Paul White <paul@...>
After a few weeks' experience with the RSP1A during VHF FM BC DX-ing, I decided to look a bit more carefully at the nature of some possible spurii that *seemed* to confuse me when evaluating a few weak signal candidate catches.
Bear in mind I'm not very experienced with traditional receivers (that definitely have problems to trap the unwary), and am quite new to SDR (that apparently introduces a whole new "spectrum" of gotchas to make lives difficult).
Here's a selection of observations and measurements from my first look at RSP1A spurii as seen through SDR Console. This is just "baby" stuff - Leif's science is all just Black Magic to me.
Just one preliminary remark: birdie hunting produces horrific-looking baddies, but I do realise these are msotly swamped by the wanted signals in real life.
1. You need a clear head for this kind of study... the first thing I found is that there are real receiver (hardware) birdies and there are all sorts of artefacts that arise from SDR processing. Don't say "I told you so", because I haven't found a handy guide online to explain all these pitfalls!
2. You can't even begin to understand a receiver without trying to eliminate all the other sources of confusion - especially real signals (hence the 50 ohm terminator on antenna port) and software artefacts (notably the weirdos that come and go as you tune through them).
3. It's not at all obvious which receiver/software configurations make a significant difference. It's *beginning* to look as if receiver bandwidth (0.5, 1.0, 2.048, 3, 4, 6, 7, 8, 10 MHz) is only a secondary factor. I'm a bit surprised because I assumed, perhaps wrongly, that different sampling rates were involved here. I'm also a *little* surprised that LO mode selection (120, 144, 168 MHz) has next to no effect on birdies.
4. The biggest surprise was that sprurious signals *seem* to arise from more than one source (don't yet have any idea how many). There's seemingly a significant contribution from mixer leakage (harmonics of 24 MHz), but other patterns predominate in different spectrum ranges.
5. Receiver configuration was set with RF gain 7 dB (default), IF gain manual -20 dB, though settings do not seem to be critical.
6. The first set of measurements looks at harmonics of 24 MHz that we made up to the 10th at 240 MHz. Throughout this range the noise floor is roughly constant at about -133 dB (variation seems to be mostly less than +/- 2 dB, but we did see differences up to about 10 dB). You can see from PDF1 that most spikes show at about 40-60 dB above the noise floor (we did *not* check if absolute values were even more consistent). There is no systematic difference in these harmonics, by frequency, or by receiver bandwidth.
7. In the course of testing, one phenomenon was especially noticeable: transient peaks at apparently random locations in the displayed bandwidth during panning or other types of display refresh. Obviously processing artifacts, and something we have noticed many times before, but now particularly obvious in the absence of real signals. We wonder if there is any prospect of reducing their intrusiveness?
8. Another side note - another well-known feature, we believe - a centre-span dip, or broad notch, that seems to arise with all bandwidth settings over 2.048 MHz. This too seems to be a processing artifact and we wonder if that would be difficult to correct?
9. Moving on from (but associated with) the 24 MHz harmonics, an oddity we first noticed above 120 MHz was the presence of lower-level spurii a few kHz higher in frequency. Initial findings are that their visibility is more dependent on receiver bandwidth and tuned frequency, but one pattern is peaks of 10-20 dB above noise at slightly varying offsets around 3.2, 6.5 and 10.5 kHz higher than the harmonic. Usually the ~6.5 kHz peak was more prominent, but we were concerned to note that results may not be very repeatable. That may imply variable interference from outside the receiver and will be investigated further.
That's all for now.
It has been the start of an interesting learning process for me, and apologies if this is boring for experienced members.