Elekta Neuromag NM21283A, Руководство пользователя

Страница: 45 / 85

Manual tuning 7
NM21283A-A 39
with an unsymmetrical distribution, probably the offset needs read-
justment
7. If the noise level is still higher than in other channels, check the current
vs. flux characteristics again and try readjusting the bias (step 3).
7.7 Typical sensor problems
This section describes several types of sensor problems and the proce-
dures how to deal with them.
Tuning the sensors in presence of strong disturbances, may cause the
noise level measurements to be affected by them. This appears as abnor-
mally high average noise level on magnetometer channels. This external
noise masks the sensor noise, and makes the tuning difficult and inaccu-
rate. In such conditions, SSP should be used in noise measurements. See
Section 11.8 on page 59 and Section 8.6 on page 47.
If the current v.s. flux curve shows hysteresis, this is most often due to
trapped flux in the sensor. To cure the problem, heat the sensor. If this
does not help, try repeating the heating a couple of times, or try changing
the heating time temporarily to 2 seconds.
Sometimes the current v.s. flux curve is tilted, that is, one end of the curve
is higher than the other as if the zero line was tilted. This is most likely
due to moisture in the preamplifier connector on the top flange of the
dewar. To fix this problem, the connectors must be dried. Tuning is not
likely to remove the problem. This kind of phenomenon may also be
result of other types of hardware failures.
If the real time signals (on raw data display) in the normal measurement
mode show jumping between two levels (also called “popcorn noise”),
most likely reason for this is trapped flux. To cure the problem, heat the
sensor.
If the real time signals show low frequency drifting in a small set of chan-
nels, this may due to oxygen snow in the dewar. If this is the case, only
way to remove the problem is to heat the whole dewar. This can be per-
formed only by skilled maintenance personnel. Note that this kind of phe-
nomenon may also rise due to bad SSP vectors. Check carefully the
channels both with and without the SSP and compare the behavior to good
channels.