Since no one answered this question, I decided to do so myself. My Spyverter R2 arrived this week in the mail and I ran some experiements. The first was to test it to determine its noise floor. Figure 1 shows the result of putting a 50 ohm terminator on the Spyverter input port and connecting the output port to a Siglent SSA3032X spectrum analyzer. As is apparent, the noise floor seems to be around -110 dBm.
Figure 1 -
However, I am also interested in the Spyverter's conversion loss, especially at very low frequencies. It is advertized to upconvert a signal in the range DC to 60 MHz to 120-180 MHz. Is this a resaonable claim? Well, yes and no.
There is not a great deal of documentation for the Spyverter R2; about the only official information is in
Airspy's product blurb. This stipulates the the "typical" conversion loss is 5.2 dB. To be fair, the Technical specifications given there indicate that the RF input frequency lower bound is 1 KHz (not DC as some of the marketing babble claims). So, is this accurate?
I connected a Rigol DG1022 to the Spyverter R2 input and connected its output to a Siglent SSA3032X and then varied the input frequency to see what sort of performance the Spyverter R2 would deliver. The amplitude of the input sine wave was kept constant at 700mV (.8814 dBm).
Figure 2 shows the results of a broad sweep of frequencies betwwen 1 MHz and 25 Hz. The y-axis is in dBm, while the x-axis is in log(frequency).
Figure 2 -
As is apparent, the conversion loss seems quite a bit greater than 5.2 dB. In fact for this particular input signal amplitude, it seems to be around 11 dB (-10 - .88 dBM). Perhaps for inputs of smaller amplitude (which is the real target of the upconverter) the conversion loss would be less. Here is a table that records the exact values obtained by this broad frequency sweep.
| Hz | dBm |
| 30 | -41 |
| 50 | -37 |
| 100 | -30 |
| 200 | -25 |
| 500 | -18 |
| 1000 | -14 |
| 10000 | -10.08 |
| 50000 | -10.07 |
| 100000 | -10.07 |
| 200000 | -10.09 |
| 1000000 | -10.07 |
Since from this broad sweep, the conversion loss appears to degrade around 10 KHz, I decided to explore the region between 9 KHz and 30 Hz in finer detail. The result of this narrower sweep are shown in Figure 3. For this particular figure, the y-axis is in dBm, while the x-axis is in linear frequency.
Figure 3 -
The conversion loss versus frequency curve seems to be either logrithmic or, perhaps, of the general form x/(x+1). In any case, at around 3 KHz, things start going rather badly and get worse as frequency decreases.
Here is a tabluation of the data shown in the plot.
| Hz | dBm |
| 40 | -40.39 |
| 50 | -38.3 |
| 60 | -36.58 |
| 70 | -35.13 |
| 80 | -33.93 |
| 90 | -32.9 |
| 100 | -32.07 |
| 200 | -26.07 |
| 300 | -22.76 |
| 400 | -20.49 |
| 500 | -18.82 |
| 600 | -17.56 |
| 700 | -16.56 |
| 800 | -15.92 |
| 900 | -15.24 |
| 1000 | -14.84 |
| 2000 | -12.04 |
| 3000 | -11.15 |
| 4000 | -10.85 |
| 5000 | -10.62 |
| 6000 | -10.59 |
| 7000 | -10.44 |
| 8000 | -10.43 |
| 9000 | -10.5 |
I was interested to test the hypothesis that the curve was growing logrithmically with increasing frequency. So I plotted the same data using a log x-axis. Figure 4 shows the results.
Figure 4 -
It is evident that the curve does not perfectly represent logrithmic growth (this is also apparent in that it approaches an asymptote at around -10 dBm). However, it obviously is smooth and well-behaved.
I am very interested in how the Spyverter R2 performs at very low frequencies (in the 10s of Hz). I know it was not really designed with this in mind, but I have an application that requires the upconversion of these low frequencies to the 120 MHz band. Figure 5 shows the spectrum of 30 Hz upconverted by the device. The interesting part of this image is the difference in power between the upconverted 30 Hz signal (marked with "2") at -41.5 dBm and the next highest power signal (ignoring the LO power at -45.43 dBm), which is at -62 dBm (for some reason the Siglent didn't mark this peak, which is between the peak labled 5 and the one labeled 2). This is a difference of around 20 dB, which is pretty impressive for frequencies so close together (about 20 Hz appart). Now, there may be some reason this isn't as interesting as it seems to me. These frequencies are very near the RBW of the measurement (set to the lowest value supported by the Siglent), so some experienced spectrum analyzer expert may later disabuse me of my surprise.
Figure 5 -
Figure 6 pushes the upconverted frequency down to 25 Hz. Things are starting to get fuzzy, since the difference in power between the input and LO frequency is now only 1 dB.
Figure 6 -
Figure 7 pushes the upconverted signal even closer to the LO frequency, putting it at 20 Hz. It is apparent that the Siglent is mashing together power from the LO and input, which is understandable since, as mentioned previously, the RBW is 10 Hz.
Figure 7 -
Figure 8 shows the spectrum when the input is 10 Hz. It is clear that the Siglent cannot resolve the power difference between the LO and input, since the input frequency now equals the RBW.
Figure 8 -
I hope there are some out there who find these results interesting and also someone with more expertise than I who can provide insight into upconversion of very low frequencies.
Home
Help
Search
Login
Register
