[SOLVED] RF Power Amplifier Tuning Tips for Low Gain and Gain Flatness

[mistyme]

To our RF experts,

Kindly share your experiences on this: I did a MWO simulation for a 20W UHF broadband amplifier (470MHz to 900MHz). The simulated linear gain is around 17 to 19dB. Available test eval board of the transistor in standard Push-pull class AB config has a gain of around the same range.

After verifying my prototype design (balanced class AB), the gain flatness is really in a mess and the small signal gain dips reaches as low as 3dB.

I attached the plots. The red plot is the test eval board in Push-pull Class AB and the blue plot is for Balanced Class AB.

Any additional insights on solving the gain and the flatness issues?

 

[maniana]

Hi, you didn't show your board. Transistor type would also help.
But I can tell you some issues I had in the past with my designs.
I had some dips (although not that many) in the gain when there was a parasitic resonance.
This can come from many sources. Your decoupling capacitors, board design etc...
With some transistor packages you have to experiment with the leads and board layout, especially at the gate side.
I had a design at 2GHz where transistor package had 3 leads at the gate side. The outside leads were not internally connected, but still grounding them caused parasitic resonances and dips in the gain. Leaving them floating solved the problem.

 

[mistyme]

hi mariana,

thanks for the response.
the transistor is an LDMOS plastic package with 2 die transistors inside. their common source are internally connected to its ground flange.

 

[maniana]

You still didn't tell the transistor type/package. look for example at package type sot1244B from NXP. The two narrow leads on the gate side are not connocted but can generate resonances depending on your board layout. In your case those dips can be caused by gate side matching and resonances very likely also in the biasing/decoupling.

 

[mistyme]

please see the picture below for the package


another thing i notice is with the amplifier DC turned-on (no RF input) @ Idq=400mA, when i connect a 50-ohm RF cable (hanging, no termination) at the output SMP port, the Idq increases to around 1040mA (cable length=320mm). At 270mm cable length, current increase to 1800mA. But at longer rf cables (370mm,400mm), the Idq return to 400mA.

 

[maniana]

Are you using your amplifier without the load?
Then it is unstable.

 

[biff44]

that funny gain glitch at 300 MHz might be an oscillation. When it is running, carefully use a spectrum analyzer to see if there are any spurious oscilations going on. Remember that a spurious oscillation in a 20W amp might actually be oscillating at up to 40 watts or so, so use pads!

 

[mistyme]

that funny gain glitch at 300 MHz might be an oscillation. When it is running, carefully use a spectrum analyzer to see if there are any spurious oscilations going on. Remember that a spurious oscillation in a 20W amp might actually be oscillating at up to 40 watts or so, so use pads!

hi biff44,

i could not see any spur or oscillation in the spectrum analyzer even at -80 noise floor.
however on my setup of small signal S-parameter in vector analyzer, mu seems to be <1 on some frequencies which should indicate some instability, right?

in that sense, any tips or suggestions how to improve the mu or stability? i already have 100-ohm in my gate and seems like feedback gain reduction of series resistor/cap is not an option for my power class. i will look into the impedance response of my drain bias feed.