The trigger statistics display is a good tool for uncovering certain varieties of operator error, detector or acquisition misbehavior, and bad beam conditions.
The panel may be reached from the Index panel, the System Status panel or the Just for SLC panel. The Just for SLC panel itself has some of the 120 hertz buttons on it, including most of the ones discussed below.
In order to get a display from the panel, you must select
These may be selected in either order. The display will actually be generated when you push Make display. (This is not quite true with Ungated -- see below.)
The accumulation type selected determines exactly which pulses will be in the data sample you see. Running includes pulses which occurred while we were (no kidding) running. It is cleared at the beginning of a new run. It does not include pulses while we were paused in the middle of the run nor pulses immediately after a FASTBUS reload (though it should not take long after a reload before pulses again accumulate into Running).
For diagnosing problems Small Sample and Ungated are usually more useful than Running. Small Sample contains statistics on the last 10 or 20 seconds only. For a larger sample use Ungated. In this case you have control when the statistics are cleared by pushing the Clear Ungated button. This button must be pushed at least once in the life of the SCP before you can see a display. Ungated statistics are accumulated independent of run state.
Detailed descriptions of each accumulation type and data type exist online. First push the help button. This puts you into help mode. Now pushing any other button causes help text on that button to be written to your current display. For uninteresting technical reasons help on all accumulation buttons will be displayed no matter which one you push; similarly help for all data types will be displayed no matter which data type button is pushed. You can page through the help displays the same way you would with any other display (e.g., use icons in the display window). To get out of help mode, push the Help button again.
Before proceeding any further be sure the 120 hertz statistics are updating properly. (But don't get the idea this is a common problem; it isn't.)
If the Last update occurred at date isn't current (to within a minute for Small Sample or Ungated) perhaps the S120 process isn't running. Go to the Status Panel and bring up the Check Processes display.
If the Last pulse occurred at date isn't current, check the Monitor Etc window on the BEL workstation to see if monitor records are flowing to the VAX. If not, and there is no reason not to start them (most common reason to leave the monitor stream off would be reloading slaves or aebs), go to the BTC window (also on the BEL workstation) and type
BTC> open monitorMonitor records should always be flowing to the VAX during logging.
Once you know you are accumulating data properly, look at the OR'd Statistics to get an idea of overall rate and investigate anything that shows up red on the display. For a complete run-down on all reasons for red, see the online help; some of them are discussed below. You will find some pointers on how to examine rates at the end of this document in the section Understanding Rates.
If the standard triggers have not been armed, the Disarmed row of the display will have some red numbers in it. Under normal circumstances none of these numbers should be red. To see if triggers are currently disarmed, use the Small Sample accumulation.
When conditions are extremely noisy it may be appropriate to disarm triggers; in this case you should also pause the run. There are some other circumstances in which disarming the triggers is reasonable; however if triggers are disarmed people on shift should know why.
For an independent check on which triggers are armed, go to the BTC window on the BEL workstation and type
BTC> show slots allThe armed triggers should include at a minimum all the ones whose names appear on the 120 hertz display and the one called CST, a non-physics trigger which won't be discussed further here.
Statistics kept by time slot (e.g., True TS1 and True TS4) will turn red if the differences between the time slots for that trigger are large. Usually this is an indication that the beam is not well-tuned on at least one time slot. If there is such a difference, be sure that MCC operators are aware of it (e.g., ask the Liaison to pass it along).
When some resource needed by a trigger slot (e.g., a subsystem read out for that slot) is unavailable, the slot is declared busy. This is independent of whether the slot "wants" to fire or not. If the percentage of pulses for which a given slot is busy is large the number will be red on the display. In this case first determine if the number is simply large or is 100% (compare to the number in the # enabled row). If the two numbers are equal, there may be a problem with Below-line acquisition and an appropriate expert should be called. Otherwise, go to the section Understanding Rates below.
This is not straightforward. Occasionally rates can be too low (i.e., zero for some or all trigger slots); this is usually some form of operator error or "stuck" acquisition. If the solution isn't something obvious, like arming triggers, call an acquisition expert.
High rates are more common. First look at OR'd Statistics towards the bottom of the display. During good running all the rates will be similar and all will be under .5 hertz. If any of these rates are above 1 or 1-1/2 hertz for extended periods, especially if most of them are triggers that read out the full detector (currently anything other than Bhabha) that's high and you should investigate further.
The usual reasons for high rates are problems with the detector or noisy beams (or both). Look at subsystem 1-event displays and histograms if you suspect the detector. Possible remedies are calibration (if the offending subsystem has not been calibrated recently or you have reason to distrust the last calibration) and/or burying noisy channels.
If the detetector and data acquisition are operating correctly, high or unbalanced rates are a sign of noisy beams. Even if the overall trigger rate (i.e., Triggers in the OR'd Statistics part of the display) is acceptable it may be that this rate is being kept down by vetoes.
To study beam conditions look at the top part of the display where statistics on individual trigger slots are kept separately. The rows
True TS1 (true, time slot 1) True TS4 (true, time slot 4) Vetoed TS1 (vetoed, time slot 1) Vetoed TS4 (vetoed, time slot 4) True, ~veto (true and not vetoed)are more useful than the rate of triggers (top two lines). In quiet conditions for each trigger slot ((True TS1 + True TS4) approx. = (True,~ veto)). If these numbers differ significantly (not sure how to quantify this; maybe #true > 3 * (true,~veto) ?) for ENERGY, HADRON, or TRKCDC, MCC should be informed. Let them know which subsytem is getting swamped. A veto of the ENERGY trigger occurs when too many LAC towers are over threshold. A veto of TRKCDC or HADRON occurs when too many DC cells are hit.
Most of the additional information on triggers available through SCP displays such as the Run Summary Display comes from event data. There are some advantages to this approach (e.g., ability to correlate trigger slot with event size) but, except in very quiet conditions, such statistics are likely to be biassed.
Of the other data types available through the 120 hertz panel, the most generally useful are the energy spectrometer histograms and the luminosity statistics. Be aware that at times the overall scale of the luminosity statistics is suspect, but the percentage of luminosity in each category is reliable. (Currently - November, 1994 - the scale is believed to be correct.) For a full description of these displays see the Online Help.
The 120 hertz displays, especially the trigger slot statistics display, are a sensitive measure of performance, but some care must be taken in determining the source of an anomaly. There are many such sources, and symptoms may be subtle. After you've verified that the 120 hertz information itself is updating properly, look for problems in three major areas: