[hpsdr] VHF/UHF downconverter for HPSDR

[Ante Vukorepa]

On utorak, 4. listopada 2011. at 03:09, Ante Vukorepa wrote:
> On 3. 10. 2011., at 14:10, "Phil Harman" <phil at pharman.org (mailto:phil at pharman.org)> wrote:
> 
> > Hi Ante,
> > 
> > Yes, you could use an image rejection mixer. This is basically what we do in Mercury - but since it's done digitally the image rejection is effectively perfect.
> > 
> > The practical issue with an analogue image rejection mixer is this. In order to get 40dB of image rejection then the I and Q channels need to be matched in amplitude to within 1dB AND in phase within 1 degree. For every 20dB increase in image rejection these values increase by a factor of 10. So for 60dB of image rejection the figures become 0.1dB and 0.1degree etc. These figures needs to be maintained over the entire frequency range.
> > I'm not trying to discourage you, just applying the "get it to work on the whiteboard before spending any time or $'s" approach that I normally take.

After doing some reading, spec sheet browsing and some discussion on another (non-HAM) forum (with a HAM interested in the topic), i'm beginning to think using a "canned", of the shelf quadrature demodulator, combined with a suitable hybrid and a little bit of filtering, might be the most practical way about this.

Furthermore, the aforementioned HAM suggested chopping up the problem into bands and having multiple sections, each dedicated to a certain band of spectrum (i.e. one for VHF, one for UHF, one for near microwave) and each with its own demodulator.

Poking about various spec sheet repositories, it seems something like 0.5 deg of phase error and a few hundred mdB of amplitude error can be expected from most demodulators. The spec sheets, however, often don't give error vs. signal level plots, though.

Would that make sense?


-- 
Ante Vukorepa
Sent with Sparrow (http://bit.ly/sigsprw)


PS: Apart from the basics, the Hartley and Weaver architectures, i've been reading about some more recent, more complicated image rejection techniques, including some closed-loop, self calibrating ones, that easily achieve 50-60 dB IRR, but, frankly, seem too complicated to pull off and without a good source of suitable off-the-shelf building blocks.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20111008/d2b55ced/attachment-0004.htm>