After O4a, we succeeded to reduce IPC glitches drastically in klog#29110.
But glitch rate increased gradually with many updates of real-time model after O4a.
Reduction of data rate is one of the most simple solution for this issue.
(I have some other ideas but they probably need a down time of DAQ system in several weeks.)
So I listed up fast channels and show channels which may be unnecessary during observing run.
After the end of O4 they can be backed to the channel list.
-----
A current number of channels of each sampling rate is as follows. The 1st and 2nd columns are the sampling rate and a number of channels, respectively. A channel number equivalent to the 16Hz sampling is also shown in the 3rd column.
| fsample[Hz] | # of chan | 16Hz equiv. |
|---|---|---|
| 16 | 179826 | 179826 |
| 32 | 0 | 0 |
| 64 | 0 | 0 |
| 128 | 0 | 0 |
| 256 | 2044 | 32704 |
| 512 | 1309 | 41888 |
| 1024 | 592 | 37888 |
| 2048 | 888 | 113664 |
| 4096 | 29 | 7424 |
| 8192 | 0 | 0 |
| 16384 | 203 | 207872 |
| 32768 | 0 | 0 |
| 65536 | 10 | 40960 |
Below is a list of those that may be removable (I only checked 1024-65536Hz sampling). If we can remove all of them, it becomes ~15% reduction of data rate. Attached file is all channels for each sampling rate.
65kHz channels
They are related to the beacon system and they are mainly used for Q-value measurements. Q-value measurements will not be done during observing runs probably. We can removed them during O4c. When we will do these measurements in post-commissioning, they can be restored.
16kHz channels
14 GIF related channels:
Fringe analysis on GIF model is no longer working for a long time due to an absence of management people. So they should be removed.
6 OPS related channels:
They are the test channel for calibration stream. So they are sometimes used during non-observing period but not used during observing period. There is no problem to remove them.
5 PSL-REFCAV and PSL-TTFSS related channels:
There is no longer REFCAV. We may be better to ask IOO team to remove the model itself after O4c. Anyway these channels can be removed form DAQ list.
4 CAL-INJ related channels:
There is no HW injection plan during science-mode (at least I haven't heard). So we can remove all data with the excitation flag from science-mode. In this situation, we don't need to record injection waveform itself and recording excitation flag is enough.
4 CARM, SRCL IN/OUT channels:
They don't use and can be removed during observing-mode. When we will do RSE trial in the post-commissioning, we can restore SRCL related channels.
4 Type-B BF_LOCK_L channels:
We don't have BF damper for TypeB.
3 PEM-PORTABLE channels:
If no sensor is connected to these channels we may be able to remove some of them. I'm not sure the current status of PORTABLE channels. So we need to ask PEM team.
2kHz channels
89 PEM-PORTABLE channels:
Same situation as 16KHz PEM-PORTABLE channels.
22 AOS-NAB channels:
They seems to be old informations. Current NAB has only baffle PDs and the are recorded as K1:NAB-.*
11 VIS-GIF channels:
Same situation as 16kHz GIF channels.
2 SRCLFF channels:
Same situation as 16kHz SRCL IN/OUT channels.
1kHz channels
45 Type-B BF related channels except BF_GAS:
Same situation as 2kHz BF_L channels.
