Long Precision Drift Tube Production at Michigan E. Diehl, S. Goldfarb, D. Levin, S. McKee, H. Neal, H. Schick, G. Tarle, R. Thun, C. Weaverdyck, Q. Xu, Z. Zhao, B. Zhou Department of Physics University of Michigan Ann Arbor, MI USA May, 2002 Abstract: The University of Michigan ATLAS group is constructing 31000 long precision drift tubes for assembly into 80 Monitored Drift Tube (MDT) chambers for the endcap of the ATLAS muon spectrometer. These chambers provide a precision measurement of muon momenta in the ATLAS endcap toroid magnet. This paper describes the assembly techniques, quality-control tests, and data from 16,000 ATLAS drift tubes produced at the University of Michigan. 1. Introduction The University of Michigan ATLAS group is constructing 31000 long precision drift tubes, for assembly into 80 Monitored Drift Tube (MDT) chambers for the middle layer of the endcap of the ATLAS muon system ([1]). These chambers are part of the ATLAS muon spectrometer and provide a precise measurement of muon track sagitta in the ATLAS endcap toroid magnet. The tracking resolution per tube is approximately 80 micron. The expected muon momentum resolution is about 2% at a transverse momentum (P_T) of 100 GeV/c, and 10% at a P_T of 1000 GeV/c. The basic detection element is a cylindrical aluminum drift tube with a W-Re central wire. The tube is operated at a wire potential of 3080 V with Ar/CO_2 gas (93:7) at 3 bar absolute pressure for reduced diffusion and ionization fluctuations. This paper will focus on our precision drift tube production techniques, quality-control (QC) tests, and mass production experience. Individual drift tubes are assembled into muon chambers consisting of six tube layers with 64 tubes per layer. The layers are divided into two 3-tube multi-layers (ML) which are glued on either side of a spacer frame. The chambers are 2 m wide, 0.35 m high, and with lengths from 3 to 6 m. The endcap chambers are built in a stepped trapezoid shape with 8 tubes per step. Michigan is making 16 each of chamber types EMS5, EMS4, EML3, EML4, and EML5. For these chambers, 40 different tube lengths must be produced. MDT chamber production has been reported separately ([2],[3]). The drift tubes are referred to as ``monitored'' drift tubes (MDTs) because the chambers contain optical systems to monitor the position and distortion of the chambers during data taking (see [1] for description). The ATLAS group at the University of Washington pioneered many of the endcap tube assembly techniques ([4]). We have implemented and adapted many of their tools and methods in our tube assembly line. 2. MDT Tube Components A monitored drift tube consists of three major parts: an aluminum tube of diameter 30 mm and 400 micron wall thickness; two endplugs which hold the wire in a precise position and which provide electrical and gas connections; and a 50 micron gold-plated tungsten-rhenium wire. Figure 1. shows an exploded view of all tube parts. figure{tube_diagram.jpg} Figure 1: Exploded view of monitored drift tube. The raw tubes arrive from the factory cut to size (accurate to +- 0.5 mm) and cleaned. Each chamber type requires a total of 384 tubes in 8 different lengths. The tubes are stamped with bar codes indicating the tube ID and tube length. The MDT central wire is located with a precision of better than 25 micron footnote{The gravitational wire sag, while typically much larger than 25 micron, can be determined precisely from the known length and tension of the wire.} by endplugs made of injection-molded Noryl with an inner brass tube and an outer aluminum cylinder (see Figure 2 The outer aluminum cylinder has a 4 mm wide precision surface with diameter 30.01 +- 0.01 mm , which is a reference surface for positioning a tube. The inner brass tube is precisely machined to be concentric with the outer precision surface. The brass tube houses a wire locating device called a ``twister''. A twister is a 10 mm long, 5 mm diameter, brass spiral whose inner surfaces define a 50 +15 -10 micron central hole to position the wire. The twister positions the wire to an accuracy of 10 micron with respect to the outer reference surface of the endplug. The twister is held by a small plastic clip. The inner brass cylinder protrudes from the back of the endplug into a threaded tube, on which is attached the gas manifold, and ends in a crimp tube that holds the wire tension of 350 g. The crimp tube is made of copper with dimensions length 11 mm, O.D. 1.2 mm, and I.D. 0.35 mm. The outer aluminum cylinder provides a surface on which the aluminum tube is swaged to provide mechanical and electrical connection between the tube and endplug. An O-ring provides the gas seal between the tube and endplug. The electrical connection between endplug and tube wall is established by the physical contact between endplug and tube. A corrugated stainless-steel ``spring'' between the endplug and tube enhances their electrical contact. The endplugs must be cleaned and assembled before tube manufacture. Endplugs and twisters are visually inspected for chips left over from machining and then given a 10 minute ultrasound bath in ethyl alcohol. Endplug assembly entails inserting the twister and retaining clip into the endplug and installing the O-ring and stainless steel spring. figure{endplug_text.jpg} Figure 2: Monitored drift tube endplug, unattached (upper), and finished monitored drift tube (lower). 3. Tube Assembly Facilities Tube assembly takes place in a class 50,000 clean room with temperature controlled to 20 +- 1 degree C. The room is entered from another room where workers change into clean room garb. The clean room contains equipment for both wiring and testing of assembled tubes. This equipment includes a complete tube assembly station, a gas leak test station, and a wire position measuring station. The tube assembly station consists of two optical tables connected end-to-end to form a single 7.6 m table as shown in Figure 3. The two tables are coupled together with a 7.6 m aluminum U-channel that ensures that temperature expansions or contractions of the tube are matched by movements of the optical tables. figure{wiring_table.jpg} Figure 3: Wiring table. Movable platform is close to the camera, and fixed platform and control computer are at far end. Two platforms on the optical table hold equipment for inserting and attaching endplugs and wire. One of these platforms is fixed (see Figure 4), and the other is movable and rides on a 2 m linear actuator, whose position is accurate to 1 micron (see Figure 5). Mounted on each platform is a 30 cm vacuum chuck which holds the tubes during wiring. Also located on the vacuum chuck is an electromagnetic coil for swaging the endplug onto the tube, and an air-driven endplug insertion fixture. A pair of precision crimping jaws are directly mounted on the endplug insertion fixture to crimp the wire into a small crimp tube. The fixed platform also supports a wire tension sensor and a 30 cm linear actuator to control the tension on the wire. The tension meter makes contact with the wire via three offset wheels. The tension on the wire exerts an upward force on the center wheel which is used to measure the wire tension. The spool of wire is held in a fixture behind the movable platform. A magnetic clutch prevents a rapid unrolling of the spool. A computer and associated electronics are mounted on one end of the table. The computer runs control software to operate the actuators and sensors during the assembly process and creates a database entry for future reference. Each assembly and test station records various parameters and measurements of tube production into the database and tracks the status of each tube. This database is described in [5]. figure{fixed_platform.jpg} Figure 4: Fixed wiring platform figure{movable_platform.jpg} Figure 5: Movable wiring platform 4. Tube Wiring Procedure Tube assembly proceeds in the following steps: o Production is initiated by scanning the bar codes of tube ID and length into the control computer. Based on the tube length, the computer will set the movable plate to the correct position by moving the long actuator. Typically, only one length of tube is made per day, so the length adjustment occurs only once per day. Raw tubes are stored in in a rack adjacent to the table. o The raw tube is then placed on the two vacuum chucks and held firmly by the vacuum. o The wire is threaded through the tube by attaching it to a plastic plug which is pulled through by applying a vacuum at the opposite end of the tube. During threading the wire is fed over rollers and care is taken not to scrape the wire on the tube edge. o The wire is then cut and threaded through both endplugs. This threading can be accomplished easily with no special tools, despite the small size of wire and hole. o The endplugs are then placed into precision holders on the vacuum chucks. The endplugs have two threaded holes for establishing ground connection. These holes must be oriented correctly to allow installation of electronics later on. A spring-loaded pin on the endplug holder is inserted into one of these holes to set the endplug orientation to an accuracy of 0.1 radian. The endplugs are held in place by pressure-backed O-rings (100 psi) in the holders. o The wire is attached to a clamp at the movable platform. At the fixed platform the wire is threaded through the wire tension meter and attached to the linear actuator. The wire is then pulled by hand to an initial tension of about 200 g. The wire is not crimped yet. o The endplug holders and swaging coils sit on pneumatically operated rails. The rails have been machined to be collinear with the tube vacuum chuck to insure that the endplug is inserted into the tube parallel to the tube axis. To maintain an accurate wire position, the endplug and tube axes must be collinear. The endplugs are are now inserted into the tube ends by activating the pneumatic cylinders. o Next the electromagnetic swager is activated causing the tube end to collapse around the endplug providing a rigid and gas-tight connection. The swager uses a seven-turn coil of 2 mm square cross-section copper through which a 10 microFarad capacitor charged at 9000V is discharged. The coils and swaging trigger box were provided by the University of Washington ATLAS group. o The wire is crimped to the endplug on the movable platform. The wire crimp tube protrudes from the back of the endplug holder and into the jaws of the wire crimper. The crimping is accomplished by a pair of vice jaws which have been mounted on the outside of the endplug holder. The crimper is operated by turning a handle with a cam controlling the crimp thickness of the jaws. Crimp thickness must be carefully controlled: too small and the wire breaks; too large and the wire slips. In our setup, the 1.2 mm diameter crimp tube is crimped to a thickness of 0.68 mm. o The wire is tensioned by the linear actuator with feedback from the wire tension meter. The wire is pretensioned at 450 g for 15 sec and the tension is relaxed to 350 g. The pretensioning stretches the wire and minimizes later de-tensioning. After tensioning, the wire is crimped at the fixed platform end. o Finally, the tube is released from the endplug holders and vacuum chucks. Signal caps are screwed onto the endplugs and a red plastic cover placed over the end to protect the signal cap. The cover also protects workers from the pointed signal caps. The whole tube assembly process is done by two workers in about 6 minutes per tube. Allowing for daily calibration, setup and breaks, we make about 50-60 tubes per day in full production. The quality of the wiring has been checked using test stations as described in the following sections. 5. Tube Quality Control The tubes must meet the specifications which are summarized in table 1. This table also indicates the measurement techniques used to test these specifications. The tube quality specifications are exacting, in particular, the position specification of 25 micron. In addition, there are tight specifications on tube leak rate, dark current, and wire tension. The tube quality control requires three additional workers. Each test is done at a specified ``station'' with computer and testing hardware. Each station records data for the computer database and for tracking each tube status. Tube are transported on rolling racks to the different rooms where the tests are performed. If a tube fails any of the tests, it is immediately pulled from the stream of good tubes and placed in a special rack. The ``bad'' tube is also marked with black end covers (replacing the red) to designate its status. The tube testing takes place shortly after tube manufacturing so that any manufacturing errors are quickly detected and fixed. The following sections discuss each of the test methods. Item Specification Test Method ------------------------------------------------------------------------------ Tube Length +- 100 micron Linear encoder Wire tension 350 +- 15 g Measure fundamental vibration mode Leak rate < 10^{-8} bar-l/s Helium leak detector with fixture Wire position <25 micron off-center Electromagnetic micrometer Dark current < 2 nA/m Precision ammeter Wire tension final check 350 +- 15 g & Tchange < 5 g & Measure fundamental vibration mode. ------------------------------------------------------------------------------ Table 1: Quality control specifications for MDT tubes 5.1 Wire Tension and Tube Length Measurement The tube length and wire tension measurements are performed by the two tube assembly workers in a 7.6 m long aluminum V-channel located on the side of the tube assembly table. One end of the tube is pushed against a stop and the position of the other end is read out by a linear encoder, accurate to 10 micron, to determine the length of the tube with endplugs. The linear encoder is read out via computer. For the wire tension measurement a wire clip is attached to one endplug. An oscillating current with a frequency near the expected resonant frequency of the tube is fed into the wire via the clip. The oscillating current on the wire induces wire vibrations due to the the presence of a U-shaped magnet at the center of the V-channel. The induced signal is read out via an amplitude-to-digital converter and Fourier analyzed to determine the resonant frequency, and hence tension, of the wire. figure{wire_tension_meas.jpg} Figure 6: V-channel for tube length and wire tension measurement. The tube is then placed in the outgoing bin of the assembly table. Another worker takes the tube for leak testing, wire position measurement, and dark current testing. 5.2 Tube leak testing The tube is placed on a leak-testing rack (see Figure 7) where it is pressurized to three atmospheres with helium and sealed. Two different fixtures are used to test for leaks. One fixture fits on the tube ends to check for leaks at the joint where the endplug is swaged onto the tube (Figure 8). Inside the fixtures holding the tube ends are two pressure-backed O-rings which seal off a short section of tube on either side of the endplug joint. This fixture is connected to a helium mass spectrometer and evacuated to determine if helium leaks out of the endplug O-ring. The fixture can also detect leaks through cracks in the the Noryl plastic of the endplug. A separate device is used to search for leaks over the length of the tube (Figure 9). This fixture consists of two aluminum halves which fit around the tube. In one half there is a small hole through which nitrogen flows to purge air (which contains small amounts of background helium) from the space around the fixture. The other half contains a small hole which is attached to the "sniffer" of a helium leak detector. This fixture is slowly moved over the length of the tube via a motorized track to check for leaks over the entire tube length. figure{leak_detector.jpg} Figure 7: Tube leak detection station figure{leak_endplug.jpg} Figure 8: Fixture for endplug-tube leak detection. figure{leaktest_tube_body.jpg} Figure 9: Fixture for tube body leak detection. 5.3 Wire Position Measurement Wire position is measured with a device called an Electromagnetic Micrometer, or EMMI ([6], see Figure 10). This device measures the position of the wire relative the the precision surface of the endplug. The endplug precision surface is placed onto precision spheres to hold the tube. At each end the tube is placed between two coils. A small AC current is injected into the tube wire which induces a signal on the adjacent coils. The differential signal from the two coils is sensitive to the wire position. Measurements of the differential signal from the 2 coils are made at 90 degree rotations which are combined to determined the wire offset. Initially, all tubes were tested for wire position. However, it was found that very few tubes failed the test, so the testing is now only done on 10% of the tubes. figure{emmi_stand.jpg} Figure 10: EMMI (wire position) test stand. The EMMI is mounted on the same cart as the leak test station to simplify laboratory layout. 5.4 Dark Current Testing Tubes are taken to another room and put in a rack to test the leakage current (see Figure 11). The rack has space for 48 tubes to allow overnight testing of all tubes produced during the day. The tubes are placed in the rack and filled with Atlas MDT gas (Ar/CO2) at 3 atmospheres. The tubes are connected to a HV supply set at 3400 V, and tube currents are read out via scanning ADCs attached to a 10 MegaOhm resistor between the wire and ground. The estimated dark current due to cosmic rays is ~40 picoamperes - much less than the nanoampere level thresholds used for tube rejection. Tubes which have a high initial current (> 20 nA) are treated with reverse high voltage (HV). For these tubes a reverse polarity voltage is slowly increased until the current is a few hundred microAmpere. This voltage is maintained for ~2 minutes and then turned off. This procedure fixes the vast majority of high-current tubes. After the reverse HV treatment, all tubes are ramped back up to 3400V and monitored overnight. Tubes may have initially high currents that diminish over a long duration ``burn in'' (hours or even days). Such a conditioning period during which the tubes are monitored serves not only as a diagnostic but may be considered part of the tube production regimen. figure{dark_current.jpg} Figure 11: Dark current testing stand 5.5 Follow-up wire tension measurement The final tube quality-control check is a second measurement of the wire tension a week after tube construction. The measurement tests for wire slippage or possible breakage (though breakage has never occurred). We have made a separate station for this test shown in Figure 12. The tension is measured to ensure that it is still within specifications, and to check that that the tension has not changed by more than 5 g since tube wiring. If the wire tension has been found to have changed, the tube is put aside for another week, after which another tension measurement is made to see if the tension has stabilized. If the tension is now stable and within specifications, the tube is passed. Also at this test station, a check is made with the tube database to confirm that a tube has passed all previous quality-control tests. The tubes are then available for use in chamber gluing. figure{retension_station.jpg} Figure 12: Wire tension testing stand 5.6 Tube Components Quality Control We have developed a number of devices and procedures for checking endplugs and twisters. These checks were made chiefly at the start of production to insure that parts met specifications. Recently (December 2001), we found problems with twisters leading to errors in wire tension. In response to this we conducted more tests on twisters. 5.6.1 Endplug Tests At production onset we made measurements of endplugs to ensure they met specifications. We developed two techniques to measure the precision surfaces. One of these consists of go/no-go gauges to check that the outer precision surface is within tolerances (see Figure 13). We also developed a test to measure both the inner 5mm hole (in which the twister is inserted), and the outer precision surface. In this test one end of a 5mm rod is inserted into the inner hole of the endplug and the other end is placed on a V-block (see Figure 14). The endplug is then rotated and a micrometer measures the position of the outer precision surface of the endplug. This provides a precise measurement of the difference in the inner and outer radii. Figure 15 shows the results of this measurement for a sample of endplugs. After testing an initial batch and finding no problems, we decided that further testing was not required. figure{endplug_nogo.jpg} Figure 13: Go/No-go gauges for testing endplug precision surfaces. figure{endplug_dia.jpg} Figure 14: Setup for measuring endplug precision surface. figure{endplug_check.jpg} Figure 15: Results of endplug precision surface micrometer measurement. figure{twister_survey.jpg} Figure 16: Results of twister hole concentricity measurement. 5.6.2 Twister Tests We have devised several tests of the twister outer and inner diameters. In one test, we mount a twister in an endplug, and mount the endplug in a holder where it rests on its precision surface. The endplug is back-illuminated and the twister hole is imaged via a lens onto a CCD (see Figure 17). A computer program analyses the image and finds the center and size of the twister hole image. The endplug is then rotated through 360 degree, while taking several images, defining a circle about the endplug center position. The radius of this circle is the offset of the twister hole from center. We tested 571 twisters in this manner and found that they were within specifications as shown in Figure 16. Another test we developed for checking the twister hole size is to thread various diameter wires through a twister and seeing whether the twister moves freely on the tensioned wire. figure{ep_survey1.jpg} Figure 17: Setup for measuring twister center hole concentricity. 6. Tube Production Statistics Michigan MDT production started in October, 2000, and is now half complete. About 16,000 tubes have been wired and tested and glued into 40 chambers. Figure 18 shows the Michigan tube and chamber production rates since production onset. After an initial ramp-up period production rates stabilized at 200-250 tubes per week and 3 chambers per month. figure{tube_prod.jpg} Figure 18: Tube and chamber production rates. Production was slower at first as we debugged our system. Production stopped in June, 2001 to change tooling from the EMS5 to EMS4 chamber type, and in January 2002 to change from EMS4 to EML3. figure{tube_fail.jpg} Figure 19: Upper plot: Tube failure rates by QA test type (wiring = failure in tube wiring; Len = bad tube length; Tens. = bad wire tension; Leak = bad leak rate; Pos. = bad wire position; DC = bad dark current; Misc = other failures. Lower plot: Total tube failure rate by month. figure{tube_qa.jpg} Figure 20: Plots of QA measurements on MDT tubes. Arrows indicate tolerance limits. In addition to meeting our production rate goals, we have been able to maintain very high quality standards. Figure 20 shows the results from tube QC measurements demonstrating the very high quality of MDT tubes produced at Michigan. Our overall tube failure rate is 2.0%. Figure 19 shows failure rate by QC test type, and total failure rate over time. Tube failure rates have fluctuated over time. Most failures are due to variations in the quality of raw tube components. For example, a recent rise in wire tension errors is caused by faulty twisters. Some of the twisters have undersized holes or burrs which may snag the wire and cause wire tension errors. On other occasions raw tubes have not been properly cleaned, leading to large dark currents due to tube contaminants. In some cases, equipment malfunctions have led to tube manufacturing errors. For example, the electromagnetic swager for attaching endplugs to tubes has had several difficulties, occasionally resulting in endplugs not being properly swaged and leading to tube leaks. Bibliography [1] ATLAS Muon Spectrometer Technical Design Report, CERN/lhcc/97-22, 1997 [2] Michigan ATLAS MDT Chamber Mass Production, E. Diehl et al, ATLAS Internal Note ATL-MUON-2002_006, 2002 [3] The First Precision Drift Tube Chambers for the ATLAS Muon Spectrometer (F. Bauer et al), Nucl. Instr. and Meth. A478 153-157 (2002) [4] ATLAS Muon System, Tube Assembly Manual, Henry Lubatti, MDT Production Readiness Review, 17/18 June 1999, (available at on the web at http://hpfrs6.physik.uni-freiburg.de/~herten/prr/prr32_tube_assembly.pdf) [5] Michigan ATLAS Monitored Drift Chamber Production Database, H.A.Neal, S. McKee, H. Chunhui, ATLAS Internal Note ATL-MUON-2000-012, 2000 [6] An electromagnetic micrometer to measure wire location inside an ATLAS MDT drift tube, Cambiaghi, M., Cardini, A., Ferrari, R., Gaudio, G., ATLAS Internal Note ATL-MUON-1998_259, 1998