[hpsdr] New HPSDR card for VHF/UHF ?

[Grant Hodgson]

Phil, Shel, Mike et al

There are obviously two ways to implement a VHF receiver - direct to baseband
or downconvert to HF.  There is a third way, which is to undersample, but my
understanding is that an undersampled ADC system has an even higher noise
figure than a conventional DDC, which makes achieving wide dynamic range
difficult or even impossible - please somebody correct me if this is wrong.

These solutions present quite different problems from the point of view of
trying to achieve a truly high-performance receiver - which is of course what
drove HPSDR in the first place.

I/Q
---

My belief is that biggest problem with I/Q is probably the LO.  It would be
almost trivial to modify the uWSDR 2.3GHz PLL with the addition of a divide by
16 prescaler to give 144MHz with fairly low phase noise - if -120dBc/Hz @
10kHz is not low enough then I might be able to do some work on an even lower
noise version.

I think that phase noise is more of a concern than I/Q balance - surely the
I/Q balance issues have been resolved for high performance HF receivers, in
which case the same results would be achieved at VHF?  I'm assuming that the
I/Q drift of a VHF downconverter such as the LT5517 is stable with time and
temperature which I think is a reasonable assumption, but I haven't yet
verified this.

Downconvert to HF
-----------------

If the Mercury/Hermes 122.88 clock is used as an LO then the IF becomes
21.22-25.22MHz (I'm only concerned with 144MHz for now - one band at a time
unless more RF developers can be found!).  All that is then required is a
simple high-level buffer to drive the mixer.  The simplicity of the LO is what
makes this approach attractive.

Obviously Mercury/Hermes will have no problem providing a top-notch 21-25MHz
IF back-end.  

The problem is what happens between 144MHz and 21MHz.  There are two elements
which are critical to achieving top-notch performance - the mixer and the 1st
IF sub-system.

The mixer needs to be a high-level type such as a level-17 or even a level-23
- I haven't done any research into this but a browse through the websites of
Mini-Circuits, Hittite, Linear Technology, Analog Devices and maybe Avago,
TriQuint, RFMD, Skyworks and others should yield something suitable.

But in order to get the best out of the mixer it is necessary to terminate the
IF port in a good, broadband 50 ohm load - this is done either with a diplexer
or an attenuator (or both).  Then a filter is required to remove the image
response at 266MHz - this might not be too difficult, the filter has to pass
21-25MHz so a simple low-pass arrangement may suffice.  And then some sort of
gain stage will be required - if Mercury/Hermes has a 30dB noise figure then
maybe 30dB of gain would be needed at 21MHz - not difficult, but needs to be
done with care, and requires a very linear amplifier in order to avoid
degrading the strong signal handling performance.  The issue of correct gain
distribution is crucial to achieving best dynamic range, which becomes quite
challenging one the spurious-free dynamic range exceeds 90dB (I would aim for
100dB otherwise you might as well buy an FT225RD + Mutek).

So if we can ignore the issue of whether the ADC is done by Janus/AK9354A or
Mercury/Hermes, then I believe that the issue becomes that of a complex LO
(for the I/Q option) versus complex 1st IF (for the HF downconverter option).

I don't think that there is a right and wrong, and this is one of the reasons
that VHF receiver design is still - er - interesting (read - not easy).   It
almost gets religious - I can almost hear cries of 'I love Mercury - go HF!'
and 'Janus works great - go I/Q!' 

There are few that would doubt that DDC is the obvious way to go for HF
(DC-50MHz or so)- the advantages are so blindingly obvious that there is no
real point in considering anything else.  But high-performance DDC isn't
available (yet) for 144MHz at a reasonable price, and it may be that just like
clock speeds for Pentium microprocessors there are limits to speed - which
means that conventional VHF front-ends may be with us for some time yet.

I have block diagrams for each of the approaches above and will try and post
them on the web soon, or can email direct if someone else can host for now.

Also note that I haven't considered the Tx side - this is relatively easy, as
transmitters don't have the same requirements for dynamic range, and generally
speaking have almost identical levels of phase noise as the receiver. 

regards

Grant  G8UBN


On Tue, 7 Sep 2010 09:02:15 +0800, Phil Harman wrote
> Hi Grant,
> 
> Thanks for your input. I feel sure that VHF/UHF Atlas boards are in the
> HPSDR future so your input is very much appreciated.
> 
> If you could find that time the draw a block diagram I think that would
> progress the design discussions significantly.
> 
> I understand that you are proposing a I/Q up/down converter system using
> the same ADC as we used on Janus.
> 
> How would such a system compare with a transverter feeding Mercury/Penny?
> 
> The reason for my question is the DDC/DDU techniques we have been using
> for the past few years overcomes the I/Q phase/amplitude correction issues
> required with direct conversion techniques - particularly with the 
> sort of dynamic ranges you are proposing.
> 
> Whilst the art of correcting such errors has advanced in the past few
> years, if there was not a significant performance difference then my
> preference would be for a transverter approach.
> 
> I look forward to your comments.
> 
> 73 Phil...VK6APH
> 




 1283986112.0