A basic CLI for the Jetperch Joulescope JS110 under native Win32 libraries. The current goal of this project is to provide a driver for EEMBC's benchmark framework, so it may appear a bit opinionated (e.g., timestamp behavior). This may (or may not) change in the future.
Starting the program initiates a simple command-line interface. It is intended to be used through a bidrectional pipe/IPC, rather than a user typing instructions. Here are the commands:
deinit - De-initialize the current JS110.
exit - De-initialize (if necessary) and exit.
help - Print this help.
init - [serial] Find the first JS110 (or by serial #) and initialize it.
power - [on|off] Get/set output power state.
rate - Set the sample rate to an integer multiple of 1e6.
timer - [on|off] Get/set timestamping state.
trace - [on [path prefix]|off] Get/set tracing and save files in 'path/prefix' (quote if 'path' uses spaces).
voltage - Report the internal 2s voltage average in mv.
The output energy file format is:
1 UInt8 - Trace version
1 Float32LE - Sample rate, in Hz
N Float32LE - N energy samples, in Joules
The timestamp format is a list of JSON array of floating point times in seconds.
The file device.cpp is effectively a line-by-line translation of the Matt Liberty's Joulescope Python-driver for Windows, the same is true of the raw_processor.cpp file. The joulescope.cpp file slims down the functionality provided by the Python driver. The main.cpp file controls the CLI and issues commands to the driver object, and writes to the output files.
The CLI downsamples based on samplerate, which is a command that can get or set the final rate in Hertz. The current_lsb is used as a falling-edge counter, which generates an m-lap-us-\d+ message on each falling edge. Be sure to ground gpi0 when developing to avoid spurious messages.
The sampling code spins in its own thread until the command line parser thread recieves an exit or stop-trace command.
As the sampling thread spins, it calls into a RawBuffer for initial 2Msmp/s data storage. The data callback hands the RawBuffer a number of packets. These packets' indices are checked, and any missing packed IDs are replaced with bad packets of 126 bad samples (to maintain the correct # of samples over time). Later on these become NaN values during raw processing.
When the sampling thread calls the RawBuffer's process callback function, the RawBuffer sends the samples to the FileWriter by way of the RawProcessor. The callback for the RawProcesser calls into the FileWriter. The FileWriter then downsamples the calibrated I/V values by accumulating (and listens for an IN0 timestamp), and then stores the accumulated energy sample in a ring buffer. As each ring buffer fills, it is writen asynchronously (overlapped) with Windows WriteFile. Another thread waits for completion of these overlapped writes and then advances the tail pointer of the ring buffer.
This complex process is needed due to some slower media or heavily IT-managed systems, which can severaly slow down synchronous file I/O and cause loss of samples.
Any time a packet index is missing, a dropped samples value is updated in the RawBuffer which is reported at the end.
All joulescope-win32 code is released under the permissive Apache 2.0 license. See the License File for details.
The original Python code is Copyright (c) by JetPerch and released under Apache 2.0. Many thanks for Matt Liberty's feedback on the port and general camraderie!
The Json library used is Copyright (c) 2007-2010 by Baptiste Lepilleur and The JsonCpp Authors