AnsweredAssumed Answered

How to analyze audio frequencies with a RF spectrum analyzer

Question asked by Frank_BR on Oct 3, 2007
Latest reply on Oct 5, 2015 by AndreaCW
Most low and medium cost SA’s (spectrum analyzers) have a lower frequency limit equal to 100 kHz, or so. It is the case of my N1996A. Sometimes you need to analyze audio frequencies and it is good to know that it is possible to use such SA’s for the task. The “secret” is to use an up-converter to shift the audio frequencies to higher frequencies. Most SA’s have at the rear panel a BNC connector with a 10 MHz output that can be used as the LO (local oscillator) for the mixer. Best, the “Reference Out” usually has a level appropriate to drive directly the mixer. In my N1996A, the 10 MHz Reference Out has +5dB amplitude, what is almost ideal for a Level 7 mixer.

I have used a homemade mixer based on the discontinued Mini-circuits RMS-2 (see picture), but you can use the ZAD-1+, or an equivalent mixer. The Mini-circuits ZAD-1+ come complete with three connectors, so it is a question of connect-and-use.

The mixer connector labeled “RF” must be connected to the SA input by a cable or an adapter, whereas the “IF” connector receives the low-frequency signal to be analyzed. Of course the “LO” of the mixer is connected to the “Reference Out” at the rear panel of the SA. For protection of the SA input, it is a good idea to use a fixed coax attenuator between the “RF” mixer output and the SA input. A 10~30 dB attenuation should be enough.

The impedance of the source of the signal to be analyzed should ideally be 50 ohms, a value that is normal for RF and microwave systems, but not for low-frequency circuits. I have used an oscilloscope as the signal front end, since the vertical amplifier of my oscilloscope has a 50 ohms output connector at the rear panel. This solution is ideal because the 10x oscilloscope probe of can be used to pick the signal, and the oscilloscope’s sensitivity control can be used to accommodate signals which amplitude varies from mV to dozen of V. This configuration is very convenient because the oscilloscope displays in time domain the same signal that the SA shows in frequency domain.

An alternative to an oscilloscope that has a vertical amplifier output connector is to use an operational amplifier as front end (see schematic diagram). The circuit uses a TL074 FET-input opamp with a voltage divider R4-R5, which gives an output impedance of 50 ohms, approximately. The total gain can be controlled by the input attenuator or by varying R2 or R3.

The two screenshot show the spectrum of beat signals of a FMCW radar. The RF carrier is FM modulated by a 1 kHz triangular-wave. The first screenshot uses a log scale, whereas the second, a linear scale. The beat signal is periodic, with the same period of the modulation. Its spectrum shows many harmonics of 1 kHz. Note that all the frequencies are up-shifted by 10,000,000 Hz. The 10 MHz Start-frequency corresponds to 0 Hz for the low frequency signal which is been analyzed. Also note that the “Ext. Gain” parameter was adjusted to show the correct values of analyzed signal.

A final comment about dynamic range. The dynamic range depends on several things. First is the level of the signal at the mixer IF input. It should be –30 dBm, or less, to maximize the dynamic range. Second, the dynamic range depends on the SA resolution bandwidth. Use the smallest RBW compatible with a reasonable fast sweep time. It is here that a SA with digital IF like the N1996A is better. Third, the dynamic range depends on the phase noise of the SA local oscillator. The N1996A doesn’t have brilliant specifications for phase noise but it is better than –80 dBc/Hz, even for frequency offsets as low as 100 Hz. With my N1996A, I get a dynamic range that is better than 70 dB for a 10 Hz RBW, what is sufficient for many applications.

P.S. Mixer in action picture in the next post