Need help on BIAS circuit design for broadband amplifier

In narrowband case,we use 1/4 wavelengh high impedance line as RFC to isolate bias circuit from RF signal.
For wideband case(>40%), 1/4 wavelengh(for higher freq) high impedance line can not work as RFC at lower freq.

Please give me some instruction or paper on this.

Thanks in advance!

have you considered the use of bias tees?

Thanks for your suggestion.

I cannot use bias tee module in my circuit,because I should not make my circuit too big.

Use resistive biasing then, stabilises the PA and also currents....

eep2nw,thanks for your suggestion.

Pls show me detailed description or a paper on this.

Look up the chapter on Amplifiers in MMIC Design, Published by IEE and edited by I.D. Robertson. It discussed several biasing techniques...

I think there is no one solution.
40% (1/2 octave) is large BW but not so large to be difficult to bias.
I may suggest the following solutions:
1) Have you tried to use (Highest Zo lambda/4) followed by a radial stub?
2) have you tryed a cascade of inductors? (rising valuse from RF to DC), it's very important that Inductors have extremally low shunt capacitance.
Have a good job.

at all, AC and DC must be separated , so we can place 1/4 waves to isolate the RF signal, and power edge placed two capacitors which one is smaller (for high freq.) and the other is higher (for low freq.) they avoid noise come from the power supply.

sergio,
I'v applied first solution.Does 1/4 lambda refer to up-freqency of needed band or center-freqency?

chinaing,
These components are in my circuit.

Hi STD,

I believe what sergio means is that the quarter lambda line calculated by your centre freq where it is a high impedance line followed by a low impedance radial stub.

is it possible to use high Z quarter lambda line shorted with capacitor

Generally speaking 1/4Lambda should be calculated at geometrical center freq. But in pratical cases, often, other parameters should be taken in account.
My opinion is that you can move a little away from the nominal lenght of 1/4 Lambda to maximize performances in terms of gain flatness, Return Loss, Noise Figure or Power, etc. and last but not least, ... pay particular cure to the PCB artwork. Let be the PCB artwork in a similar way that Miteq (or similar factories) do.

Note:
A good guide to learn how to produce useful PCB artwork may be found on pictured advertising (Microwave Journal , Microwave & RF) rather than text books.

If the bandwidth that you are looking for is from DC to 10 GHz, you can refer to the the paper attached in http://www.elektroda.pl/eboard/ftopic65524.html

The design demostrated is using lumped element to replace radial stub. The insertion loss is reasonably low as the author uses FR4 in the design.

I have made PCB using the proposed method. It works well :P