[hpsdr] VHF/UHF downconverter for HPSDR

[Ante Vukorepa]

On 3. 10. 2011., at 15:46, Steve Bunch <steveb_75 at ameritech.net> wrote:

> Another issue is the harmonic and spurious content of your LO.  Since the output of the SI570 comes from a buffer fed directly from a digital divider, it's more like a square wave than a sine wave, at least at lower frequencies where the output buffer isn't limiting it, so it will have strong odd harmonic content.  Each of these harmonics, though weaker than the fundamental, will still mix with everything in the mixer input signal, and any mixing product that falls below your LPF cut-off will be admitted.


From what i've read on the operation of frequency mixers, the square LO should actually be beneficial with switching mixers (i believe most wideband, linear Mini Circuits mixers fall under that category). There is, of course, the issue of harmonics leaking through into IF. The LO-IF isolation for the chosen mixer is about 40dB through almost all of the frequency span of interest. I don't know whether that's enough or not, though...



> If you can easily move the LO injection and IF frequency in sync, as you can with HPSDR and an SI570, it is easy to distinguish an image from a wanted signal -- the wanted signal won't move, but an image will.  Images caused by harmonics of the LO will move much farther.  This 


That sounds very much like the basic approach used in Signal Hound that Phil's mentioned (at higher frequencies at least), sans FFT analysis. Or, rather, with an "eyeball-based FFT analysis".  Sounds interesting, i'll look into it.


> Spectrum analyzers also sometimes have tuned preselectors on the input that track the LO change and will knock down out-of-band signals by perhaps 30-40 dB.  (These are generally YIG-filter based.  You can often see YIG preselectors advertised on ebay, seldom ranging below 1GHz, but going up to tens of GHz.  They are often used in wideband microwave test equipment.)  A friend of mine used to design TV tuners for a living in pre-digital TV days.  He designed varicap-tuned tracking preselectors into the front-ends of the tuners, which was 


I've thought of trying to build a tracking preselector (something similar to some of the ideas from the Anicetus wiki page, construction-wise), but decided it's way over my head, especially at UHF and near microwave frequencies...


> It's not hard to pre-compute some of the "troublesome" spots, like the FM broadcast band, commercial radio bands, etc. and pro-actively work around them.  It's also not hard to jump-tune and find combinations that give relatively image-free reception around a single frequency of interest, though as Phil points out, some interference sources are pretty broad so it's good to have a lot of choices in LO-IF combinations.  Many commercial radios (e.g., cellphones) use these tricks - the technique is well-known now, but was novel in the not-so-distant past.  You'll find it described in patents from not that many years ago.


So far, it sounds like using some form of IQ-based rejection and on-the-fly LO-IF shifting, combined with a pre-computed "lookup table" designed to "avoid" those troublesome spots would probably be the way to go about this "project". Or that's the impression i've gathered so far, at least...

Thanks for all the info and insight and please, do keep it coming :)


Ante Vukorepa
----
Sent from my iPad
 1317691689.0