Rev B
26
Here the system response is slower, and detectability of repetition coded
transmitters (with their obligatory longer group size) is especially poor. What is
happening is that so many noise events are being interpreted as impulses (i.e.,
ignored) that the local gain reduction algorithm has failed to reduce the gain
sufficiently to clear away the other (non-impulsive) noise components.
Disabling the noise blanking algorithm altogether had this effect:
10/11/89 14:45:11 151.450*64*78 151.138*96*107
10/11/89 14:46:03 151.450*64*92 151.138*96*91
10/11/89 14:47:03 151.149*2R*183
10/11/89 14:48:34 151.450*63*71 151.138*96*79 151.149*2R*164
Again system performance is slow, and the 3 beater at 151.158 is not detected at all,
this time because the strong interference could not be sufficiently suppressed by
gain reduction methods alone.
With the noise blank level = 64,
10/11/89 14:51:15 151.450*64*90 151.138*97*98 151.149*2R*178
151.158*3R*96
10/11/89 14:52:07 151.450*64*101
10/11/89 14:53:05 151.149*2R*173 151.450*64*82 151.138*97*77
10/11/89 14:54:09 151.149*2R*186
10/11/89 14:55:14 151.450*64*98 151.138*83*75 151.158*3R*82
the 3 beater is back. Trying blank level = 54,
10/11/89 15:01:05 151.149*2R*169 151.158*3R*135 151.450*58*102
151.138*96*87
10/11/89 15:02:05 151.149*2R*173 151.450*64*95 151.138*96*98
10/11/89 15:03:03 151.149*2R*164 151.158*3R*132 151.450*64*66
10/11/89 15:04:25 151.450*64*89 151.138*96*84 151.149*2R*172
151.158*3R*80
10/11/89 15:05:04 151.450*64*88 151.138*96*92 151.149*2R*175
10/11/89 15:06:08 151.158*3R*120 151.450*64*79 151.138*96*78
151.149*2R*189
This is quite good performance for the conditions.
On the assumption that starting with a lower baseline gain should speed up
acquisition we next tried setting the global noise threshold so that the system would
settle at a lower gain. In this case the target was 56, the highest gain without noise
triggering in
SIGNAL
/Interval
. Because the noise blanker is working, we must
expect to set the global noise threshold fairly low (since noise rejected by the noise
blanker is not logged, and therefore not counted by the gain reduction algorithm).
Trying a value of 4, the gain settled at 58, giving
10/11/89 15:34:06 151.450*66*120
10/11/89 15:35:10 151.149*2R*172 151.158*3R*127 151.450*64*87
151.138*96*93
10/11/89 15:36:05 151.149*2R*163 151.158*3R*97
10/11/89 15:37:11 151.450*64*103 151.138*96*79 151.149*2R*181
10/11/89 15:38:14 151.450*64*103 151.138*96*85 151.149*2R*172
151.158*3R*123