Phase Noise measurements in VFO and Crystal Oscillators
Document last updated: 02/01/2012.
This document presents some comparative measurements for Phase Noise in oscillators.
At this stage, a small numbers of oscillators have been tested (three VFO and one Crystal Oscillator).
The tests were performed on a FSUP8 Signal Source Analyzer made by Rohde & Schwarz.
One of the main functions of the family of FSUP instruments is Phase Noise measurements.
The power supplied to oscillators came from a NGSM adjustable power supply, also made by Rohde & Schwarz.
In the hindsight, I should have used one the low-noise DC supplies available on FSUP8, which is optimized for low noise and excellent stability.
Next time when I will perform measurements, I will use the integrated source and I will redo the phase noise measurements for some of the oscillators mentioned in here.
For reference purposes, I took measurements of a high quality Signal Generator, renown for its low phase noise (100kHz – 4.3GHz) made by Rohde & Schwarz, namely SMHU58.
It is an older product, already discontinued and replaced by SMA100A. The typical phase noise at RF < 16MHz, offset 20kHz is around 146dBc/Hz
I set it up on 4.26MHz, 7dBm.
Here is the phase noise chart, measurement performed in crosscorelation mode. I decided to set the marker to 10kHz offset, for comparative measurements between different oscillators.
The SMHU58 has a phase noise of -141.51 dBm. If we repeat the measurement, usually the value is within ±1dB. The residual FM is almost 22Hz.
Next, the have the results of the BITX oscillator designed by Ashhar Farhan (http://www.phonestack.com/farhan/bitx.html). Despite its simplicity, it was the best performer of the oscillators that I built at home.
The module I built contains only the first two transistors Q5 and Q6. The level reached 0dBm and the second harmonics is quite high, at -9dBc.
I supplied only 9V supply instead of the recommended 12V simply because I couldn’t remember this detail.
I expect it would not affect significantly the phase noise. Later I will retest the phase noise of this oscillator with 12V supply provided by FSUP8 and I would expect some (slightly) better results.
You can see the result is very comparable (and in fact, even better!) with SMHU58 for a carrier offset from 500kHz to 10kHz, while for offsets less than 10kHz, SMHU is a bit better.
Next phase noise measurement was on an oscillator inspired by the design of Iulian Rosu (http://www.qsl.net/va3iul/Very_Low_Phase_Noise_VFO/Very_Low_Phase_Noise_Vackar_VFO.pdf).
Instead of the recommended transistors BD135, BD136 I used BC337-25 and BC327-25, which are also high current and low fT.
I was puzzled by the very low collector current of one of the transistors and I contacted Iulian Rosu for clarification. He reassured me that the circuit diagram is correct and indeed the very low collector current was purposely chosen.
However, my experiments were performed with R4 with 1kohm instead of original 100ohm. This changed the current through Q2 emitter from virtually zero to a bit over 1mA. Both simulation on LTSpice and real measurements agreed about the current through Q2 before and after modification.
I also decided to modify the output filter and made a few other changes around Q3 in an attempt to increase the output level. Considering all these changes, the design is only based on that of Iulian Rosu.
The results, although not too bad, are shy in comparison with Farhan’s oscillator as phase noise is concerned. Instead, Residual FM is less than half.
Since I trust the experience of Iulian Rosu, I will have to redo the measurements based on the original diagram.
Another VFO was measured and the results are below. Since the results are clearly exceeded by Farhan’s oscillator, no details about it will be provided at this stage.
If demand exists, I can publish the details for ideas of what not to do…
Next, a 15MHz Quartz Crystal oscillator was measured.
The phase noise at 10kHz offset is very good, at -152dBc/Hz, but there are some unpleasant humps toward 100kHz and 1MHz.
Probably a better circuit diagram should provide phase noise better than -150dBc at any offsets over 10kHz.
Please note that it is possible that the results described in here may depend significantly with the quality of pcb layout, construction, shielding, power supply, components, warm-up time, etc.
The measurements here were performed without any shielding, not the best possible power supply and with very little warm-up time (< 5 minutes).
I can provide gerber files and circuit diagram for the oscillators I tested in here, although I expect people will be interested in a VFO which performs better than Farhan’s oscillator and a better quality Crystal Oscillator.