QS1-R vs SDR-14 vs Perseus -- Direct Sampling HF Receivers

[wb4jfi]

Mike O'Dell wrote:
> thinking hard about how far i wish to put my foot
> into the SDR pot, I've been doing a lot of reading,
> trying to distinguish data from opinion from venom.
> i'm recording them here to elicit commentary from
> the knowledgeable chaps around the table.
>
> if one is willing to spend a grand on an SDR receiver,
> that puts one in the realm of the "direct sampling" receivers.
>
> after comparing these three at some length (reading a lot
> of extremely animated conversations, esp on the E-K3 list,
> of all places), the Perseus appears to have the edge if
> what you want is a full-up *receiver in a tin*. The SDR-14
> seems also to be "dedicated receiver" but a bit more
> bare-bones (or purist, depending on your religion).
>
> However, if one is looking for a receiver which can be the basis
> of a *platform*, then the QS1-R appears to have the edge
> as it was clearly designed as such, and new companion
> boards are coming along.
>
> It should be noted that the Mercury receiver card and the
> RX section of the Hermes one-board xceiver from the "OpenHPSDR"
> project are essentially identical to the QS-1R.
>
> As for sampling performance and BW (sans undersampling),
> the QS1-R samples at 125MS/s and goes to 55MHz (hello 6M!)
> the Perseus samples at 80MS/s and goes to 30MHz (limited by preselector)
> and the SDR-14 samples at 67MS/s and goes to 33MHz
>
> the QS1-R and the Perseus have a much
> newer A/D part than the SDR-14.
> That's one of the huge paradoxes in the
> electronics biz - the guy who starts 6 months
> after you has a big advantage because he usually
> has better parts to pick from. The cellular radio
> base station business is driving the evolution
> of digital RF hardware really hard. The LTE/WiMAX
> MIMO air interfaces assume massive digital
> processing.
>
> the other possible advantage the QS-1R and Perseus
> have over the SDR-14 is that all of the DDC
> processing (Quadrature NCO + Mixers + Digital Filters)
> is done in an FPGA while the SDR-14 uses
> a fine, off-the-shelf DDC from ADI. The QS-1R
> has a very large FPGA while the Perseus seems
> to have been cost-optimized (not a bad thing!).
>
> the net result, though, is that in line with
> being a platform, the QS-1R has lots of room for
> further expansion of the digital processing.
> (one option being explored is a true multi-channel
> receiver).
>
> In all 3 cases, the raw A/D samples
> are available, as are the I/Q signals from the
> DDC.
>
> so my take is that if one were to take the plunge
> for a "big bite" direct-sampling receiver intended
> to be the basis for AMRAD-style geekery-hackery, the
> Quicksilver QS-1R is the best of 3 quality alternatives.
>
> again, this is my digestion of the current state of things
> and the whole point of posting this is to have my
> understanding improved by facts known to others.
>
>     cheers,
>     -mo
>
>
I could be wrong (again), but I think there is a design difference 
between the openHPSDR Mercury/Atlas/Ozy/Magister boardset and the 
QS-1R.  Early on, Phil Covington was talking about some data path being 
parallel on his QS-1R, while it is serial in the Mercury design.  This 
would lead to less throughput on the samples.  I believe Phil's point 
was that his board could do more multiple channels/receivers from the 
same data samples than the Mercury.  But, I think the Mercury has a 
front-end amp, and the QS-1R does not (right now).  You should probably 
plan on putting bandpass filters in front of either device, as a 16-bit 
A/D does not quite cut it for total HF dynamic range, especially near 
broadcast stations.

I ahve the OpenHPSDR board set, but not the QS-1R.  I think the QS-1R is 
a slightly better unit, but it cost more all-at-once, and I could not 
afford it.  I could afford to purchase a board here and there, therefore 
the openHPSDR project fit my wallet better.

To me, Phil's overall plan seems more better (!) in the long run.  I 
believe both projects have the software & firmware freely available.  
The openHPSDR servers have been restricted lately, due to too much 
traffic.  That surprised some folks.

Terry