This note summarizes the work I have done in the last few weeks
hunting down and categorizing WFD problems caused by the STOP-masters.
In summary, I will describe three problems, one of which is understood.
All of these problems happen rarely(<.5%) and they are recognizable so
that they can be filtered out of the data stream. The result of these
errors are that the WFDs are stopped at the wrong time and the WFDs do
not contain data where you expect.
It should be noted that this note does not describe errors associated
with the WFDs themselves. These errors are only ones resulting from
the STOP masters or the interaction of the STOP-masters with the
acquisition system.
It is important to filter out these events since they result in random
WFD data in the time windows you expect your signal to be in. This
can result in fake signals for LIPs, monopoles, etc.
Problem #1: Input missing resulting in no WFDs
------------------------------------------------------------------------
In some periods of the data it may be the case that one or more inputs
to a STOP-master have either become unplugged or broken. This can
have two possible effects:
1) If there are other triggers in the event which are plugged into
the STOP-master the STOP will still be issued. Because of this all of
the WFD channels will be present. The data may however be in a
different place then where you expect.
2) If the only trigger in the event is the trigger which has the
broken STOP-master input, the STOP-master will not be stopped, and
when the acquisition tries to read out the STOP-master there will be
several CAMAC errors generated. Then, when the acquisition tries to
read out the WFDs there will be VME errors generated since the WFDs
were never stopped. The effect of this is that there will be no WFDs
in the data of the event.
In the case of 2) you need to filter your data looking for the CAMAC
errors from the stop master. One thing I learned is to make sure you
are looking in the data for your errors. When you are debugging
problems do *not* rely on the LGB program to look for these errors.
It does not reliably record errors.
Here is a request: I have code that decodes the error equipment and
gets out the CAMAC errors(God apparently told Bob how to do this.) I
don't know how to get the VME errors however. If you understand the
format of equipment 999 please post a description to the list.
In addition, you should always check to make sure that the STOP master
input for the trigger you are interested is working in the data you
are looking at. If the trigger you are interested in has a high
overlap with other triggers the error condition may only happen
rarely.
Problem #2: STOP master stopped by CIT but there is no CIT trigger.
------------------------------------------------------------------------
This problem is not understood. There are some events where although
the CIT trigger does not exist in the event, the STOP-master has been
stopped by the CIT monopole trigger. It looks as though somehow the
CIT trigger has caused the stop but is ignored by the acquisition.
Because of this, the WFDs are stopped too early, and when the next
event comes and attempts to read out the data it sees data which does
not at all correspond to what you expect.
This problem manifests itself in two ways:
1) If you look at the stop master data you will see that the CIT
monopole input has a time of zero(the system is common start) and all
other inputs have the overflow time(1ms or 10000 counts.) Note that
since the two stop masters in each micro-Vax are connected to
each-other this stop can happen in either of the STOP-masters in the
pair. In addition the acquisition trigger pattern will not include the
CIT pattern.
2) The WFD data in the event will not have what you expect in it.
Because the WFDs were stopped some unknown time before the event the
time period where you expect your signal may contain radioactivity or
nothing at all.
In order to filter these events you need to look for the configuration
I described in the STOP-master.
Problem #3: The STOP-master counts for more then 1 ms.
------------------------------------------------------------------------
This problem is also not understood. Sometimes the STOP-master counts
to an overflow value of more then the expected 10000 counts(1 ms). The
overflow ranges from 1 usec to 1 ms. The effect of this is that the
data is not where you expect and there may be radioactivity or nothing
in the expected region.
I guess that this is a problem that can sometimes happen in the
synchronous counters which Rongzhi designed in the ACTEL chips. In
order to filter out these events, look for the value of the overflow to
be greater then 10000. I find the easiest thing to do is to look at
the unused streamer-tube monopole input. This is effectively a
hardware OR of all the possible inputs. Of course if we start using
this input this won't be possible anymore.