Michael,
Thanks for your thoughtful comments. A few responses:
"One possible answer is that in the artificial lab environment,
we have eliminated atmospheric noise, so we are able to
listen down to the receiver MDS where we can hear the
broadband noise from a strong signal, whereas in real life
this broadband TX noise gets covered up by atmospheric
noise which can easily be several orders of magnitude
stronger than a receiver's input thermal noise (especially
on the lower HF bands)."
Several people have suggested this, but as mention further down in
your email, we're not talking about spurious signals that are 20 or 30
dB above the receiver MDS. Phase noise and intermod products from
signals that are 20 or 30 dB above 50 µV should be close to the middle
of the dynamic range of the receiver. And our on-the-air experiences
don't always corroborate this.
I hear the clicks and mush that Tom's descirbed, but at other times I
also seem to be able to hear and work stations that should be buried
under the spurious stuff emitted by these radios. Why is this?
"Another explanation (which Tom has offered) is that the
signal we normally receive just don't get that strong (e.g.
a real S9+50dB signal is rare)."
I agree with the comment. I've measured the behavior of the S-meter
on my Mk V and found that in the middle of its range--S7 to about 30
over S9, it's fairly accurate. Above and below these levels, it isn't
even close. Still, a signal that runs the lighted bars all the way to
the end of the scale is very strong--at least 40 dB over S9, which is
on the order of 5 mV. Again, it shouldn't be possible to operate
close-in to a signal when its sideband energy is over 150 µV.
"Another possible explanation for the seemingly dirty
signal is measurement error. One difference between a
real RF link and a simulated link is proximity. In a real
RF link, the stations are miles apart. There are no sneak
paths. In a simulated link, however, the transmitter and
receiver are in some cases right next to each other.
They probably share a ground connection vis-a-vis
the shield of the coaxes and attenuators connecting
them. There is also the common connection to
the AC mains."
This is the reason that I've switched to making measurements on one of
the metallic surfaces we use for military and aircraft system tests.
The metal top of the bench goes to a ground system through 5 inch-wide
copper strap. The instruments and EUTs are all bonded to the metal
surface with copper tape, and there's an attenuator in every RF path
to improve return loss at least a little. I haven't been able to find
big differences between the numbers I take at the lab and the ones I
take at home. But I'm sure that measurement error is lower and
repeatability is better when I work at the lab.
The aim of all this is to measure behaviors that predict how radios
will respond when they're connected to antennas. I believe I can hear
the differences between radios that do well in IP3, IMD, and BDR tests
and those that don't. Friends have been kind enough to let me borrow
their rigs for days and sometimes weeks at a time so that I could use
them under a variety of band conditions. For the most part, what I
hear corroborates what the numbers, particularly BDR, tell us about
the radios.
But in extreme circumstaces, I have not always seen the very bad
receiver and transmitter behavior that the numbers predict. At our
old QTH, I was two houses away from Alan, K5AB, who confirmed over 100
countries on 160 using a 1000D, a Henry amp, and an inverted V. We
could operate simultaneously if we stayed 15 or 20 kHz apart and I was
using my TS-930. No amount of separation was adequate when I used the
TS-440.
"If the signals you are describing really were
S9+60dB, then that would imply a receiver input level of
-10dBm (a very very strong signal). If we go out on a limb
and assume an accurate s-meter, then at 6dB per S-unit,
your S3 noise floor would be at -106 dBm. This implies
a blocking range of 96dB which is nothing special at all
(at least at 5 KHz spacing). In the example you gave,
the signal you were working was S7, which in the case
of an S9+60dB adjacent signal would require a
dynamic range of 72dB. Again, nothing extraordinary, so
I don't think the examples you cite make a strong case
that IMD and BDR specs are not good predictors of
real world performance."
I think your comments underscore the point I was making. If copying a
relatively weak signal next to an S9+60dB signal on a quiet evening
only requires a 72 dB receiver dynamic range, then pressing for much
higher IMD and BDR numbers may not be so important. I hadn't thought
about this in numeric terms, but this does explain why I was able to
have those contacts. The situation we've outlined is realistic as
concerns the weak signal and the noise floor, but it's unrealistic as
it concerns the adjacent channel signal--under most circumstances it
would not be 60 dB over S9. It would be considerably weaker, and the
dynamic range required would be less than 72 dB.
So are we really interested in high numbers so that we can copy weak
signals among strong ones? Or do we need the high numbers for
immunity to atmospheric and man-made noise? Or is there another
reason?
Brad, KV5V
_______________________________________________
Topband mailing list
Topband@contesting.com
http://lists.contesting.com/mailman/listinfo/topband
|