IV.8 - Noise Diodes?
The avalanche breakdown process in diodes is inherently noisy (or random). Some diodes are designed to have a very well controlled avalanche breakdown characteristic; these can be used as white noise generators. If you aren't looking for something particularly fancy, a normal avalanche zener diode (not a tunneling zener diode) will work quite well as a noise source when biased in breakdown.
Personally, I wondered what noise diodes were used for.
This is what Marshall Jose, Marshall.Jose@jhuapl.edu told me:
- Hams use noise bridges to cheaply determine the complex impedance of a component or antenna.
Recall that a Wheatstone bridge involves a variable resistor in one leg, and an unknown resistance in another leg. An excitation voltage is applied to two opposite corners of the bridge, and voltage is measured across the two orthogonal, opposite corners. The variable resistor is adjusted to produce a null in the measured voltage.
In a noise bridge, the variable resistance is replaced by a variable resistance and variable reactance in series. The excitation signal used is broadband noise, and the null measurement is performed by a receiver. In this way, reactance of an unknown (e.g., an antenna) can be measured at a given frequency by tuning the receiver appropriately, and adjusting the resistance and reactance alternately to obtain a null in the noise heard from the receiver.
Typically, the noise source (the zener diode and succeeding amplifier) is also switched on and off at an audio rate to make it easier to distinguish the bridge's noise from thermal noise.
Most hams never bother with a noise bridge. Every so often, though, a ham will put up an antenna which ought to work but doesn't, or doesn't work at the proper frequency, and a noise bridge provides information on why it doesn't work and how to fix it. Most importantly, it does so without causing the inadvertent transmission of signals on prohibited frequencies.
The ARRL's Handbook or their Antenna Handbook discuss construction of noise bridges.
This is what Bob Underwood, bobu@msm.com, told me:
- I agree with what you have on Noise Diodes, but I think you overlook the main use: measurement of noise figure. To characterize, for example, a receiver, one adds noise from a noise generator until the observed noise just doubles (or changes by some other convenient factor). Shannon's theorem says that noise always adds (Murphy's law says the same thing). Since the noise generator would typically make a relatively high level of noise that would be attenuated down by precicesly calibrated attenuators, one can know how much noise was required to equal the intrinsic noise of the receiver, and thus infer its performance.
And this is what Gabriel Paubert, paubert@iram.es, told me:
- Another use of the noise diodes is in BITE (built-in-test-equipment). In this case the noise source is used
to inject at the input of a potentially complex system some signal either through a directional coupler or by
switching the input. This provides a rather crude but cheap and simple way to check a system for major
malfunctions. Building in a synthesizer would be much more expensive and self test would take much longer since
the noise source can generate simultaneously the whole frequency range of interest.
In some systems, like correlators used in radioastronomical interferometers, this is used to calibrate the systematic phase or time delays between different hardware paths. These are frequency dependent over the bandwidth of the instrument but a single measurement is enough to measure it thanks to the broadband properties of the noise sources.
I am not sure but it is likely that some commercial or military systems who require accurate phase control over significant bandwidth (phased array radars ?) use similar techniques.
But wait! There's more!
- Matthew J Bennion, mjb296@soton.ac.uk, pointed out that noise sources are important as random number generators, for situations where you really do need truly random numbers (as opposed to the pseudorandom number generators implemented in software). An example where this is necessary might be encryption and decryption.
Here are some manufacturers:
- Noise/Com, E. 49 Midland Ave., Paramus, NJ 07652
Phone (201) 261-8797
Fax (201) 261-8339
- Micronetics - 26 Hampshire Dr, Hudson, NH 03051
Phone: 603-883-2900
FAX: 603-882-8987