I don't want to mill smaller than 1.5mm in diameter and not to big since its a round pipe 63mm with high pressure water inside).

In my simulation (see attached file from HFSS) I have done the design flat (and not round like the 63mm diam. pipe) and I only look at S11 for the antenna optimization.
I have tried everything and I can not get a good resonance in this structure. Why? I only get pretty much -0.1dB over the band and in some areas down to -4dB. Is there an error of some kind that's not just a little design flaw?

In the attached file I have chosen the channel (water normally) and the substrate to be rubber as well for simplicity. Feel free to play around if you want.

The slot is half a wavelength and that is λ/2 = (c0/(2*(fc*√(2.5))) = 37mm. This is also the distance between the metallic walls inside the substrate (vias in the future) that reflects the substrate mode to have a zero at the slot. The width of this box is only chosen bigger than the width of the slot and is not important.

The feed is grounded directly after the slot (could also be and open quarter wavelength stub) and it is offset from the center to match the feed to 50Ohm. The stripline width is optimized for the substrate height and dielectric constant.

I have read articles of cavity backed microstrip feed structures like mine where the cavity introduces a second resonance and hence increases the bandwidth. I cant have that since the pressure in the water is around 5 bar and needs mechanical support.

I have a design now that should introduce a good match at around 2.5Ghz and the only non optimized thing now is the feed offset but I can't get it to resonate nicely.

A couple of ideas and questions:

Can the steel slot be used to increase the bandwidth? I know that with only one resonance it will be hard (impossible?) to achieve a fractional bandwidth of 40%.
Can the pipe thickness be used as a quarter wave transformer between the antenna impedance to the rubber and the impedance in water?
The substrate dielectric constant can be chosen higher minimize the size of everything and perhaps match it better to the water (εr = 80 and high losses).

OK, that was about it. I have probably forgot a lot of things but if anything is unclear, I will clarify it for you.

I normally design microwave electronics (and also a lot of digital circuits) so this is a bit new for me (summer project...) and so is HFSS but man it's fun! Hope someone can help me a bit or at least steer me in the right direction. Good related articles is also welcome.

So, let the braincells free!

Over and out.

Joakim, Sweden