Page tree

Description: page reserved for preparation, task list, and records of the test beam in April 2019 at DESY T24/1.

0. Preparation



  • trigger and clock line 20Ohm damping resistors added: on both cassette boards
  • upstream cassette:
    • sensors are now glued onto the frame
    • the previous S2 is not S1 and the S2 is now S#55.


  • Azalea

  • DAQ: Data taking and Producer run from fhlrcazalea, but RunControl connecting to aida2020-kpix1
  • Threshold:  threshold4, see Daniel Pitzel's slides
  • TLU: Ethernet to HDMI converter and on AIDA-TLU HDMI port 4; run under EUDET mode – handshake busy mode.
  • Movement:
    • rotation around Y axis, i.e. using the black fine thread screw
      • one turn is 0.5mm
      • clockwise  
      • anti-clockwise


  • our own AIDA TLU, firmware never changed after received it from Bristol;
  • DAQ: with aida2020-kpix1
  • Scintillators: 
    • 1 * Vertical in front of Mimosa plane0
      • chan0 at Eudet-TLU, chan1 at AIDA-TLU
      • powered 850mV, threshold -100mV
    • 1 * Horizontal in front of Mimosa plane0 after the Vertical
      • chan 1 at Eudet-TLU
      • powered 800mV (default), threshold -40mV (default)
    • powered through fhlrcazalea using
      .EudetTluControl -p1 850
    • Threshold set in .CONF file at aida2020-kpix1


Power Supply


How-to run aida TLU + mimosa + KPiX


  • TLU Clock is always sent out via HDMI_3, where lycoris should be plugged in;
  • AZALEA is supposed to be on HDMI_4;
  • Shutter mode (kpix acq. start signal) delay w.r.t. E_min is set for: 4.4 GeV, kpix data taking BCC from 1000 to 4000;
    • START of TLU active delay w.r.t. Shutter is set to be 6 ms (so bunchclkdelay 3000 * acq.clksel 320ns + triginhibitdelay 1000 * bunchclk 8*320ns + extra delay to avoid time edge problem 2.5ms i.e 1000 * 2.5us)
    • Period of TLU active is set to be 4.5ms (active for 1800 bunchclkcount, i.e. trigger sent out from 2000 to 3800);

How-to control Wiener PS

power/power-cycle the DAQ board

cd workspace/Wiener-PS
./ [on/off]

power the cassette in T24

./ on
./ on

power the 2nd cassette for PCMAG

./ on
./ on

check current on terminal



  • This is only valid if you are using the LV at Channel 0, 1 and 304, 305;
  • HV at channel 100, 101, 102, and 103, 104, 107.
  • DAQ board at channel 4.

How-to run KPiX

On aida2020-kpix1 PC

cd workspace/kpix/software

# Do a normal RUN in GUI-less mode python scripts/ --config YOUR_YAML_CONFIG_FILE -o YOUR_OUTPUT_FILE_DIR -r NUMBER_OF_RUNCOUNTS_NEEDED

# Do a Calibration RUN in a GUI-less mode
python scripts/ --config YOUR_YAML_CONFIG_FILE -o YOUR_OUTPUT_FILE_DIR


  • If you wanted to run TLU CLK: ensure you have TLU INIT+CONF from EUDAQ2;

How-to Analyse data

On aida2020-kpix1 PC

There are three types of data and 4 types of analysis.gear/gear-mimosa-255-b.xml

  1. Self triggering data:
  2. External triggering data
  3. Calibration data

cd KPiX-Analysis










./bin/ymlCalibrationFitter  YOUR_DATA_FILE





While most analysis is generalized to provide debugging info. pedestal_tree and cluster_analysis are specialized and provide only the bare essentials to ensure the analysis is relatively quick.

1. Lists of Task & Shift

See all the data TODO from google spreadsheet:

1.1 The prioritized data to take

  • External data with 2 Cassettes and Mimosa close to the Strip sensor
    • T24 with S55 for week of  
    • T24 with S43 for week of  
  • Self trigger data which is missing for Uwe's thesis
  • To study the signal aptitude w.r.t. the trigger delay — take data with different extra delay of the TLU trigger.
  • To study the Z-axis determination of our tracking – take angle run i.e. both Mimosa and Lycoris rotate along Z axis
    • ± 1 degree → 5mm movement for each direction for Mimosa stage
  • Power cycle disabled run. DisPerPwrCycle = False

1.2 Shift plane for  

Overall: we take data in 2 shifts per day, morning and afternoon.



Changing the sensor of upstream cassette and redo calibration.

Uwe Kraemer

Took multiple external trigger runs to determine beam location (which has changed) and after determining the correct position taking a larger statistics run with both TimeSetX1 and TimeSetX3

Mengqing Wu

start self-trig threshold scan

2. Day Log

Mar 9, Monday

Nothing happened, waiting for cassette board to be finished; afternoon,

Mar 10, Tuesday

Move in and all installed, all tested – good. But PI stage can not be moved. so we took some calibration.

Mar 11, Wednesday

Got PI stage working but it can not hold our sensor in the beam.

X-axis can make Lycoris in the beam however Y-axis barely make top part of the sensor in the beam.

Mar 12, Thursday

  • Morning: Swap the X-axis and Y-axis of the PI stage: so our sensor can be fully in the beam;
  • Afternoon:
    • External trigger data taking – at telescope lower part
    • Found out the previous conclusion on S43 and S55 was not fully true – that S55 is better with less bad-slope channels
      • because S55 is more noisy than S43
      • So Uwe Kraemer went to FE clean room to glue the S43 to frame – went good

Mar 13, Friday

  • External trigger data taking
    • normal gain at lower part
    • move to the center - especially check data taking with SetTimeX3 (looked like it has lower noise in the center)

3. Setup Explanation

4. Photo gallery

5. Setup Geometry

5.1 Sensors to Layer Number


Layer 0 = S59
Layer 1  = S1st
Layer 2 = S55


Layer 3 = S48
Layer 4 = S47
Layer 5 = S46

5.2 T24 Geometry

6 Analysis results

  • See Uwe's slides about the noise situation of the sensor : uwe_2020_03_19.pdf
  • See preliminary GBL results from different runs:
    • based on run 897 with S43 & Timesettings X 3
    • based on run 869 with S55 &

  • No labels