Index of /people/artem/projects/tools/rt_control

 Name                    Last modified      Size  Description
 Parent Directory                             -   
 communicate.c           27-Jan-2006 12:15   12K  
 control.c               27-Jan-2006 12:15   11K  
 control.h               27-Jan-2006 12:15  1.2K  
 rt_control.c            27-Jan-2006 12:15  7.4K  
 rt_control_report.m     27-Jan-2006 12:53  472   
 rt_control_saved.m      27-Jan-2006 12:53  801   
 rt_info.h               27-Jan-2006 12:15  1.5K  

rt_control.ko

To be used with sub-kernel real-time linux (RTAI). Uses Comedi library to listen on input channels and provide real-time feedback on output channels of National Instruments DAQ board. The core module (rt_control.c, rt_info.h) originally written by Alan Chen. In the present version the control algorithm is defined in a separate file (control.c, control.h) to simplify user implementation. Other modifications include multichannel functionality and extra fifos for interprocess communication (comminucate.c).

To start the sub-kernel module:

root@RTAI_host:/root/rt/bin# /sbin/insmod rt_control.ko

To start the user communication process:

root@RTAI_host:/root/rt/bin# ./communicate outfilename

Output:

The process displays event statistics every 30 seconds in the following format:
timestamp    trig_events    thresh_events    reported_events    baseline_voltage_mV
Same information is written to a outfilename.report, except that the timestamp is saved in seconds without hh:mm:ss formatting

User interaction:

'Ctrl-C' to bring up the prompt.
'?' to list available commands.

Present control algorithm:

A longer 'baseline' window is followed by a shorter 'probe' window. When the difference in baseline and probe average crosses the specified threshold the 'ON' state is noted. If this 'ON state' is sustained for a specified duration, an 'event' is 'detected' and a command output is issued.

Hardware connections assumed by control algorithm:

connect the signal of interest to AI0 (analogue input #0)
optional: connect external trigger to AI1 (see 'g' below)
optional: connect external threshold level control to AI2 (see 'e' below)
read feedback command voltage pulses from AO1 (see 'a' and 'w' below)
read timeout indicator from AO0 (see 'b' below)
optional: add probe voltage average to AO1 (set 'm=2' or 'm=3', see below)
optional: read averaged baseline voltage from AO0 (set see 'm=1' or 'm=3', see below)

Interaction with rt_control module (commands supported by communication interface)

t - threshold voltage, millivolts, negative value for downward polarity
o - detection timeOut, milliseconds, if we do not want events too close in time
p - probe time window, milliseconds, use this to average sampling noise
b - baseline time window, milliseconds, use this to average baseline level
d - minimum duration requirement for reportable events, milliseconds
g - attention span after external triGGer, milliseconds, '0' to turn off
s - save traces of detected events, limited to one event per report period
e - turn on/off threshold level readout from an external analogue input
a - command amplitude, mV
w - command width, us
m - extra output monitors: 0='none', 1='baseline', 2='probe', 3='both'
l - write one-line note to a log file, nothing is sent to rt_control process
q - save statistics, close connections to rt_control, quit

Capturing traces of events (see b,o and s commands of communication interface)

We continuously keep record of all the samples in 'baseline' and 'timeout' time intervals. After an event is reported, 'baseline' samples are frozen and 'timeout' samples are collected to allow the 'capture' of input signal trace for the 'event'. Saves in microvolts as 32-bit integers.

Loading recorded data into Matlab:

rt_control_report.m loads reported statistics:
rt_report = rt_control_report('20060118_rt/20060118_test1.report')

rt_control_saved.m loads captured traces of detected events:
rt_traces = rt_control_saved('20060118_rt', '20060118_test1.00*', 10000)
     where three parameters are: folder, filename pattern for multiple traces, sampling frequency (Hz).

Plot like this:
plot(rt_traces.time_ms, rt_traces.voltage_uV/1e6)