The 1600 Local Stations [LS] of the south site are composed of :
- a 12 m3 tank filled of water in which charged particles give Cerenkov light,
- a 'liner', inside the tank, which reflect and diffuse this light,
- 3 Photo Multipliers [PM], which collect the Cerenkov photons and convert them in an electric signal which is amplified by an electronic located at the base of each of them,
- an electronics and its associated real time software (see below),
- two solar panels (60 W, 49 V) coupled by a regulator connected to two 12 V batteries,
- a GPS and its antenna to allow a precise particule datation ,
- a radio and its antenna to link the LS to a Central Station [CS]. The communication (Suscriber Unit or SU), is connected to the station electronics by a serial link (RS232 type) modified to include a wire for the 12 V alimentation, and others for the initialization signals and GPS synchronization.
The FrontEnd board receives the PMs signals, digitizes, selects and numerizes them. There are six analogic inputs (three anodes and three amplified dynodes). Data are bufferized in two FIFO memories (1024 octets by channel) in order to minimized the dead time.
Data selection is realized by several triggers (time and amplitude selection). When a set of data is selected, an interrupt signal is send by the FrontEnd to the Unified Board [UB] processor. The UB starts a DMA to transfer the data to dynamical memories located on this main board.
The station controller is the intelligence of the Unified Board. It controls all the Local Station. It is composed essentially of four main fonctions :
- the power supplies,
- the processeur, memories, ...
- events time tagging,
- slow control.
This LED flasher is used for tests, calibration and PMT linearity measurement.
Unified board descriptionThe electronic fundamental contraints are the following:
- functioning temperature from -20° to +70° Celcius,
- the experiment will last about 20 years,
- maximum total electrical consumption : 10 Watts,
- mean hygrometry : 30 to 80%; 70% of the site is floodable; saline environment.
- printed circuit : 240*340mm in epoxy glass; 4 layers ( 2 for the internal power supply distribution and 2 for the external signals);
- processor IBM Power PC 403 GCx, 80 MHz internal, 40 MHz on the bus;
- bus 24 bits address et 32 bits data;
- operating system : Microware OS9000;
- 32 Mo RAM EDO dynamic memory;
- 8 Mo Flash EPROM (write protectable) to store the operating system as well as all the acquisition, calibration, monitoring and test programs;
- 128 ko of PROM OPT memory (write protected) to allow the Flash Eprom reprogramming using an Ethernet link in case of corrupted program;
- a Programable Logic Device [PLD] for system configuration. It manages most of the controls signals of the UB and stores its identification number;
- DMA channels are used for the Front End fast data tranfer;
- 4 RS232 serial ports (in addition of the Power PC one) are used to exchange informations with the external elements (GPS, radio, terminal and test);
- a connector allows to plug an external Ethernet board (realized by our laboratory and used to test the UB and for program developments);
- the board power supply control : this function can automatically switch off the UB if the external power supply is too low (to protect the electronic componants and the batteries). This system sends an interrupt to the processor to inform it that the switch off is imminent; a message is then sent to the Central Station [CS]. If, after a minute, the voltage is still too low, the power supply is cut for about 3 hours and 30 minutes and then switch on automatically. Such a system must have a very low consumption and an excellent fiability.
The event time taggingWhen an event is detected (generation of a trigger), an interrupt signal (EventClock Fast or Slow) is sent to the time tagging function which detremine the precise start time of the event using a 100 MHz clock. The GPS module (Ground Positioning System), receives signals from the satellites and generates a PPS (Pulse Per Second) whose precision is, after corrections, better than 10 nano secondes. This allows to correct the 100 MHz clock drift.
The GPS is a commercial Motorola UT+ module linked to the Unidied Board by a connector Dubox 10 points. To determine the precise time, the time tagging counts the number of clock cycles (100 MHz) between two consecutive 1PPS and detremine the correction to apply. Besides, the GPS sends, on request, the correction to apply to the 1PPS (saw tooth) which may vary from -50 to +50 nano seconds. All these corrections, associated with signal processing technics, leads to a precision of 8 nano seconds.
The first implementation of this board was based on an ALTERA FPGA (10K50EQC208-1) and was successfully tested on the Engineering Array. An ASIC version was then designed and tested by our laboratory and used on the Unified board. The 2 versions have the same functionnality. The board records the time of arrival of the two triggers generated by the front-end board, with a resolution of 10 ns:
- EVTCLKF (Fast trigger, active low): the time is recorded at the high-to-low and the low-to-high transitions of EVTCLKF.
- EVTCLKS (Slow - or muon buffer- trigger, active low): the time is recorded only at the high-to-low transition of EVTCLKS.
The time consists in two parts (naming of counters and registers refers to the block diagram figure ):
- One coarse time, handled by the 1PPS (1Hz) clock from the GPS receiver and counter C3 (24 bits ). The 1PPS signal is the time reference for the whole experiment.
- One fine time, handled by a 100MHz clock from a crystal oscillator and counter C1 (27 bits).
At each EVTCLKF transition the fine and coarse times are stored in two 4 word fifos, M2 and M3 respectively.
At each EVTCLKS high-to-low transition the fine and coarse times are stored in two 1 word registers, M4 and M5 respectively .
The 100MHz clock is liable to have medium and long term variations due to different factors , and also have skew between the different stations. So in addition , a calibration ( monitoring ) of the 100Mhz clock is locally implemented. This is done by recording the C1 (100MHz) free-running counter at each 1PPS signal , in the register M1 (27 bits).
In order to monitor the possible variations of the 40Mhz used for flash ADCs , another calibration channel has been added , identical to the one described previously. This frequency is also the one for the PPC micro controller chip of the station controller. The 40MHz ( FCPU or SYSCLKBUS) is recorded by the C2 (40 MHz) free-running counter at each 1PPS in the register M6 (27 bits) .
The analysis of the shower signal (on both anode and last dynode of each 3 photomultipliers), measured by step of 25 nanoseconds on all hit detectors, associated to the precise time measuring, leads to the direction of the incoming cosmic ray with a precision better than one degree.
The correction to apply to the measured time to take into acount all these corrections is :
t(n)= c0*( 109+ ST( n+1)- ST(n)) / Cn+ ST( n)
With: t(n): nano second in the nth second
C0: nth second cycle
Cn: 100 MHz clock counts between two 1PPS
ST(n) : nth second saw tooth (value given by the GPS).
Slowly varying parameters or Slow ControlThis function is responsible of the control of the detector : command of the tension applied on the photomultipliers, LED flasher control, measure of the slowly varying parametres like PMT currents and voltages, temperatures, etc...
For the Slow Control, there are 29 analogic measures (inputs) on a range of 0 to 2.5 Volts and several numeric inputs and outputs.
A 12-bit Analogic Digital Converter (ADC) numerizes the analogic measures. 24 analogic input channels are multiplexed in order to be read by the 3 inputs of the 8 channels converter. Five other inputs allows the direct reading of the PMT voltages as well as the board temperature.
The main analogic measurements are the temperature, voltage and current of each PMT, the UB temperature, the battery and solar panels temperatures currents and voltages, as well as all the power supplies values. There remain two free inputs, reserved for the water temperature measurement.
The four analogic output voltages are delivrered by a 12-bit Digital Analogic Converter [DAC] and amplified at the output by operational amplifiers. They drive the voltages applied on the 3 PMs as well as the one on the LED Flasher. Last but not least, the numeric inputs and outputs allow, among other things, external triggering of the FrontEnd, to detectet the presence of the Ethernet Board, to generate the contraol signals of the SlowControl multiplexers, to manage the control signals of the TPS function, etc... acquisition software which is currently running into the Local Station [LS], is started as soon as the LS reboots.
During the first years of functionning, the LS software has been strongly modified. It's likely that the part related to the LS-CS communication is now stable, but that the acquisition tasks will still need some improvements.
So, for fiability reasons, the software has been split into two parts.
- The first one, called Service, concerns the communication with the Central Station [CS] and the synchronization :
- The message servers :
All the dialogue between LS and CS is managed by two message servers :
- MsgSvrIn receives the commands from the CS and pass them to the concerned process,
- MsgSvrOut receives from different processes the messages to be sent to the CS, with a given priority. The role of this server is to split these messages in packets and send one of this packet every second to the CS via the Suscriber Unit [SU] (the radio module).
- the GPS and 1PPS servers :
The GPS server receives informations, from the GPS module, of the 1PPS state. The messages are sent to the CS only if the 1PPS state is correct. Every process can ask to receive an interrupt from the 1PPS server, every second.
- the OS9 Command server :
The CS can send OS9 commands to be executed by the LS. It's also possible for the CS to download/upload files to/from the LS.
- The message servers :
- The acquisition tasks
This part of the software is still likely to change over the next two or three years. Physicists are very inventive people and like to change running conditions, for example the triggers.
The CS send a command to start the acquisition. This message is sent, via the message server input to the OS9 command server. When a task has to be launched, the system will look first into the Ram Disk, then into the Flash Eprom to find it. Then it will execute it.
When a new version of the acquisition software has to be tested, it is first downloaded into the Ram Disk. Then the CS sends a command to start the acquisition. If this new version crashes, the Station will reboot, start the Service part and wait for a command from the CS. The CS can then remove the bugged software from the RamDisk, reboot the Station and start the acquisition again. The system will then starts the old version of the acquisition from the Flash Eprom.
When a new version had been successfully tested, it is copied into the Flash Eprom and becomes the current aquisition software version.