Configuring the LGD

Configuring the ADC's

The ADC's are configured using "/home/mpsdaq/e852/daq/bin/setupLGD -t". This will scan the crates for ADC's, ROC's, and the MAM, and configure them properly. It will issue warnings or errors when something goes wrong. When run with the "-DAQ" option, it will also send errors to the alarm server about problems. These errors are VERY important! If any errors are encountered, the LGD will not perform correctly. The output of setupLGD should look like this:


SG:\mpsdaq\>setupLGD

Reading map...done

Scanning crate for ADC's...done

 0 : 24 ADC's at PA 0x10000018

 1 : 24 ADC's at PA 0x11000018

 2 : 24 ADC's at PA 0x12000018

 3 : 24 ADC's at PA 0x13000018

ROC at 0x10000019

ROC at 0x10000019 will control segment 0 at 0x10000018

ROC at 0x11000019

ROC at 0x11000019 will control segment 1 at 0x11000018

ROC at 0x12000019

ROC at 0x12000019 will control segment 2 at 0x12000018

ROC at 0x13000019

ROC at 0x13000019 will control segment 3 at 0x13000018

Configuring segments as MDT groups...done

Building channel ID's...done

Downloading channel ID's...done

Downloading pedestal trim values...done

Downloading discriminator values...done

If FASTBUS errors are reported, the first thing to try is simply re-running the program. If it succeeds, everything is probably ok. If it continues to fail, there is something wrong in the hardware. If there are no ADC's found in a particular crate, the first thing to try is to reload the SI routing tables using /home/mpsdaq/e852/daq/bin/si. If this does not solve the problem, a DAQ expert needs to be called in to debug the problem. If individual ADC's or ROC's are missing, they are probably bad and need to be replaced. The same is true for the MAM.

FASTBUS errors reported during the configuring of the modules can indicate a problem with DAQ or a problem with the module. If the errors allways occur at the same place (i.e. loading the channel id's to the ADC at 0x12000015) it is probably the module that is bad. If they move around, there is something wrong in DAQ, and an expert should be called in.

Map errors generally occur for one of two reasoons:

E852_CONFIG is not defined correctly.
The map file is corrupted. This is really bad, and an expert needs to be called

Note that to configure the LGD for testing with the local trigger, you need to run setupLGD with no arguments.

Pedestals

The IU ADC's have user setable pedestals. There are 32 CSR's (0xc0000020 - 0xc0000039) that control the pedestal value for each channel. The possible values for the "pedestal trim" are 0 - 0xff. These values correspond to a DC offset applied to the integrator. With the value set to 0, the pedestal are actually negative (however, they still show up as zero when read out because of the digitation process). As the trim value is increased, the pedestal increses. For E852, the pedestals are set at a nominal value of 15 ADC counts.

There is a Motif based program for trimming the pedestals : /home/mpsdaq/e852/daq/bin/lgdPed. On-line help is available for this program, so not much detail will be presented here. The basic theory of it's operation is as follows:

Scan the crates for ADC's, and allow the user to select which ADC's will be trimmed.
Repeat the following for each ADC to be trimmed:
- Set the pedestals trim values to 0xff, causing the pedestals to be too high.
- Take a user settable number of pedestal events (usually 20). Find the average value for the pedestal. If the average is greater than the target value, lower the trim. If it is more than 2 counts less than the target value, raise the trim. If it is within this window, call it good.
- Load the new trim values, and repeat until all channels in a module are ok, or 80 iterations have been done. If, after 80 iterations, a channel has failed to trim, an erorr is reported and the program moves on to the next module.
  - If at any time the trim value reaches 0, the trimming is aborted for that channel and an error is reported. A trim value of 0 means that the trim has been lowered as far as it can go, but the pedestal is still to high. Notify and LGD expert of the problem.
  - If, after the first iteration, the trim value is 0xff, the trimming is aborted for that channel and an error is reported. A trim value of 0xff means the trim has been inscreased as much as it can , but the pedestal is still too low. Notify and LGD expert.
  - If the width of the distrubution of pedestal values for a channel during any one iteration is greater than 4.0, a warning is reported. Wide channels are genrally caused by noise on the signal cables. We have tried to fix this problem in the past, but we have been unable to do so. Generally only a handfull of channels are effected (less than 100). Unless the problem becomes a true epademic, it can be safely ignored. If it does get out of control, there is no idea how to fix it.
  - For more information on debugging problems with the ADC's, see the note " IU ADC maintainence "
Save the new pedestal trim values to the map. A summery of problems encountered is given after all ADC's have been trimmed.

The pedestals do not need to be re-trimmed often. They should be re-trimmed when an ADC module is replaced, however.

Note that the pedestals need to be trimmed seperately if you are using the local trigger to debug the LGD. This is done by clicking on the `Use local trigger' toggle.

Discriminators

Each ADC channel has 2 discriminatos, designated " high " and " low ". The low discriminator is used to control zero suppression on readout. The high discriminator is used to control zero suppression on MAM readout.

Low Discriminators : Only channels whose low discriminator has fired are read out for data events. Unfortunatly, there is some jitter in the discriminators that causes too many channels to be read out. This problem has been investigated, and no solution has been found. Instead of performing a full zero suppresion on readout, these discriminators suppress about 2/3 of the channels at pedestal on any given event. If the discriminator settings were raised enough to give true zero suppression, some data would be lost (up to about 20 counts over pedestal could be suppressed by mistake). The correct values for the discriminators are found by running /home/mpsdaq/e852/daq/bin/discSet -fmap. This will find new values for the discriminators and save them to the map. They are downloaded by the LGD setup code, /home/mpsdaq/e852/daq/bin/setupLGD. Note that setupLGD must be run after discSet for the LGD to function correctly.
High Discriminators : Only channels whose high discriminator has fired are read out by the ROC's for the MAM. The value of this threshold is set by the trigger control code, and is generally 75 to 100 counts.

** 1/10/95 R.L.