Try making everything 10% longer. It won't increase the bandwidth but it will move the frequency about 10% down.

Dear Malcolm White ,

I actually did increase the length from 150 mm to 300 mm but I just got more ripples and still no improvement!

Actually, I'm trying to enhance the bandwidth near cutoff frequency. There are many papers improving bandwidth after cutoff frequency (e.g. for X-band with the cutoff frequency of 6.55 GHz, bandwidth improvement has been applied for 8.2 GHz to 12.4 GHz)...

Without actual design information it's difficult to tell how to improve bandwidth. But irrespective of your design there are several ways to increase SIW antenna bandwidth like;

1. By exciting two hybrid modes in a SIW cavity and merging them, impedance bandwidth is enhanced.

2. In SIW, usage of a thinner substrate would result in higher Q-factor and then lead to narrower bandwidth. The bandwidth of a SIW cavity-backed slot antenna through viahole located above the slot can be enhanced. Using metalized via holes in the middle of the cavity, resonate frequency of TM010 mode is shifted to near that of TM110 mode. This creates an additional resonance at a higher frequency by shorting the effective length of the slot

3. SIW stacked patch antenna structure also increase bandwidth

4. Transition is another component which a broadband design should be. The transition should have a low insertion loss and wide bandwidth. A stepped impedance transformer is an effective method to enhance the operating bandwidth of transitions.

You needed to make it only about 10% longer. That will move the good match on the right of the picture to the left of the picture. Making it twice as long is too much in one go.

Dear Malcolm White

I just did what you said but I just got very slight shift in bandwidth!

In the first place you said make everything 10% longer. I mean what do you mean by everything? Am I supposed to increase the width of the microstrip feedline as well or just the length of both feedline and SIW is enough?

Thanks,

You need to increase the length of the resonant components - you should not necessarily increase the width because it will change the impedance.

If you made everything 10% bigger, including the thickness and the width, the wavelength will scale and the shape of the frequency response will be scaled by 10/11. Usually you don't need to change the thickness or the width. You need to increase the length of transformers and stubs too. It won't scale completely, because length adjustments due to fringing fields won't scale (they would if you changed the width and thickness as well as the length).

That is one reason for not changing the length by a big ratio in one go - small changes mostly preserve the shape of the return loss, which can then be corrected by small changes to individual lengths. Changing by a big ratio can completely change the shape so that you can't tell what to do to recover it.

When you say you only got a small change in bandwidth I'm not sure what you mean. I would have thought that the first thing you want is a shift in frequency, not bandwidth, because there is a wide bandwidth but at too high a frequency. Scaling will move the frequencies down but probably reduce the bandwidth (in Hz, but not in percentage). Once you have the bandwidth roughly in the right place you can then try to increase it if necessary.

From response, it is very clear that Subtrate Integrated Waveguide design has resonance near 9.1 GHz either low impedance with ground or high impedance in output based on type of design.

Pl check type of impedance effect and modify design accordingly.