Tony Brown's comments on PIMs are excellent and Zvi Frank and
undoubtedly others fighting PIM have a real problem. I would like to
add further experiences to the cooking-pot that may be helpful in this
tough business.
The JPL work was difficult in that the "dynamic range" of diplexed
transmitter power and receive sensitivity (a 10% band separates the
T-R functions near 2 GHz) was well over 250 dB (2 transmitted CW tones
each of 40 kW peaking at +82 dBm with a receiver threshold well below
-170 dBm [in a1Hz bandwidth]). The simultaneous transmit-receive
system was mainly of Copper waveguide with some machined Aluminium
parts, with this arrangement "feeding" a large two-reflector ground
antenna. The power amplifier was a single CW klystron operated in a
linear region. The twin carriers were combined at low power prior to
the final amplifier. PIM (the distant ones as will be explained) were
undetectable at the output of the final amplifier.
The troublesome PIMs were roughly the 30th spectral lines away from
the twin carriers, falling within the receive band. In the literature
one must take care not to confuse the spectral line index "n" used
above with the definition of "intermod order". The relationship
between PIM frequency and "intermod order" is (for two carriers) :
f = mf1 - nf2 where f1 and f2 are the carrier frequencies, m and n are
positive integers, and the intermod order is then m+n. For PIMs
centered about the carriers (instead of centered about harmonics),
m - n = 1. So the "intermod order" for twin carrier systems is equal
to (2n + 1). The 30th spectral line in the JPL work corresponded to
an "intermod order" = 61, a very high mix.
The conversion efficiency for such high intermod orders is low but can
still be troublesome; in the JPL case the average efficiency was about
-220 dB (the transmitter peak power of +82 dBm causing -135 dBm
average received PIM with peaking 30 dB stronger. This level and
general behavior of PIM was experienced on two otherwise good
everyday-operating antenna systems with remarkable similarity in
levels and other characteristics (variance, bandwidth, simultaneous
noise bursts and others). It was chaos repeated. It was as though a
deaf chorus was practicing without benefit of a leader.
The PIM as generated from extremely pure transmitter tones had about
10 Hz or so bandwidth. The hypothesis is that the noise bursts
modulate the PIM. Another clue but not understood lies in consistent
observations that twin 40 kW carriers (160 kW peaks, amplifier still
linear) are more effective in stimulating noise bursts than a single
tone of 400 kW (amplifier still linear). The effectiveness is evident
in both amplitude and time-density of the bursts.
It seems that the incredibly thin skin depths at high frequency cause
extremely high and perhaps not generally appreciated current
densities. These currents forced across microscopic imperfections
must cause microarcs and/or tunneling and perhaps other diode effects,
(MOM, metal-oxide-metal), all nonlinear, leading to noise bursts and
PIM. Perhaps lower power does not cause "arcs" (and thus not the
wideband noise bursts) but that tunneling or other "rusty-bolt" diodes
continue to cause PIM. At the microscopic scales in play, the world
is a filthy and corroded place and difficult and sometimes impossible
to manage.
With a very considerable effort including patient management support
over a 3 year period some 30-45 dB improvement was realized, every dB
gained as in "pulling teeth". At one point a sensible manager
requested a down-to-Earth, practical, "farmboy" explanation as to what
was going on for so long. The manager was told the situation was as
though the farms' pond was overrun by 10 000 croaking frogs and it had
to be "cleaned up". (This, prior to the more recent emphasis on
maintaining the natural environment). Fully the first 5 000 frogs
have to be removed before a 3 dB improvement will be noticed; a barely
discernable difference. The manager then expressed thanks for the
clarity and requested to be informed when we got 9 000, and then 9
900, etc. eliminated. Note that a 40 dB improvement means there's
still a croaker out there, probably two, and "regrowth" would seem
inevitable.
Much of the work followed the mantra: "If you can't preferably weld
it, insulate it". Some success with joining dissimilar metals used a
very thin layer of silicone grease on each (flanged) surface to be
connected. Apparently this prevented at least macroscopic oxides.
Choking and/or shielding gave some help. Absorber, though sometimes a
fire hazard, was tried and was not effective.
The final conclusion of the JPL work was that, while we could contain
the PIM's (the old acronym was IMPS) in a controlled setting at 40 dB
or so below our ordinary stations, we could not come close to
guaranteeing "regrowth" with time in the real-dirty-world and hectic
operational scenario.
The requirement for simultaneous twin carriers in a practical
environment could not be met. Time-domain multiplexing finally turned
out quite sufficient for the task.
Early T-R ground antenna systems operating in narrowbands near 890 and
960 MHz used large (3-1/8 inch as I recall) rigid coax, the kind that
requires the center conductors to be joined using "bullets". Bullets
are short slot-compression center conductor "bridges" between adjacent
pieces, the outer conductor being flanged. Well before 10 kW single-
tone, the bullets would "arc and spark" as heard in the receiver,
particularly as temperature varied and the bullets slipped a little.
The narrowband solution was a clever non-contacting bullet based on
"resonance" developed by the RANTEC company in the early 60's.
The first experience with PIM seems to be from the late 20's or early
30's. Powerful Radio Luxemburg was mixing with another European
broadcaster, the mixing occurring up in the ionosphere. The BSTJ has
articles by Babcock (1953) and Acompora (1972) and by Rootsey of
Philco (1972) and there are many others. PIM are found from the decks
of solid ships to flimsy spacecraft. PIM has been predicted to be
"The most serious problem limiting future comm systems" and I believe
it.
Tony Brown's advice about "painful measurements" is not to be taken
lightly. In fact I do not know of any measurement in all of physics
that covers a dynamic range in the 200-250 dB arena (10E20, 10E25 are
Really big numbers). Does anyone know of one??
Dan Bathker
At 11:04 PM 9/21/2000 +0100, Tony Brown wrote:
>
>We have had a lot of experience in the past with PIMs at microwave
>frequencies in waveguide arrays and reflector structures, particularly
>satcom, also some at UHF with broadband structures, both printed and
>otherwise. It's difficult to generalise but the following I hope is of
>interest....
Received on Mon Sep 25 2000 - 03:08:54 EDT
This archive was generated by hypermail 2.2.0 : Sat Oct 02 2010 - 00:10:40 EDT