Having fun with a low noise, high power RF oscillator
Today (13-Jan-2013) I built the Low Noise Oscillator, designed by Linley Gumm, K7HFD. The design is presented, amonst other places, in the AARL 2009.
Here is the circuit diagram (image belongs to ARRL 2009):
I didn’t have any problems with the construction, apart for soldering two 2N3906 (PNP) instead of 2N3904 (NPN). Mistake was obvious once I measured the voltage BE on transistors and it was like 7V instead of 0.7V.
Next, I wanted to check if L2 has an effect if it is soldered one way or the other (with the pins reversed). It only works one way.
The output signal was really large (I got at some stage 20dBm, from a supply of 10V), but it was far from a sinewave. First time I didn’t use any of the five ferrite beads from the schematics, just to see how the signal looks like.
It was quite bad in my opinion, because I expect to build as pure sinewave oscillators as I can. The ferrite beads improved slightly the waveform, especially the high frequency ripples visible on the waveform (high frequency).
I built the transformer on the left on a T50-2 core and I varied the number of turns (2 turns and 1 turn). In the end I left them with few turns.
For the output transformer I didn’t have a toroid core T50-2 left, so I decided to make a transformer on a ferrite bead. The ones I have have a high AL. In the end I left 5 turns in the primary and 2 turns in secondary.
The photo below is the oscillator (board on the left, followed by a band-pass filter, described below).
This is a screenshot of the oscilloscope at the output of the filter. Before I made changes to the output transformer and number of turns in the first transformer, it looked much worse than this:
The output, this time as seen by the Spectrum Analyzer:
Since I wanted to get a nice sinewave at the output, I used the free filter calculator (kindly provided by Neil from Almost All Digital Electronics) and I calculated a filter with a center frequency of 7 MHz and a span of 6 MHz. I opted for a 5th order Butterworth filter, for no special reasons (I remember the response looks quite nice and the components values were common). I notice that the Center Frequency is reported as 6.32MHz, while I specified 7MHz. I am not really sure what the reason is. I decided to do a quick and dirty construction, Manhattan-style and to spend as little time as possible to make the coils. So I used some of the binocular cores I have and checked how many turns do I need for obtaining around 2.14uH and 1.93uH. I found that only one turn was enough. Since there were slight differences in the AL of the cores, I built 5 coils and selected some with higher values for 2.14uH and lower values for 1.93uH. The coil which is 597uH I made it on air and at the moment has a slightly larger value.
For capacitors, I used some ceramic 300pF (not very stable with temperature) and some 330pF silver mica. I took a leftover piece of PCB, cleaned it and soldered the components.
This is how the board looks like:
Below is the frequency response. It can be seen that from 70MHz, it behaves badly.
This is another view, for a smaller frequency range. It definitely needs some work done to make it better, but this was the first build, without optimising any of the components. It has 3dB loss at 7MHz, but there is plenty of power at the oscillator output. All I want is above 7dBm.
Finally, here is the output of the oscillator using the bandpass filter:
And here is the spectrogram:
I am pleased that despite a not-so-good band-pass filter, the output is now clean and I am now pleased to watch the oscilloscope display 🙂
I built the PCB using all SMD resistors size 1210 and SMD capacitors size 1210. I used FreePCB to design the PCB and I used Press-n-Peel to etch the board.
I still haven’t learnt how to use properly Press-N-Peel, after I iron it on the board, sometimes there are heaps of pores and then I have to go over the board with permanent markers…