Most receiver designs use mixers. Most receiver designs use RF and audio
amplifiers.
Pure analog receivers use mixers because its most convenient to get
selectivity and gain at fixed frequencies. That's the fundamental of the
super heterodyne receiver invented by Armstrong during WW1. Some analog
receivers have DSP added for flexible filtering, often at audio or near
audio. Like the Omni VI+, and most modern Yeacomwood.
There are at least three SDR designs on the market today.
One, like the Orion, is much like an Omni VI+ with more digital options
for filtering, AGC, and controls. It uses at least as many mixers as the
Omni V/VI. It has at least one, probably two, AGC loops, one to keep the
input signals from overriding the D/A range, and another to keep the
audio level relatively constant. At least one varies the RF or IF gain
to protect the D/A. So the first mixer has to handle the entire receiver
dynamic range, while later mixers and amplifiers may be inside the AGC
control loop and be less critical.
Then there's the direct conversion I/Q receivers like the Flex and the
SoftRock. Where a local oscillator sets the center of the listening
frequency range and the DSP using a speedy PC does the fine tuning, the
filtering, the gain controls, the detection, and the presentation. This
takes two mixers, running in parallel with LO split into two signals 90
degrees apart in phase. And takes two channels of D/A but with limited
bandwidth under 200 KHz. Some use the sound card, some like the high end
Flex uses a dedicated D/A for greater performance. All the AGC is done
digitally, probably at audio. So the mixers and and RF stages have to
handle all possible signal levels from the antenna. There is minimal RF
selectivity and there might be an RF stage.
The third that I'm aware of is the SDR-IQ. It uses no mixers. It needs a
robust RF stage because it takes in all the RF from a few Hz through 30
or 50+ MHz and samples it fast enough to have then entire HF spectrum in
digital data. Then it uses dedicated digital hardware to tune and select
the desired individual signals, and a FPGA to further filter and detect
those signals.
Ever since Armstrong invented the superhet, mixers have been the weakest
stages in the receiver. They have to be nonlinear to accomplish the
mixing, yet in the ideal world they would not add noise or have signal
strength limits. But a mixer is always used to convert a frequency band
at its input to another frequency band. Trouble is most mixers also
convert an image frequency that has some noise and sometimes unwanted
signals and they always have two fundamental outputs, LO-RF and LO+RF.
Receivers use just one of those outputs and the other must be terminated
so the signal isn't reflected back to the mixer. Indeed signals just
outside the IF bandpass need to be terminated to prevent that
reflection. The down part of that reflection is that adds signal level
to the mixer and noise and limits the dynamic range at both ends. Many a
receiver design has ignored those termination needs. Phil Maas book on
Mixers says improper terminations can cause as much as a 30 dB reduction
in mixer dynamic range.
It is possible to extend the top end of a ring type mixer with more
diodes and more rings. The trade off is the mixer that handles more
signal requires more LO power. The standard mixers take 5 mw, or 7 dBm.
The most robust mixers take up to +27 dB, 1/2 watt LO power. One common
receiver need is to not radiate the LO, so that added 20 dB LO power can
mean an added 20 dB LO signal on the receiver antenna connector
compromising the neighborhood (really crucial when its a military
receiver) and imperfect shielding gets LO to stages that don't need to
see it.
The most robust mixers today use FET based signal switchers, like the
Softrock and achieve better performance, especially strong signal
handling than the ring mixers. The FST3523 is a favorite right now.
Fundamentally its much the same as the diode ring mixer because the LO
switches the diodes in the ring on and off acting as imperfect switches.
The FST3523 comes loser to an ideal switch.
It is possible to use two mixers with quadrature LO to get SSB or image
reject mixing, but its not clear whether the added complexity improves
either noise or signal handling. The unwanted input gets cancelled when
the outputs of the two mixers are combined, so each mixer still has to
handle all the signals in a single mixer. The unwanted products reach
the combiner out of phase and so are cancelled if the gains and phases
match. But those unwanted signals still are at full level in the mixers.
And the internal mixer noise at the unwanted image frequency isn't
correlated, so won't cancel. If the RF stage noise is stronger it will
get canceled but not the mixer noise.
The transformer feedback circuit of the Corsair family has been Ulrich
Rohde's favorite (receiver design guru who has multiple books out, the
main one it at least its third edition) but its patented by Anzac so may
be expensive to design into a receiver. Its benefits are that though it
uses a bipolar transistor it has low noise and very good strong signal
handling when that transistor is a RF power transistor. Hence the desire
for other circuits, like the large JFETs (no longer available), or
several JFETs in parallel grounded gate.
MiniCircuits is a common source of packaged diode ring balanced mixers
and has several useful application notes on line.
http://www.minicircuits.com/
73, Jerry, K0CQ
On 12/17/2010 8:28 PM, CSM(r) Gary Huber wrote:
Jerry,
Would you care to comment on
http://lists.contesting.com/_tentec/2002-10/msg00602.html (diode
ring mixer and transformer feedback circuit of the Corsair) compared
to the recent receiver designs of the SDR and hybrid SDR radios of
the past few years? Thanks.
73 es DX,
Gary - AB9M
|