Feedthrough Meeting 18/12/98 present: P. Birney, A. Dowling, M. Fincke-Keeler, R. Keeler T. Hodges, M. Lefebvre, P. Poffenberger, M. Rensing, G. Vowles - Opening Comments Many thanks to all for your help in setting up the ATLAS Celebration. Judging from the many comments I heard, I believe it was a success! - Welding Status Rowland Roper performed welds in Victoria on Fri Dec 4th. After the welding of blank pin carriers in a blank flange, Glasseal pin carriers were welded in an ambient flange. No leaks were detected. Glasseal pin carriers were also welded in a cold flange. No leaks were detected. A cold box assembly was welded together, then welded to the cold flange. Two pigtails have been installed on the assembly. Rowland was happy with the welder and the welding environment. Paul to communicate the welding parameters to Terry, so that we have them for our meeting in January. Paul commented on the deflection measurements. Extensive measurements were made. In summary, the deflection on the blank flange (thinner than the real ambient flange) was about 250 microns. The deflection on the ambient flange was about 60 microns. The deflection on the cold flange was about 50 microns, which became about 100 microns after the welding of the cold box. Terry's FEA note stipulates that flange deflections above 200 microns should be avoided. Thus the measured deflections appear to be acceptable. - Status of Blank and Flange Annealing Paul reported that one blank ambient flange set, as well as one ambient flange were sent for annealing. They are expected in a few days. Upon reception, deflections will be measured and compared to the values before annealing. - Pin Carriers and Temperature Cycling Status Only three more temperature cycling are left to be done on the last 5 Glasseal pin carriers. - Status of Leak Tester The sensitivity of the He leak tester seems to be decreasing. Attempts have been made to try to compensate for this, but without success. After discussion with the tech rep, it appears that the head may need cleaning. We agreed to have the He leak tester sent back for repair as soon as possible. Meanwhile, the RGA commissioning is under way. The goal before Xmas is to be able to use the RGA for the leak testing. calibration can be made using our 3 calibrated leaks. - Bellows We agreed to send back two of the three bellows as soon as possible. Special jigs for the leak testing of one of the bellows require more work due to the shape of the bellows. - Plans for Assembly We discussed a possible scenario for our plans in the near future. Assuming that the leak tester is operational, and that pin carriers have been re-checked: - warm and cold test of the cold flange and cold box assembly - installation of temperature probe on one of the vacuum cables - welding of bellows - warm and cold test of the whole assembly - annealed blank welding, measurement of deflections - welding of Glasseal pin carriers in annealed ambient flange And assuming that the annealing has no detectable effects: - welding of 8 PCT pin carriers in an ambient and cold flange. - assembly of two cold boxes Before the next step, we need to think about which pigtails to install: - welding of cold boxes on two cold flanges Paul to contact Rowland about the welding schedule. - The Schinzel X-Files Dieter Schinzel has communicated his comments to us by fax. This will be further discussed in January. Terry summarized the issues in three points: 1- Steel Quality The point is made that we should attempt to use the best possible steel for the pin carriers. So-called vacuum-remelt 304L should be used. Some is available at CERN. The cost increase for the use of such steel is negligible, compared with the overall pin carriers cost. 2- Leak Scenarios Dieter argues that we should be able to detect whether a leak is from the argon side or from the air side. He argues that to be able to detect this, virtual leaks inside the bellows need to be avoided. This has impacts on many welds of the assembly. Terry and Roy have commented on this, and we feel we have enough counter arguments not to follow Dieter's proposal. 3- Pin Carrier Design Dieter proposes to modify the pin carrier profile. Roy and Terry have commented on this, and we feel that there are no compelling reasons to follow this proposal. - Meeting of January 4th, 5th Richard pointed out that electronics considerations should be discussed at the meeting in January. We all agreed. These should include the specifications of the vacuum cable, the specifications of the tests, and the a discussion of how the tests are done. - News from Orsay Richard and Margret had a very fruitful meeting in Orsay with Christophe De La Taille and collaborators. Richard and Margret have prepared very useful and informative detailed notes, appended below. Margret made a brief summary of some of the important points raised at the meeting. The issue of the screws and the tap holes was discussed. It appears that the virtual leaks that could be caused by the cable connector tap holes are negligible compared with the one expected from the HEC. Teflon could be used on the screws; the HEC also makes use of Teflon. The electrical measurements for the feedthrough chain were discussed. One criteria that was retained for the vacuum cables: the vacuum cables servicing the calibration lines should have the DC resistance of their channels within an acceptance band (total dispersion) of width 50 mOhm, while the remaining vacuum cables should be within 500 mOhm. We could envisage selecting vacuum cables for this purpose. The 100 vacuum cables that we have ordered will tell us what kind of dispersion we have in practice. [Note that the number of connectors reserved for calibration is - Standard EM Feedthrough: 2 for the EM - Special EM Feedthrough: 0 ? - HEC feedthrough: 4 for the EM and 2 for the HEC - FCAL Feedthrough: 1 for the FCAL so when Christophe says "2 per feedthrough" he talks about the barrel feedthroughs, which happens also to agree with the bulk of our feedthroughs.] It is not clear if such selection would need to be done for the pigtails. Margret pointed out that the pigtails should now all be "calibration worthy" in their specifications. The calibration cables must maintain the signal shape integrity and reduce the cross-talk to a minimum. A detailed testing scenario has been discussed. It allows the simultaneous measurement of the cross-talk, the amplitude, the peaking time, the delay time. Michel pointed out that apart from the DC resistance, we have not yet defined what values of these measurements that constitute a valid cable channel. This will have to be defined. Margret reported that Richard agreed to look into the effect of the impedance mismatch (25 Ohm or 50 Ohm on 35 Ohm) on the signal. Aboud Fallou has also raised the question of what are our plans upon reception of the feedthrough at CERN. Roy to discuss with Aboud the cut length definition issues. It is not clear either if an RF gasket is required at the bottom of the aluminum skirt (chimney). - Low Voltage Vacuum Cables Six low voltage vacuum cables have been received. It is not clear yet how to test the effect of these cables on the HEC feedthroughs. This requires more thoughts. - Signal Vacuum Cable Margret reported that the flex circuit and the spring clips are expected Dec 23rd. Paul has recently made extensive transient analysis of the vacuum cables we have had in Victoria for some time. Broken traces were found on many cables. One connector has been disassembled and the broken traces have been identified. Margret has couriered a broken cable to Ernie with comments. The cables received from Strataflex should be verified for this fault. Margret to ask Ernie about the copper traces. Are they rolled- annealed or electro-deposited? Paul commented that rolled-annealed is stronger. - ATLAS Meetings and Other Important Dates - Dec 17 10:00 : ATLAS Celebration in 022 - Jan 4 ?? : PRR Preparation meeting in Victoria - Jan 11 - Jan 15 : LAr week - Jan 29 : Feedthrough Production Readiness Review - Jan 31 : ATLAS NSERC Review (to be confirmed) - Feb 12 - Feb 14 : Western Nuclear and Particle Physics Conference - Feb 14 - Feb 20 : Lake Louise Winter Institute - Feb 22 - Feb 26 : ATLAS week - Mar 15 - Mar 19 : LAr week - May 17 - May 21 : LAr week - Jun 07 - Jun 11 : ATLAS week - Jun 16 - Jul 07 : HEC testbeam - Jul ?? : LAr week - Aug 18 - Sep 08 : HEC testbeam - Mar 15 - Mar 19 : LAr week Draft notes of the meetings at ORSAY LAL Electronics Division These are draft notes. After discussion with Margret, a final version can be released to the Victoria group. December 8, 1998 9:30 AM Office Christophe de la Taille present MF-K, RKK, CdlT, (Nathalie Seguin, Walter Bonivento, Emannuel ?,Pierre Imbert: some of the time) Organized an agenda for the 2 days. 1. Testing equipment and procedure 2. Connectors 3. Mechanical aspects (? Fallou) I) Spring clip contact discussion. (PI +W?) We should avoid tin plating of spring clips. There is a commercial material called ARCAP. Pierre Imbert was looking for what it was but we did not find out. There was a question on the tolerances on the width of stainless steel slot the connectors must fit into, 0.13 mm was read off of a diagram. The ORSAY people were working on the insertion force for the cold cables and were worried about the tolerances. they wanted to know what the tolerances were on the distance between the screw holes to fasten the cables and on how parallel to each other and perpendicular to the plate the holes were drilled. Three vacuum cables tested at ORSAY as bad, one recovered after the feedthroughs were shipped to CERN. Maybe a loose slot? Comments were made by MF+CdlT that Don's vacuum cables only made ground contact through through-plated holes. This was a worry. II) It was decided to divide cable measurements into R&D first and to discuss production measurements later. A) It was further decided to divide the cables into calibration lines and ordinary signal lines. The calibration lines have VERY stringent requirements but the signal lines have quite "relaxed" tolerances. CALIBRATION LINES: The striplines have a DC resistance of 0.7 - 1.0 ohms measured by MF. The total calibration uniformity is required to be 0.25%. Therefore the tolerance allocation on individual lines is allowed to be 0.1% of the cable impedance ( 0.001*50 Ohms = 50 mOhms). (WHY?) This is the full width of the variation. There are two calibration cables per em feedthrough. They will have to be handled separately. They will require 4-wire resistance measurements. ACTION: Get a current stablized power supply and a Keithley DVM. (not very expensive). SIGNAL LINES: The signal lines should have a Dc resistance which is 2% of the termination of the preamplifiers (WHY?). This is equal to 0.02*25. Ohms= 500 mOhms. Our DC resistance cable tester has a precision of 50 mOhms. Therefore it will be more than adequate for testing the signal cables. It can be used to screen the cables looking for ones that are apparently equal to within 50 mOhms. These can go on to be tested by the 4-wire system to see if they are good enough for use as calibration cables The 100 prototype cables will make an excellent test to see if we can find a few percent of cables that are very close in resistance and that are uniform across the cable to 50mOhms. The vacuum cables have a large attenuation factor (R/(2*Z_0). MF pointed out it is 1.4 dB/metre. CdlT says this is not a problem because the cable is short. He said he assumed," a 10% dispersion on resistance." (0.10*1.0 Ohm <= 100 MOhm)(? ??) He also said the skin depth on the striplines was saturated ad therefore not a problem (WHY?). (I think I might have an idea why, but need to think it through first - mgf.) B) CdlT noted that during the cold cable PRR one reviewer was worried about all the cable impedance mismatches for the calibration. CdlT suggested we should look at this with a SPICE simulation. RKK can do this because of what he has learned in B'Ham. ACTION: Purchase ORCAD PSPICE - not expensive ~300 pounds. M. Rensing is looking into it. III) Production measurements with scanner A) Overview of the LAL test setup for cables. PULSER - Voltage Ramp (They use a Phillips - MF has model number) | action: Berkeley Nucleonics one may be ok - check MULTIPULSER FANOUT | - We have a 32 channel manual version from UBC | - Upgrade to a 64 channel switchable opamp parallel | port computer controlled version ACTION: AXEN/UBC | short warm-cable - reliable multiway cable with AXON or MF connector. | pin carrier | Vacuum cable - test item | pin carrier - can just loop the vacuum cable around so only one actual | pin carrier is needed. cold cable | INPUT SCANNER - We have the plans. ACTION: AXEN/UBC have one made. | - Many more details below. | | PULSE SHAPER - Nathalie Seguin modified some existing shapers and | has shipped them to us (hopefully with plans). | Digital Scope - Should be ok with existing scope. The above system can be used in several ways. It will measure everything we need except the characteristic impedance (or equivalently the capacitance). B) Operation One can operate the above system in 3 ways. A single channel can be pulsed and the same channel captured on the digital scope. - measure peak height == DC resistance - peak time == weakly connected to capacitance. Measures dielectric constant. A broken ground is only a 4mOhm effect and will not be seen. All the channels can be pulsed and a single channel read. - measure extra ground resistance from pulse height shift from single pulse case. Each channel pulsed and every channel read - 64*64 measurements. This gives the crosstalk. It is sensitive to broken grounds and resistive grounds. C) We ran and saved files from each of the three operations. The cross talk measurement took more than an hour and did not finish completely but we have some data. Note the cross talk measurements had 40dB between the input signal and the measured response. Signal drive is 6 volts. Transient is 500ns full scale. IV) Characteristic Impedance Z_0 There was a long discussion about how to measure the characteristic impedance for the production case. There was no solution that did not require changing the setup. The TDR would do the job but is too complicated. The propagation time measured by the pulser method is no good because the cable is shorter than the peaking time. It just looks like a capacitor. The best bet is to measure the capacitance. This can be done with out a large change to set up but requires a capacitance 4-wire metre and special probe. CdlT suggested a special connector that summs the signals and do a single capacitance measurement ACTION: Price cost of reactance/impedance meter + probe. CdlT suggested a SPICE simulation to determine the sensitivity to the Z_0. This was mentioned in an earlier action item. ==================================================================== Wednesday, December 9, 1998. V) Cross talk analysis: signal pulse cross talk pulse + . + . + . + . + . + . + . + . + . + . ========> cross talk max + . + . + . ========= + . =====> cross talk at peak + . . + . . + +. . + . + . . + . . + .+ . Peaking time is 5% above trigger time to peak time Peak voltage is 5% above trigger voltage to peak voltage rise time is 5% to 95%. The cross talk from capacitive or inductive pickup is the derivative of the signal. It is characterized by the time and voltage of the cross talk maximum and the voltage at the signal peak (should be zero in the ideal case). Cross talk can be negative if there are mutual inductances of opposite sign. Cross talk can be multilobed if there is another strong pole created by an RC or an RL. VI) Scanner board. We met with the technician (Emanuel) to discuss the scanner board design. Margret has the schematics. The board is driven by a 64 way signal connector. We saw a microD and and a very nice homemade "pincarrier" design. MF has sketches. A 6 bit binary number is written to a TTL demux that drives the on/off inhibit on 64 Comlinear CLC 410 opamps. An equivalent will work - they are used as switchable unity gain buffers. The switching is done in a 3 level tree of 4 positions at each level The demux takes the binary input and drives 3 sets of switches that drive the opamps. Each binary number corresponds to a different combination of switches being set. I hope MF has some idea how this works because I cannot reproduce it all!!!! (yes, I do - mgf). Maybe I can figure it out from the plans. All I can remember is there are 4 rows of 16 opamps. The switching turns on quite a few at once. However only one path is continuous all the way through. The tree reduces the time difference and attenuation between channels. =====> I will think about this. VII) SPICE models Christophe suggested a number of SPICE models for us to look at. VIII) Mechanics Fallou asked us a number of questions about the electrical properties of the chimney, and the electrical tests to be made at installation. MF has notes on this. (Aboud had a number of questions about how to proceed once the feedthroughs arrive at Cern - what measurements to take (mechanically and electrically), and how to accomplish the actual installation in the cryostat. He gave us a couple of drawings, one of which deals with the welding mandrel, which is supposed to facilitate the final weld of the funnel to the cryostat cold vessel. The insertion of the mandrel will be very tight. Mgf has passed on the drawings to Paul Birney and to Roy. We might have to do some mock-up tests on this after the PRR is through. - mgf)