Unterschiede

Hier werden die Unterschiede zwischen zwei Versionen angezeigt.

Link zu dieser Vergleichsansicht

lehre:ws18:fsm_18ws:group_e:2019-02-04_more_testing [04.02.2019 14:54] (aktuell)
Thomas Oswald Erstellt mit dem Formular lehre:ws18:fsm_18ws:group_e
Zeile 1: Zeile 1:
 +---- datatemplateentry blog ----
 +template ​           : ..:​tpl:​display_blog
 +title               : More Testing
 +description ​        : We are testing our calibration processes and overall measurement with different approaches.
 +date_date ​          : 2019-02-04
 +bygroup ​            : E
 +tags_               : blog,
 +---- 
  
 +# Measurement
 +
 +Added a new calibration method. The step size starts at 1 and is increased when the number of repeated calibrations per step size doesn'​t match the events received. Fifty repetitions per step size are found to be a reasonable size. The calibrated step size is afterwards used for the measurement.
 +
 +After each measuring process the stepper moves back up saving us time for the next measurement.
 +
 +A second calibration (case 6) was added to narrow the time, when the click event happens, down.
 +Moves down in single steps till button is pressed. Logs time when stepper is done moving down one click and the event time. Substracting these results from one another gives back the latency.
 +
 +To compare the two methods moving the stepper at a stroke and lowering it in single steps we implemented case 7.
 +In a first round the stepper moves down in single steps till it gets contact. Then he saves the pressure to a log file and moves the same distance back up.
 +The second loop instructs the stepper to move down in a stroke till measures pressure. The value is also saved in a different log file. After the measurement the stepper also moves back up the same distance.
 +Both modes are repeated a hundred times to compare if there is any difference in pressure.
 +
 +Adjusting the method for moving the stepper allows us to eliminate generated latency up to 4ms.
 +The waiting time added in this method doesn'​t influence the overall latency any more due to adding a return statement. The "​moveStepper"​ method now informs when it's ready.
 +
 +
 +
 +# Logging
 +
 +Now the moment after lowering and before elevating the probe tip is logged as "​middle time". Drawing the pressure curve doesn'​t show explicit points where the click should happen. Pressure curves are therefore for now not significant enough to determine the moment of clicking.
 +
 +Logfiles come now with headers containing important information about device type, button, iterations and stepsize. Every measuring process now creates his own folder holding pressure and event data. Also each single measurement is saved in a separate file.
 +
 +Log instructions are now outside the measurement loops to reduce latency.
 +
 +Log files of different cases are now saved to different directories giving us a better overview.
 +
 +
 +# Data Analyzation
 +
 +We created a jupyter notebook to readout the generated log files. The data is mapped on a pressure curve containing click and event time. We use pandas, seaborn and scipy to visualize and test our data.
 +
 +
 +# Further strategy
 +
 +In a first step we will test the different stepper moves (one flow or each step on its own) and its effect on the measurement.