PA design: issue with parasitic capacitance (bondpad+bondwire+pack pad+pcb)

[analog_ambi]

Hi,
I am designing a PA as shown in schematic. Some specs are:
2.4 GHz
ultra low power (<1mA)
moderate linearity (modulation is QPSK, moderate PAR, either class A or class AB)

The problem is my gain falls by 2dB (w.r.t mid band gain) at 2.4GHz when O/P is loaded with 200f parasitic capacitance. The O/P shall drive offchip matching network and then 50 ohm. I have the following questions:

1. How do I get fair estimate of parasitic cap Cpar (bond pad+bond wire+ package pad+ pcb)
2. Is it possible to tune the parasitic cap out by including it in the cap (C) of matching network.
3. If yes then onchip or offchip cap for matching solves it best? The matching network upconverts 50 ohm to 500 ohm.

 

[chuckey]

Xc200 @2.4 GHZ = .3 ohms. very high?,2. YES its always tuned out. 3.Off chip I would have thought. 1. See what other people have done (previous designs). can you alter the series inductance (multiple leads?). My 2 pence worth
Frank

 

[albbg]

Just a clarification: 200fF are Xc=330 ohm @ 2.4 GHz.

1. use a simulator to exctract the parameter of your parasitics

Very very rough estimate, I think you can consider the PADs as parallel plate capacitors. Thickness and epsilon of substrates are known. Probably the real capacitance will be higher due to fringe effects. Bond wires could be very roughly seen as a 0.75 nH/mil inductor (for 1 mil diameter gold wires)