Why do the harmonics power levels increase when radiated?

Normally the phone transmitted path is as following:
PA output--> Antenna
At the PA output, conducted output 2th harmonic is at -45dBm.
But in EMC test, the phone radiated 2th harmonics is about -30dBm.
I don't understand why the power level increase? Is it caused by the mobile phone antenna gain? I think antenna has -4dBi gain at operating freq.
Does any other reason result in that?

It is very likely that the antenna is more efficient at the higher frequency. Another possibility is that the harmonic is being radiated from the wiring on your circuit board. Try putting a resistive load on the antenna connector and measure the two radiated components.

Isn't there a circulator between the PA and the antenna?
It would produce even-order harmonics.

Z

Was happened to me many years ago when designing a dual-band GSM phone using an embedded dual-band antenna.
I discovered that the radiated fundamental signal (900MHz in that situation) was reaching a non-linear component on the main PCB. This component was connected to a trace which length was quarter wavelength at 2nd harmonic (1800MHz). I found that the 2nd harmonic was not only radiated by that PCB trace, but also was coupled to the DCS1800 antenna which helps to increase the radiated 2nd harmonic even more.

Hi, vfone, zorro, and flatulent,
Thank you very much.
There is no components between the PA and the antenna, just some LC components.
So some traces or non-linear compoents on main PCB radiated.
I have the chance to insert a LPF between the PA and antenna, so let's see what will happen.

Antenna is not wideband termination for PA and is causing missmatch at 2nd harmonic. It can produce detuning of PA and harmonics are higher. Try a circulator instead of LPF first.

But we have already tested the load pull of the PA with VSWR=3/4/5, and in any condition, the 2H is less than -47dBm.
So we concluded that it is not caused by the load pull of PA, so I guess that it's no use to insert a circulator.

Terminate with a 50 Ohm resistor instead of antenna. Make a small loop sniffer, 5-10 mm dia., 1-3 turn. Hook it to spectrum-analyzer.
Guess it is a resonant inductor somewhere between PA and ground that shows highest level of magnetic field.
Reducing this current will also reduce radiated harmonics.

"But we have already tested the load pull of the PA with VSWR=3/4/5, and in any condition, the 2H is less than -47dBm."

This should have provided you the right answer. I'm curious to know how you measured the power level of the 2nd Harmonic while the PA was miss-matched. Did you use directional couplers into a spectrum analyzer? edit: Did you test different phase angles for each VSWR test?

There is no components between the PA and the antenna, just some LC components.
So some traces or non-linear compoents on main PCB radiated.
I have the chance to insert a LPF between the PA and antenna, so let's see what will happen.

Click to expand...

Any results?

I think you probably misunderstood vfone's comment. The non-linear component must not be necessarily "between the PA and the antenna". It can be an arbitrary circuit part that picks up sufficient power from the radiated field and re-radiates the harmonic, e.g. a diode with a PCB trace connected to it. It's particularly likely to happen with embedded antennas.

There are however several alternative scenarios that should be checked as well:
- structural radiation. It would be also present if the PA is terminated with an artificial load.
- selective antenna gain, as already discussed. I think it's unlikely with well designed antennas. In case of doubt you can measure the antenna gain separately.
- PA harmonics caused by load mismatch. Can be checked by providing an attenuated monitor output

know how you measured the power level of the 2nd Harmonic while the PA was miss-matched. Did you use directional couplers into a spectrum analyzer? edit: Did you test different phase angles for each VSWR test?

Click to expand...

Yes, for each VSWR, we tested different phase angles, that is 30 degree apart to cover all the 360 degree. My company has a auto-test station for load-pull test.

Next week we will test the 2H with LPF and I will post the result here. Anyway, if the radiated spurious is from the PCB, then it's no use for adding a LPF. If radiated from antenna or its peripharal circuit, then the LPF may help.

Our phone has already passed the relevant tests, and I am just curious about why the 2H being increased.
I will do the test to find the root cause. Thanks every guru's comments.