Phase noise has multiple characteristics. When coming from an analog
oscillator its pretty much random noise, causing a spreading of the
signal. And that can lead to strong adjacent frequency signals being
mixed into the IF and sometimes stronger than the desired weak signal
despite the "roofing" filter and mixers being capable of handling the
strong signal with intermod. That's reciprocal mixing. I look at the
transmitted phase noise plots and presume they are the same for the
receiver oscillators.
Phase noise from a synthesized radio tends to have peaks of noise often
corresponding to the PLL reference frequency and its harmonics, though
fractional dividers can add their own peaks. That type of phase noise
has discrete tones. When I put a clean CW signal on the steep slope of
the CW filter in my FT-726r that steep slope converts the FM noise to AM
noise and there's considerable buzz amplitude modulation to the CW
signal. Clearly the products of a PLL synthesizer.
In measuring a receiver, the effects of reciprical mixing and mixer
overload have the same result, that the strong unwanted signal reduces
the S/N of the detected weak signal in the pass band. With filter skirts
and the tendency for the "roofing" filter to be 14 or 20 KHz wide, both
tend to have more effect as the unwanted signal gets closer to the
passband. Some of the European ham magazine receiver checks do that
blocking or IMD test continuously over a range of +/- 50 KHz. In most
cases a combination of reciprocal mixing and second mixer overload are
very pronounced at the smaller frequency separations. Most modern
receivers tend to not change between 20 and 50 KHz because of the slope
of phase noise and the bandwidth of the "roofing" filter.
The roofing filter could be narrower bandwidth except for a few things.
One is the FM receive option needs the wider filter. Another is that its
inconvenient to make a SSB bandwidth filter at 45 MHz or 75 MHz, the
typical first IF of a continuous coverage receiver these days. And I
suspect that the first conversion oscillator has large steps (10 KHz)
for easy PLL synthesis and the second conversion oscillator has fine DDS
steps over just that 10 KHz range for the fine tuning, that even for
narrow modes the entire roofing filter bandwidth is required. Which
means that the first IF gain stage and the second mixer need to have far
better IMD characteristics than they typically do. That won't cure
reciprocal mixing problems though.
73, Jerry, K0CQ
--
Entire content copyright Dr. Gerald N. Johnson. Reproduction by
permission only.
|