Trace Tools

A python library for parsing and working with trace files from Siemens Sinumerik CNC system.

You can parse:

• Autotuning results - After finishing an autotuning, you can save the "auto-tuning result", this file contains the sugested control parameters and the frequency responses.
• Frequency measurements and step responses - From the "Setup/Optimize" meny
• Trace files - That you configure under the "Diagnostic" meny

Note that this library only works with the newer version of Sinumerik operate, not sure which exact version, but if it produces xml-files when saving, then it should work.

Also this is not affiliated with Siemens, I leave no quarantees for the accuarcy of the parsed data, use it at your own risk!

Repo structure

• TraceTools.py Main file, import this and use the functions to parse different type of files.
• tracetypes.py class definitions of what the measurment files are parsed to.
• b85.py used for decoding their non standard implementation of ASCII 85.

Usage

Download the repo and place the tracetools folder in your working directory, then use the library as in the examples further down this page.

Examples

This section will present some simple usecases of parsing each of the file-types supported.

Automatic Servo Tuning files (AST-Files)

AST Files can be saved at the end of performing an auto-tuning of an axis. These contains the frequency response of the mechanical system and current control. It also contains all the control parameters for the servo-drive.

Example code for plotting the mechanical frequency response.

import tracetools as tt 
import matplotlib.pyplot as plt

ast = tt.parse_autotune_file(r'traces\ast_file.xml')

tt.plot_freq(ast.plant_freq_response)
plt.show()

You can also read out the control parameters computed by the AST.

import tracetools as tt 
import matplotlib.pyplot as plt

ast = tt.parse_autotune_file(r'traces\ast_file.xml')
print(ast.speed_ctrl_params)

SpeedCtrl(Kp=0.2160315604218958, Ti=0.01, Ts=0.00025, 
            ref_mdl_active=True, ref_mdl_freq=146.35482877622718, 
            ref_mdl_d=0.7071, ref_mdl_delay=0.0, 
            act_value_filters=[], 
            current_setp_filters=[PT2(freq=2000.0, damping=0.7071)])

To see what is parsed from the AST-file, check the tracetools/tracetypes.py file.

Frequency measurement files

Appart from auto-servo-tuning, one can also do stand-alone frequency measurements. These can also be parsed, but they do not contain the servo-drive parameters. However, they do contain the timeseries signals that are used to compute the frequency measurement.

import tracetools as tt 
import matplotlib.pyplot as plt

freq_meas = tt.parse_freq_meas_file(r'traces\freq_trace0.XML')
tt.plot_freq(freq_meas.plant_freq_response)
plt.show()

One can also read out the time data used for computing the frequency measurement.

import tracetools as tt 
import matplotlib.pyplot as plt

freq_meas = tt.parse_freq_meas_file(r'traces\freq_trace0.XML')

ts_in = freq_meas.input_time_series[0]
ts_out = freq_meas.output_time_series[0]

plt.plot(ts_in.t,ts_in.values,label='Input')
plt.plot(ts_out.t,ts_out.values,label='Output')
plt.xlabel('Time [s]')
plt.legend()

plt.show()

Servo trace files

Normal servo trace files can be parsed with the parse_trace_file function. Its then possible to plot the traces or do some signal processing of them.

import tracetools as tt 
import matplotlib.pyplot as plt

traces = tt.parse_trace_file(r'traces\trace4.xml')

torque = traces[0]

tt.plot_trace(torque)
plt.show()

Here I show how you can smooth the signal using a noncausal lowpass filter, i.e. running it forward and then backwards. Google filtfilt if you want to learn more.

import tracetools as tt 
import matplotlib.pyplot as plt
from scipy import signal

def filtfilt(x,alpha):
    # Lowpass smoothing
    # alpha \in (0,1)
    a = [1,-alpha]
    b = [0,(1-alpha)]
    return signal.filtfilt(b,a,x)

traces = tt.parse_trace_file(r'traces\trace4.xml')

torque = traces[0]
torque_f = filtfilt(torque.signal,0.95)

plt.plot(torque.time,torque.signal,label='Sampled signal')
plt.plot(torque.time,torque_f,label='Filtered')
plt.legend()

plt.show()