Ole allowed us an interesting final stage – OZ2OE - translate and publish on CQ.sk. The original in English can be found at http://hjem.get2net.dk/ole_nykjaer/oz2oe/mrf317/317.html. we thank you!
Amplifier on 144 MHz with built-in source Where are the times, when allmode 2m RIG had 10W and who wanted more, you usually built a 100W linear? When the MRF317 transistor got into my hands (100W/28V) – was decided – build a PA.
The amplifier is built into a metal box with dimensions 240x160x160mm. Most of the space is taken up by a 28V / 7A power supply. There is a large radiator on the rear panel. The amplifier itself is built on a piece of printed circuit board, approximately 140x140mm. It has only a few areas, which I cut into the board with a knife. Base and collector circuits are soldered to them. This is a fast and efficient way, while also having the advantage of doing so, that most of the printed circuit board is ground. A few remaining components above the printed circuit board, as can be seen in the photos.
It's very simple. Input and output circuits were designed according to catalog data, which, however, are for Class C, while the amplifier operates in class AB, so the input circuit was optimized during recovery. Is important, so that capacitor C3 is as close as possible to the transistor and withstands a large RF current.
To make the amplifier linear, it is necessary to apply a direct voltage to the base of the transistor 0,6 – 0,7V. The exact value depends on the collector current measurement. I adjust normally 1/40 to 1/20 maximum collector current at single-tone excitation at maximum power. In my experience, the exact value of the quiescent current is non-critical.
The bias circuit uses a diode as a voltage source. When changing the current through the diode (by changing the value of resistor R1) it is possible to fine-tune the base preload. This very simple involvement does not confirm its bad reputation. The problem occurs, if the bad bias circuit is not able to supply sufficient current during the peak. In this case, the base bias voltage drops and the transistor enters class C, which will cause the deterioration of IMD products and the emergence of splatters.
All you have to do is select the correct diode and sufficient current for the transistor. My rule is primitive, but working: use a diode with a current up to half the collector current of the transistor. So I chose a 3A diode.
Switching RX / TX is done by applying 28V to R1, which opens the transistor. Upon reception, the collector can be under a voltage of 28V, making switching relatively easy.
Preload circuit test:
Testing the bias circuit is not complicated. Measure the voltage at the base (behind the choke). For example, there will be 0.65V. Now get the signal from the driver and check the voltage drop. If it is less than 50mV, It's delicious, a drop of 100mV is acceptable. But a larger decline is unacceptable. Maybe it will happen, that after the introduction of HF, the voltmeter "goes crazy". It becomes a digital device. Not a classic analog, HI.
Increasing the base current requires a larger diode, or two connected in parallel. Set the quiescent current again and repeat the test. As can be seen in the pictures, I used two diodes connected in parallel, until I have achieved satisfactory results.
With 10W from the driver and about 26V, I reached 90W at the output. The transistor allows 28V power supply, while the power should exceed 100W, but I didn't need that. The reason was, that the components in the collector circuit were slightly undersized and I did not want to risk their destruction.