Noise generator substitutes for tracking generator

When your spectrum analyzer lacks a tracking generator, you can use a low-cost noise generator to characterize RF components. Here’s how.

Low-cost spectrum analyzers often lack tracking generators, preventing you from having a signal source that tracks the analyzer’s frequency sweep. Not having a tracking generator can make some measurements difficult. You can, however, use a low-cost noise generator to characterize filters, amplifiers, and coax cable loss. In effect, they serve as a scalar network analyzer (no phase information).

In this article, I’ll describe how to use a noise source as a substitute for a tracking generator.

Figure 1 shows a typical broadband noise source. Available from Amazon^[1], the broadband noise source will produce a wide range of RF signals from near DC to 2 GHz. This board is designed to run on 12 V, but I found it also operated well at 5 V. I did notice the amplifiers get quite hot.

noise generator board — Figure 1. This board produces broadband RF noise. (Image: Kenneth Wyatt)

The noise is generated by biasing a Zener or Schottky diode, which creates low-level broadband noise. The board then runs this noise voltage through three broadband amplifiers and terminates it into a 50 Ω matching network. In Figure 1, the noise diode is in the board’s lower-left corner, and the three amplifier stages lie between that and the RF output port on the right.

Noise voltage is difficult to capture on an oscilloscope because of randomness. I was, however, able to trigger on one cycle that appeared to have about 500 ps of rise time. I was using a R&S MXO38 (8-channel, 12-bit, 1 GHz bandwidth) oscilloscope and the spectrum plot showed usable broadband emission all the way past 1 GHz (Figure 2).

Noise generator spectrum response — Figure 2. An oscilloscope shows broadband noise voltage and spectrum response. (Image: Kenneth Wyatt)

Noise generator applications

I switched over to using my Siglent SSA3032X spectrum analyzer so I could capture multiple spectral plots. We’ll measure cavity resonance and a couple of filter responses.

Cavity Resonance: We’ll use the “cookie tin” cavity resonance demo (Figure 3) we used to demonstrate ferrite absorber performance earlier^[2] to demonstrate how the noise source can identify structural resonances. Recall the cookie tin lid had two BNC connectors attached with short (1 cm) stubs soldered to the center conductors. The position of these is not important. We’ll drive one with the noise source and connect the other to the analyzer input port.

noise generator oscilloscope — Figure 3. I used this setup to measure cavity resonance of the cookie tin. (Image: Kenneth Wyatt)

The resonance equation was covered in^[2], and for a circular cavity, the fundamental resonant frequency was 1.275 GHz. Figure 4 shows the resulting screen capture with the marker at the peak of 1.23 GHz. The yellow trace was the baseline measurement with the noise source off.

cavity resonant frequency — Figure 4. A screen capture shows the cavity resonance of 1.23 GHz. (Image: Kenneth Wyatt)

Scanner filter: The radio scanner filter is designed to allow VHF Low Band (30 to 50 MHz), VHF High Band (140 MHz to 174 MHz and UHF Band (420 MHz to 512 MHz) to be heard and to block other strong ambient signals such as AM/FM broadcast, TV, and other high-powered signals. Figure 5 shows the general test setup using a noise generator to drive the filter, which is connected to the analyzer input port.

noise generator filter — Figure 5. The noise generator lets you characterize a scanner filter. (Image: Kenneth Wyatt)

Figure 6 shows the two bands being blocked with markers at the band edges of the VHF Low Band and VHF High Band. The notches are about 20 dB down. The yellow trace indicates the noise floor of the measurement.

notch filter VHF spectrum response — Figure 6. A close-up of the scanner filter spectral response shows the areas notched out. (Image: Kenneth Wyatt)

FM Broadcast Notch Filter: This will be a similar measurement to that above, except we’ll use a Mini-Circuits ZX75BS-88108-S+ FM Broadcast Band Stop Filter^[3]. This would be a good filter to use when operating a spectrum analyzer near high-powered FM stations to prevent overload of the front end. Figure 7 shows the test setup.

Noise generator filter response — Figure 7. The filter attenuates signals in the FM broadcast band. (Image: Kenneth Wyatt)

In Figure 8, you can clearly observe the band-stop filter edges at 81 MHz and 108 MHz. The yellow trace is the measurement noise floor.

filtered noise generator spectrum — Figure 8. Markers show the band edges of the FM band stop filter’s frequency response. (Image: Kenneth Wyatt)

Summary

For spectrum analyzers lacking a tracking generator, this broadband RF noise source is affordable and will help measure a host of RF applications, including filter and amplifier responses. While missing the dynamic range of a vector network analyzer, it can serve in a pinch to help identify or confirm RF characteristics of many devices.