| What's New in Test |
| Announcements |
|
|
|
|
|
|
|
|
| Interviews and Forums |
|
|
| Magazine Articles |
|
|
|
|
|
|
|
|
|
|
|
|
|
| Product Releases |
|
|
|
|
|
|
|
|
|
|
|
| Reports |
|
|
| Standards |
|
| Web postings |
|
| |
| Selecting a Noise Generator |
Patrick Robbins, Micronetics, Inc. - Enon Microwave and Noise Products Divisions
There are many attributes to noise generators. Sorting through these and matching them to current and anticipated requirements will avoid costly but too often made mistakes. For the purposes of this article, a noise generator is defined as an instrument that you plug into the wall that contains an amplified noise source. In simplest form, there is an on/off switch and an adjustable attenuator to control the amplitude. Calibrated noise sources used for noise figure measurement and noise diodes are sometimes generically called noise generators, but these are not included in this article. Noise is defined as thermal or Gaussian noise. Noise generators typically have a white spectrum (flat output over frequency) but sometimes have a pink spectrum ( 3dB/octave roll-off over frequency).
Most tests using noise generators involve adding noise to a signal of interest (SOI) to simulate some sort of real world or typical operating parameters. Satellite and wireless receive system test platforms are perhaps are the most common systems to use noise generators. This is because receive signals in the real world are weak requiring substantial amplification which adds a tremendous amount of noise.
One of the first selection questions to ask is: Does the test require an exact known amount of noise or more specifically an exact Signal:Noise (S/N) ratio or not? A figure of merit for digital radios is bit error rate (BER) vs. S/N. Setting up an exact S/N requires a sophisticated brand of noise generators called a carrier/noise or C/N generator (Click for an
example). These range in price from $20 k to $40k. These are microprocessor controlled instruments which have power meters built in. The user simply connects a “clean” signal to the input port and the output port is connected to the DUT which has this same signal but with noise added to it to a user set S/N. The most important criteria for selecting one of these instruments is S/N accuracy, Noise stability and Gaussianity, ease of use, built-in algorithms (to eliminate the need to externally manipulate noise equations) and the ability of the built-in power meter to handle complex signals.
It is possible to use a simple noise generator ($5 k - $10k) and manually set up the calibrated S/N. However, this requires a fairly high degree of RF test expertise and external RF components, equipment and cabling. Simple noise generators are more commonly used for less demanding applications where an exact amount of noise is not required. The user can simply “dial in” some amount of noise until the desired effect is realized. Simple noise generators often come with standard options such as built-in Signal+Noise Combiner, which can simplify the test set-up.
Between the simple noise generator and the full-up C/N generator is a microprocessor controlled noise generator ($10k - $20k). These instruments can be remotely controlled via GPIB and/or RS-232 bus. Some of these have limited functionality - on/off switch and control of the attenuator following the noise source. Others such as the Micronetics
ANG series of instruments allow the user to set an exact calibrated noise amplitude and can be configured with a suite of standard options for an instrument which contains only the necessary features while not paying for others.
Noise generators, like all instruments are defined by frequency range. However, noise power is proportional to bandwidth so it is unusual to find an instrument such as from 10 MHz – 26.5 GHz. However there are several multi-octave generators. It is important to look at the noise spectral density output to see if this enough for the test application. Noise spectral density (No) is defined as power normalized to a 1 Hz band typically in units of dBm/Hz. Noise generators are often around +10 dBm total output power (N). When divided by the bandwidth (BW), one can arrive at the spectral density. In log form the spectral density of a +10 dBm noise generator with a 1 GHz operating band is
No = +10 dBm – 10*Log(1 GHz) =+10 – 90 or –80 dBm/Hz. As a reference, ambient thermal noise is –174 dBm/Hz. To figure how much noise power this spectral density is in a communication channel (say 1 MHz) simply use the equation in reverse
N = -80 dBm/Hz +10*Log(1 MHZ) = -80 + 60 or –20 dBm. The very broadband noise generators have lower spectral density, so it is important to examine ahead of time if they have enough amplitude for the test requirements.
Lastly, arbitrary waveform generators can be programmed to output Gaussian noise. The noise is generated digitally and is called pseudo-noise. Whether this can be used is often a question of BW. The wider the band, the more processing power is required so this is a huge figure of merit which isn’t an issue with analog generated noise. Secondly, noise has a very high peak:RMS ratio called peak factor or crest factor in dB power. The Gaussian distribution is infinite, but in practice, 5:1 voltage (translates to 14 dB in power) is considered a minimum. This further challenges the processing power of the digitally generated noise hardware. Due to some deviation from Gaussianity and the fact that the noise is actually made up of discrete instead of agile frequency components (resolution limits), the most demanding applications such as satellite communication modem testing generally don’t lend themselves to pseudo noise. However, audio signal testing these days is typically performed with pseudo noise as bandwidths are small, and it is quite easy to change from pink to white spectrum.
|
| |
| |
| Next Issue's Product/Service Focus |
In our next issue of Product/Service Focus we will cover Testability and Built-In Test Products/Servic/Scan Analyzers and Sythesizers.
You can add or upgrade a listing before the next issue comes out.
If you would like to include an exclusive article on how to best select Testability and Built-In Test Products/Servic/Scan Analyzers and Sythesizers, please contact LouisUngar@ieee.org.
|
| |
- If your friend forwarded this newsletter to you, please register as a member and receive The BestTest Newsletter -- absolutely free!
- If you wish to update your news preferences or cancel the subscription, please unsubscribe.
- If you have any questions, please email experts@BestTest.com
|
Newsletter
