Project: 7.023Mhz PIXIE_V3 QRP CW Tranceiver Kit

After I stupidly decided to push the limits on the PIXIE_4.1 and subsequently destroyed its amplification, I wasn’t too concerned. Being the prepared sort, when it comes to low-cost kits, I sometimes buy two in anticipation that maybe some parts are missing. In the case of the first PIXIE_4.1 kit, there wasn’t any missing parts which was a delightful result.

I grabbed the second kit from the same order, intending to build it. All of the bagged components were identical, except the PCB was different.


This wouldn’t have been another post if it were not for the fact that the order was shipped with the older version of the board without the beeper/buzzer functionality, and with a different power/antenna footprint configuration. Because it didn’t meet the item description, I’ve contacted the seller to see if there’s anything that can be done – but regardless, I went on building it anyway.

The Kit


Much of what I said about the PIXIE_4.1 will apply to this unit as well, because the schematic is identical with the exception of the buzzer which is not accommodated on this version. We can see that this board has only light scuff marks and is in good condition. It has tinned pads, with solder mask and silkscreen printing. The size is pretty much identical, and vertical component mounting is used to save space as much as possible. It is a double-sided board with plated-through vias.

One significant difference is that the PIXIE_V3 has dual footprints for power and antenna connections, allowing the use of the “low cost” JST headers, or more appropriate barrel jack for power and BNC connection for antenna. As my kit had the PIXIE_4.1 components included, I had the JST headers (and a buzzer with no place to mount it). Because of the larger footprint of these connectors, a large amount of board space is dedicated to them, thus the other components are pushed out very close to the edge of the board.

The PIXIE_V3 also doesn’t have a lad for grounding the can of the crystal – I suspect it’s not a vital thing, but it’s always interesting to note the differences.


As with the other board, the underside only has one trace connection, and otherwise is just completely ground plane. However, there is one major difference with the boards and that is that the pads on this board are not well optimized for hand soldering. The pads, especially for the ground plane, are directly connected to the plane with no “bridges” to isolate them thermally. As a result, it works as a heatsink, sucking away the heat from the soldering iron and making it challenging to get the pad hot-enough for good solder adhesion. Further to that, the solder mask layer comes very close to the pads, and the pad area isn’t particularly large, resulting in difficulties in transferring heat to the pad.

So even though all the components are through-hole, and the holes are large and easy to populate with components, soldering the PIXIE_V3 was much less enjoyable than the PIXIE_4.1. In fact, it could be a challenge for many lower powered irons. To try and overcome this, I turned up the temperature of my iron slightly, and this resulted in more flux spattering and flow-through of solder to the other side. It still required a deal of patience to heat up the plane-connected pads. It doesn’t look as good, but at least, the construction finished without any major hassles.


At least the solder managed to adhere to all the pads.


Testing Again

As it turns out, the kit worked first time, and the performance was practically identical to the first, from what I could tell. Instead, I decided to get a second opinion as to the output power and purity from a Rigol DS1102E 100Mhz digital storage oscilloscope (which I now own). The probe was set to 10x and the compensation was tuned in just prior to analysis.

idle-12vWhen supplied with 12v and just idling unkeyed, the unit is putting out 138mV (rms) into the 51 ohm dummy load, or a power of about 0.37mW.

The purity of the oscillation seems fairly clean, although it’s probably important to note that the FFT is quite limited in bin resolution on this oscilloscope and is in linear scale.

12v-txKeyed up at 12v, the wave isn’t quite perfectly sinusoidal, but fairly close. The unit achieved 5.37v (rms) into 51 ohms or about 565mW of transmit power.

The FFT does show some evidence of higher order harmonics, although this is hard to resolve on a linear scale.


12v-tx-fftdbUsing a decibel scale, we can see the second harmonic peak being about 20dB down, and the third harmonic peak about 30dB down. The rest of the harmonics are into the noise, mainly as this oscilloscope doesn’t have the necessary dynamic range (it’s only 8-bit from memory).


9v-txBack to the linear scale, and dropping the input voltage to 9v, the unit managed 4.27v (rms) into 51 ohms, so about 357mW of transmit power.

Of note is that it looks less sinusoidal compared to the 12v input, and appears to be because the harmonics appear to remain the same in power, but the fundamental is smaller.

5v-txIf we continue outside the intended operating range and drop to 5v (USB) levels of power, we see the wave shape gets even worse, and the harmonics appear to be a larger fraction of the power. The output is 2.13v (rms) into 51 ohms, or about 89mW. As a result, you can see that the transmit power varies significantly as a function of input voltage – especially important for battery powered transmitters.

3v-txBeing even more “evil”, I dropped the voltage to 3v, or the voltage you get from two alkaline AA cells. In this case, the wave has a very asymmetric shape, and the harmonic is now close to half the amplitude of the fundamental. The total output is 0.996V (rms) into 51 ohms, or 19mW, but a good fraction of that would be wasted in the harmonics and not the intended signal at the fundamental.

2v-txPushing the input down to just 2v, at the point where any discernible oscillation begins to form, we can see that the wave retains its “ugly” shape.

The output is 239mV (rms) into 51 ohms, or about 1.12mW. A very small amount of power indeed.


14-5v-txSo what if you want more power? Well, I followed my own advice in the previous post where I killed the last unit by turning it up to 14.5 volts.

Its output was 6.29v (rms) into 51 ohms or approximately 775mW. A little more than three-quarters of a watt for under AU$5 is pretty cool. I’m not sure it will be happy operating at 100% duty though.

Regardless, testing with the Rigol DS1102E was a good exercise in verifying the results obtained for the PIXIE_4.1 from the PicoScope 2205A. As it turns out, the two kits perform practically identically, and the same sort of observations are made in terms of harmonic levels. Very good to see consistency in the results, and first-time-working construction.


I mentioned in the previous posting that there were variants, but I didn’t know at the time that due to some mispacking at the dispatch centre resulted in me getting an older variant of the same design. When building it, I encountered significant frustration due to the small pad size, and lack of optimized pad shape for soldering, resulting in difficulty in soldering to the plane pads. While the provision for proper sockets is nice, it did cause the layout to get closer to the edge and slightly more haphazard. Sadly, a proper barrel jack and BNC connector were not included in this package, but at least no components were missing. I’d definitely say, the upgraded PIXIE_4.1 to be the much more enjoyable build. Both boards performed practically identically.

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Electronics, Radio and tagged , , , , , . Bookmark the permalink.

Error: Comment is Missing!