We are trying to design a PCB antenna for wifi application. This is a verified PCB antenna design for 2.4 GHz band from texas instruments. antenna Their PCB's stack up is like this:

--------- 15 um copper layer

1.47 mm substrate FR4 with 4.2 Dielectric coefficient

--------- 15 um copper layer

when I simulate this exact design in HFSS with exact PCB model of TI's, it gives fine results. s11 has a down peak at 2.45 GHz and VSWR is below 1.2 between 2.38 and 2.52 GHz.

However, when I use the same antenna design with our PCB stack-up:

--------- 18 um copper layer

0.87 mm FR4 with 3.95 Dielectric coefficient

--------- 18 um copper layer

the center frequency of the antenna shift very much above 2.7 GHz. Of course I figured it is related to difference in substrate height and dielectric propoerty but I don't know how should I modify this. I feel like I have to decrease the width of the lines somehow, so I did. But it did not wok, infact, it shifted the center frequency to 2.8 GHz. Can you suggest me some intuitive approach to modify this antenna so I can use it in my wifi application? I would not ask this if I have much time but I don't. I need to figure it out in a week I guess.

  • \$\begingroup\$ was it the right term? I can change it again. \$\endgroup\$
    – Alper91
    Commented Apr 4, 2017 at 13:47
  • 2
    \$\begingroup\$ It seems to me that if narrowing the tracks moved the frequency the wrong way, the obvious next thing to try would be widening them. \$\endgroup\$
    – Dave Tweed
    Commented Apr 4, 2017 at 13:49
  • \$\begingroup\$ It put a smile on my face but in general it is not advised to refer to one self in third person. Don't know why my comment just disappeared. \$\endgroup\$
    – winny
    Commented Apr 4, 2017 at 14:01
  • \$\begingroup\$ From: hackaday.io "One thing I failed to mention is how forgiving the TI antenna is. It has greater bandwidth than other common pcb antennas, is reasonably omnidirectional, and the only board parameter it cares about is copper thickness." \$\endgroup\$ Commented Apr 4, 2017 at 16:12

1 Answer 1


The results you have are a bit odd, due to the lack of a ground plane under the antenna (as the datasheet for the antenna suggests http://www.ti.com/lit/an/swru120c/swru120c.pdf) the PCB thickness should not matter, there isnt a reference plane that will modify the antenna characteristics themselves (you did remove the ground plane from underneath the antenna right?)

What does change however is the feed line from your tranciever to the antenna, which will need to be modified to maintain the same impedance as before, as your layer stackup is different.

Also, as SteelRat pointed out in the comments, you might want to use the same copper thickness as the TI stackup, as that has some effect on impedance, he indicated it is the copper thickness that might make or break the antenna. So if you follow TIs recommendations with copper thickness, shape, ground plane recommendations, and modify your feed trace to 50 ohms, then your design is quite likely to be good and stable.

  • \$\begingroup\$ Unfortunatelly eventhough TI suggest that PCB thickness doesn't matter, in HFSS when I inceare the Substrate height, it comes down to 2.5 GHz. It is very confusing. \$\endgroup\$
    – Alper91
    Commented Apr 9, 2017 at 7:56
  • \$\begingroup\$ My hypothesis then would be that it is a issue with HFSS or how you set up the analysis. The thickness of the dielectric shouldnt have such a strong effect according to theory. \$\endgroup\$
    – Stonie
    Commented Apr 10, 2017 at 11:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.