Summary of Drift Chamber Wire Issues K. McDonald Jan. 16, 1997 1. Aluminum wire We have received the entire order of Al wire from California Fine Wire. However, essentially all spools fail our creep criterion: that the loss of tension for a wire strung over a fixed length be less that 10% in 10 days. All other tested characteristics of the Al wire are excellent. Loss of wire tension is bad in at least 3 ways: a. The wire sage varies, so the exact wire location is uncertain, which affects position resolution. b. If the loss of tension is great enough the electrostatic instability sets in. c. As the Al wire lose tension the endplates will deflect, thereby increasing the tension on the tungsten wires (which show no creep). CFW is looking into modifications to their processing that might result in lower creep. The wire they produced around March '96 had only about 1/3 the creep of the wire they produced in Dec. '96, so it is likely the they can find a beneficial modification. So far they have concentrated on a post-anneal process: cooking the wire in an oven at 200-400 deg (F or C?) for 3-24 hours. We have received one sample, cooked for 3 hours at ~200 deg. The creep appears to be reduced by about 1/3. Several samples of variously annealed wire were reportedly sent to us yesterday, 1/15/97 The advantage of the annealing approach is that it can be done retroactively on the existing batch of wire. I am also trying to get CFW to look into modifications earlier in their processing sequence. They are dragging their feet on this a bit, as the implication would be to junk the existing batch and make a new one for us. An ambiguity will remain: Perhaps the rate of creep can be slowed. But then does the creep last longer, until the same total loss of tension occurs? There is partial evidence along these lines. The wire from March '96 with the lower creep rate went for ~150 days before the creep rate flattened out, while the June '96 wire with a higher creep rate showed a flattening of the rate around 100 days.... Only very long term studies will clarify this point. 2. Tungsten wire Several arguments were summarized in Princeton/BABAR/TDNC-96-53 (Dec. 16, 1996), available in the Princeton Technical Notes Web Page. Numerical simulations by Dave Coupal and Dave Nelson are now available in file high_ohm.ps of the Current_Meeting Web Page. Recall that we are considering the merits of 20-micron diameter gold-plated tungsten wire from Thermionc Products and from LUMA. The quantifiable mechanical properties of these two wires are nearly identical, but the surface quality of the LUMA wire is superior. The debate also includes consideration of 25-micron vs. 20-micron wire. a. Would the higher resistance of a 20-micron 3%-rhenium Luma wire be disadvantageous? [Measurements at Princeton gave 281 ohms for the LUMA wire and 187 ohms for the Thermionic Products wire.] Dave Dorfan and Dave Nelson appear to agree that the higher resistance does not lead to noticeable increase in electronic noise in the drift chamber preamp. See Figs. 1 and 2 of the note by Coupal and Nelson. However, the increased resistance does increase the signal dispersion, so the pulse height is reduced somewhat. The pulse-height reduction factor is greatest at the far end of the wire, but the smallest pulse comes from the middle of the wire because the reflection at the far end is more helpful for pulses orgination near the end. Figs. 3 and 4 of Dave Coupal's note indicate that the smallest pulse is from the middle of the wire and would be about 15% smaller still if the 20-micron LUMA wire were used (without raising the chamber voltage to compensate). This would lead to about 5% worse momentum resolution in the central region. Of course, raising the chamber voltage by about 10 volts would restore the lost pulse height and lost resolution (see Fig. 6 of TNDC-96-51 by M. McDougald). b. We could adopt a 25-micron 3%-rhenium wire from Luma, for which the resistance is measured to be the same as for a 20-micron Thermionic Products wire. In this case there is no electronic issue. However, the multiple scattering due to the wire would be increased -- a 5% increase in total radiation lengths inside the chamber volume by my calculation. Roughly, this corresponds to a loss of momentum resolution of 5% for momenta below 400 MeV/c (Fig. 6 of Coupal's note) and smaller loss at higher momenta. Such a loss would have little effect on the analysis of modes like B_0 -> J/psi K_short or B_0 -> pi pi, but would have some effect on decays with low-momentum charged particles, most notably D^* -> D pi. c. The multiple-scattering issue is related to another issue that has not been discussed sufficiently in my opinion: the choice of chamber gas. Advocates of the dE/dx measurement want a high density gas, while advocates of good momentum resolution want a low-density gas. Apparently a gas mixture was chosen a long time ago that favors the dE/dx mesaurement. [I do not consider it wise to compromise the primary mission of the drift chamber, momentum measurement, for a secondary issue, dE/dx measurement, that in effect is the primary mission of another subsystem, the DIRC.] So if a 5% change in the number of radiation lengths is really a major issue for momentum resolution, the gas mixture should be changed! Dropping from 20% isobutane to 17% isobutane will drop the number of radiation lengths by 5%. The effect of such a change (and larger changes as well) on both momentum resolution and the dE/dx measurement should be investigated. [We are making studies at Princeton that will help resolve this issue.] d. There is a debate as to whether a larger-diameter wire is more robust against breakage. There is no clearcut data or model for this. The issue is not the wire tension, but localized excess strains from the crimping process (or from kinks in the wire as it comes off the spool). Many people feel that bigger is safer, but there is not a complete consensus on this. [SLD used 25-micron 3%-rhenium Luma wire for this reason.] e. Another issue is the chamber operating voltage. For a given gas gain, a larger wire requires a higher voltage. Going from 20 to 25 micron wire would require increasing the chamber voltage by about 100 Volts (5%). [However, reducing the isobutane content of the chamber from 20% to 17% would permit a 5% drop in chamber voltage at fixed gain.] [The eventual chamber operating voltage (gas gain) depends on the noise performance of the preamp. The nominal specs of the Dorfan preamp should permit use of considerably lower chamber voltages (10-15%) than those used thus far on Prototypes I and II, which were equipped with less sensitive preamps. However, channel-to-channel variations in the preamps are likely larger than the few percent effects related to wire choice. Tests of this are desirable soon!] f. Delivery of the tungsten wire appears favorable: 7 weeks. The cost is low: $8k for 20-micron Thermionic wire $12k for 20-micron Luma wire with 3% rhenium $18k for 25-micron Luma wire with 3% rhenium (compare $130k for the Al wire from CFW.) [If samples of tungsten wire are needed in the near term for commissioning of the wire-stringing operation, delivery from Thermionic Products is very rapid.] ---------------------------------------------------------------------------- I prefer Luma wire over Thermionic Products because of the superior surface characteristics of the Luma wire (the brand used in most drift chambers). I am favorable to use of 25 micron wire rather than 20 micron wire, but do not feel as strongly about this as about the choice of Luma.