Summary of the FT Upgrade Meeting held on 19th and 20th June. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Present: Steve Burke, Don Clarke, John Dainton, Joe Foster, Erwin Gabathuler, Tim Greenshaw, Rob Henderson, Theo Holtom, Jim Kay, Steve Maxfield, John Morris, Girish Patel, Dave Sankey, Terry Sloan, Geoff Tappern. Current FT ---------- Girish explained the status of the FT monitoring and calibration. All is in very good shape with the exception of plots of current drawn against time. These are available, but not in a format which allows them to be displayed as part of the web monitoring. Attempts are being made to rectify this! The FT chambers seem to be performing very well. John Morris showed the results of a study of the planar chambers. He showed plots of gain (corrected for pressure changes) and resolution as a function of time (1993 to 1997). There is a clear case for flushing the gas through the chambers rather than recirculating; the resolution is above 200mu in the first case and about 175mu in the second. There is no significant evidence of ageing over the time period studied. Rob reviewed some of the evidence that the data from an entire orientation of the planar chambers are sometimes shifted in a correlated way. These shifts are up to 1cm in magnitude and appear to occur in cosmic and ep scattering data and in both directions in drift. Significant variables which can be used to tune the correlated effects across planar cells have been established for use in tuning the MC to simulate the data taken to date. Questions were raised about whether the material was well simulated. It was suggested that the effects of lumpy material could be investigated by adding material to the simulation in a non uniform way. Physics with current FT ----------------------- Theo Holtom showed his p_T spectra which involve the use of the FT to extend the pseudorapidity range covered. The advantages of further increasing the pseudorapidity range studied were discussed. Dave Milstead's work was shown. He has taken CJC tracks, projected them into the FT, and looked at the efficiency for finding those tracks and the p_T resolution obtained (1994 data). The MC seems to be underestimating the efficiency of the FT in this region. The p_T resolution in the data is somewhat worse than the MC prediction, but no systematic shifts are observed. This is in contrast to the work of Kirstee Hewitt. She has compared the FT and CJC momentum measurements in the overlap region and sees systematic, charge dependent differences (1995, 1996 data). W.r.t. the CJC, the FT underestimates mom. for -ive tracks and overestimates for +ive tracks in 1995. In 1996 it overestimates mom. for -ive tracks but +ive tracks appear correct. These results are supported (with low statistical significance) by the Fmu system momentum measurements when corrected for energy loss in the LAr and iron. They require clarification! ACTION GDP,JVM,SB Repeat Kirstee Hewitt's plots plotting as a function of q/p as well as in the way Kirstee plots the data. Decide what the problem is. Dave is looking at p_T spectra as a function of Q^2. He has found that the MC incorrectly simulates the FT efficiency particularly in photoproduction events (MC underestimates efficiency) leading to obviously incorrect results (step in spectra at FT CJC boundary). Motivation for upgrade ---------------------- Terry confirmed that the FT region is important in the study of leptoquarks of mass about 200 GeV as in this case either the scattered lepton or the quark jet is to be found in the FT region. The case for intrinsic charm studies is less clear. Terry's work is described in more detail in: http://hep.ph.liv.ac.uk/~green/ftupgrade/notes/lqkine.ps Performance of current/upgraded FT ---------------------------------- Steve showed first results from the programme of studies he started following the previous meeting. He has looked at the purity of samples of primary hadrons obtained using the current FT and the efficiency with which they may be detected, using first the high Q^2 MC data. Rough numbers are 40% efficiency and 75% purity, with Steve's track definition which required one planar and one other segment. He then studied the effect of improving the FT by decreasing the amount of material between the CJC and the FT to a minimum. Both the efficiency and purity of the sample increased a little. A larger improvement was made by increasing the planar segment finding efficiency by removing the correlated data loss. The results on the performance of the current FT were confirmed by John Morris's work using lower Q^2 MC data, with the exception that John had made the assumption that the amount of material in the CJC end-wall is the same as that distributed throughout the forward tracker, leading to his expectation that a large proportion of secondary vertices are to be found in the CJC end-wall, this will be discussed further below! (see http://dice2.desy.de/~jvmorris/FT2ndaries.html for more on John's work.) Steve then looked at the scattered positron in his high Q^2 MC data. The performance of the current FT, using the same track definition as above, was poor. Removing all possible material from the CJC end-wall made little difference. Even removing the correlated data loss left an FT capable of measuring the charge of the positron in only a few per cent of events in which the scattered electron passed through the forward tracker. Furthermore, the mean energy lost by a positron of energy about 200 to 300 GeV in moving from the interaction point through the FT is around 75 GeV. Discussion of Steve's results centred on two important facts: An interaction length is typically a factor of 5 to 10 larger than a radiation length (eg. Cu, lambda=15.1cm, X_0=1.4cm.). A radiation length is the distance in which an electron's energy is reduced by a factor 1/e=1/2.7=37, ie. 0.63 of the electron's energy is lost in the material on average. (As a reminder, the conversion length is 9X_0/7=1.3X_0.) Study of the amount of material in the CJC end-wall (probably too low in MC) and in the FT makes Steve's numbers regarding the energy loss plausible and, given that the FT is functioning as an electromagnetic calorimeter (with a relatively high proportion of active material), explains also the results as regards the charge determination. The difference between the interaction and radiation lengths explains why the FT functions better as a detector for hadrons. The low track finding efficiency for hadrons is also affected by the amount of material in the FT and CJC endwall, however. This is clear from work done by Rob. He showed how the segment finding in the planars depends on the track density in the FT. The number of spurious segments found starts to rise, and the efficiency for finding the correct segments to fall, for track densities of the order of 15 over the entire FT. (In an electromagnetic shower, local track densities rapidly exceed this value.) The number of photons traversing the FT in an event is (on average) the same as the number of charged tracks. If 1/2 of these convert, as is likely given the amount of material in the CJC and FT, this doubles the number of charged tracks in the FT, even before account has been made of the secondaries that scatter into the FT from the collimator etc. (MC studies have shown that typically, the number of secondary tracks traversing the FT exceeds the number of primaries by a factor of three or more.) Again, the segment finding limit is reached. It appears that it is impossible to significantly improve performance by modifying Rob's software. Improvement for hadrons could be made by tightening segment finding criteria if the chamber resolution were effectively the measured single hit resolution of about 175mu. The track density limit is then reached at 20 tracks over the entire FT area. Correlated shifts of groups of hits on an orientation in the planars mean the effective resolution is nearer 400mu, however. (Note, that the showering observed for positrons is so dramatic that even with the 175mu and tightened segment finding the improvement is unlikely to be large.) These results have somewhat changed the emphasis of the upgrade. It is obviously of paramount importance to decrease the amount of material before and within the FT. Roughly, over the angular range 7 to 17 degrees, the material presented to a particle passing through the CJC/FT in units of radiation lengths is: CJC end-wall 0.3 (underestimate?) 3*planar 0.32 3*MWPC 0.1 (underestimate?) 3*radial 0.1 ---- 0.82 Outside this angular range the amount of material climbs steeply due to flanges etc. (Data from Dave Sankey's thesis.) ACTIONS Repeat the analysis with chambers made from material totalling 10%X0 thick. (SB) Check the material budget (TS). Get Dave Milstead's study of pairs (JVM) Study the composition of the shower ie. electrons accompanied by 1 gamma, 3 e etc. (SB) Silicon/FT ---------- Steve Burke (yes he is doing most of the work!) described preliminary studies of FT/FT Si linking. Projecting FT tracks into the Si allows unique linking of hits. The momentum resolution obtained is better than that using just the FT alone and slightly better than that obtained using the FT plus vertex constraint. Using the vertex, Si and the FT allows a further improvement. Planar HV check --------------- Joe Foster described a device that may be inserted in the planar high voltage lines to allow a search for spikes indicative of breakdown somewhere in the chambers. This could be a cause of the shifted hits mentioned above. Tim will take this to DESY next week and try and get some measurements done. Feasibility studies ------------------- Don Clarke, Jim Kay and Geoff Tappern discussed the feasibility of the various proposals made to modify the forward tracker. Jim showed possible designs for options A and B proposed by Rob (see http://hep.ph.liv.ac.uk/~green/ftupgrade/notes/doweupft.ps) with the MWPC in fixed z positions to maintain the trigger function. Jim will estimate the cost of these two options. Geoff showed a solution with a shorter tank and with extra planar orientations in place of the transition radiators. The MWPC would need to be redesigned and remounted elsewhere and the feedthroughs used for the new planars. A ball park cost for this was about 300k pounds with roughly 200k pounds needed for the extra planar chambers. This does not include the cost of the new MWPC trigger which would need to be provided from another source. Trigger studies --------------- Tim explained how the z-vertex trigger functions. The amount of effort necessary to rebuld the trigger logic would be considerable. The following options are available: Keep the current chambers and logic, then first MWPC must be at its current position to within about 1cm. Keep logic but build new chamber, then chamber ``pads'' must cover same (theta,phi) regions as those of current chamber, perhaps possible within this group, but stretching things! Rebuild the lot, then we almost certainly require additional effort. The forward ray trigger was also discussed briefly, but at the time of the meeting little information was available on this. One point to note is that Christian Pascaud has expressed concern that the chambers may not survive through all HERA's lifetime. Rutherford Electronics are likely to be involved in efforts to build a L2 CJC based z-vertex trigger. This was described by Dave Sankey, as was the current status of the DAQ upgrade proposal. Perhaps the FT trigger logic problems could be solved with Rutherford's help? Proposal -------- The current status of the proposal was shown. A subset of this document can be given to Phill at the forthcoming H1 meeting. Conclusions ----------- It has now been shown that the efficiency problems in detecting hadrons in the FT are related to the amount of material before and in the FT. Reducing this is of great importance. Without a significant reduction in the amount of material, measuring the lepton charge in high Q^2 events is impossible. Given the mechanical constraints (minimal modifications to tank possible), the feasibility of the following possibilities should be considered (Don Clarke, Jim Kay and Geoff Tappern): A) Remove all MWPCs. Remove TR volumes. Remove old planars. Add one new 4 orientation planar (x, y, u, v) per supermodule with 6 wires per orientation and total thickness < 3% X_0. If possible, shorten the tank to leave space for trigger chamber. B) Remove all old planars. Remove all MWPCs. Add one new 3 orientation planar (x, u, v) per supermodule with 4 wires per orientation and total thickness < 3% X_0. Add one new MWPC per supermodule, with total thickness < 3% X_0. Note added by Rob; this is a high risk option since there is no added redundancy over the present system. We know that such redundancy is important to sort out hodoscope ambiguities. C) If in B) the positions of the new planars and new MWPCs can be switched, investigate adding further planar orientations of the type used in B) again ensuring that the amount of material involved is an absolute minimum. We should attempt to find out if these options are capable of giving the required performance. We need to know: What is the purity/efficiency for scattered electron/positron detection if the CJC endwall is minimised, the amount of material in the FT is minimised (by making bits of FT air in simulation), the planar resolution is set to 150mu and the segment finding cut is set to (about) 3mm. (This is probably the closest we can get to a simulation of the performance of our improved FT.) (Steve Burke) The same assuming all improvements except the decrease in thickness of the CJC end-wall. The same for hadrons in high Q^2, medium Q^2 and charm events. (Steve Burke) The charm selection power (consultation with FSi group necessary). Assume best FT and see what can be done with rudimentary linking that Steve has already looked at to some extent? (Steve Burke and Karin Daum) What is the likely effect on segment finding of the planar configurations described above? (Rob Henderson) Goals ----- For the FT we now redefine these to be: Over 90% efficiency for hadron tracking. Reconstruction of the charge of the scattered electron for 90% of electrons which do not undergo a secondary interaction before entering the FT. We must examine whether the latter is sensible: if too many electrons shower before entering the FT even with a minimised CJC end-wall there is no point in pursuing this goal with the FT alone. Further things to do -------------------- Trigger options (Tim) Proposal (Tim) Presentation to H1 (Tim(?)) Next meeting ------------ FT upgrade meeting, RAL 5th and 6th August 1997 FT meeting, DESY 3rd and 4th September 1997 Thanks again to all contributors for what was a most instructive meeting!