Present: Andy Bacher, Kirk Bays, Chris Lavelle, Hermann Nann, John Olmsted, Paul Pancella, Mark Pickar, Adam Smith, and Ed Stephenson
Abstract:
The run from June 3 through 24 had two objectives:
Both objectives were met. By the end of the run we had missing mass spectra containing several dozen "alpha + pi0" events that were associated with two gamma rays recorded in the Pb-glass detectors. The pi0 peak had a width of about 700 keV. There were a comparable number of alpha + 2*gamma events, presumably from double radiative capture.
Details:
Startup followed a machine down time during which work was done on the RFQ and the electron cooling system. Aging tubes in the RFQ were replaced. The transfer energy between CIS and the Cooler was not exactly the same as on our previous runs, with the result that the cogging system for extraction did not work properly. Eventually, we got stable beam, but there appeared to be a limit of about 1.7 mA on the circulating beam at the top of the ramp to 228.5 MeV, independent of how much extra beam was available from the fill.
Setup again started with the 25-degree detectors in place. A check with an H2 target showed that we were again able to obtain luminosities that agreed among the three measurements, one with the forward counters and two with the 25-44 degree scintillators as situated on either side of the beam. This was done at a lower rate where L ~ 2E30 /cm2/s. So we proceeded with the calibration by running for a few hours each with H2, HD, and D2 targets. We conserved HD gas by roughing the gas feed lines rather than using gas to purge the system. The only thing to observe during the HD target runs was the presence of two distince peaks in the silicon energy spectrum. The question of whether these two luminosity systems give consistent values for subsequent d+d runs was left to offline replay. Once completed, the 25-degree detectors were removed and the Pb-glass counters rolled back into position.
It was decided to retain the flattop time of 103.5 s from previous running. A scan was made of the data acquisition rate as a function of the pushing pressure in the target. This showed a maximum that was associated with a lifetime of 94 s, so operation near this point was the goal for the rest of the run. A scan of the horizontal position of the beam revealed that on one side of the point of best ring acceptance we had a background of deuterons that made triggers in the channel and significantly increased the rate. So we ran with the beam tuned a little to the inside of the ring about 1 mm from the threshold for this background. The beam stayed stable throughout the run to within a small fraction of a mm and this background was never a problem.
The major difficulty with early D2 production running was keeping the second Pb-glass high voltage supply running. The controller would shut off the supply after minutes to hours of operation. We ordered a replacement controller from Brookhaven, but it did not arrive as soon as expected. The temporary solution of putting all 256 high voltage lines into the first supply worked and was left for the rest of the run.
A second issue was the question of whether random coincidences in the veto detectors were removing a significant number of good events from the data stream. A rough estimate put this value near 10%. So Mark Pickar assembled two "out-of-time" coincidences where the delay was set equal to one rotation period of the beam. These were scaled and showed typically several percent random losses due to spurious vetoes of good events.
The major issue for alpha+pi0 identification early in the run was whether the time-of-flight information on which the missing mass calculation is based was in good shape. Everything in the reconstruction was based on the TOF from earlier 3He + pi0 missing mass calculations and the presumption that nothing had changed since. In fact, over time the gains of the various channel PMTs had changed, and our response had been to adjust the high voltages to maintain consistent gains. This process carries the possibility of changing the transit time of signals going through the PMTs. In addition, if the gain sags, there is also the possibility that this time varies during running and smears the time signal.
By looking at 3He + pi0 data from old commissioning runs, we learned that unless the TOF is within a range of +- 2 ns or less of the right answer, the alpha + pi0 signal will be obliterated.
The first thing we did was to search for a "marker" of TOF that we could monitor comparatively quickly. After trying a number of "features" in the channel Z=1 data, we settled on a group of deuterons that range out in the E-detector. By picking this group and also choosing a trajectory through the channel using MWPC gates, we could generate a group whose TOF was a narrow peak 2-3 ns wide. We then took data with the current scintillator voltages and also with the voltages used for the 3He + pi0 running for the 0.95-degree cone (since this has the best missing mass resolution). Various PMTs showed time shifts up to 10 ns. Once an offset for the trigger shift was removed, the relationship between voltage change and channel shift was nearly linear. Thus we felt confident that we could approximate the shift due to PMT voltage changes. Next the 3He + pi0 data was re-examined to make sure that the TOF starting point was optimal (it wasn't). Then the time shifts were calculated and added to replay software that was used during the production run. The only confirmation that this procedure is working is that a peak in missing mass appears at 135 MeV and that this peak position is insensitive to the initial TOF input. It took about 1.5 weeks of running before enough statistics were accumulated to have some confidenct that the time offset was correct for each PMT (there are 6 that are critical to TOF). The signature of an incorrect offset is a tilting of the locus of pi0 events in a missing mass by TOF plot. On the basis of such plots, several small adjustments were made to the time offsets in order to obtain "optimal" results. The small number of 2-gamma candidate events and the presence of double radiative capture background made this process subjective. The only cure is more statistics. For the delta-E1D scintillator, the base had been changed before the run, and this produced an additional 8 ns time shift that was recovered only because the time shift was different from the other delta-E1 scintillators when compared with Pb-glass timing. In the end, we obtained a trial missing mass plot with a peak at 135 MeV whose width was about 700 keV. This contained between 60 and 90 events, depending on the Pb-glass cuts used to select 2-gamma events. A comparable number of background, or double radiative capture, events were present in the spectrum down to about 120 MeV.
Beginning with the swap to a single Pb-glass high voltage supply on Sunday, June 9, production running was generall smooth. Runs lasting 2-5 hours were recorded for most of each day. There were interruptions totaling usually no more than a few hours each day. Most of these were for regenerating the cryopumps located near the target. The stage-2 pumps would load up after about 1 day of operation. The gas jet nozzle was warmed twice, but it was not clear whether this helped. Fortunately, we were able to maintain sufficient target thickness through until the end of the run.
There were several machine problems, including poor lifetime for various reasons, regenerating the beam line 9 cryopump, fixing the extraction phase for CIS, cleaning the filters for the CIS chilled water system, and dealing with power line glitches. Incidents in which the beam would suddenly vanish from the ring were traced to a bad rampDAC card. The usual mechanism is an unstable bit that changes the regulation voltage on the card output. These losses show up as a sudden spike in the trigger rates for events 4 (cosmic) and 6 (luminosity). In addition, the front two wire chambers in the channel shut off from overcurrent.
Part way through, it was noticed that the event 6 rate would drop to zero for a few seconds on a random basis. This was fixed by changing the width of the logic module that comes just before the prescaler; perhaps this pot was noisy. The event 6 rate is recoverable from scaler information. We occasionally lost processing of event 5 data because the FERA module lost its unit number. Again, this data is recoverable.
Leakage current in the silicon detector wandered during the run and was occasionally tracking the changing particle flux as the target when on and off. This posed no problem except for a slight lowering of the silicon energy peak when the current was near 5 microA. The long-term trends in this were not understood.
The voltage on the front MWPC was slowly lowered during the run. The efficiency for Z=2 remains good, but there are losses for Z=1 (about half for Y1). This chamber was cleaned before the run, and will be cleaned again before the run in July. Keith Solberg has suggested that we operate at a lower voltage from the start to forestall any deterioration.
A logging scheme was put into place for tracking several items in the data acquisition, T-region vacuum, and beam properties. This was useful in spotting a number of problems and making sure that no large amounts of useless data were taken because of an undiscovered fault.
The run ended at 8:00 a.m. on Monday, June 24.
BETWEEN-RUN JOBS
We reminded everyone of the between run tasks on the job list. The only ring access will be July 3-8, with Monday, July 8 being startup day. We will need to change some gas bottles and install the detectors at 25-degree for a second luminosity calibration. The first MWPC detector will be mounted and all chambers checked with an electron source.
SHIFT SCHEDULE
The shift schedule will keep the pattern already established. Off times were noted for Paul, Hermann, and Ed. Jack Rapaport will be here on shift during July 16-21.
OFF-LINE ANALYSIS QUESTIONS:
CHECKLISTS
Some modifications to the checklist system were noted at the end of the run. To this list we decided to add MWPC leakage currents.
APS ABSTRACT
We decided to submit a "progress report" abstract to the fall DNP meeting in East Lansing. Ed Stephenson will present.
RUN PLAN
The initial run plan is a duplicate of what we did this time. We will start with a repeat of the luminosity calibration, followed by production running.
We discussed at some length the question of whether we should take time from the next running period to measure d+d elastic scattering with polarized beam using the A-region detectors. The purpose of this data would be to check the calculations of the deuteron entrance channel distortions.
Changing the source and optimizing the setup for polarized beam would require about 3 days. There would be another day of setup for the experiment. The run about 1.3 years ago to look at deuteron polarization in the ring required about 4 shifts of d+p data to obtain a useful check at several angles. So altogether, it seems that a week would be needed at a minimum to obtain useful data on d+d.
With the need for additional alpha + pi0 statistics, we do not want to devote more than a week to d+d elastic running. If there are problems, then the d+d would be skipped.
There was some consideration given to normalization of the cross section. Again, the only scheme is to run for some time with HD gas. This means that some gas should be held back in reserve from the luminosity calibration at the start of the run.
So the run would consist of doing H2, HD, and D2 targets in sequence. The first confirms the polarization of the deuterons in the ring. The second calibrates the cross section. The third accumulates d+d cross section and analyzing power data. Switches to H2 could be made periodically to check the polarization. The goal would be 10% precision.
For d+p scattering, the appropriate trigger is two forward prongs. For d+d, we would need to install the silicon detector array and use one forward and one backward prong as the trigger. In both cases, there would be a requirement that the prongs appear on opposite sides of the beam.
There was concern expressed at the meeting that this scheme offers the chance to obtain d+d data relatively easily and in a timely manner. A proposal at another laboratory later would delay this data for 1-2 years even if the proposal were accepted.