Two-Watt Driver Amplifier


Since my final amplifier needs at least 1 W input, I looked around for some options on a driver amplifier and decided on a fairly common broadband approach using bipolar transistors and binocular core transformers at the input and output. The circuit is loosely based on Motorola application note AN779, the famous 20 W amplifier that you can buy as a set of parts from Communications Concepts, Inc. (CCI). I stole the first stage, used some generic CB transistors, and customized the transformers a bit to get “good enough” performance. This was my first attempt at a broadband amplifier of this type...

The first task was to try and optimize the winding of the transformers. For a constant turns ratio of 2:1, which is better: 2, 4, or 6 turns on the primary? Using 2843000202 binocular cores and 28 ga wire, I made three test articles and connected them via SMA connectors to my spectrum analyzer and tracking generator. Here are the frequency responses:

A Two-Watt Wideband Driver Amplifier



    16 dB max, down 2 dB at 3

    and 30 MHz

Max Power at 7 MHz:

   +34.7 dBm, 3.0 W

Input return loss:

   >11 dB, 1 to 30 MHz

(SWR < 1.2:1)

Distortion at 2 W output:

  HD2  -48 dBc

  HD3 -25 dBc


Full power at 7 MHz is about +34.7 dBm (2.95 W) when higher-order harmonics begin to jump up. At 2 W, HD2 = -48 dBc, HD3 = -25 dBc. Here is a typical distortion spectrum.

I never did get a current draw measurement. It runs off of 12 V and the PC board was fabricated by the Sharpie method. A full ground plane is used on the component side. There’s an overkill heatsink, about 2 x 3 inches and 0.5 inches thick and is the same size as the board. It runs comfy-cool and is totally stable and blowup-proof.

Here is the final frequency response, which is reasonably flat though not spectacular. Considering how un-flat the power amp turned out to be, this is more than sufficient.


With too few turns, there is insufficient inductance and low frequency response suffers. With too many turns, a premature high-frequency rolloff occurs. The optimum appears to be the 4:2 turns ratio, so I went with that for the input transformer. It was flat within 0.3 dB from 3-30 MHz. The output transformer was a 4:3 ratio, plus a single turn for the feedback network.

Since the specified transistors were not on hand, I used some fairly generic low-cost devices (2SC2075), which meant that the input matching and feedback parameters from AN779 were invalid. So the tweaking began, attempting to flatten the response while maintaining a good input and output match. Having the tracking generator and analyzer, and a return loss bridge, were incredibly valuable.