FMS New Run Display HTML

This note describes a new Run Display 

To see an example of my new display, look at

When this page loads, it loads about 10,000 small plots, (15 Mbytes).

If you have a slow link, there may be some delays and it may take a while before the page operates smoothly.

For a slower connection, you may want to open this in a separate tab and read on for now. When you get around to looking at the example, it will likely be fully loaded.

I use Google chrome but FireFox is fine. Other browsers may have some problems.

Also, the browser display should be as wide as possible, >1000 pixels.



To assist with the trigger remapping, I wrote a program that makes a HTML based display of the FMS for the purpose of graphically showing the mapping between (detector - PatchPannel - Qt crates). This is posted here.

The essential feature is a graphical map that shows the connections for a particular cell when the mouse is moved over the location of that cell. On this map, each QT crate is indicated by a color and if one clicks a cell for one QT crate, then a table for that crate is filled in (at the top of the page). After a QT crate is chosen, clicks on other elements of the crate are added to the table. (Clicks on other QT regions are disabled until you click on the top line of the page to clear the QT crate).

The program that makes this html file has access to all the information that is specified in the root12fms/SetFMSEnv file. Making and pointing to a new "qtmap2pp.txt" file provides the remapping and leads to a corrected html file

Playing with this, it became clear that this method of showing detector data could be very powerful. Below, I descripe work I have done over the last 2 weeks to create a more general tool for looking at the large amount of data that is required to study the performance of the FMS detector.

 Need for Real Time Data Display for new Runs.

There are more than 1200 FMS cells and when we want to evaluate the performance of the FMS in the early stages of a run, it is important to look at several displays for each of these 1200 cells. That means an accurate snapshot of the FMS performance involves ~10,000 plots. Sometimes we want to look at several plots associated with a particular cell together. Sometimes we want to look at a plot in the context of what plots of neighbor cells are doing. Sometimes when we look at a plot, we would like to check out what the high voltage setting was or how the gain shifts are set or what the current gain settings are. I find that this involves lots of jumping from one stack of information to another. Even a spreadsheet can be time consuming in going between different types of information.

Making some of these plots require a little analysis and making others involve a lot of analysis. It should be a goal to have a display online for many eyes to look at as runs occur but organizing these plots for online presentation in a reasonable way is not so easy.

I have put together a package that creates plots based on "fast" analysis of date (time<1 hour). In addition, we would like to see plots based on "slower full" analysis (time~0.5 day).

The fast analysis takes advantage of a fast and simple clustering algorithm that I wrote last year and is in the current root12fms analysis package.

HTML Data Display Features

The new display is an extension of the image map HTML example described above.

New information, beyond just mapping,  is added to the text display. Now we include the current High voltage settings, the current QT shift settings, the current gain and gain correction information.

When we move the mouse the across the FMS image, these fields are updated to correspond to the cell that we point to. Also updated, to the right of the FMS image are several plots. The plots shown are 1), 3) and 4) below. 

If you click on a cell, two things happen. 

  • The plots 1) and 2) are shown to the right and stay in place until another click.
  • A 5x5 array of cells within a single NSTB detector is selected, and a rectangle is shown. On the image map, the 25 selected cells are covered by the rectangle until another click in that region occurs. Then that again uncovers the 25 cells and reduces the rectangle to a small point on the originally clicked cell.  

    Now scroll down on the web page. At the bottom of the web page, 25 ADC plots "plots 2)" that correspond the the selected 5x5 array are shown. There are seven choices as to what these 25 plots can actually be. The list of plot types is shown and if you click a type from this list, the 5x5 array of cells will update to the selected plot type.

Simple Clusters

The first 5 plots shown below require running programs that do simple clustering. 

This fast clustering analysis involves a clustering algorithm that defines a cluster as a contiguous (manhatten style) set of towers. The high tower (towers) is identified and each tower is included in an ordered list. Cluster energies are calculated as the sum or tower energies. The energies are based on a "current" set of gain files.

On a single PSU computer, with a 12 thread processor, the analysis of a 1.5 million event Run 11 run (example Run 120090042) takes ~ 1/2 hour.

The result of analysis is to create several plots per event. The plots are.

plots 1) A plot involving the energy distribution of 

  • (BLACK) Cell energy distribution for cells that are the high tower cell in a cluster, where the cluster is not the lead cluster of the event. (Thus probabily not related directly to the trigger).
  • (RED) Cluster energy distribution for the clusters that are ont the lead cluster in the event.
  • (MAGENTA) Cluster energy distribution for the cluster that is the lead cluster of the event. This is probabily the trigger cluster.
  • Text showing the total number of events in the MAGENTA category. This may be the number of triggers centered on this cell.
  • Text showing the number of events per GeV @3 GeV. In a follow up blog, I will discuss why 3 GeV is chosen (it has to do with triggering and with minimum ionizing particles in the 0 to 1 GeV region.)

  These plots have binning that reflects several things. First the binning is based on the calibration (Energy /bin) and on the state of the shift registers on the QT setup. This point here is that the energy binning reflects the actual step separation between measured ADC levels. Every plot has a different binning but the quantities plotted are rescaled to reflect counts per unit energy. (The shift level is detected in the ADC data for the cell in question).

plots 2) This plot is the ADC distribution for the selected cell.

plots 3) For this plot, clusters are selected that are:

  • High tower at least 15 cm from another cluster,
  • Fewer that 20 towers,

 This histogram provides the average ratio of high tower energy to total cluster energy.

plots 4) For the same group of clusters as shown in 3), this histogram provides the average ratio of energy in cells that are next to high towers to the total cluster energy. The average ratio in the 3-6 GeV region is printed on the plot.

plots 5) This plot shown the ADC count for LED events as a function of time into the run. The average ration in the 3-6 GeV region is printed on the plot.


Full analysis

Additional plots are added with further analysis.

plots 6) This is the distribution of 2 photon masses in the low energy (40 +- 20 GeV for small cells ) that results from full analysis of data using current calibration files.

plots 7) This is the distribution of 2 photon masses in narrower energy region (40 +- 10GeV for small cells) that results from additional iteration. Fits are shown and mass and width are printed on plots.