|
) ^1 K; E; h5 u8 a5 \0 t
Cross-talk between FM Broadcast Radio Transmitters (88-108 MHz)
5 F: s+ i7 |4 \: T& [ r5 J# {and NMR Spectroscopy: A recent experience
& z- {& G1 P4 C1 b) E& w ( J# [5 _8 j8 ~2 m% f
Recently I had to install a 400 MHz (9.4 T) NMR Spectrometer. The
) j! `3 ]/ K" O2 m' gsystem worked fine and, using an Indirect Detection Probe , met quickly
* x2 J# j5 s; e( tand effortlessly the specifications. But later on, when the customer
: V; d/ f- O- O2 c: Ninstalled a 13C direct detection Probe, the S/N ratio turned out to be
: I7 ?" @. j yquite low and, on top of it, the sensitivity was subject to erratic and. s# }0 T+ _& |; }" D* q
very large variations from 50:1 to 130:1 (manufacturer's specs give: D; d% m3 g( l# ~: j5 z0 V
155:1), without any apparent reason.
- [9 P0 C. ~) }5 ] Since the nominal 13C observe frequency at 9.4 T is 100,568/ v2 f7 U) F" e
MHz, right in the middle of the range of commercial FM broadcasts, I
$ l2 J1 j6 ~ V1 g& ^. g; ]have immediately suspected that the spectrometer was picking up one of! R& h- z0 G6 ?. _7 G0 B y
those radio stations. In fact, using a cheap FM radio receiver, a
0 l( W; ^- G, v7 Jstrong station was quickly found at 100,60 MHz. At this point, I have, D0 j3 _8 `8 a' |4 U* B# P- l* d C
connected a simple audio amplifier ending with a loudspeaker to the$ f5 v6 `$ `# g/ K/ R! |
output BNC of the observe receiver which was there apparently just for
8 W5 c- r) g5 r' q$ e( F0 m2 x/ g# Hthis purpose, and all of us were listening to the radio using a 200.000- R. o: r6 }9 p3 ^) \; I ], }
Euro NMR spectrometer, except that the audio quality was really poor,
v7 o$ i8 f* I( Tmuch worse than from the above-mentioned gadget radio (a shopping mall: G2 U6 F; f6 o8 C2 Z# ~
gift).8 ~/ C, m( B' Z. m' |$ q
The problem is well known from the old times, when the highest
! q$ n+ `+ |6 J. X9 D+ efield was 2.45 T and the nominal H1 frequency was close to 100 MHz. One: z" B, U1 g" J) A9 ~3 d- ^/ |
of the first Italian NMR spectroscopists to experience it, back in
1 i* ~0 k& d9 I- ~1 I1974, was Prof. L.Lunazzi at University of Bologna, on his brand new
$ a Q4 E0 r# wVarian XL100 spectrometer, and the radio station was Radio San Luchino,
0 X! z+ o% c8 S* m* g/ q0 [. F2 Ewell known to anybody living in Bologna, which broadcasts from the top0 A1 u- A) E6 v. w
of the nearby Saint Luca hill.1 T* g0 A! S! K' X6 U
The obvious solution is to change the magnetic field, and thus
8 |6 @- ]5 z. T5 aall resonance frequencies, in order to get out of the modulation+ H+ Q& s. M- r# }) H2 |: Z' v
envelope of the interfering transmitter. But this is not always easy," F4 ?# M% M. q6 X
since the range by which one can move the magnetic field changing just1 N5 N7 {$ }$ f! V2 C; f
some software parameters is usually limited to a few tens of kHz in the
( `" P; w/ d: k. g/ A0 d/ f3 Z3 ifrequency domain. If larger variations are required the poor engineer
1 w K' Q! s: c% y( C$ mhas to work on the superconducting coils of the magnet, which is a
5 v9 I# b5 P. y* K) c, O$ Uno-trivial job entailing the risk of a total or partial quench.
! r W! j# }- N+ ^; }1 n Bitter experience shows that persuading the involved radio, a4 w7 V3 `" q" q1 a
station to change its operating frequency is a time consuming,
1 i% e3 J$ H" k% ^+ \1 dfrustrating, and apparently quite impossible task.
& r5 Q+ @+ g. N Being well aware of the problem, my preliminary spectrometer. v. o! y, ]2 q4 Z& T4 Z3 C
checks always include some blank acquisitions taken before running up
6 X. P$ S5 u3 V1 s, i, t% _2 ythe magnet so that there is no chance to observe an NMR signal. The
! l3 g9 @- V7 ?resulting dataset should be pure white noise, without significant
) l' j2 K8 T$ |8 S$ Y$ Cspikes. This was done also in this particular installation but, as
$ D5 f9 e- s6 X) P! S) Jusual, in the days following the energization the magnet drifted a bit,
( e- A' m1 g+ |" `' hgetting closer to the radio station carrier. Furthermore, the usual 13C: g7 G. d# i2 W2 F4 |
spectral widths are quite wide which makes things even worse. Murphy's" ^. L. v) }# W- D* C1 T4 d6 @
Law has no exceptions!: C* c! x! V/ D# h2 D4 A
But we are just at the beginning of my real troubles. Before
- ^0 f/ p3 J$ t3 u+ Cputting one's hands on the magnet, one should better know how much, in* c7 M! i5 M; J% |8 F
which direction, should the field be moved. I have therefore used a" u2 O* ]. c* s5 z8 ~
good Spectrum Analyzer (Tektronix model 2710) to check the frequency
# d7 X: L# N) v Kspectrum around 100 MHz, ready for the worst. And the worst was what I
7 d: @: ^5 D% S# C( dgot! The band was filled with FM signals, evenly spaced by 250 kHz and1 f Y' V# \9 |$ W# L- g
with modulation envelopes as wide as 100 kHz, so that when I got far, ~$ H/ g% |: z& m
from one station I started receiving the next one; accounting for7 F. e v( E f/ N0 z% m
folding and aliasing effects, there was no chance! The only somewhat
! s. p3 n# Q+ t; R8 t7 P4 Kfree region was at 100,120 MHz, but this implied proton frequency of
* Y& W' e/ K. y4 O1 q398.100 MHz. So now the spectrometer is no longer a "400"!
* E: w& @, ^/ u8 I) ` Before installing a spectrometer, you better get a Spectrum. x% @; ]8 m% a# y
Analyzer and check for the presence of RF fields in the instrument4 E' X- W! u0 A7 n3 f
room, taking care to explore the areas close to the observe frequencies
9 x1 B7 Z9 D0 `9 i Jof all the most important nuclei. Don't forget the lock: at 14 T3 O7 j" ?2 H2 W) u5 o
(nominal 1H frequency of 600 MHz) 2H resonates at 92,095 MHz, once! E0 _5 i% B4 e! c, S
again in the FM broadcast band. The lock channel receiver has quite
- ~9 ^: S9 i; f" w5 q. @3 c6 n1 Fnarrow bandpass filters, so hitting a radio is a really bad luck, but
* B- r( C$ }# ?; m' |( ]it had already happened, resulting in fast lock level variations and
/ I% J r" K! [* i' wtotally malfunctioning Gradient Shimming which uses deuterium as
' v! I- L6 W8 E1 M$ g5 w- iobserve nucleus!
% I3 K1 x) x, s7 Z* j4 K( b( p5 R Needles to say, the extremely high sensitivity of an NMR5 l0 V2 @8 {3 x
Spectrometer shows up. The signal from the guilty radio, as observed on
2 q/ K! b- l1 G5 S/ f& u Ethe spectrum analyzer inside the spectrometer room, had very low, j+ h; a: c5 s/ D: u; Q
intensity level of about -70 dBm, some microvolt/meter, but that was
; e$ \- U C/ ]. B$ _enough to almost completely hide the quite strong 13C signal from the
% C7 t H" b; @3 x. k5 r" `ASTM sample! Y: n% L0 b$ ]: E: W* \
The radio was clearly picked up by the Probe (closing the
9 V" t( F9 P) m- S- v# bPreamplifier input with a shielded 50 ohm RF load, all signals
, x4 m) y- W% ?1 T' Z6 Gdisappear) but, quite surprisingly, there is almost no shielding effect
; D5 e1 A1 v+ Y- Mattributable to the metal body of the magnet, which is after all an
& n$ q/ Y: w; kalmost completely closed cylinder all around the Probe. Most probably a+ b5 v+ D* k( y% } l
good deal of the signal leaks in through the Shim Coils which are
" e4 s; O4 ?$ V7 J! J3 }mounted very close to the Probe and, together with their connection
) g" ]! I$ u+ Q2 j2 }cables to the Console, constitute a quite good antenna./ C# v# f ]* [/ V9 A
Too bad the Shim Coils are essential, and effective shielding
* J' V5 ~+ Y& zof the instrument with a Faraday's cage is always difficult and
" L7 {' X6 H9 V# n) s0 yexpensive (it may be almost impossible once the spectrometer is: ]; S! N0 S( \$ d
installed).2 a8 e% ~1 T0 L3 z* A9 y
a: t4 ?9 q) N. H- y
Before concluding, let me venture some additional advice based on my experience:
) l a0 X" ?- l& m+ Z, C1 c ; ^1 e6 G$ t: i" X! o" g: {% r% d
= Install the spectrometer in the best shielded room
% B# P9 G- Q$ \' `8 W; Cyou can get; the best choice is once again in the basement, where you
1 P8 m4 D' C/ |" X5 ?/ q' Chave the whole building above the ceiling and its [grounded]
# q/ C: D& Q) ?$ Q" Bfoundations all around the rest, done in iron-reinforced concrete,
) x8 F1 O3 k' z$ J" Z3 }7 G! Namounting to a good Faraday's cage at no extra cost.
- U6 N& L- c5 n! }
9 l- {% O) w6 ~ j( J= If possible, avoid top floors. If you can't avoid# g. L7 N0 @& P- x' |9 u
going upstairs, take a good look out of the window: if you see nearby
g5 f9 c1 u5 c9 [0 ztransmission antennas, get ready for troubles proportional to their
7 e9 W8 l7 O! hdimensions and closeness (to my knowledge, however, mobile telephony+ ]$ Y5 {; T4 ]' i5 p( P- I6 h
antennas cause so far no harm).2 d* `/ y u: A
* z. p' G3 U+ Q1 k& d8 O/ x= I'm sure that an exchange of experiences and/or' p( S5 b; S1 [9 e9 B+ R
suggestions regarding this matter would help a lot to solve many2 B B N: O3 B" I/ Z$ d
existing installation problems and prevent ones yet to come. Stan's Blog is an ideal location and, needless to say, I will be absolutely glad to cooperate.
f) o; I" F) n' ~/ Z. K
( v1 q* t4 ~, ?! ~, A1 T, d0 WVanni Piccinotti, Firenze, 11 April 2008
摘自stan' NMR Blog.
| 
发表于 2008-7-1 02:51:48
