Noise Measurement

“Noise Measurement and Generation” by Paul Wade, QEX Nov 1996 is a very good starting point.  Also Owen VK1OD has a theme on his website devoted to noise and noise measurement that is well worth reviewing, take a look at his excellent NFM program.

Some time ago I built a simple noise source as described in Paul’s article, except I based it on a BFR92 silicon transistor junction. It provided a readable amount of noise at 10Ghz, thus my interest was piqued. What I needed was a good reliable noise source for making measurements and adjusting front ends and LNA’s.  Good quality sources are still in excess of US$1k on ebay.com ie HP346, but for Amateur applications we don’t need this level of precision instrument.  Simple home brew noise sources are fine so long as the ENR is repeatable and we have them “calibrated” for each of our bands of interest.

At left is a BFR92 and associated components mounted on an experimenter board, the attenuator is (MidWest Microwave) 20dB to 18Ghz.  Below is the basic circuit. At 10Ghz I can audibly detect the change in noise and suspect an ENR of around 2-3dB at 10.3Ghz after the 20dB of attenuation, but more on that later. A surprising result for a handful of junkbox components.

Some time ago I purchased a couple of NC303 diodes from NoiseCom with the intention of building a noise source, just haven’t quite got there yet but it’s on the list of projects.

When I obtained these diodes in early 2008 NoiseCom were still honouring the sale of two diodes of choice to “experimenters” for a modest fee. You simply had to call NoiseCom during business hours and have your credit card ready.  NoiseCom are to be applauded for their generosity to the Amateur community.

A Noise Source

I found that Franco Rota of Italy had already written articles on sources using specific noise diodes.  My version of Franco’s design using a NC303BL diode from NoiseCom is shown below. Initially I fitted a 4dB attenuator (0603 SMT part) on the board and a good quality 20dB SMA external attenuator. A DC feed of 28v makes it compatible with a range of noise figure meters.  The article by Franco can be found on his website and is toward the end of his diode catalogue.

Come to the rescue Andy VK2AES, Andy was able to calibate the output of my source through 12Ghz. The graph below is the output with a 10dB external attenuator, and as a comparison that of an HP 346A with a very flat response around 5.5dB. It appeared the sources built by Franco had a flatter response even though I followed his design quite closely. Clearly something is different.

For amateur purposes the flatness of response isn’t such an issue, so long as we have a stable and consistent output calibrated against a range of frequencies of interest, then it’s usable. Whether the steep drop off above 10Ghz I’m experiencing is a result of resonances in the housing or characteristic of the NC303BL diode I don’t know. The diode is spec’d to 8Ghz but I suspect it will perform above that in the right configuration.  I will fit some absorber material to the milled housing at some stage and revisit the output response.

It’s easy to see why HP sources are the price they are, there’s clearly a lot of development effort goes into getting a flat output response.