TX Spectrum Analysis Methodology Introduction

A Rigol DS1054Z with some Python scripts is being used to do spectrum analysis of the Hermes-Lite TX. This page introduces the methodology used. Please right click on an image and open in a new tab for a detailed view.

Unless otherwise noted, all experiments on this page are with a Hermes-Lite v1.22 with 73.728 oscillator, built in the standard way, unfiltered, unamplified and operating at full power of ~20 dBm. The tone was generated with Quisk's spot function. Tones generated with PowerSDR Tune or Quisk with CW input did not show any difference. Measurements were made of the voltage across a 50 Ohm resistor (Z) on the antenna output. The scripts accept a value for Z, and the power measurements were validated by other means.

160M TX Spectrum

The various sources of Hermes-Lite spurious emissions are best seen when transmitting on a fairly low frequency, such as 160M as shown below.

First, notice the 19.0 dBm carrier at 1.810 MHz and the expected harmonics. These can be isolated as shown below.

You can see the harmonics taper off as expected until no more are displayed as they start to blend into the noise. Also note that odd harmonics are worse than even harmonics except for the second harmonic. Here are the numeric values for the harmonics:

MHz	dB
1.810000	0.00
3.620041	-48.46
5.430041	-59.36
9.050083	-65.86
12.670082	-63.84

We can eliminate these harmonics except for the fundamental from the plot as seen below.

Of the remaining emissions, note the 498.190 MHz. This is the expected image of the fundamental from the next Nyquist zone given the 1 GHz sampling rate of the scope. It is diminished in amplitude given the LPF effects of the scope and probes. Power above ~200 MHz is probably not accurate although emissions are still quite visible.

The scope appears to be sensitive at frequencies of 62.5, 125, 250, and 375 MHz. These mix with the Ftx fundamental to produce artifacts at plus or minus Ftx. This is a price to be paid for using an oscilloscope as a spectrum analyzer. These artifacts and the the alias at 498.190 MHz are isolated below.

As seen above, after removing these emissions, the remaining large spurious emissions are real and have been the most problematic. They occur at at multiples of the sampling frquency, Fs=73.728, plus or minus multiples of the transmit frequency, Ftx=1.81. They are isolated and shown below. As expected for the Hermes-Lite which interpolates by 2 for a DAC running at 2Fs, the strongest harmonics are at 2Fs - Ftx and 2Fs + Ftx. This illustrates the requirment for a reconstruction filter. These emissions are isolated below, and then removed.

After all these spurious emissions are accounted for, there is no emission stronger than 70 dBc. Consider amplifying this remaining signal to 62 dBm or 1500 Watts, the maximum legal limit in the US. 70 dBc is -8 dBm, or around 200 microwatts. Even WSPR would have a hard time hearing that!

Below is a table of the remaining emissions down to 80 dBc. Note that there appears to be frequencies of 200 and 300 MHz also introduced most likely by the scope that mix with Ftx.

MHz	dB
0.065000	-75.13
0.130125	-79.42
1.301792	-71.42
1.745042	-78.80
1.810000	0.00
1.875000	-79.04
2.307083	-78.76
2.318250	-70.47
33.426247	-78.57
41.630288	-76.97
45.250330	-74.42
48.870371	-73.42
52.490371	-76.82
59.730453	-79.92
63.350453	-74.03
66.970494	-72.82
70.590494	-77.03
76.866577	-77.72
80.486577	-74.74
84.106618	-75.31
90.694826	-79.85
90.697117	-78.03
90.697284	-78.01
90.703784	-77.77
90.704034	-79.72
90.707492	-78.63
92.267034	-79.61
92.340826	-79.78
92.356242	-78.88
94.966700	-79.62
98.586700	-78.05
102.206741	-78.54
195.689984	-79.52
198.189983	-70.77
199.310025	-79.74
201.809983	-71.00
223.189981	-79.05
226.810023	-79.32
236.501730	-79.60
298.189975	-79.78
301.809975	-79.60
380.338760	-79.86
383.958801	-79.38
432.764422	-79.63
436.384464	-79.08
440.004463	-79.70