Highpass microstrip power problem

Hello everyone.

I seem to be facing an interesting problem.
I have an S-Band highpass filter made with microstrip.
It requires to take some power, 10+ W.
The filter itself came out pretty smooth, the shorted resonator lines have an approximate impedance of about 80 ohms and vias of about 16 mil dia.

Everything looks good with the unit. Insertion loss and VSWR look perfect to use.
As soon as I crank up the power, I get a 30% reflection of the power.

Before this one, I had a highpass with wider grounded resonators and they were ok as well.
I crank up the power and it reflected.

I proceeded to make the lines thinner to increase the impedance thinking that it would fix the problem
but the situation stays the same.

If any of you has ever faced this situation before, I would greatly appreciate if you could explain to me how is it that such situation is happening despite the fact that I am trying to get rid of any possible variable in this system.

Thank you very much.

To determine if the filter can handle the power, you'll calculate current densities and peak voltages. 10 W sounds generally feasible for microstrip.

I don't exactly understand the problem description. Do you say, the filter impedance changes temporarily? It mighte be a measurement or PA problem.

The max power I am looking is around 40W.... But I can start with 10 and work my way up. In the VNA across the entire band, my vswr is at least 1.5 to 1. Now, I know that small vs large signal behave differently, but I can't figure out why the power is reflecting. You mention calculation of current density and peak voltage. Is that just to find out the breaking point or is also for possible mismatch?
The funny thing is that I was trying to make a different microstrip highpass for broader coverage and tried the same test and it did not reflect any power at all. NONE. Another reason why I am puzzled.

If the curent density is too high,copper losses may create some temperature issues consequently it will turn out into substrate unstabilities.Is there any overtemperature observation ?
Current density can be too high under low voltage supply conditions to achieve the same amount of power.

A microstrip filter is linear, so there should be no difference between small signal and large signal response ... up to the point where you permanently damage (burn) lines or vias in the filter by excessive current.

Yes. Unfortunately the problem description is too vague to understand if the filter had been permantly damaged.

I disagree. If you have structures that can breakdown with voltage, there might not be very big damage. A breakdown in voltage will arc and cause a nearly total reflection of power.

After you go down in power is the reflection gone, or is there permanent damage. If there is permanent damage, look for arc trails (little black lines cause by carbon)

what exactly is the microstrip structure? are there gaps, series coupled lines? Maybe they need to be wider gaps. Maybe you need an overlay structure with mylar in between two broadside coupled lines, or high voltage series capacitors.

and yes, a filter can heat up and cause power handling problems that way....if you blow a fan at it during testing, does the problem go away?

look at anyplace there are sharp edges, such as where a coaxial connector meets a board edge--the sharp edges multiply the electric field.

Yes, a microstrip structure can have a nonlinear behavior vs input power, and not necessarily to reach the point where get a permanent damage of the structure.
Just decrease the power and reflections are gone.
This happen especially when cheap materials are used at microwave frequencies and high power, corroborated with poor layout design (as was already mentioned, sharp edges, low spacing between resonators, etc).

Partial dielectrical breakdown can of course happen. If it lasts longer than a few seconds, we can expect a circuit detoriation that shows up in small signal S parameters. (unless you managed to design a perfect spark gap).

I suspect it may be something beyond the filter. I think the amplifier I am using to test it is not behaving properly.
I have no way to prove this because I don't have another amplifier in that range. But I think the filter itself should not have a problem.
The reflection happens right away beginning from the lowest power, 1 W, so I think the amp could be at fault. The filter is not arching at all.
Because the power doesn't even go through it. I'll have to find another way to test it.
The material I am using is RO4003.

How are you measuring reflection? If you have a bidirectional coupler and you're measuring both forward and reverse waves, then it shouldn't matter that the amp is nonlinear.

put a HIGH POWER attenuator at the output of your amplifier, and see if the filter works. If so, it was your amplifier tripping out due to bad load VSWR.

If the filter still stops working at 1 Watt into the filter with a 10 dB pad there....it was not the amp.

Hello to all. I had the filter tested on a separate location and found no problems at all.... I checked the amplifier with other parts in the same range....My suspicion was correct. Amplifier was behaving..problem solved....thank you very very much for your comments!