gsm harmonic testing

In GSM 900, The GSM modem I am using has a radiated second harmonic more than -30 dBm (violates ETSI). Testing the modem revealed that the conducted second harmonic at 1800 was -42 dBm. My antenna gain at 1800 is 3 dBm. I am not sure where the 9 dB gain coming from in radiated measurements. I was reading something about the matched circuit.
I am doubting that, when I connect an antenna to the GSM modem there is something that is amplifieng the unwanted second harmonic. this could be bcos of the impedence mismatch at the port of the modem bcos of the antenna?

Help required !

I think you would check your matching circuits , especially for PA firstly. layout for RF should be concerned.

Hi Raja,

Maybe you first should go back to check conducted harmonic level. If already violate spec, pls re-tune your PA output matching circuit and also be sure of
switching spectrum is pass safely margin. If this is done or conducted harmonic
already pass safely originally, you may go check if harmonic would goes down
by backing of power control level, say GSM PCL5 to 6, would let wireless harmonic
improve some dB quantity. If the level downwards obviously, then maybe your
antenna effective SWR at fundamental frequency that PA sees is away from best
power contour and the loop tried to drive PA hard to deep saturation region.
Layout concern would be more like RF leakage problem. You can go back to check
conducted harmonic leakage from you ground of board or Tantanium capacitor pad with that cap take-off on conducted testing. You can solder RF cable directly
from the points I mentioned and connect the cable to spectrum analyzer. If that level shown on SA is already BAD, then that's the harmonic noise walking through
the Vbatt trace and even power cable. Pls add bypass capacitor to terminate that
harmonic leakage guy.

Hi XinXin,

Have you any points want to go complement for me or I mislead your points
for layout concern? Pls kindly comments, thanks!

But this modem passes the conducted measurement...wouldn't this fail if the PA was over driven.

Added after 3 minutes:

Verified conducted measurements and these pass...I am looking at the SWR of the antenna...PA is on the modem and I cannot change anything on the modem, I have the freedom to change the design on the circuit so adding a mtached circuit or matching the impedence would make a difference? Please advice.

I am not sure what you are saying. Are you saing that when you connect a spectrum analyzer to the antenna port you see -42 dBm 2nd harmonic, but when you connect an antenna to the antenna port, it looks like -30 dBm?

If so, the answer if fairly obvious: your enclosure is leaking 2nd harmonic energy. It is likely that there is a 2nd harmonic filter placed between your output amplifier and the antenna connector. However, there is probably plenty of 2nd harmonic coming out of the amplifier. It hits the lowpass filter and bounces off of it, perhaps making an excellent antenna at 1800 MHz out of the pcb trace.

You can try better enclosure shielding, such as a metal can covering the amp/filter. You could also try an absorptive lowpass filter, which will not cause a 2nd harmonic standing wave on the pcb trace.

Also, really check the physical layout design of the pcb in the area of the amplifier and antenna connector. If some looney pcb layout guy cut your ground plane in this area, all bets are off! You will have huge circulating ground loops of RF energy.

Yes, you can pass 2nd harmonic conducted and could fail radiated harmonics in the same time.
Are a lot of factors that can make these phenomena.
First could be the antenna gain at harmonic frequency. At that frequencies the antenna gain shall be low enough, to do not allow propagating the residual conducted 2nd harmonic on the air.
Sometimes when using embedded antenna, is possible to get a radiation of the fundamental frequency, inside of the circuit board, and some components start behave as rectifiers and generating radiated harmonics, this with the help of the bad layout design (grounds, submultiples of lambda transmission lines, etc).

I just became facing the similar problem, and the problem is not such in the GSM modem. It turns out the supporting modules I use picks up the 900 MHz signal, retectifies it through the pn juctions and re-radiates as harmonic emissions. The worst part is the supporting modules do not even have to come in contact with the GSM module and even in off state. As long the PCB is eploxed in close proximity of the GSM modem and with the right dimensions in any trace length, wires, connectors, they pick up the 900 MHz signals and propagate to pn juctions; there's the frequency doubler! So make sure all electronic modules nearby are shielded.

PIM is one of the cause to get 2nd order even the PA is tuned to reject 2nd harmonic.
Be careful with the cascaded system design.

Hi Kspalla,

May I know what's PIM refer to?

By the way, is this issue closed?