Minutes of DAQ/Online Calibration Meeting 02/18/00, 9AM
Agenda:
=======
Friday, February 18, 9:00 am, B0 Theater
Please contact Dee Hahn (dhahn@fnal.gov) for Video connections
Minutes of last meeting Jack Cranshaw 5
(see below for corrections so far)
News
Round Table (5 minutes each)
-----------
SVX: Steve Nahn
COT/Muon: Ashutosh Kotwal
Database: Jack Cranshaw
CLC: Alexei Safonov
Consumer: Kaori Maeshima
Access Methods: Rick Vidal
Calorimeter/DBana: Vaia Papadimitriou
ShowerMax: Steve Kuhlmann
Presentations
-------------
CalibConsumer Base Class
Progress Jack Cranshaw 10
Discussion
----------
Next meetings
Milestones
AOB
Attending:
R. Vidal
A. Kotwal
J. Cranshaw
J. Patrick
S. Nahn
D. Box
Due to snow there was a late and small start to the meeting.
The only corrections to the minutes were the ones listed in
the agenda announcement.
ROUNDTABLE
COT:
Ashutosh gave a report on setting things up to measure the
delay of the 1260 strobe cables for the COT. The delays will be
measured to an uncertainty of <100 ps. It now takes approximately
2 min/cable including measuring and messing with connectors.
It will therefore take of order a few days to do the measurements.
To do the measurements the cables need to be in place which is
roughly 1 month away. It was suggested that Ashutosh contact
Young-Kee + Avi to check with Dervin that the cabling be done
one side at a time, so that as soon as they are finished with the
cabling on the first side, the timing measurements can begin.
These constants will then be used to correct the t0 values written
to the database. Bill has already set up the table and consumer
code to handle this.
Muon:
Ashutosh has not heard anything recently from the Muon folks, so
it was suggested that we tweek Kevin to check the status.
He did so and here was the result:
According to Kevin Pitts, the muon guys are using the calib system only
as a diagnostic. They will live with the timing spread as is. They will
not do any further calibration of the calib system.
Database Server:
Rick V. gave a short report on the arrival of the production
database server in B0. It arrived without any software, but Sun
should come on Monday to rectify that. It has 10 ethernet ports
instead of 18. There needs to be a decision on which Oracle version
to install: 8.0.5 (like on b0dau30) or 8.1.6. The machine is b0dau35.
User accounts will be granted on a need-to-use basis. It is expected
to be ready for use during the first week of March.
SVX:
X-mode pedestals have been taken. Steve has been testing with
overloaded crates, so the events are even slightly larger than they
will be in the final configuration. It takes roughly 3 minutes, which
it might be possible to shorten. One possibility for shortening it
is to get faster CPU's. It ends up that the crate CPU's for the
FIB crates have not been purchased yet. We could get boards with
300 MHz instead of 200 MHz processors with double the memory. There
are naturally subtleties in this decision which must be considered.
Steve was taking these events with a software event builder running
on Linux.
The event sizes becomes 1.5 Mb for a D-mode event where all of the
channels of the silicon return a value. For X-mode it depends on the
type of pedestals: 1/channel? 1/BEstate? 1/cell?
There is also an upcoming confusion creator over how to index layers.
The offline reconstruction is now moving to shifting all the layers
by one to accomodate Layer 00.
Consumer:
Kaori was unable to attend the meeting, but sent apologies and the
following:
We are very active updating the consumers and consumerframe
work these days. But I don't thing anything is directly
affected to calibration....
Calorimeter:
Jack and Jim reported that there was a
problem with different results between qietest and the run control
calibration code because they use different code. Vaia, Hao Wei,
and Frank C. were tracking this down and may have found the problem.
Here are the details from Vaia's email:
Summary: No data during past week. There have been problems with running the
consumer that puts data in the database and problems with the consumer
server logger. Before I left on Wednesday I think I understood what was the
reason the consumer got no data so I will plan to fix that for Monday's run.
The data now go tp the right partition area at b0dau31 (partition 5 in my
case) and the consumer was still trying to get them from partition 1.
This was a recent change from event buildr side. The main problem though
we are trying to debug during past two weeks is that when I plot
slop error/slope in hadron calorimeters the quantity decreases as a staircase
as I go through sequential admems. We have traced it to the fact that the
qietest code that runs in run control does not produce that same results
as the same code running stand alone. We know this as from past week
but it is not absolutely clear yet ow this pattern is generated.
We have to understand that before we are able to take meaningful
calibartions. This past week the hardware was also in bad shape.
There are in principle 4 crates available now in central but the two
were without power on wednesday and the two were under tests and the only crate
I could test some things with was a plug crate. I hope that on Monday we can make
more progress with this, at least understand better the che differences in the
two codes.
Banks:
Jack reported that there had been a barrage of mail messages during
the week about how best to add banks/StorableBanks. Dave Dagenhart
is in charge of 4152, so descriptions of bank data should be sent
to him. The most workable solution for adding banks classes to the
Banks and StorebleBanks packages seems to be 'Hand-Holding' where
the people who need to create a bank and accessors submit their
definition to Dave and then contact Liz K. and Rich Glosson who
will work with them creating the classes.
Shower Max:
Steve Kuhlmann has been writing code and got caught in the banks
confusion. This led to the above proposal -- and hopefully resolution!
Luminosity: Alexei Safonov:
(email)
We have made some progress recently. In particular, we obtained
the signal from LED source (the only CLC-specific kind of calibration),
we successfully wrote amplitude D-banks and were able to read the
data back. We used coinsidence to trigger the TS (TS has external
trigger input, and generates L1/L2 accepts after a preset delay).
We managed to adjust all our internal delays to synchronize L2
accept with the signal arrival on the end of the L2 buffer. In our
immediate plans: we want to include CIC (COT calibration card, we
use it too) to generate the LED signal from the real TS CAL_ENABLE
signal and to repeat the whole sequence of timing-adjustments.
After that we plan to include timing measurements (creation of
timing D-banks and reading them back using the same infrastructure
we use for the amplitude banks).
CALIBCONSUMER BASE CLASS
Jack reported that Olga has finished most of the work for an
intermediate class for X-mode consumers. This class sits between
AppModule and the X-mode module which should involve just writing
the code for decoding the C-bank and stuffing it in the database.
Generic functions readDB and writeDB which handle any database
manager or container have been incorporated into the intermediate
class. Both the COT and Calorimeter X-mode consumers have been
upgraded. The next hurdle is to upgrade these consumers to EDM2
which Rich Glosson is currently doing.
Steve suggested that these generic readDB and writeDB functions
may actually belong at a higher level since consumers in general
may want to use them.