***** Error History *****
1. Magadan induction magnetometer has a GPS problem from Sept.22, 2009.
The 64Hz sampling may not be corrected by GPS signal. The PC clock
is corrected by GPS. So timing accuracy would be about 1-second.
This problem was fixed by a new GPS receiver on Jan.12, 2010.
2. Fluxgate magnetometer at Moshiri was damaged by lightening from
June 2009. A new fluxgate magnetometer was installed at Moshiri on Oct. 15,
2009.
3. VLF antenna construction occurs near Kagoshima fluxgate magnetometer at September 28
- November 2009. The data may become noisy sometimes due to the construction.
4. Time tag of Moshiri induction magnetometer occasionally become unstable and
shift maximum 10-100s for the data before October 26, 2009. (See (b) below).
5. A 4-28MHz transmitter was working at Magadan for Jan.12-Mar.7, 2010,
Jan.13-Feb.29, 2012, and May 2-31, 2012. This causes pulse noise
with an interval of 5-min on the induction magnetometer data.
6. Sept.7, 2009-June 2 0110UT, 2010: RIK Time tag data was shifted
a few min to 2-3 hours due to the lack of clock correction.
7. April 1-30, 2010: PTK, Time tag is somewhat irregal??
8. July 13-Sept.14, 2011: MSR, GPS time correction did not work on the
PC for the induction magnetometer. The time was shifted 9 second
when we fixed this problem on Sept. 14, 2011.
9. Moshiri: Sinusoidal noise with a period of tens of seconds sometimes
appears in all three components after January 2010.
10. Moshiri: Z-component of fluxgate magnetometer data shows strange
offset variations, possibly because offset adjustment circuit is
not correctly working, noticed on May 30, 2013.
11. At Athabasca, we noticed that the windows clock run without
GPS clock and network clock. It was 3.5 sec different from
correct clock at Sept.12, 2013, and corrected using the network clock.
This may be due to windows did not refer to network clock since
July 2013. The GPS clock could not generate time correction signal
since September 2012. Thus we set the windows to refer to the
network clock since then. The GPS clock was replaced to a new
GPS clock at 19 UT on November 19, 2013.
12. At Paratunka, water leak happened in the H and Z sensors of the
the induction magnetometer. Because of this leak, the data become
highly noisy. We opened these sensors and dried them from Sept.11
to Nov.20, 2018. During this period, the H and Z sensors may be swapped
sometimes, mainly to keep the H-component measurement. Detailed history
of the sensor swap was recorded.
(a) Caution about the time tag of 64Hz sampled digital data
(Oct. 12, 2009 K. Shiokawa)
13. We found that the clock of the fluxgate magnetometer and 3ch photometer
at Rikubetsu, Japan, is 28-min delayed on Feb.14, 2022 due to error of
NTP time correction for long time. We could not identify when this
error has started. The clock error was corrected on Feb.14, 2022.
According to the recording history, this error seems to start
from the second half of January 2022.
The recording of the induction magnetometer data is done by a Windows PC,
which has a clock with a unit of 1-second. The PC clock is corrected
through a network or GPS every 10 minutes. The PC clock runs freely during
the 10 mintues. So if the PC clock proceeds/delays for more than 0.5
seconds in that 10 minutes (which usually do no happen), the time tag may
become wrong with a multiple of 1 second.
The sampling of the magnetometer data is done by using 64Hz trigger signal
generated by a GPS receiver. If the GPS receiver fails to receive the radio
wave from the GPS satellite, it runs freely by an internal clock.
But usually the receiver receives GPS radio wave.
The PC recording is first into a PC memory and then written into a
hard disk every 4 seconds. Since the sampling is not by the PC clock
but by the GPS clock, the start of the time tag cannot be exactly
at 00h00m00sUT.
In the 4th channel, the 1PPS signal from GPS is recorded as a time standard.
If the timing accuracy better than 1-sec is needed, this GPS 1PPS pulse
should be used to correct the time tag of the data file.
(b) Further Caution about the timing error of 64Hz sampled data
(Oct.26, 2009 K. Shiokawa)
On Oct.26, 2009, we found a problem in the Windows PC clock at Moshiri.
The Windows PCs automatically correct their clock to Windows time server every 7 days.
We found that when this correction by the Windows finish successfully, the
Windows clock becomes unstable and start to shift about 10ms per 4 seconds.
(about several tens/hundreds seconds per day). Usually we turn off
this automatic correction (click the clock of the windows and see the
internet time option). But at Moshiri, this was ON by some unknown reason.
Once the clock becomes unstable, it lasts until we restart the PC.
Since the Windows clock is corrected by GPS clock every 10 min via
RS-232C, the time tag of the file may not shift much. But if GPS
satellites was not available for a certain time, the time tag error
becomes large because of the unstable clock. The 64Hz sampling
and the GPS 1PPS signal at the 4th channel
should be correct. You may find the occation when the GPS
correction was made on the clock, by checking a sudden jump
(more than 2 seconds) of 4-sec time tag in the file.
This error would not happen at Magadan, since the PC is not connected
to the network. At Sata, the network condition is not good, so this error
would occur less frequently. For Paratunka, we confirmed that the
automatic correction of Windows clock is off (Oct.29, 2009). So this
error would not happen.
Please be careful if you want to compare the data with accuracy of
less than 10-100 sec.
ISEE Magnetometer Home Page .
> Division Home P
age .