I'm preparing to build my 40m version of Roger G3XBM's XBM80-2 QRPP transceiver into a little box but before I start I have one thing more to do. I have to design an output filter for the transmitter.
Roger didn't bother. Comments from him suggest that he was aiming for a low parts count and assuming that any filtering would be provided by an external ATU. However Alan VK2ZAY comments that the output was more like a square wave. The second harmonic of my 7.030MHz signal would be on 14.060MHz, also a QRP frequency and I can't be sure that my antenna won't radiate this too. So I think an output filter is a good idea.
Alan used a two-stage pi network, each stage consisting of two 820pF capacitors and a 2.2uF inductor. I built this up on my breadboard, terminated the end with a 50 ohm load (actually 47 ohms, the nearest resistor value I had) and checked it with my SWR analyzer. The picture above shows the result.
Out of interest I also checked the response of a single stage of filtering using the same values, shown in the next picture. You can see that the cut-off frequency is a lot shallower than when two stages of filtering are used, so clearly attenuation of harmonics would not be as effective.
My problem is that I need a filter for 40m not 80m. I tried searching the web to see if I could find some online design tools, and I did here, here and here. However when I fed the design parameters for 80m into the calculators the results they came up with were a lot different from each other and also from the values VK2ZAY used.
In the end I decided to use trial and error. I figured that since the frequency is higher the values I would need to use would be lower. The next lower value inductor I had in my parts box was 1uH, so I don't have a lot of choice in the matter. I substituted that for the 2.2uH. The result shown by the SWR analyzer was promising, but there was a big hump below the cut-off frequency and the SWR at 7.030MHz was a bit on the high side. So I then tried lower values of capacitors. The lower I went, the shallower the cut-off but the lower the hump as well.
In the end, I settled on 680pF capacitors with the 1uH inductor. There is still quite a steep hump below the cut-off frequency but I'm only concerned with the performance at 7.030MHz where the SWR is 1.2:1, which I don't think is too bad. The cut-off curve of this one stage filter seems as steep as the two-stage one VK2ZAY used on 80m so I think one stage of filtering will do.
After I'd written the above, the thought occurred to me to try the low pass filter in the circuit while it was on the breadboard. I'm glad I did, because I'd have been disappointed to find after soldering everything into place that the output power had fallen from 100mW to 25mW. After a bit of experimentation, it appears that connecting the low pass filter directly to the collector of the first transistor as VK2ZAY did in his version of the transceiver loads the circuit and reduces both transmit power and receive sensitivity. The solution seems to be to couple the transistor to the filter using a small value capacitor - I found 180pF gave the best results. This removes the need for the DC blocking capacitor on the output. The power is still down quite a lot, but part of this may be due to all the harmonic energy which is now not making it through to the power meter.