To obtain the best coupling it is necessary to adjust the impedances, adjust first C1 and C2 to obtain the maximum output signal on 50 Ohm load, then also C3 and C4 are set to obtain the best output coupling.

The output power should be read with a 50 Ohm input power meter, even only an anti-inductive resistance followed by a capacitor diode circuit and the tester may be sufficient to evaluate the power increments. Even with a 250Mhz oscilloscope it is possible to read the output signal, clearly the ideal instrument is the Spectrum Analyzer with at least 1000Mhz of band.

The filter composed of L2, L3 and C6 has the task of reducing the harmonics output, as low pass will let only the fundamental pass up to 110Mhz the higher frequencies are attenuated a lot.

To increase by about another 10dB the output power, equal to almost 1W, you can pair another equal stage, in cascade of this maybe powering it to 12V, in this case there is a greater heat sink and increase the regulation of the bias. The low pass filter eliminates harmonics and must only be output before the antenna.

This is the measurement of the output power of the LX359, we have seen that taking the signal directly from the collector of 2N2222 you get a higher output power, you see the two compensators for impedance matching, the value obtained with new 9V battery is equal to 11.22dBm which corresponds to 13.2mW. The output from point B even with the best adaptation provides us no better than 7.8dBm. I remember that the measure of power includes the fundamental as well as the energy of all subsequent harmonics. A good low-pass filter will eliminate a large part of the harmonics, allowing a more correct measurement of the fundamental.

This is the signal of the LX359 applied to the spectrum analyzer, the greater output power is obtained from the lower part of the output frequencies, ie at 80-87Mhz we get + 3dB compared to a 108Mhz calibration. You can see the subsequent harmonics that should be attenuated with the use of the low pass filter at the output.

The output of the LX359 with the SMA connectors, the output is taken directly from the case of the metallic 2N2222A.
Note the 2N3866 transistor mounted with the brass heatsink, you need to keep it a few mm away from the metal bottom to reduce capacity to ground. The low pass filter is made with 2 identical coils wound with 4 turns of 1mm wire on 6mm diameter.

The power output after optimizing the couplings, reaches 100mW, the Bias trimmer must be adjusted to have the minimum current and maximum output power by adjusting multiple couplings and Bias. In my case the current absorbed with 9V is between 35 and 50mA I advise not to increase the current as it could create instability in the transistor.

The tests with a 1/4 wave antenna gave excellent results and good quality of the voice heard well in a room even 4 meters away from the microphone.

By integrating the final stage close to the LX359 the reduced dimensions can be maintained by using small components, moreover, once the couplings have been optimized, the compensating trimmer can be replaced by fixed SMD capacitors. For a good realization, the bug, the linear stage and the low pass filter, should be inserted in a metal box and separated from each other with a screen between each stage.
With 9V battery the autonomy is reduced to 4-6 hours, if you want to have more autonomy, you need to use a good 9V power supply from the mains, or switch to a 12V battery but must be revised and adapted for a voltage highest.

Final version, mounted inside the plastic box, collector coupling with fixed divider 22 + 22pF, adjustment of coupling only for output, low pass filter with 0.3mm wire spools on 3mm diam.

In this configuration to the best coupling with 87Mhz we get 280mW output, they are too many, so we go down Bias (blue trimmer) to have only 100mW and 42mA absorption.
With a little 'patience you can fix everything without PCB, so there are no missing capabilities and the operation is very good and stable. With a GP antenna on the balcony you can comfortably reach 3KM, clearly on a frequency not to disturb commercial radios, type 86.0Mhz.