1、MEMORANDUMpage 1 of 18DATE: Revision -01: September 20, 2005 (incorporating SUS/UK comments)Rev-00 TO:Rev-01 FROM:AL SUS Design Team:Stuart Aston, Dave Hoyland, Nick Lockerbie, Ken StrainRev-00 FROM:Rev-01 TO:AL SUS/UK Electronics PDR Review CommitteeBen Abbott, Rana Adhikari, Dennis Coyne, Peter Fr
2、itschel, Gabriela Gonzalez, Brian Lantz, Virginio SannibaleSUBJECT: Questions regarding the Advanced LIGO (AL) Suspensions (UK scope, SUS/UK) Electronics Preliminary Design Review (PDR)Refer to: LIGO-L050039-01The committees review of the SUS documentation, and the presentation on July 12th, has gen
3、erated the following questions. We would like to convene a meeting with the SUS design team to follow up on these questions in the near future and before finalizing the committees review report. The questions are organized in the following subsections:1) General Electronics Requirements2) OSEM head3
4、) OSEM emitter driver this matters, since some of the SUS eigenmodes that will be damped are below 1 Hz.Comments (from K. Strain): Input to OSEM downselect document target graph there evaluated with respect to the resulting RMS. See section 4.3 of T040110-01. The noise model is 1e-10 m/rt Hz white n
5、oise “plus” 1e-9/frequency m/rtHz (added in quadrature). This level was slightly above the worse case sensor noise at that time. Since then the sensor noise has been slightly improved. There have been no important changes to the suspension model since that was done.2.2 PTFE AlternativesAre there alt
6、ernatives to PTFE in the areas this is used? Doesnt PTFE absorb water more readily than other insulators?Preliminary answer (from D. Coyne): PTFE probably does absorb water more readily than polyimides, but the quantity use is so small compared to the LIGO water pumping capacity (and compared to oth
7、er sources of water pickup in the LIGO vacuum system), that this does not seem to be a significant concern. Comments (from S. Aston): Need to confirm that the only elevated temperatures experienced by the PTFE parts after installation, will be the non-operational in vacuum bake-out at 460C. We may w
8、ish to verify the functionality of the PTFE parts (e.g. interference fit of adjustment nuts) are not compromised by the elevated temperature. These tests can be conducted at Birmingham if deemed necessary. Questions regarding the AL SUS/UK Electronics PDR L050039-01page 3 of 182.3 Axial adjustment s
9、ystem.Range is awfully large (1 cm) - is this needed? What about the mechanical rigidity of this assembly, supported by two somewhat long-and-skinny screws?Comments (from S. Aston): The noise prototype OSEM axial adjustment range of approximately 10mm has been driven by the existing controls prototy
10、pe (Hybrid OSEM) design. The Hybrid OSEM actually allows for 10mm (approximately 15mm) of axial adjustment.It would be trivial to reduce the axial range of the noise prototype OSEM to that which is required. However, we need to know what range of operating points we are likely to encounter and there
11、fore tolerate for OSEMs situated in various suspended and non-suspended locations.The overall mechanical rigidity of the assembly is maintained by the two screw fixings mentioned, but also by the additional support provided by the fit into the coil-former clamp. The effective section thickness of th
12、e coil-former clamp has been increased in recent designs, with the addition of protruding guide features. These aid the OSEM axial adjustment and reduce the tendency of the coil-former assembly to jam/gall in the clamp. Fabrication and test of prototype units has enabled us to verify the rigidity an
13、d robustness of the adjustment assembly.2.4 OSEM Cable InsulationOSEM cable harness insulation is specified as Teflon - why not use the polyimide (higher bake temperature and cleaner) insulated wire used for the coil?Comments (from S. Aston): The Cooner (medical) wire has been previously selected, f
14、or use on the controls prototype and in initial LIGO, due to its UHV compatibility and its inherent low stiffness (multiple core construction). It can also be easily braided into twisted pairs, whilst maintaining excellent flexibility for cable harnessing and routing.2.5 Magnetic coupling of package
15、sIts stated that the magnetic coupling force from the magnetic device packaging must be less than 5 mN. How/where is this number derived? Is this the only thing driving the large magnet/device separation? (ie, long flag)Comments (from S. Aston): A maximum magnetic coupling of 5mN between the magnet
16、and sensor devices was considered to be acceptable (given that the actuator can apply 50mN). Measurements of the coupling force were made and the coil-former geometry lengthened (axially) to accommodate i.e. the magnetic coupling is the only driver for the dimensions of the coil-former and hence len
17、gth of the flag.Comments (from K. Strain): One of the few problems with the new sensor design is that only (ferro) metal-encapsulated emitters deliver on performance. We must therefore separate the magnet and the sensor to limit interactions. The effects of this were analysed both in terms of static
18、 force and unwanted coupling. If the separation is increased to the point that the static force Questions regarding the AL SUS/UK Electronics PDR L050039-01page 4 of 18is less than 10% of the 50mN adjustment range, the coupling is also negligible. (Addendum 1 contains an excerpt from an email conver
19、sation concerning this issue).2.6 Lateral to Longitudinal CouplingLIGO OSEMs design is by nature are not very degree-of-freedom (DOF) selective, possibly leading to unwanted DOF coupling to the length (longitudinal) signal. In T050111-00-K, the picture of the final OSEM assembly looks as if the orie
20、ntation of the mask and size of the flag are such that it will continue to be susceptible to lateral translation coupling into longitudinal. The imposed flag size and geometry limit design options, but some improvements are possible. The actual design shows a flag 3mm in diameter and a LED window 4.
21、5mm tall. Making the flag significantly larger than window height will reduce the coupling with the two directions orthogonal to the OSEM sensitive direction. The cylindrical shape of the flag could be changed to minimize scattered light trapping within the cylindrical cavity of the OSEM head. Have
22、any tests on the prototype OSEM head been done to determine the level of cross coupling to lateral motion of the flag? Has any testing or analysis been done to optimize the flag size or shape to eliminate this lateral motion coupling into the signal?Comments (from S. Aston): No testing to determine
23、the level of cross-coupling has taken place during the sensor design and implementation. However, a general design goal of the noise prototype was to reduce the cross-coupling effects compared to the controls prototype and initial LIGO OSEMs wherever possible. For example, the surface mount design u
24、tilized a lens integrated into the emitter package and 2mm diameter flag. Our noise prototype design incorporates an additional collimating lens and mask to select only paraxial rays emitted by the IRLED (also reduces noise). As noted, we also make use of a larger 3mm diameter flag. Measurements cou
25、ld be taken with the noise prototype design to determine the level of cross-coupling, if required. However, has the cross-coupling been previously determined for the Honeywell sensor configuration? We would anticipate that the noise prototype design presented would reduce any such effect. Experiment
26、s with various flag geometries have been carried out during the “sensor study” by our collaborator Nick Lockerbie, at the University of Strathclyde. These measurements are outlined in document T040136-00. Addendum 2, shows alternative flag geometries under consideration (due to possible sensor and f
27、lag clearance concerns that have been raised by the US/SUS team). Note that, should any of the proposed alternative flag geometries be implemented, the flag will be required to be assembled/mounted in a known orientation. At present, it is uncertain if this approach will need to be taken. Further to
28、lerance analysis should help justify the path forward. 2.7 New Vacuum MaterialsThere are some new materials that appear iffy from the vacuum standpoint: hysol epoxy; LCP - what is this? What do we think about this?Preliminary answer (from D. Coyne): There are a few materials which have recently been
29、 provisionally qualified for vacuum use which have yet to be presented to the LIGO Vacuum Questions regarding the AL SUS/UK Electronics PDR L050039-01page 5 of 18Review Board and be incorporated into the approved LIGO Vacuum Compatible Materials List, E960050-B. Hysol epoxy was qualified for use by
30、SEI for use with the ADE capacitive displacement sensors. LCP is used in the GlenAir micro-D connectors.2.8 EMI Shielded Cabling this does not appear to be the case.Comments (from K. Strain): We design all magnetic actuators to work at the maximum of the force. Actuators not at the top stage are to
31、have a rather flat field gradient and minimise coupling. T050122-00 discusses some of this in outline, but the magnetic design is not yet done in detail. Questions regarding the AL SUS/UK Electronics PDR L050039-01page 6 of 18The local control actuators are designed for larger force, but will still
32、be operated at the maximum, for the coils proposed, and 10mm long magnets, the force is flat to 1% up to more 1mm away from the operating point. The sensor range is much less than that, so operating within sensor range implies operating in the correct field range. Our rough calculations were checked
33、 against Mark Bartons Mathematica model for magnet design, but the designs are not finalised and this has not been written up in detail (part of the mechanical design). 3 OSEM sensor electronics3.1 Need block diagrams.Every subsection in T050112 that is titled Circuit Description should also contain
34、 a block diagram of the circuit, in accordance with Drawing Requirements, E030350-A, section 3.3.2. (Well actually it says that block diagrams at the module/board level are optional. However it is good practice and highly encouraged.)Comments (from D. Hoyland): Bock diagrams can be added, but I thin
35、k this should wait until we have a clear idea of scope of the diagnostic circuitry to be added. Diagnostic issues will be discussed with Jay Heefner during his visit in October, and diagrams can be created thereafter.3.2 Origin/derivation of the 8 V/rtHzWhat is the origin of the 8 V/rtHz number?Comm
36、ents (from D. Hoyland): This figure simply relates the measurement noise requirement for the sensor (m/rtHz) to the noise content of the sensor electronics output using the OSEM sensor and electronics transfer functions (at 10Hz). So the origin and derivation is as follows: The references given to t
37、he figures below are the relevant section of E050160-01 (The SUS-US/ SUS-UK ICD)The required measurement limit is 3E-10m/rtHz. (Section 1.1.2.2.6 )The required OSEM sensing range is 0.7mm pk-pk(Section 1.1.1.2.2) The required Electronics output is 20v pk-pk (10V/0V- 0V/10V) (Section 1.1.2.2.2.2)Syst
38、em gain is therefore 28571V/mThe required measurement limit at 10Hz referred to the electronics output is therefore 8uV/rtHz (actual figure 8.57uV/rtHz)3.3 Output Voltage Noise SpectrumWe need plots of the modeled voltage noise spectrums to better determine what whitening is necessary to read the se
39、nsors.Comments (from D. Hoyland): Currently our specification for noise is only at 1 frequency (10Hz) and therefore noise performance was derived by hand calculation. What do you actually want us to provide here. I could develop models in something like SPICE, however I would want Questions regardin
40、g the AL SUS/UK Electronics PDR L050039-01page 7 of 18to verify them against actual measurements, therefore I would suggest that we provide noise curves which are derived from numerical fits to the experimental data (ie the model would consist effectively of a noise function defined in terms of pole
41、s and zeros). Such issues will be discussed with Jay Heefner during his visit in October, and models will then be created as necessary.3.4 LED BalancingThe issue of balancing the optical outputs from each LED emitter can be addressed by buying a bunch of these things and hand picking enough to last
42、until eternity with a simple biasing and detector fixture - unless they are prohibitively expensive. This would result in better symmetry in terms of individual currents through diodes.Comments (from D. Hoyland): This is a fine solution if we were making 10, but the actual figure is closer to 1000.
43、This approach is far more robust and repeatable than the solution you suggest because the devises are optimized in-situ (i.e. in their final operating position & orientation in the OSEM). This would not be the case if they were screened in a test jig prior to being installed into the OSEM. The appro
44、ach should produce OSEMs whose spread of outputs over all units is better than 10% (tighter if required) so you do not sacrifice too much dynamic range on the emitter/detector tolerance. Also, the approach naturally provides some compensation of the emitter output intensity vs temperature coefficien
45、t. 4 Coil drivers (two styles)4.1 Derivation of RequirementsWe have lots of questions about the origins of many of the numbers, and problems with the way requirements have been derived. There is much reference in T050112 to G010086, which are some viewgraphs on the control hierarchy that Peter Frits
46、chel made 4 years ago. This was a fine start, but it needs to be revisited, with the current suspension model. (For example, in G010086, it looks like a different pendulum model is being used than the high Q quad transfer functions that Norna Robertson has been showing.)The first thing we need to do
47、 is to agree on and document the actuation requirements for each stage: control range vs. frequency for each stage, for both acquisition mode and science mode allowed noise spectrum for each stageThen we can come up with an input signal range and noise level for the coil drivers. The coil driver des
48、ign would then need to connect these input specs with the above range & noise specs.Comments (from K. Strain): Refer to T050122-00. The suspension model does not have a version control system in place. But I am always most careful to get the latest version from Norna, and compare. Many effects are o
49、nly subtly different and the results you point to above are not applicable to our present work (as Norna has pointed out in an email). Note too that I reviewed G010086 to see if it was still applicable, and concluded that, until we see E2E results regarding Questions regarding the AL SUS/UK Electronics PDR L050039-01page 8 of 18acquisition it is a very good basis for this planning in terms of Global control actuator strength. A few details of the suspensions lead to the minor changes mentioned in T050122-00. The uncertainty remains large and we have thus
