GaN Amplifier Microstrip Impedance Matching and Smith Chart Analysis Question

[nealkendrick]

For this question, please reference the datasheet for the Nitronex NPTB00025. Freq of interest is 3 GHz.

**broken link removed**

On page 4 of the datasheet, Nitronex provides the geometries of the microstrip impedance mathing network. Using AWR's txline program, I calculated that the 265 mil x 441 mil trace is 18.2 ohms with an electrical length of 71 degrees. The impedance that should be presented to the output is 7.4 - j5.8 (see page 2 of datasheet).

This does not appear to be a quarter-wave transformer, being that the electrical length is not 90 degrees -- 600 mil would be about 90 degrees. And my understanding is that you need to get to the real axis on the Smith chart before implementing a QWT anyway.

So what is the effect of this 265x441 mil trace on the Smith Chart? Note that there is a similar trace implemented on the input side of the GaN amp.

 

[Element_115]

For this question, please reference the datasheet for the Nitronex NPTB00025. Freq of interest is 3 GHz.

**broken link removed**

On page 4 of the datasheet, Nitronex provides the geometries of the microstrip impedance mathing network. Using AWR's txline program, I calculated that the 265 mil x 441 mil trace is 18.2 ohms with an electrical length of 71 degrees. The impedance that should be presented to the output is 7.4 - j5.8 (see page 2 of datasheet).

This does not appear to be a quarter-wave transformer, being that the electrical length is not 90 degrees -- 600 mil would be about 90 degrees. And my understanding is that you need to get to the real axis on the Smith chart before implementing a QWT anyway.

So what is the effect of this 265x441 mil trace on the Smith Chart? Note that there is a similar trace implemented on the input side of the GaN amp.

To get similar values you must make sure you have the same substrate info

"Taconic RF35, t=30mil, er=3.5"

The 265x441 mil trace is to transform the Low impedance O/P to 50 Ohms, It could have been optimized by trial and error.
Note there is a DC feed trace that invalidates the QWT scenario.
Cheers

 

[tony_lth]

ou can't ignore the another trace at the output port, which is equal to shunting to GND.