BaBar Drift Chamber Current Problems

This page summarises the known problems we are experiencing with the Drift Chamber. The problems are being worked on, and will hopefully disappear before I am forced to document it as a permanent feature. At the bottom of this page is the complete list of past "current problems" for archival purposes.

Problems? What Problems?

TSF-TIOM GLink problem

On 06-Feb-2002 the red GLink Ready and Synch LED on TSF Y0, connected to TIOM 0 went on. This coincided with damaged events and 100% dead time from the TSFY crate. The cable's length and reflectivity and the TIOM power output were all within spec. Power-cycling the DCH low voltage "fixed" the problem. On 11-Feb-2002, the TIOM (#660) was swapped with #669. The TSFY problems returned. The fiber from the patch panel to the TSFI - Y0, Y7 was swapped and the problem went away.

EPICS does not display last discharge channel

On the EPICS panel DCH-> HV-> Spike Ack.-> Last Disch. Times the "Last Discharge Ch." always reads (-1, -1, -1) indicating "none".

Calibration channels that fail

With the accelerator turned back on, we're back to having about a dozen failing channels, mostly channels /2/0/d/*/*, /2/1/1a/*/*, and /2/2/2b/18/*.

Online calibrations currently have about half a dozen channels that fail regularly, primarily on T0, /2/0/d/0/*, /2/1/1a/12/1, and /2/2/2b/18/* are the most common.

Data acquisition failures ("Xilinx reset")

This problem was fixed during the Winter 2000-2001 shutdown, by replacing the fiber optic circuitry (Finisar laser and GLink chip) in the affected Data I/O Module. If the problem comes back, we need to know about it!

The problem came back beginning 25-Feb-2002. The Data Flow software that forces the electronics reconfiguration through a PAUSE and RESUME has been reinstated. However, since swapping ROMs (crate 13 #1 and #2) on 7-Mar-2002, the problem has completely disappeared.

Section of chamber OFF

Since 03-Nov-1999 we have reconfigured the HV segmentation in the problem area of the chamber. Two HV boards (16 wires each) from superlayer 6 are now off, as well as 5 HV boards in superlayer 5 that encompass the swath covered by the superlayer 6 boards above. The HV segmentation in that area is currently :

Only one dead channel

A lot of work was done during the October shutdown to repair all the dead channels that we had up to now. We currently have a single dead channel, superlayer 7 wire #34. It shows up in Fast Monitoring plots as well as in the list of dead channels.

End of current problems list.

Obsolete problems. If they come back, let us know!

Data acquisition failures ("Xilinx reset")

Starting around 11 August, the front end electronics for quadrant 1 (DCH crate 13 ROM #2, FEE's 16-31) have been resetting themselves in the middle of normal data taking. This causes a characteristic pattern of missing channels, because the default readout configuration has some ELEFANT chips enabled and others disabled. Several runs, about 15 pb^-1 of luminosity, were lost to this problem.

Software is in place in DataFlow to both catch the problem early, and to reconfigure the electronics to eliminate unusable data. When the problem occurs, a fake "DCH Crate Full" error will be asserted, and the DAQ shifter must PAUSE and RESUME the run to force the electronics reconfiguration.

Calibration channels that fail

Online calibrations typically have between 10 and 20 channels that fail, mostly on t0, but a couple on analog gain (AGain). The stealth box (/2/0/c/*/* and /2/0/d/*/*) has not been optimized, so there are more failed channels there, and channels /2/2/2b/18/[012] nearly always fail.

Calibration channels that fail

Since the October shutdown there are only a few channels that fail the online calibration. Two channels fail essentially all the time, they are /2/1/1b/18/0 and /2/2/2b/18/0. On occasion you will also see /2/2/2b/18/1, /2/2/2b/18/2 and /2/1/1b/18/2 fail.

Water concentration is 3000 ppm, old sensor removed

Since about 23-Oct, the water bubbler has been turned on. Right now the water concentration is around 3000 ppm, and is expected to stay that way. The concentration measurement is available in EPICS. Go to the DCH gas system panel, and right click the "Control panels" button to get a menu. The menu item "New water sensor" gives the measurements.

The old water sensor that we had been using up to this point has been removed. Therefore, the readings in EPICS are nonsense. Use the readings from the new sensor as described above.

V1 nominal setpoint is 1900 V

Since 23-Oct the nominal setpoint for V1 is 1900 Volts, instead of 1960 Volts, which we had been using up to this point. This is about a factor of two less gain. The change was done to try to prevent further damage related to the problem described below. The setting for V0 is still 1200 Volts.

Bad HV Sector

One HV sector in superlayer 6, outer quadrant has a problem. We cannot turn on this sector for the moment. The V1 setting has been modified for this sector so that it will remain at 1200V when the DCH is brought up to the Runnable state.

This sector shows up in the Fast Monitoring histograms. In the hit map, 1/4 of superlayer 6 on the righthand side appears empty. This also affects the shape of the FEE hit map, etc. Look at recent FastMon histograms to see what to expect.

The inner quadrant sector in the same superlayer (6) has a different kind of problem. It shows current spikes 4-5 times more frequently (a few times a shift) than any other HV sectors. It has been the dominant source of DCH HV trips. In order to localize the problem into either the chamber or the power supply, the HV cables have been swapped at the output of the HV filter box. The mapping of the HV channles is therefore wrong for the moment: the top and the inner quadrants are exchanged.

Water Concentration

We have started adding water to the gas to prevent aging. The water concentration is currently around 400 ppm, but will be rising slowly for the next few weeks. Experts are paying close attention to this, so please ignore whatever the water concentration you see during the checklist.

Dead Channels

In addition to the 2 channels (one in SL8, the other in SL10) that have been dead for quite a while, we have now 8 bad channels. They are all on one Elefant, wedge 6, FEE 1, chip "b". Four of them show up as holes in the hit map (hard to see) in SL3, around 10 o'clock. They are also reported at the end of offline Fast Monitoring. We will fix them when we have other more important reasons to take an access to the electronics. Eight out of 7104 channels is just not serious enough to open the detector.

Online Calibration Failing

The above-mentioned bad Elefant is causing the online calibration to fail for the entire ADB. At the validation phase of the calibration, you should see a lot of errors now. Ignore all error messages that come from channels /2/1/18/[23]/[023467] or /2/1/19/[89abcd]/[01234567]. The remaining errors must be from the 3 known bad channels /2/1/17/18/0, /2/2/2b/18/0, and /2/3/3f/1c/2. If there are no other channels failing, validate the run. (The calibration for the bad channels will not be changed.) If there are other channels failing, don't validate the run and send e-mail to Gabriella and me.