how to measure s11

I am looking at S11 on the network analyzer to be as minimum as possible or at least below -20dB at the specific freq: 2.4GHz, and I am using a 59 inch long RG400 coax cable.
However the rezonance is shifting if I move the antenna far or near from Net ****.
I assume I am missing something or I am doing something wrong.
I am sure I have reflections from the walls since that room is a shielded room, but I decreased the S11port power to -35dBm and the shifting is still there.

Where is the problem?
I don't know anymore when I am measuring correct and when I am measuring in error.
Do I need to build a small anechoic chamber using flat panel aboserbers ?
Do I need to reduce the power of port 1 ? Right now I am using 0dBm but I don't see a difference between 0 and-35dBm.
Do you know a site or an application note which address this isssue ?
Thank you in advance !
HS

I can guess that when you move the antenna, there is some standing wave issue, probabl that is the cause.

1. Reflections
I think you are getting reflections from the network analyzer and the metallic
wall of the shield room. I would recommend you take the measurement out
of the shield room and direct the antenna into an empty space. If directing toward
the sky is not feasible in your location, long corridor may be good enough.
Anechoic chamber is the best, but maynot be easily available.
2. Power
Power change does not affect the result, as the network analyzer corrects
the reading automatically. At low power levels, noise will make the measurements
less accurate. So, 0dBm is a good level to use.
p.s. I assume you performed calibration at the end of the RG400 coax.

Regards,

Hi HS,
Can you tell how you've performed the calibration and where is the reference plane? What analyzer do you use? Thanks.

/pisoiu

Hi Pisoiu
I am using an E5070B netw. an. and 8033D calibration kit.
I calibrated just port 1, using 1.5 meter RG400 coax cable.
I soldered an SMA just bellow the feeding point (a small 1mm hole) on the oposite side of the FR4 board taking care to have a good ground too and cut the legs of the SMA to have a good contact to the ground of the board.
Now the cable is perpendicular on the board (heading down) and the FR4 board is sitting on my bench horizontally.
Maybe I should use ferite bits on my coax cable to cancel the ground curents and eliminate the influence of my coax cable to antenna measurements.
But this RG400 is already double shielded.
I don't know anymore what to do.

HS
P.S. Of course the ref plane is at the end of the cable.

Hi,
Thanks for the answers. I will have to perform these operations on a PIFA and this is why I've asked you about the calibration. I suppose the SMA will affect the measurement, and I don't know how can you deembed it. My idea is to use U.FL connectors since they are much smaller. The only problem is the calibration in this case.

/pisoiu

I don't have flat panel absorbers, but that might give me a good feeling about what is going on. I cannot find any absorber tuned on 2.4 GHz here in Arizona.
With them I can build a small box and place the antenna inside. I hate reflections.
I don't know how much they affect my measurements but I don't feel comfortable.

I still have to investigate how big that box should be and how the distance to the walls is affecting or not the measurements.
Right now I am out of shielded room and I am still facing problems with the impedance as magnitude, but less drastically compared to before and the shift in frequency is not anymore. Thanks GOD ! !
I am at -24dB ( almost ) on a 100 MHz bdw and I can see the cable still affecting my measurements.
(still dreaming at some ferrite beads ).
Please comment my lines. I like corrections !:D and don't forget I am a newbie.
H(umane)S(ociety)

As HS described "PCB antenna", I assumed it is a patch antenna which
is unbalanced.
HS, can you explain the geometry of your antenna?
Regards,

No, it is something close to an inverted F.
I don't have ground underneath.
HS

Just extend your RF cable to the place where you will be doing the measurement
and do your calibration of the Network Analyzer. When you move the cable
you are changing the impedance values that you just calibrated to.

With that long of a cable you can introduce a lot of errors when you move it even
a few inches(And avoid sharp bends).