24 May, 2011

QRPp: Ultra low power operation with the Pixie 2

QRPp is radio amateur jargon for communication using a transmitter with less than 1 Watt power output. That’s less than a handheld GSM mobile phone (max wireless range ~35 km) or a flash light. My Pixie 2 transceiver has so far contacted 15 different countries on 3.5 MHz CW. The first version from 2002 was made in an Altoids tin (picture later), but in 2010 I repacked it in one of the nice clear top tins from the 4SQRP (picture below).

My repackaged Pixie 2

It is based on the KnightSMiTe which is a surface mount version of the Pixie 2. Like the original design it consists of only two transistors and one IC. I have made the Pixie in two different versions:
  • An 80 m Pixie 2 Plus with only a few additional passive components. It has variable frequency, much more audio gain, CW audio filtering, and an improved muting scheme.
  • A Pixie 2 Plus with a sidetone circuit which is the one pictured here 
The various additions and modifications are:
My original Pixie 2 in Altoids tin 
  • Frequency agility to cover the entire 80 m band. The pictures show a version where I can choose between three different ceramic resonators: 3.58, 3.69, and 3.84 MHz and use a variable capacitor for tuning. Combined this gives coverage from 3495 to 3830 kHz.   
  • The LM386 has been boosted from 46 dB to more than 70 dB gain and a bandwidth of a few hundred Hz with no additional active components. It will directly drive a loudspeaker.   
  • Reduced broadcast interference due to an improved scheme for muting the LM386 which is just as simple as the original one. 
  • Single-transistor sidetone oscillator on an extra surface mount board.   
The receiver has a sensitivity of around 100 microVolts and draws about 10 mA @ 9V. It is very wide and there is no discrimination between upper and lower sidebands. If you tune to the incorrect sideband, the other station won't hear you as you will be transmitting on the wrong frequency. The sidetone is about 1.1 kHz.

Some data for receiver frequency vs resonator frequency as the variable capacitor is changed, transmitter frequency offset, current draw and output power:

The first contact was made with LA3NCA, Kjetil, on 15 April 2002. On the third attempt I was able to tune my frequency so that he could hear my response, resulting in a 569 report over a distance of 30 km (!). I thought that was about as much as I could get out of a few hundred milliWatts, but I was lucky one weekend in the spring and stumbled upon a contest in the 80 m band. To my astonishment I have now contacted many other countries with it: Sweden (SM5 and SM6), Denmark (OZ), Finland (OH1), several German stations (DL), Estonia (ES), Latvia (YL), Lithuania (LY), Kaliningrad Russia (RV2), Poland (3Z), Czech Republic (OK), Scotland (GM), England (G), Belgium (ON), the Netherlands (PA). My record is TM5Y on the Island of Yeu, south of Brittany in France – almost 1700 km and with 0.3 Watts, this is 3473 miles per Watt.

The key to obtaining many contacts with QRPp equipment, in my experience, is not higher power, but the ability to tune. The usual mode of operation is to answer a call, and being locked to a single crystal’s frequency severely limits the number of contacts.

I have also tested a 30 m version of the Pixie 2, built Manhattan style. It has an input tuned circuit just like KX7L’s 20 m Pixie 2 in order to reduce broadcast interference. It can tune all over the 30 meter band since the oscillator uses a 10.2 MHz ceramic resonator pulled with a series capacitance. I have not been able to get its frequency stable enough that I dare use it on air. This is unfortunately in line with GM0RWU’s recommendation not to use ceramic resonators above 6 MHz due to thermal stability problems.

Main Pixie resource: The Pixie File, (from the pages of SPRAT - The Journal of the G-QRP Club).

My Pixie 2 modifications have also been published in SPRAT:
My Pixie2 with improved muting, more audio gain and sidetone seems to have inspired others:
This text was originally written in 2002 on my web page and has been updated several times since then. This blog post was taken from that page in order to enable discussions relating to the modified Pixie2. The first comments are all based on emails I have received over the years with questions and comments,  therefore you will see me as the author of these comments.


  1. Comment: An interesting point is, that Peter, VK1PK have used two crystals in an Colpitts-Oscillator with the same frequency which obtains a considerable swing. (perhaps 27 kHz on 7 MHz by using HC18/U crystals). I have used this oscillator-circuit to combine it with the rest of the Pixie-trx.

  2. A: At 3.5 MHz, a ceramic resonator as I used, gives even more frequency range than the two crystal super-VXO circuit. But I don't think you can find ceramic resonators for 7 MHz, and also the oscillator may become too unstable at that frequency with ceramic resonators

    1. I have sfe7.02mu 7.02 resonators weak_signals@yahoo.ca
      Swing 7.00-7.04 ish
      These are rare CRs

  3. Q: I have completed the sidetone oscillator from your webpage and it works well according to my scope. A very nice, clean, sinusoidal wave, adjustable in amplitude from the 10K pot. I am ready to install the sidetone oscillator on the Pixie 2.

    When I remove the 1N914 diode to pin 6 of the LM386, do I replace the diode with a solid piece of wire or just leave it out???

  4. A: When you remove the diode, you have to replace it with a wire, as pin 6 is the voltage supply for the LM386 chip.

    Good luck with your Pixie!

  5. Q: I was looking at your Pixie2 Plus sidetone schematic. I like the fact that you can tune your pixie over a wide range of frequencies.

    I want to put a pixie on our SSB net frequencies of 3.920 and 3.933 MHz. I have been to Mouser.com and I cannot seem to find a ceramic resonator other than 4.0 MHz.

    Two questions:

    - What resonator did you use in your Pixie and where did you get it?

    - Will a 4.0 MHz resonator tune down to 3.920? What is your opinion?

  6. A: I used three different ceramic resonators: 3.58, 3.69, and 3.84 MHz. I found the 3.58 and even 4 MHz ones at www.elfa.se. But as our 80 m band is only 3.5-3.8 MHz, the 4 MHz one is not so interesting.

    I can find 3.58, 3.68 and 4.0 MHz resonators with two legs at Digikey.com by just entering the frequency in the search field. I cannot recall where I got 3.84 MHz from, but searching mouser.com I find 3.58, 3.68 and even 3.84.

    It is important to use the ones with two legs as they are pullable in frequency, the ones with three legs have built-in capacitors and are hardly pullable at all.

    I may have found some resonators in old computer mice also. 3.58 MHz has to do with the NTSC color-burst frequency, one of the others I seem to recall is related to serial interface data rates.

    I was able to pull the 3.58 MHz resonator down to at least 3.50 MHz, the 3.68 MHz resonator could be pulled from 3.60 - 3.73, and the 3.84 MHz from 3.71 to 3.83 MHz. For 3.92 MHz I guess a 4 MHz resonator should be usable. I don't think a 3.84 can be pulled to 3.92, as it is my experience that it is easier to pull them down in frequency than up.

  7. Q: PIXIE 2+ W/ SIDETONE - FANTASTIC RIG !!!!!!!!!

    I am ham and in need of the technical information about circuit. I love QRPp circuits and need modification of the circuit for 40 meter, 20 meter and 15 meter. If you have the modification.

  8. A: you will find many versions of the Pixie2 for 40 meters in the GQRP Pixie files. There is also a FOXX2 with filter values given for 80 m, 40 m, 30 m and 20 m.

    Most people seem to use them on 80 m and 40 m, I don't recall seeing any Pixie for 15 m and up.

  9. Q: I see that you have operated with a PIXIE on 80M. Do you know of a little QRP rig like the Pixie for 160M?

  10. Q: A Pixie for 160 m should be simple to make. 1.8432 MHz is a common crystal frequency (e.g. Digikey), but it would be better with a ceramic resonator as you can tune it around more. I did that at 3.58 MHz and it should be even more stable at 1.8 MHz.

    However, I searched both Digikey and Murata for anything in the right frequency range and I cannot really find anything. I believe a 2 MHz resonator is too far away in frequency, as a comparison a 4 MHz resonator cannot really be shifted all the way down to 3.5-3.6 MHz.

    If 1843 kHz is fine then it should only be to redesign the output filter with increased C and L values in the pi-filter compared to 80/40 meters, and that's it.

  11. What resistors can I use for my pixie 2, 1/4W or 1/2W...1%, 5%, 10%?

    1. That's not critical at all so the smallest and cheapest are fine: 0.25 W (or less), 10%. Better tolerance is also fine if that's what you have, but it's not worth paying extra for here. Half Watt resistors are of course also fine, but will take up more space and may cost more.

  12. Hi, this is great site! You posts encourage me to build own Pixie. In particular I am interested in building "Pixie 2+ w/

    sidetone" version. However I am bit confused about values of the elements used in you modifications. Capacitors C7, C6, C5 on

    Fig. 5. (http://la3za.blogspot.com/2003/04/using-pin-7-of-lm386-to-reduce-bci-and.html) are 10 nF, but the same capacitors in PDF

    (http://folk.uio.no/sverre/LA3ZA/blog/Pixie2PlusSidetone.pdf) are 0.1 (uF?) = 100 nF. Moreover capacitor connected to pin 2 in

    PDF is 0.01 (uF?) while in "Pixie 2+" PDF (http://folk.uio.no/sverre/LA3ZA/blog/Pixie2Plus.pdf) is 0.1 (uF?).

    Could you please take check and correct if necessary the values of passive elements in "Pixie 2+ w/ sidetone" PDF?
    Is it possible to get your PCB drawing? I don't have problem with drawing simple PCBs for uC project, however here with high freq

    i am affraid that PCB alignment might be critical.

    Thank you for your time

    1. Thanks for the careful reading! I just checked the sidetone circuit. It comes from the second last page of The Sprat Pixie file, and the three capacitors should be 0.01 uF (10 nF). The pdf of the Pixie2 Plus Sidetone has now been corrected.

    2. The PCB for the basic Pixie2 comes from the KnightSMiTe website. I also made an add-on board for the sidetone circuit, but I am afraid that I cannot find the pattern anywhere on my computer. It shouldn't be too hard to make on a perf-board.

  13. Thank you for a quick replay. I definitely try to build your "Pixie 2+ w/ sidetone" version. It overcomes most of the problems reported for a Pixie 2.


  14. Thank you so much for all the help figuring out the sidetone circuit. I will try the audio filtering as well in later versions as I find that to be really nice. Next Pixie will be a scratch build. Easier to make changes in the circuit, not as hard as it is to modify the existing PCB that comes with the kit. 73 DE YO6DXE