Single tube counters and monitors are handled through the EL737 counter box. This seems simple but the EL737 counter box has some quircks.
The EL737 counter box has in interesting output: this is the TTL trigger signal if data acquisition is active or not.
In order to understand neutron DAQ with area detectors let us follow the path of a detected neutron through the system.
Even with an area detector an EL737 single counter box is necessary. This box handles the monitors and through its TTL output triggers neutron data acquisition.
Just another time, to make the triggering process crystal clear. When data acquisition becomes active, the EL737 counter box sets its TTL output. This is fed into the MDI and is represented is one of the sync bits of the neutron event message. The histogramming task tests both the event type and the presence of the sync bit against a mask. Only matching neutron events with appropriate sync bits are histogrammed. The question may arise why this is so important. Now, SINQ is an unreliable source compared to a reactor. The beam may go down for milliseconds to days. But when the beam is down, the detector and electronics continue to produce noise. You do not want this noise on your data. Thus DAQ needs to be interrupted by way of the threshold in the counter box and the sync bit.
In time-of-flight (TOF) neutron data acquisition not only the positional information of a neutron event is measured but also the time it needed to cover a certain distance before it arrived at the detector. This is a measure for the energy (speed) of the neutron. In order to do this, a starting point in time is needed when all neutrons started to travel towards the detector. At pulsed neutron sources this starting point in time is provided by the source. At SINQ it is provided by the chopper. This is a fast rotating disk which cuts the neutron beam into pulses by way of slits in the disk.
TOF neutron DAQ is very similar to area detector data acquisition:
Besides the normal sync inputs there are additional ones. Most notably an input from the chopper which indicates if the chopper is running at the desired speed and phase. If any of this is not the case, then the neutron event is useless.
Another complication comes from the fact that most choppers have two slits (for symmetry) but deliver only one signal per chopper revolution. The signal needs to be duplicated. This is the purpose of the Emmenegger electronics. This is a rather obscure bit of hardware. Mr Emmenegger has left PSI a long time ago and no one knows any details about it anymore.
Yet another complication comes from the fact that the data width for the time stamp is only 20 bits. In order to increase the available time resolution a delay time is subtracted in the MDI. This delay time is the time which the fastest interesting neutrons need to travel from the chopper to the detector. The delay time is set in the MDI by way of a RS232 connection from SICS. Unfortunately the implementation of the RS232 interface is dubious. RS232 is connected via a fibre optic link. Thus there is a fibre optic to RS232 converter. Some of those converters work, some not. One has to try out several ones when this link needs to be replaced until a converter is found which works.
Summing it up, the MDI in the TOF case needs the following inputs:
CCD cameras are like digital cameras in that incoming photons increment a charge in a detector pixel. After exposure, the charge map of the detector is read out. CCD detect photons: neutrons are detected by looking at scintillator screens which essentially converts neutrons into light. This explanation serves to give the background for the fact that the electronic noise suppression which works for normal detectors at the neutron event level does not work for CCDs. Thus one needs to rerun the exposure if SINQ goes down to long during an initial exposure. SICS checks for this and does the reexposure when necesary.
Now, commercial CCDs come from companies living in the dark ages IT wise. Which means that they usually get deliverd with some windows system to run them. The better ones come with a Linux API in a library. This is what we use. On top of the library a litte WWW-server, ccdwww, is run which allows SICS to expose images and download data, configure the camera and such. This WWW-server is run at a separate computer. For a good reason, these CCD cameras require a special hardware connection. Which often fails. With a different computer the CCD camera computer can be restarted without having to take down the rest of the data acquisition system.
At AMOR, the Emmenegger electronics died. Fortunately, the data for a full revolution of the chopper could be fitted into the 20 bit time stamps. Thus there are actually two pulses in the neutron event data. This is corrected for by a special histogramming process on the HM computer for AMOR which merges the two pulses. This makes the HM configuration for AMOR a little complicated.
POLDI works with frame overlap. This means that the data from different neutron pulses overlap in the data. POLDI has a very complex chopper: the disk is split into four quadrants with each 8 slits. And there is only one reset signal from the chopper per revolution. Fortunately all data fits into the 20 bit time stamp. The raw data from the POLDI HM contains data from all four quadrants after each other in time. SICS sums the four quadrants together again.
HRPT is the first instrument having the second generation data acquisition electronics. Namely the prototype. It replaces the MDI and front level electronics. This electronics has an own TCP/IP address and its own console. No documentation available, naturally.