One of the big considerations in receiver design is image rejection.
When the superheterodyne technique was first employed, the received
signal was down converted to a lower frequency at which hi-Q, stable LC
tuned circuits, with fairly critical adustments, could be built and
aligned for that fixed frequency. Since this intermediate fixed tuned
amplification stage was at a frequency in between the received RF
frequency and the detected audio output frequency, it was called the
Intermediate Frequency, or IF. Better selectivity AND frequency agility
could be realized by having fixed tuned filters in the IF, so that they
are set when the receiver is built and aligned, and not adjusted during
operation. Only the local oscillator and perhaps some fairly broad
bandpass filtering in the mixer or first RF stages was adjusted during
operation of the receiver. This was usually done using multiple ganged
variable capacitors. Typically the intermediate frequency was in the
range of 50 kHz to 500 kHz. Probably the most common IF frequency ever
used is 455 kHz.
This scheme works quite well for medium wave AM broadcast receivers. To
down convert a 1 MHz signal to 455 kHz a local oscillator frequency of
1.455 MHz can be used . (Ever notice how a lot of multiple gang variable
capacitors have one section with much smaller plates? That is for the
local oscillator operating at a frequency above the receive frequency)
With an LO or 1.455 MHz both 1.000 MHz and 1.910 MHz could be down
converted to an IF of 455 kHz. One is the desired signal and one is the
"image" response. The image is 2X the IF frequency away from the desired
signal. In this scenario the image is almost a full octave away, so the
tuned mixer or RF stage can easily reduce the image response to an
acceptable level. When the same low IF frequency is employed in a
shortwave receiver, the image is still 910 kHz away from the desired
frequency, and 910 kHz is a much smaller proportion of, say, 7 MHz.
Nowhere near an octave away. The tuned RF and mixer stages would have to
be much higher Q to acheive the same image rejection ratio. Much more
difficult to have the tuned circuits using multiple ganged capacitors
track each other precisely enough when the Q gets higher. (Ever notice
how some multiple ganged variable capacitors have slots in the outer
plates of each section, that can be bent to make fine adjustments of
that section for a particular shaft position?) So cheap shortwave
receivers using to old MWBC standard IF usually had very poor image
rejection.
One solution is to use a higher IF. Since the image is 2X the IF away
from the desired signal, it can be rejected more easily when the IF is
higher. If you use a high enough IF, and get the image far enough away
from the desired receive frequency, you can use fixed tuned bandpass
filters in the RF and mixer stages, or broadband transformers, and throw
out those multiganged vairable capacitors. (Really they go in the junk
box and later become loading capacitors for 500 watt amplifiers) With
much higher IF the construction of sharp tuned IF filters becomes
different. Using LC circuits it is more difficult to get the same
selectivity when the IF is several MHz instead of a few hundred kHz. So,
enter the dual conversion superheterodyne. The received signal is first
converted to a fairly high IF, which helps with the image rejection, and
then down converted to a lower IF which helps with the selectivity. This
technique is not without it's pitfalls though. Now there are two local
oscillators (not including the BFO) and other possible undesired mixes,
or images. With the proper choice of the IF frequencies the image
response problem can be managed without too much difficultly as long as
you don't try to cover the whole HF spectrum. If you make it general
coverage there are going to be areas where the image rejection is poor,
or where one or more of the local oscillators or their harmonics
interfere with the received signal.
You can also use a high IF (such as 9 MHz) without going the double
conversion route, by using crystal filters instead of LC filters in the
IF. This will reduce some of the problems with multiple images. Still
there will be a frequency range where desired receive frequency is very
close to the image. I would guess that if 30 meters was one of our ham
bands in the 60s, 9 MHz would not have become such a common first IF.
If you want general coverage you can get around a lot of these problems
by first upconverting to a really high IF, such as 45 MHz or 70 MHz. Now
the image is so far away, for whatever frequency you are tuned to in the
HF range, that images can easily be rejected. The LO has to operate at a
much higher frequency, and could probably not be made stable enough
using a mechanically adjusted variable capacitor or permeability tuned
oscillator. Only a crystal controlled oscillator or a phase locked loop
oscillator is likely to work with this upconversion scheme. And since
the conversion is to a higher frequency thats is not intermediate
between the received frequency and audio, it isn't really and
intermediate frequency anymore. Still, the habit has been well
established, so we still call it an IF. At least it is still and
intermediate stage, so what the heck, call it an intermediate frequency
anyway. Upconversion solves some problems and creates others. More
difficult to make narrow filters, noise in PLLs, etc, etc.
DE N6KB
|