Noise sources and their use
Rohde & Schwarz SUF2
The SUF2 delivers noise in different bandwidths and at a range of levels, nominally up to 1 V RMS in 75 Ohm. The noise source inside the SUF2 is a combination of a BFY90 and a 2N708, the latter being reverse-biased. It is the base-emitter junction of the 2N708 which acts as an avalanche diode. The BFY90 sets the emitter voltage for the 2N708. The two transistors are temperature controlled in order to stabilize the noise level. Filter options were available for triangular noise, pink noise, CCIR Rec. 559 weighting, CCIR Rec. 571 Weighting, and CCIR G.227 weighting. My unit does not include any of these options, and I have never needed them. If required, you may make your own filter, and with any weighting you desire. The board just has to include a relay to allow activation of your filter when you press the "RES" button on the front panel. I have used the SUF2 for testing of PLLs and demodulators with noise added to the signal-of-interest. Such tests are done by adding the noise resistively to the signal source. The noise and signal levels are adjusted as required to provide a specific channel SNR or noise density. |
Upper screen dump: The rail voltage for the +5V supply right after power on. Lower screen dump: The same rail once the SUF2 had been powered on for a few minutes.
|
My SUF2 started to behave weirdly with a flickering front panel LEDs, a slight buzzing from the relays, and a varying degree of unresponsiveness of the front panel buttons.
The first likely explanation that I came to think of was a couple of tired electrolytic capacitors for the 5 V rail, as seen before with Rohde & Schwarz gear from the era. However, some further investigation soon revealed that the capacitors were not to blame. As shown in the oscilloscope screen dumps to the left I found that the unregulated rail's voltage would suddenly drop, and the ripple frequency halved, after the SUF2 had been powered on for a few minutes. Though not the first thing one perhaps would think of, the culprit was the BY164 bridged rectifier which apparently lost an internal connection when becoming hot. A simple replacement of the BY164 with a PBL403 did the job, and as a side bonus this rectifier does not get as hot as the BY164. Speaking of heat, the back panel would get too hot for my taste, when I left the SUF2 on. Now that I had the SUF2 open at the work bench I dediced to add a small heat sink where the three power regulators are located, and the results are shown below. |
|
A while back I measured the level of the SUF2 by a selection of wideband RMS voltmeters. When stepping down the level I observed some large, unexpected differences between the level stated on the display, and the level measured by the voltmeters. Even when the level was changed by just one dB these differences could be quite distinct, and could not be explained by a change of the meter's bandwidth with the level, for instance.
A further investigation revealed that the power spectrum, to my huge surprise, was far from flat when the attenuator stages with the largest attenuation were activated. The slideshow demonstrates the issue with the flatness at lower levels from the SUF2 noise generator, in this case for its 6 MHz setting, going from -65 dBV down to -100 dBv in steps of 5 dB. The measurements were done with HP 8591C with averaging, and the screen data saved with PrintCapture. In view of this, the SUF2 must operate at levels not lower than about -70 dBV, and you have to connect an external attenuator, if you need a flat response at low levels. |
(Picture to be uploaded)
|
Ailtech 7615
This is an example of a diode noise generator often used for noise performance testing of receivers and RF amplifiers. The Excess Noise Ratio (the increase in noise over the thermal noise when the diode is switched on, abbreviated ENR) is specified to 15.5 dB over a frequency range of 10 MHz to 1.5 GHz. For an introduction to noise measurements check out the application note from Keysight: http://literature.cdn.keysight.com/litweb/pdf/5952-3706E.pdf |
TTi TG5011A
As many other modern synthesized generators the TG5011A may provide arbitrary waveforms and noise in addition to the traditional sine, square/pulse, and triangle/sawtooth. The noise bandwidth is fixed in the TG5011A, so the user may only change the level. However, the noise signal can be added to other waveforms, and may even modulate them. The distribution is Gaussian, as verified by the analysis shown on the left (picture to be added). |