unreleased - Pending changes
Raw data dumps can now write text csv or netcdf, for either 2K or 4K data. There is an algorithm which tries to detect contiguous blocks of data, then adjusts the timestamps of subsequent blocks to match the sampling rate, to account for clock drift between the ADC clock and GPS. New output files are started when there is missing data (meaning GPS sync was lost or a buffer overrun occurred) or clock drift gets too large.
Use the new --precision flag to data_dump to print data values with full
precision. dump_hotfilm.py must be run with the nidas
buster branch, since
--precision only exists on that branch.
1.2 - 2023-09-07
In support of 4K sampling, constrain the scan rate option to divide evenly into 1e6 microseconds and be divisible by the allowed read rates. The read rate can be 1, 2, or 4 reads per second, in case modifiying the size of each read (which also depends on the scan rate and number of channels) improves the overall network throughput and avoids dropped scans from device buffer overflows.
1.1 - 2023-08-09
Set STREAM_BUFFER_SIZE_BYTES in the LabJack to the maximum value of 32768, about 2 seconds of scans, in an attempt to overcome probable networking delays causing a buffer overrun in the device and resulting in dummy values (-9999) in the data returned by the LJM library.
Detect -9999 in pps_count changes and log them as notices, so that lost
scans can be seen in the journal and not only by looking at the data.
Log version information on startup.
1.0 - M2HATS first release
This is the version of the acquisition code (hotfilm.cc) that has run since M2HATS started. A few changes related to installation and monitoring were made soon after the project started, but this version contains those files as they have been since then for the bulk of the project.
There are scons targets which install the hotfilm executable into the NIDAS bin directory and also set capability flags on the file to allow realtime scheduling priority without running as root.
Most of the LabJack stream configuration registers can be modified through command-line arguments, including rate, resolution index, number of channels, range, settling time, and whether to enable the PPS counter.
Viewing the data with tag information, for either the real-time network stream or data files, requires changes being committed to the NIDAS buster branch.
Uses the LabJackM streaming mode and API to accumulate scans at 2 KHz into 1-second NIDAS samples and write them to the sample outputs configured in the XML configuration file. Sets realtime FIFO scheduling priority to minimize latency. Timestamp synchronization seems to work reliably, but might need improvement.