27 - Building and Characterising a RF Two Tone Generator and a Switched Attenuator

 Building and Characterising a RF Two Tone Generator and a Switched Attenuator

I am building a uBitx SSB Transceiver and want to test each module using simple test gear. A NanoVNA can do quite a lot but sometimes you need to apply two RF signals quite close together and see what the output contains on a spectrum analyser – I plan to use my SDRPlay software defined radio receiver and run the RSP Spectrum analyser software available on SDRPlays’ website.

One of the two or three books I consider the bibles of Homebrew is “Building a Transceiver” by Eamon Skelton EI9GQ and Elaine Richards G4LFM available in the RSGB Store. It covers construction of a two-tone generator using two simple crystal oscillators, low pass filters and a simple combiner. I reproduce Eamon’s circuit here but do get the book if constructing it.

You need to build two of these, I built each of mine in an Altoids mint tin using small squares of PCB glued onto a PCB substrate – this type of bread boarding is quicker and easier than making PCBs for small RF projects and gives excellent results as the circuit is above a ground plane. Shielding is important in this circuit and you don’t want extraneous pickup from leakage.

Build two of these. I added a small inductor in series with one of the crystals. Use any similar pair of crystals.

to the Altoid tins used feedthrough capacitors and I used short lengths of Coax to take the two signal generators outputs to the combining circuit.

To test this unit, I fed the output to my SDR and used spectrum analyser software to check both for harmonics and intermodulation. However, the recommended maximum input to the RSP1A is 0 dBm (or up to 10 for short periods) and the output could have been too high so I built an attenuator and, as I would need an adjustable one for some other tests at some stage, I used a design with eight slide switches to allow setting from 1 to 81 dB. 

The circuit is similar to one found in the ARRL Handbook (a quick search of the net found https://www.arrl.org/files/file/Technology/tis/info/pdf/September1982_Attenuator.pdf). Here are some photos of it:

The table below might help you understand harmonics and intermodulation and mixing of two frequencies. (mixing is a special case of intermodulation). 

Assume each frequency generates harmonics, so f1, 2*f1, 3*f1, 4*f1 etc come out - known as the fundamental, the second harmonic, the third harmonic and the 4th harmonic. Repeat for f2. 

Now simply add each term in the first series to each term in the second series. Repeat using subtraction between the same terms. You quickly generate a lot of frequencies!

Luckily the strength of these gets less as the harmonic number goes up. Also, quite often the frequency is too far away to matter too much. 

Usually we focus on the second or third harmonic and the third order intermodulation, ie (2*f1–f2) and (2*f2-f1). Marked in red in the second table below. This is because they are close to the original frequencies f1 and f2. In a receiver, they might well be close enough to be passed through to the loudspeaker and be an interfering signal.

Incredible to think you get 60 different frequencies for just two tones generating up to the sixth harmonic, you sometimes need to consider higher, Bill Meara (of Soldersmoke.com) had to filter out an 8th harmonic lately as it was interfering with an oscillator. These frequencies can be computed by spreadsheet, which I have done for my two tones. The spreadsheet is on my blog.

In the table above note that the 3rd order IMD terms are at 13.021 MHz and 13.063 MHz. The 5th order terms are at 13.077 and 13.007 MHz. The other higher terms that are near the two tones are below the Spectrum Analyser noise floor so I need only worry about the 3rd and 5th. I need to add a bit of attenuation as I am slightly overdriving the SDR – the manufacturer says it can take 10 dBm for short periods and 0 dBm continuously – my tests show it is more linear if only driven by signals less than -10 dBm. An interesting feature of IMD is that if you drop or increase the fundamental signals, the 3rd

If I now insert 10 dB attenuation, the 5th order terms disappear and the third order terms are just under 60 dB below the fundamental tone. They have dropped 20 dB despite the fundamental dropping 10. This is evidence that it is the SDR that is generating some of this IMD – if it was all originating in the two-tone oscillator then everything would just drop 10 dB through the attenuator.

But increasing the attenuation by another 2 dB does not cause the 3rd order to drop 6 dB, so at this stage we are seeing the distortion in the two-tone oscillator itself. We need to keep the input below -10 dB and live with an IMD that is 60 dB below one tone or build a better shielded two tone oscillator and a better combiner (might have been better to put it in its own metal box…)

The fact that the 3rd order did not drop much further with additional attenuation indicates we are actually looking at the distortion of the two-tone generator. SDR third order IMD should drop 3 times faster than the amount the tones are dropping, and as the attenuator is after the tone generator and before the SDR then we are not looking at the IMD of the SDR, once the input gets below -10 dB.

In summary, this low cost test equipment will do the job – provided we work within its capabilities and doublecheck what we are measuring. Attenuators are vital, even the 0 dBm output of a NanoVNA needs to be attenuated to avoid overloading amplifiers and many circuits to be tested. 

By next month I will have tested a three transistor termination insensitive amplifier and a diode double balanced mixer (DDBM). The uBitx uses three of the TIA amps and a pair of DDBMs. 

The earlier articles described using a NanoVNA to test filters and I will apply them to three filters. That will be most of the receiver section tested.

You don’t need to test everything to build a uBitx but I want to

   (a) build a complete uBitx as it was originally designed and 

   (b) start substituting different modules to see if I can improve performance. 

Part of the fun of homebrew!

uBits is described at http://hfsignals.com and the Bitx20 groups.io website (https://groups.io/g/BITX20 ), also see http://ubitx.net/