Quadrature Sampling Detector (QSD)
By John Ackerman, N8UR
Here's a very quick write-up of the Quadrature Sampling Detector (QSD) that I mentioned at the meeting the other night. If anyone is interested in more details, I strongly encourage them to visit:
where you can see information about the SDR-1000 radio, but more importantly
download the QEX articles that describe the radio in great detail. There
are also lots of other good resources there.
The QSD works much like a direct conversion receiver, but is much more sophisticated (and has higher performance) than the traditional direct conversion circuit. It provides bandpass filtering, has no loss (in fact, it can actually show a gain), and is used in the SDR-1000 radio both as the receive down converter to generate an IF at 11kHz and as the transmit exciter to convert an 11kHz input signal up to the operating frequency. The QSD is based on a circuit called the Tayloe Detector that was published several years ago.
A picture does a lot better than words to describe the QSD, but I'll do my best here in text. It's really a
4-position rotary switch that turns all the way around once each cycle at the
RF frequency -- obviously, we're talking about an electronic rather than a
mechanical switch here! The switch connects the RF input in turn to four
grounded capacitors, which sample and hold the signal. The result is that
you have signal samples at 0, 90, 180, and 270 degrees. The 0 and 180
degree capacitors are coupled to the inverting and non-inverting inputs of a
high-impedance instrumentation op-amp, and the 90 and 270-degree caps are fed to
another op-amp. The result is a pair of signals -- I and Q -- that convert
the RF frequency to baseband and are 90 degrees out of phase. The op-amps
have such high input impedance that as a practical matter they don't load the
capacitors at all, so the circuit has very high conversion efficiency.
The I and Q outputs are sent to the left and right inputs of a sound card, and the computer then demodulates the signal however you tell it to. For transmit, things just work in reverse, with the I and Q signals charging the caps.
An advantage of the design is that the antenna impedance and the capacitance act as an RC network, and the QSD therefore has a bandpass filter built in. You still want some filtering in front of the detector, but the inherent filter can be quite sharp and tracks the RF frequency precisely. In the SDR-1000, the components are chosen so that the bandpass is about 40kHz wide and the IF is centered at 11.025kHz. (By the way, one disadvantage of this design is that if the antenna impedance is not at the design frequency, the demodulator performance falls off. Therefore, it's a good idea to have an amp stage in front of the detector, more to provide matching than to provide amplification.)
I hope this helps explain how the QSD works; I really recommend the articles at the Flex-Radio web site as they give a much better description than I can.
PS -- another interesting point is that by using an IF at 11kHz rather than directly at DC, the sound card provides much better SNR than it does in normal audio work -- using the higher frequency allows all the hum and noise near 0Hz to be filtered out and improves the phase noise.
Merle (Cliff) Rummel, W9LCE needs Help
Merle has been experimenting with the above concept, also known as a "Dirodyne", using a decrepit HB-10 Heathkit rx. He wants to push the operating frequency higher into the VHF range and needs, of course, extremely fast switches. He is thinking of microwave FETs. However, this gets to be beyond his field of expertise. Any suggestions or help in this area would be appreciated. Merle’s e-mail: Cliff@rtkonline.com. (Ed.)
For more info, please become a member
This site last updated Oct 22, 2003