The Pixie is a tiny, direct conversion CW transceiver using just a handful of readily available parts. Most people should be able to build one with parts in their junk box .
Pixie kits for 40m are available from China at incredibly low costs.
The Pixie was derived from an earlier circuit, the Micro 80 by Oleg Borodin RV3GM which used just 4 transistors. The Pixie circuit consists of a crystal or ceramic resonator oscillator feeding a PA. Depending on the PA device chosen, RF power outputs of between 100mW and 500mW may be achieved. The only components which are band critical are those in the TX PA output, so converting a version from 160m to 10m takes not more than a minute! On receive, the TX PA is used as a mixer which feeds the LM386 audio amplifier.
My version of the Micro 80 was built into a Maplin's aluminium project box. The 2 switches are on/off and TX/RX which switches the VXO between one of two frequencies spaced about 800Hz apart. This sort of gives a couple of working frequencies. The whole transceiver is powered by a single PP3 9V battery towards the rear of the enclosure.
The antenna connector was the BNC socket to the left on the back of the rig. It has managed a few QSOs on 80m into a quarter wave tuned against ground. The whole circuit is built "Manhatten style" on a small piece of copper clad board with small cut-out PCB pads to fix components to. This is a simple technique well suited to one-offs and much simpler than making a PCB. You can buy "MeSquares" to make quite neat prototypes.
A 40m kit from China was bought and built. Local QSOs were easy.
Drawbacks and modifications
This most simple of arrangements is less than perfect and results are as would be expected with this level of "non" complexity. Usual problems are the lack of decent RF and AF selectivity on receive - only the TX low pass filter protects the receiver input - as well as RF broadcast station breakthrough and mains hum pick-up. Both may be reduced by putting the unit in a screened box and powering the unit from its own battery supply as is shown in the photo. Another issue is the lack of RIT so that the transmitted signal is almost coincidental with the received signal i.e. practically at zero beat. Modifications have been published to add RIT, to add side-tone, increase the oscillator tuning range and to reduce BC interference. All these add increasing levels of complexity that rather detract from the sheer simplicity of this design. I have even seen a Pixie modified for DSB by adding an audio amp and balanced mixer!
What results can be achieved?
People have achieved some remarkable contacts with these little rigs. It is not uncommon to hear of people working many countries on both 80m and 40m. My own lash-up version, which used 2 transistors rather than the LM386 for the RX audio worked first time even without a box or morse key: I finished the rig and had it on the bench and was so keen to try it that I called CQ by tapping a wire to the copper ground plane instead of a key. Much to my amazement and surprise someone answered my call - a G at nearly 300kms distant - so a contact was made in this rather unconventional manner! My first attempt never got as far as being put in a box although a later version did (see above).
On the higher HF bands like 15m or 10m the Pixie is certainly capable of working Sporadic-E DX. Patience and acceptance of the limitations of something this simple are essential, but building and using such a rig is definitely rewarding.