Chris,
I agree with Ward. QSD mixers are very interesting devices, but have
intrinsic limitation due to the poor image rejection.
In 2006 I spent several months working on the QSD mixer. After many
discussions with friends on the Softrock and HPSDR forums (in
particular, thanks to Ahti OH2RZ), I arrived at a design capable of a
noise floor of -123/-128 dBm in 2.4 kHz, depending on the audio board
gain setting, and equivalent to an NF of 12 dB, better than many HF
receivers. At the same time, the 3rd order IMD was measured at 106 dB
at 5 kHz spacing - higher than any other mixer or receiver that I
measured, giving a 3rd order intercept point of +36 dBm. My mixer is
different from the standard QSD as used in the SoftRocks and
FLEX1000, since it has proper terminations for the IF port, and is
not balanced at the input, simplyfying the input transformer. Then I
did a lot of work on operational amplifier noise figure and input
matching - opamps are NOT ideal amplifiers as you learn on school
books :-), when it comes to noise there are some interesting points
to examine. At the end I was very satisfied about the results, after
spending about 300 hours at the work bench!
I quit the development since there were several problems and
limitations. First, such a mixer requires drive signals with
components in the GHz range - very fast signals, with extremely high
dV/dT, requiring a true microwave PCB layout (the prototypes work
quite differently on "dead bug" and on PCB). Also, being pushed to
the limits, it is sensitive to the components used (brand and batch).
Then it requires the best possible audio codec, since the dynamic
range is defined by it, after a careful design of sytem gains. Last
but not least, somebody noted that an extremely high IMD dynamic
range is not worth very much if there is no good suppression of image
signals, as the spurious free dynamic range is at least 50 dB lower
than the IMD3 DR. Also, all QSD mixers are sensitive to
sub-harmonics, and require accurate design of preselector filters.
Ah, well, QSE is, of course, the same story, with the complication
that image signals don't bother you, but other OMs on the band. 40 dB
under 1kW is a good 100 mW, and it's able to do good DXs.
The original idea was to offer a good mixer for the HPSDR Phoenix
module. The development of Phoenix basically stopped; the direct
sampling SDR receivers are clearly the right choice for performance,
so the interest on QSD/QSE architecture dropped. You may see my work
on QSD in http://www.spinelectronics.eu/ftp/SDR/IK1ODO_SDR1.html
Returning to your original question, simply I don't know. I never
seen a serious review of the Flex 5000 RX ;-) and QS1R is, AFAIK, a
still work-in-progress design.
Marco IK1ODO
On May 13, 2010, at 7:40 PM, Chris wrote:
Certainly an interes ting thread. But so far nothing about the
QSD/QSE approach. Which performs better a Flex 5000 or a QS1R?
Chris
