[hpsdr] VHF-millimeter up/down converters?

[Glenn Elmore]

Whether one does a traditional banded transverter (up/down converter) or 
one of the architecture I'm suggesting, everything still depends upon 
LO(s) spectral purity. The best one can do is to transfer the best 
aspects of the best references at various offsets to the output signal. 
Long term stability from Rubidium, GPS, quartz,  mid-offset from quartz 
or SAW and wide offset from the fundamentals of the resonator, probably 
varactor or perhaps YIG. All of these are candidates but at different 
cost points.  The goal would be to come as close as possible to 
maintaining Angelia's spectral purity to as high as possible at 
microwave/millimeter at a reasonable cost.  I think that this is a 
reachable target. The specifications of a modern spectrum analyzer, 
which are considerably better than your(Helmut) example, are probably a 
good starting place. Even economy ones can do -125 dBc in 1 Hz bandwidth 
at 100 Hz offset at 100 MHz, how they roll off beyond that, at other 
offsets, depends upon the design. Long term generally is not a problem. 
At SSB sorts of offsets, 1 to 10 kHz, it depends on the references 
chosen, implementation and architectures.

I have operated meteor scatter, weak signal and EME on bands  in the 
range  of 2m through 10 GHz. I know about at least some of the issues. 
My goal is to produce (first for myself!) an "all-band" companion for my 
Angelia to allow the benefits of SDR through at least our 5.7 GHz band. 
Having already done this in the more traditional banded way I would 
probably chose to do it in a wider-band continuous way which could also 
easily extend SA and VNA SDR applications to mid-microwave.  In a 
previous career, I designed references and converters for HP spectrum 
analyzers, some of which are still in production.

This is fundamentally an analog rather than a purely digital 
architecture. Until we all have inexpensive 16 bit converters to 50 GHz 
I think we're stuck with this!  Even so, I think much of Angelia's 
performance can be translated to/from these higher bands. Some aspects 
will not likely be quite as good but probably in use, these shortcomings 
will not be obvious. In any case, I'd expect it to work as well as the 
best available conventional banded transverters.

If I do continue on this, I will likely not be interested in spending a 
lot of effort on the HPSDR interface. It will likely be satisfactory to 
stop at I2C or similar control/monitor of the system. I can do this 
myself with Arduino/PIC sort of control. But it would probably be more 
useful to OpenHPSDR if someone else wanted to contribute to this by 
working on integrating and interfacing to the rest of the system. This 
would make, for example, extension of the VNA application to 6 GHz a lot 
cleaner and possibly faster.

If anyone else is interested in helping, contact me off list.

Best,
Glenn n6gn


On 04/19/2015 09:40 AM, Helmut wrote:
> Glenn,
>
> unfortunately things are not so easy to handle at these frequencies:  Assume
> a typical phase noise of the LO1 say -90 dBc/Hz @ 10 kHz offset. Relative to
> our bandwidth B we will get:    -90dBc + 10logB = -56 dBc . That means that
> a strong neigbouring signal – say -53 dBm- at a distance of 10kHz produces
> a noise of -109 dBm. This will flood all weak signals within this distance.
>   
> 73, Helmut, DC6NY
>
> -----Ursprüngliche Nachricht-----
> Von: Glenn Elmore [mailto:n6gn at sonic.net]
> Gesendet: Sonntag, 19. April 2015 17:52
> An: Helmut; 'Joe Martin'
> Cc: hpsdr at openhpsdr.org
> Betreff: Re: AW: [hpsdr] VHF-millimeter up/down converters?
>
> Helmut,
> I think this need not be the case. By using same-architecture LO1/LO2 along
> with wide PLL bandwidth, phase noise can correlate out far enough to cancel
> and not be an issue. You'll notice that with the mix-up mid-down approach,
> correlated noise cancels. This is the done in high performance spectrum
> analyzers and gives performance as good as the references can provide, even
> out to millimeter wavelengths some times.
> I only suggested 6 GHz as a target because filtering can be done within
> common PCB materials (stripline filters etc) and SMD parts and connectors
> can work just fine to there without heroic efforts. Parts are pretty cheap
> too.
>
> There will certainly need to be pre/post amplification provided, just as
> there is presently in Angelia. That can be done as it conventionally is.
> For EME and weak signal this may mean both are antenna mounted even.
> The goal is to get clean Angelina performance, say 25 MHz of bandwidth,
> translated to/from VHF-6GHz and let the banded details be done separately -
> akin to putting LPFs and HPFs on Angelia's in/out for the QRO arrangements.
>
> I believe it is possible and practical to do all this at reasonable cost
> - though I'm unsure of exactly what price/performance target OpenHPSDR
> targets.  Has a charter or statement for OpenHPSDR been written that talks
> about this?
>
> Glenn n6gn
>
>
>
>
> On 04/19/2015 08:39 AM, Helmut wrote:
>> Hi Glenn,
>>
>> Just my 2 cents: I think this three-mixer-design will not provide
>> adequate performance for severe weak signal and/or contest application
>> on the VHF, UHF and SHF bands. The phase noise of LOs at that these
>> frequencies is on principal too bad and dominates the dynamic of the
>> whole system. A lot of pre-amplification is necessary to meet the
>> noise figure requirements. This degrades the dynamic performance further.
>> At the present time conventional transverters are the better choice.
>> As some guys know I run on VHF  a  modified Mercury  and Penelope in
>> undersampling mode with similar performance to the HF bands (details
>> http://www.hamsdr.com/data/GlobalFileUploads/9636__VHF%20DDC+DUC.pdf ).
>> That's a nice option for the 2m band.
>>
>> 73, Helmut, DC6NY
>>
>>
>>
>


 1429474630.0