About HB3 of DCS,who can give some suggestions?

My problem is that the mobile phone of our company can't pass the HB3 of DCS.
When we test the HB on the test point of the power PCL0,HB3 is under -40dB，But when we test it in darkroom ，the result is about -30dB.When we change the power to the PCL5，the power is lower ,but the HB3 was also -30dB.
The HB2 was very well，But the Harmonic of 5.3GHz is around -30dB and have the bandwidth of 200-300MHz，It looked like miscellaneous.
I have tried to change the power by series or parallel the RCL components. The result wasn't pass.
I also packaged the pcb board by Conductive fabric，only exposed the antenna point. But also failed.
The baseband chip is made of Taiwan company---MTK，the transever is AD6548 which is made of Analog Device.The PA is made of A Chinese company---RDA.

Regard
Yang...

Seem there is a radiated issue for 3rd harmonic of 1800MHz.

When radiated harmonics are greater than conducted harmonics, usually this happened when the radiated waves generated by the antenna are going back to the PCB, and finding there some nonlinear components and/or quarter-wave traces (at harmonic frequency) which radiates the power back into the air.

It is relative difficult to identify these components/traces, one option is to use a high frequency RF probe, ideally tuned only to the 3rd harmonic, to minimize the carrier and other harmonics levels.

Thank you for your reply.Do you have some doc about it?
regards
Yang

MAybe you can use some absorb materials. Sometime conductive materials can cause self-osc. How about try to add a LPF, say LFCN-2200+(?) from Mini-Circuits, OR just use LC form a LPF?