-021 Spectrum Analyser
This is an audio frequency up to (20kHz) spectrum analyser plug in for the Tektronix 560 series oscilloscope, made by Nelson-Ross.
Coming to grips with it has taken studying the circuit and really understanding it. The manual describes a calibration methodology but it is incomplete and sometimes contradictory. What I really need is the method used by the manufacturer to set it up from scratch. Here it is displaying a series of harmonics from a 2kHz square wave. As of now, it will not resolve frequencies below 2kHz.
It is actually a scanning superheterodyne amplifier, having a sweep generator, a voltage controlled oscillator, a mixer, an intermediate frequency amplifier and a detector. The oscillator frequency is controlled by the sweep generator over the range 120 kHz to 100 kHz. This signal is mixed with the signal under analysis to produce a difference frequency at each harmonic that occurs in the range 0 Hz to 20 kHz. As the oscillator sweeps (the X-axis of the scope being connected to the sweep generator, sweeping the trace from left to right in step), each time a difference output is produced at 100 kHz, a harmonic is present that is amplified by the narrow band 100 kHz IF amplifier, then rectified and filtered to produce a DC voltage proportionate to the harmonic amplitude that is displayed on the scope.
It can also be operated in manual mode. A multi-turn knob that is ostensibly frequency calibrated in 1 kHz increments from 1 kHz to 20 kHz varies the DC voltage setting to the oscillator while a small voltage variation is superimposed on the DC voltage to sweep the frequency over a small range (called dispersion*) around the knob setting. This allows each harmonic to be displayed. Given correct calibration, if a harmonic is present at the same frequency as manual setting, it will appear in the centre of the display.
*The dispersion is variable from zero to 6 kHz. Say it is set at 6 kHz and other harmonics are present within ± 3 kHz of the centre setting, they will be visible either side of the centre harmonic.
It has been was possible to get it to work on full scan fairly well, the transform of a 2 kHz square wave looks very similar on my HP 3580A (a superb piece of kit). To get the manual tuning in the ballpark I set up an external accurate power supply to substitute for the manual tuning voltage feed to the VCO and measured the voltages required to get the VCO spot on for each frequency, noting that as I did do, the harmonics of the test waveform displayed properly. The next task was to get the manual tuning knob to correlate with my results. In the process, I discovered one leaky ceramic cap and one resistor that though it looked original, was much larger than that specified on the schematic. The specified value is closer, if fact I ended up very close to the specified value. I also replaced all the caps in the tuning circuit, not just the obviously leaky one. The overall result is that the device is capable of demonstrating spectral analysis but not of making measurements. It is a neat exhibit and rather akin to how oscilloscopes were before Tektronix in as much as it demonstrates the very limited functionality and accuracy of spectrum analysers before Hewlett Packard took the bait.
Here is a picture of it connected to a Tek 561A with a homemade plug-in test cable (just visible on the left), the 24-way Amphenol connectors are courtesy of Stan Griffiths.
Finally a comparison between the HP 3580A and the -021 using a 1 kHz square wave from the “fast rise” output, of a Tektronix 106 square wave generator; at a glance the harmonic series appears identical, however the amplitude ratios are quite different, both units are set to linear (as opposed to log).
Here is the schematic (click on it for full size):