The AP-80 An Alternative Pixie


The Pixie is about the simplest rig ever devised, using just 2 transistors and a LM386 audio amp. Unfortunetly, it's also one of the poorest performing rigs ever devised. It's deficentcies are too numberus to list.

So, I decided to try and improve the design, but that was hopeless. Instead, I ended up designing a real rig. When you look at the schematic, you might think the circuit is pretty complicated, but in fact its very straight forward. It uses inexpensive and reasonably easy to get parts and can be built small enough to fit in the QRPer's favorate enclosure, an Altiods tin.

The circuit starts with a standard Collpits crystal oscillator. In order to provide a 600 Hz frequency shift, a trimmer cap and 2N7000 FET is added in series with the crystal. During transmit, the 2N7000 is turned on, effectively shorting out the trimmer cap and lowering the oscillator frequency. During receive, the '7000 is turned off, allowing the trimmer cap to raise the frequency. The trimmer cap is adjusted to give a shift of about 600 Hz.

The only way to fix the deficiencies of the Pixie receiver was to use a real mixer. I chose a cascode j-fet mixer, as described in the book "Experimental Methods in RF DESIGN" by W7ZOI, KK7B and W7PUA on page 5.12. This mixer is shown as an IF mixer, but is easily converted to use as a Direct Conversion mixer by adding a couple of electrolytic caps for audio by-passing. This mixer is reasonably simple and inexpensive to make, and has conversion gain. Unfortunetly, it only has modest isolation between the LO input and RF input, so some LO signal leaks through. So, a jfet pre-amp was added to reduce LO leakage at the antenna terminal and add a little gain. Most of the receiver gain is still supplied by a LM386 audio amp, but with out the gain bootstrap cap across pins 1+8. Using the gain boost cap increases the noise of the amplifier and degrades the signal to noise ratio. Weaker signals are easier to hear with the lower noise level, despite the lower gain. The receiver has a MSD of about 1 uV. While this is not stellar sensitivity, its a vast improvement over the Pixie and suitable for 80M, which has a relatively high background noise level anyway. The volume control shown on the schematic is actually optional. For the most part, you'll probably run it full open anyway. It is nice to have just in case a strong station does come on frequency though.

A classic series tuned C/L with diode limiting is used to couple between the transmitters low pass filter and receiver input. A capacitor divider network couples the lower impedance input from the QSK network to the high impedance of the receiver input tuned circuit.

Several inexpensive 2N7000 mosfets are used for the other QSK switching functions. One is used to simply short the input to the audio amp to ground for audio muting. The R/C network on the gate input is used to eliminate clicks and thumps in the headphones when keying. The diode across the resistor makes the fet turn on right away when the key is closed, but allows for the R/C delay on keyup. Without this delay, there is a serious click in the headphones on keyup.

The transmitter consists of a 2N2222A (or a 2N3904) buffer amp and a 2N7000 final. With a 12 volt supply, power out is about 1 watt. The emitter leg of the buffer amp is keyed, rather than the switching on and off the collector voltage. Simply removing voltage from the collector allowed too much signal to leak through the transistor and is actually enough to make the PA put out about 10 mw. There's quite an output swing to the crystal oscillator, as the mixer needs about a 5 volt p-p signal. Keying the emittor leg solved this leakage problem. The emittor resistor is split into two 100 ohm resistors with a 4.7 uF cap connected at the junction. This slows down the turn on and turn off time of the buffer amp to provide for some waveform shaping and reduce transmitted key clicks.

The diode across the 2N7000 base, keeps the AC signal coupled to the gate via the .01 cap from going negative. This puts a charge on the cap during the negative part of the clycle, which increases the peak positive voltage on the gate so it gets turned on hard. This improves effiecentcy. At 1 watt out, the little plastic TO-92 package bearly gets warm.

Since the 2N7000s used in switching are "active high" devices, a PNP 2N3906 is added to the circuit to key those used in the T/R switching, providing standard "active low" keying.

The "Simple Keyer Chip" (SKC) I sell would make a nice addition to the rig. It not only adds keyer functions, but it would simplify the circuit somewhat. The SKC would provide side tone and since it has sequenced audio mute and transmitter keying outputs, Q6, R11, R15 and C24 can be eliminated, along with R21, C33 and D7. The gates of Q4 and Q7 would be connected to the Audio Mute port and the gate of Q5 to the transmitter keying port.



The rig can be built in a variety of ways. Here's a photo of the rig as it was being developed using real ugly, dead bug construction.

Of course, now that we know what the final circuit looks like, we can build it in a similar fashion, but doing a much neater job. Instead, I went directly to a printed circuit layout, which looks quite abit better. The board is 3" x 2", making it small enough to fit into the qrp'ers enclosure of choice, the Altoids tin.

Here's my pcb version, which is no longer accurate, as the circuit has since been improved by adding the receiver preamp, 2N7000 final and better keying. To the right is a version built by Denny, N9JXY. Nice job Denny, and he's even made contacts with it!


The most critical adjustment is the receiver offset. This is controlled by the C25 trimmer. This adjustment can be done with a frequency counter or with a 80M receiver. With the PA disabled by removing or not installing the V+ jumper to the left of T1, key the rig. Measure the frequency with a counter or find the signal in your receiver. (about 3.57800 MHz). Note the frequency and add 600 Hz to it. Unkey the rig and adjust C25 to produce the Tx freq + 600 Hz.

An antenna can now be hooked up and the Rx input trimmer adjusted for best signal. (band noise if no one is on the air at the moment).

The spacing of the turns on L2 can be adjusted to give the best power out. Making all the turns "scrunched up" instead of evenly spaced around the core generally increases the power out. You'd be surprised at how much difference this can make. L1 can also be fiddled with, but most likely will have less effect than L2.

Using the rig.

There's not much to do here. The best bet with a crystal controlled DC rig is to call CQ and hope someone answers. Since the receiver is so broad banded and hears both side bands, answering a CQ may not result in a response, as while you can hear them, they may not be tuned to your transmit frequency. Only one way to find out though, call them and see if they come back.

Direct Conversion rigs are best battery powered. AC supplies can cause hum and other problems. Hum can still be an issue when battery powered, so the rig should be mounted in a metal box. Interference from AM broadcast stations can be a problem and not much can be done about that. Least not with this rig design. Since there is no selectivity to the receiver, you'll hear everything on the air +/- 10 kHz. This can make the band seem really busy, when in fact there are only a few on. All I can say is good luck. Making contacts with a rig like this is a challenge, but makes you feel good when you do.

Parts list:

Except for the toroid cores, which you can get from Diz at, all the parts are available from Mouser Electronics. Sorry I didn't look up all the catalog part numbers for you. Before you order a few resistors, a couple of caps, one or two diodes and transistors, take a look at where the quantity price break is. Its generally worth ordering a bag of 100 resistors, or 200 disk caps. Transistors and diodes in quantities of 25. While all this adds up, the price per part will be less and you'll have stock for next time you want to build something. In the long run, you'll save on shipping and handling, which is a fair fraction of the price you pay.

RESISTORS ALL 5% 1/4W Transistors
R20 10 OHMS (1) Q1/3 PN2222A 2N3904,ect
R1/6/10/13/19 100 OHMS (5)
R2/8/12 1K (3) Q2/4/5/6 2N7000 mosfet
R18 2.2K (1) Q8/9/10 2N3819 J-310, mpf102
R14 4.7K (1) Q6 2N3906 PNP
R5/7/9/15 10K (4)
R3/4/11/17/22 22K (5)
R16/21/23 1 MEGA (3)
D1-5/7 1N4148 (5)
Capacitors Value D6 1N5817 1A shottky
C9/10/11/12/15/25 100 pF NPO (6) Inductors
C1/16 220p (2)
C5 470p L1 T37-2 23 turns
C7 680p L2 T37-2 19 turns
C22/26 1000p (2)
C6 1500p L3/4 18 uHy RFC choke
C8 0.01uF T1 T37-43 8 bifiller
C2/3/4/17/18/19 0.1uF
C24/27/28/31/32/33 0.1uF U1 LM386 8pin dip
C14 4.7 uF /16V X1 3.579545 HC49U
C20/29 10 uF/16V V1 50K audio pot
C30 22 uF /16V
C23 100 uF/16V