Questions about PA(DA) Linearity

[Henry-cool]

Dear All~!

Thanks in advance.

I designed DA with cascode structure.(see the attached file)

In two tone test, IM3 is 32dBc@4dBm(output power) (CMOS 0.18um, Vdd:1.8V)

I need IM3 >40dBc.

I think the DA is the well-known structure and many people use it.

Do you know the method that improve linearity?

Please let me know the method and related materials.

 

[vfone]

Tune the bias point, and tune the matching output (Lo, Co, Cac) for maximum available output power (shall be more than 4dBm) and after that, back-off the input power to find the sweet-spot for IM3.
Sweet spot is a deep in the IM3 vs P_inp/output curve, which depends by the bias, AM-AM and AM-PM characteristics of the Power Amplifier.

 

[kspalla]

is the topology is DA only or you can go to Doherty with pre-D

 

[radiohead]

Sweet spot is a nice thing but
1. It does not always exist because it is caused by two nonlinear IM3 generating components which generate cancelling spurious. If one is dominant there is no sweet spot
2. Generating IM products depends heavily on temperature. So the IM3 product at the output of a transistor caused by the base-emittor junction is very much temperature dependant as the VI characteristics are exponential with threshold voltage linear with T!

So if you choose this solution: do many corner simulations and temperature sweeps and make sure the thing keeps stable!