LSi Correlator

Wiki > The Backend System > Specific Device IOC > Other > LSi Correlator

The LSi correlator is a device used as part of the DLS system on LOQ. It can be controlled either with the vendor-supplied software, or using the Python, Labview or Matlab APIs.

The vendor software

Connecting the software to the correlator

The software struggles to find the correlator's IP address on our network. Instead, the IP address can be entered manually in the settings file in

<LSI Software root>\LSI Correlator Acquisition Software\LSI Correlator Acquisition Software Settings.ini

In this file, change the default setting of the correlator's IP address to the actual device IP then reboot the software.

The Python API

The python 'API' is little more than a reference implementation of a websocket client. It is installed with the vendor software, and is found in <LSI Software root>\LSI Python API.

The LSI python API has three dependencies:

1. ipaddress
   • This provides a standardised IP address object back-ported from python 3
2. autobahn
   • This is their module used to create websockets
3. twisted
   • This is a common python library used by autobahn to control low-level TCP connections

There is also a vendor-supplied example script to use with the API, ExampleMainCorrelator.py. This can be used, although it is recommended to remove the plotting from this script as it seems to be quite buggy. The ip address in the script will have to be changed to match that of the device

Note: The second argument of the LSICorrelator object is the firmware value of the device. This is shown in the LSi correlator vendor software.

LSI correlator communication investigation

Ticket 4021 was an investigation into how the LSi correlator can be integrated into Ibex. The points are addressed here.

1. I can open a connection to the LSi Correlator.

   • Using the steps above, both the vendor software and the Python API can be used to connect to the device.

2. I can configure the LSi Correlator (to the extent permitted by the API).

   • The vendor-provided example script runs without errors, which includes configuring various options. The measurementDuration and TransferRate were changed and affected the output of the device.

3. I can close a connection to the LSi Correlator.

   • There is a bug in the python API which causes a script to hang on terminating the connection with the device. This will need to be fixed before it can be used.

4. I can run a Python script demonstrating the collection & plotting (e.g. with matplotlib) of data from the LSi Correlator. Also demonstrate how to start a (neutron) run and how to generate a suitable run title.

   • The LSi-provided script can run and return some data (all zeroes as the correlator was not connected to a laser box).
   • This data can be plotted with matplotlib
   • We should clarify that the correlated data the python API gives us is actually what the instrument scientists want.
   • The script was run within genie python, so we should be able to call the API from a normal instrument script (after the dependencies have been added to genie python).
   • The run title in the vendor software cannot be changed from within the python API. The API and the vendor software are largely independent.
   • The python API can not be used to start a run which was set up in the vendor software

5. I have created appropriate documentation explaining how to use the LSi Correlator API.

   • See above for description on how to run the vendor-supplied API

6. I have produced written recommendations for how we should support this device in IBEX.

   • See conclusions

Conclusions

1. It is possible to use the Python API to get both unprocessed and 'correlated' data off the device.
   • We can add this to Ibex this by either calling the API from instrument scripts, or by using pydevsup or PCASpy to make an IOC-like device.
2. It is not possible to use python to start a run which has been set up in the vendor software (LSi Correlator Acquisition Software)
   • This is an issue, as the vendor software performs analysis on the raw data
3. In order to use the vendor's analysis tools, we would have to reverse engineer the XML save format used
4. We need to clarify that the correlated data from the python API is the 'correct' data that instrument scientists want

Conclusions of October experiments

1. Data taken from the python API are equivalent to the vendor's software
2. The python script was set up using the parameters given on the wiki
3. obj.Correlation and obj.Lags are the values of interest to the scientists
4. Python API still needs modifying for python 3 and to make sure the connections are terminated correctly after each set of repetitions
5. The scientists also want to be able to save metadata along with the data from the DLS

Next steps are written in ticket 4918

Python API vs vendor software

To test that the python API functions in the same way as the vendor software, a DLS experiment (a run with the vendor software) was repeated several times using both the vendor software and the python API. Ideally, the correlation functions produced using both these methods will be indistinguishable within an experimental noise.

The test procedure is:

1. In the vendor software, perform a run with 5 repeats of 20 seconds each.
2. After the runs have completed, export each of the data sets to CSV in the data analysis tab.
3. To perform the same experiment from the python API, run this script (replacing for appropriate file paths, IPs...):

import sys
sys.path.append( path_to_LSI Python API)

import LSI.LSI_Param as pr
from  LSICorrelator import LSICorrelator
import time
import matplotlib.pyplot as plt
import numpy as np

# Replace for IP of correlator. Can be seen in the vendor software
obj=LSICorrelator("X.X.X.X", "4.0.0.3")

obj.setCorrelationType(pr.CorrelationType.AUTO)
obj.setNormalization(pr.Normalization.COMPENSATED)
obj.setMeasurementDuration(20)
obj.setSwapChannels(pr.SwapChannels.ChA_ChB)
obj.setSamplingTimeMultiT(pr.SamplingTimeMultiT.ns200)
obj.setTransferRate(pr.TransferRate.ms100)
obj.setOverloadLimit(20)
obj.setOverloadTimeInterval(400)

deltat=0.0524288

obj.configure()

for i in range(5):
    print "Measurement "+str(i+1)
    obj.start()

    while obj.MeasurementOn():
    
       time.sleep(0.5)
       obj.update()
       timeTr=np.arange(len(obj.TraceChA))*deltat
       TrA= np.asarray(obj.TraceChA)/(1000*deltat)
       TrB= np.asarray(obj.TraceChB)/(1000*deltat)
       
    # Save the data acquired with the python API
    np.savez(path_to_data+"pydata{}.npz".format(i), TrA=TrA, TrB=TrB, timeTr=timeTr, Lags=obj.Lags, Corr=obj.Correlation)
    
# This will not properly close the connection
obj.close()

This script is a modified version of ExampleMainCorrelator.py provided with the vendor software