1. Comments from reviewers and updated plans G. Ambrosio and P. Ferracin October, 21 2013

2. General comments/requests SC – My opinion regarding end-parts criteria is that if the conductor looks good the spacers are good enough. – Having said that we should ensure that: The first turn wraps around a spacer without a gap No gap between the last turn and the following spacer Following impregnation use an internal inspection to ensure turns linearity, no or little S curves. FN – With coil winding at CERN and LARP, what is the selection criteria for a particular design considering all the various parameters like cable tension, cable type (Cu, Nb 3 Sn), cable insulation (tape, sleeve), end part configuration (slits, no slits, flex legs) etc.? No evaluation criteria was presented. PF – Establish a check list for the selection and evaluation criteria on – Establish common procedures from the beginning. Make clear to the technicians that they are not alone but another guy on the other side of the ocean is doing the same thing of him and the results shall be the same. If during agreed test it looks like a different option/solution is better, before you complete the assigned test and then you do the other one in addition. This will allow everyone to move forward together. 21/10/2013 G. Ambrosio and P. Ferracin2

3. Different codes BA – Starting on two alternative designs is useful if there is an uncertainty as to which design criteria are more important. The direct comparison brings clarity. In the MQXF project there is no such difference in criteria. Therefore, there is no clear necessity to pursue multiple options. Recommend to go with one design quickly through the necessary iterations that will yield a good enough end design. – Which tool is used is secondary. Susana showed conclusively that both, ROXIE and BEND, can yield identical designs with standard input parameters. That Susana's version is more pointed has to do with the fact that she did not use the order of the hyper-ellipse as a free parameter. – If BEND is used, either the interface to ROXIE should be revived, or the ends are reproduced with ROXIE. Both options are practicable. The first is beneficial also for other projects. The latter does not seem too time consuming either. FN – There is a lot of effort to manipulate Roxie output to look like the BEND end part design, like the winding key. Most of it is straight forward, but not the torsion parameters TORS1-TORS4. These parameters are used to insure that the surface vectors are at least close to the rulings of the BEND design. As a result, I recommend using BEND as the primary tool for end part design. Investigate the program written by GianLuca to input BEND files into Roxie for magnetic analysis and end part design evaluation. SC – Be reasonable with whatever computer program is used, it is based on a mathematical concept (zero strain) and does not take into account material properties (e.g. no stress analysis). Personally I would use BEND. PF – Use the program that you like the most, avoid calling one the BEND design and the other the ROXIE design. The key is the person using the tool, her/his understanding of the problem, what she/he wants to do. For me they are equivalent. Results matter. Nevertheless (see below) – When a design has been agreed the source file for the production and the drawings shall come only from one place and they should be stored on a common web site 21/10/2013 G. Ambrosio and P. Ferracin3

4. Rapid prototype parts BA – Where iteration is needed to close gaps, in the past we made good experience with central-post-type fast- prototyping pieces like those in the attached image that can easily be produced for any block. Test-winding without insulation gave good insight in the actual dynamics of the cable. FN – In general, make 3D printed parts with various final edge angles (nose angle) and do single turn winding tests of problematic current blocks. – It was mentioned that end parts were not inspected prior to coil winding. Recommend using QC inspected end parts prior to coil winding. – As was recommended recently in the 11 Tesla Dipole Readiness Review, do the necessary analysis and mechanical testing of laser sintered parts as part of the qualification process for the material and the laser sinter process. This could be shared/coordinated effort between the 2 projects. SC – Use rapid prototype parts as a variable. For a given choice of A-length and angle, vary the angle by +- 3-5 degrees and try several such parts during winding. The techs should be able and provide useful feedback. – Fix each part as you become aware of it, we cannot afford the time or justify the performance by trying to reduce problems one at the time. PF – Complete qualification of the metallic rapid proto material but define before the criteria and do not make them unnecessarily demanding – In order to have stable winding condition for a series of coils differences in +/- 0.1 0.15 mm on the curve part of the spacer in contact of the cable shall not affect the final results. On the other hand material thickness in prolongation to the wedges shall be in +/-0.01 0.02 range. This dimension can be easily measured by hand and will provide indication of the quality of the pieces. Personally I would do 3D checking to understand the production capability of the rapid proto, but in series this shall be a statistical check 21/10/2013 G. Ambrosio and P. Ferracin4

5. Short/long legs BA – It is reasonable to aim for short spacer "legs". A criterion is needed to define where spacers can be cut. FN – Recommend always making end part spacer legs that transition to a wedge to be as short as possible. SC – The straight-section wedges should be staggered as they are terminated, e.g. the wedge length near the pole should be shorter than the one close to the mid-plane. Avoid including a wedge-like straight section in the spacer. PF – Stagger the transition wedge –end spacer – Straight part could be included in the spacer wedge during the design and initial proto phase. Then preparing the drawings for real coil, shift them to the straight part wedges. This will also reduce costs 21/10/2013 G. Ambrosio and P. Ferracin5

6. BEND/ROXIE parameters BA – The final integrated field-quality optimization should be done once the applicable cable parameters (FAT and KEY in BEND, virtual-wedge dimensions in ROXIE) are known from winding tests. Note that variable parameters per block might give better results than a single setting for the entire end. – Problems in winding can usually be correlated to specific features in the distributions of geodesic curvature (hard-way bend), normal curvature (easy-way bend), and torsion. This offers an efficient and insightful way to improve a design. (For example, MQXC blocks that exhibited gaps had a maximum normal curvature away from the apex; 11-T blocks that gave cable stability problems had elevated overall torsion.) SC – The experience with BEND has been to reduce the “nose” angle even if strain appears. Reduce that angle but stay above the point where errors or cable reversal appears. Vary the A-length (or shift or blunt) to try and not go much above 15 degrees. – The FAT and KEY variables are empirical. I have used them to “cover up” uncertainties especially when a new cable is used for the first time. 21/10/2013 G. Ambrosio and P. Ferracin6

7. Flexible spacers BA – As for accordion slits in the selective laser-sintering process, we have made the experience that slits in conjunction with thermal stresses inherent to the process can lead to distorted geometries. Careful quality control is needed. FN – Using end parts that don’t need to filed or modified during coil winding is a must for coil mass production. Recommend testing the effectiveness of the flexibilize/accordion style end part design in a magnet as soon as possible. – Investigate the use of a flexibilize end saddle as tooling for coil winding to help protect cable insulation and replace after curing with a monolithic end saddle. SC – Flexing end-parts will make the winding process easier but may have a negative impact on the magnet performance exactly because they are flexible. As suggested they could be used during winding and later replaced after curing. Their use in the magnet itself should remain an R&D item. PF – Piece shall not be modified in production. This is for time effectiveness, reproducibility but also avoiding risks of electrical defects. Flexible part could be a tool, but here I have some concern on the possibility that they could harm the electrical insulation to be checked. – Investigate the use of a flexible end saddle as tooling for coil winding to help protect cable insulation and replace after curing with a monolithic end saddle. – Flexible end part shall be tested in HQ or LHQ before use 21/10/2013 G. Ambrosio and P. Ferracin7

8. Measurements of turn positions BA – The z-position and angle of each turn should be measured. No large changes are expected after curing. FN – Recommend measuring each turn angle and longitudinal position during coil winding. PF – Establish a unique way to control and measure turn by turn cable position and angle. This shall be analyzed to see what is happening and feedback in next generation. If reasonably well done is powerful. In addition here we are facing what it will be a multi-site production chain. Having comparable quantities to match is key to understand what is happening on each production site and understand if agreed changes produce the same effect on the product. It is the start of a common QA and QC plan 21/10/2013 G. Ambrosio and P. Ferracin8

9. Winding tension FN – Recommend back winding test coil and repeating winding test with 25 Kg winding tension. Measure each turn angle and longitudinal position. Consider making another 3D printed LE part(s) with a steeper angle and wind a single turn to evaluate how the cable responds (Block 1.b, see page 10 of Miao’s talk). SC – To be consistent maintain a cable tension that minimizes strands popping. In my experience it will have a small affect on the shape of the spacers. PF – If it would be possible back wind the coil and rewind with 25 kg, change the angle more vertical has probably to be done immediately. Nevertheless take care not to change too many parameters at the same time or the picture will get foggy (I would change only one per time). Use the time effectiveness of rapid proto to do this. Some years ago we could not do it, now you can. In key position multiply the rapid proto pieces to test different options 21/10/2013 G. Ambrosio and P. Ferracin9

10. Block positioning BA – In the final FQ iteration, avoid to have blocks in first and second layers terminate at the same z-position. SC – Place the last turn next to the shoe of layer 2 on top of the inner layer shoe. Avoid the last turns of both layers from overlapping. PF – If possible avoid superposition of the end(or beginning) of the last outer layer block and the last inner layer block. But do not look only to the section. Look at the developed heads to see if they cross/ align somewhere else 21/10/2013 G. Ambrosio and P. Ferracin10

11. Updated status and possible plan 1.28/10-01/11, fabricate short plastic poles (BEND poles), 1 set 2.28/10-01/11, Wind/cure outer layer short Cu coil 3.28/10-01/11, Update BEND inner/outer layer spacers after short winding test 4.04-08/11, Update ROXIE inner/outer layer spacers 5.11-15/11 fabrication inner/outer layer BEND plastic spacers (v3) 2 sets 6.18-22/11 fabrication inner/outer layer ROXIE plastic spacers (v3) 2 sets 7.18/11-6/12, CERN/LARP to wind 2 short coils (ROXIE/BEND) with plastic parts 1.Cu cable fabricated by 8/11, insulated by 22/11 8.9-13/12 Update ROXIE and BEND models 9.04-08/11 Start price inquiry and procurement of SQXF metal poles, 3 sets 1.LARP will procure 6 sets 10.11-15/11 Launch fabrication for IL-OL RE metal spacers (after receiving first 3 by) 1.Less than 10 kCHF 2.We need to define the cuts of the flexible spacers, the tolerances, the rounding of the corners 11.11-15/11 launch the price inquiry for 6 sets of metal end spacers 12.16-20/12 Launch fabrication of metal spacers for 6 coils (4 practice and 2 mirror) 21/10/2013 G. Ambrosio and P. Ferracin11

12. Additional slides 21/10/2013 G. Ambrosio and P. Ferracin12

13. From Bernhard Starting on two alternative designs is useful if there is an uncertainty as to which design criteria are more important. The direct comparison brings clarity. In the MQXF project there is no such difference in criteria. Therefore, there is no clear necessity to pursue multiple options. Recommend to go with one design quickly through the necessary iterations that will yield a good enough end design. Which tool is used is secondary. Susana showed conclusively that both, ROXIE and BEND, can yield identical designs with standard input parameters. That Susana's version is more pointed has to do with the fact that she did not use the order of the hyper-ellipse as a free parameter. If BEND is used, either the interface to ROXIE should be revived, or the ends are reproduced with ROXIE. Both options are practicable. The first is beneficial also for other projects. The latter does not seem too time consuming either. Where iteration is needed to close gaps, in the past we made good experience with central-post-type fast-prototyping pieces like those in the attached image that can easily be produced for any block. Test-winding without insulation gave good insight in the actual dynamics of the cable. Problems in winding can usually be correlated to specific features in the distributions of geodesic curvature (hard-way bend), normal curvature (easy-way bend), and torsion. This offers an efficient and insightful way to improve a design. (For example, MQXC blocks that exhibited gaps had a maximum normal curvature away from the apex; 11-T blocks that gave cable stability problems had elevated overall torsion.) The final integrated field-quality optimization should be done once the applicable cable parameters (FAT and KEY in BEND, virtual-wedge dimensions in ROXIE) are known from winding tests. Note that variable parameters per block might give better results than a single setting for the entire end. The z-position and angle of each turn should be measured. No large changes are expected after curing. In the final FQ iteration, avoid to have blocks in first and second layers terminate at the same z-position. It is reasonable to aim for short spacer "legs". A criterion is needed to define where spacers can be cut. As for accordion slits in the selective laser-sintering process, we have made the experience that slits in conjunction with thermal stresses inherent to the process can lead to distorted geometries. Careful quality control is needed. 21/10/2013 G. Ambrosio and P. Ferracin13

14. From Fred There is a lot of effort to manipulate Roxie output to look like the BEND end part design, like the winding key. Most of it is straight forward, but not the torsion parameters TORS1-TORS4. These parameters are used to insure that the surface vectors are at least close to the rulings of the BEND design. As a result, I recommend using BEND as the primary tool for end part design. Investigate the program written by GianLuca to input BEND files into Roxie for magnetic analysis and end part design evaluation. With coil winding at CERN and LARP, what is the selection criteria for a particular design considering all the various parameters like cable tension, cable type (Cu, Nb 3 Sn), cable insulation (tape, sleeve), end part configuration (slits, no slits, flex legs) etc.? No evaluation criteria was presented. Recommend measuring each turn angle and longitudinal position during coil winding. Recommend back winding test coil and repeating winding test with 25 Kg winding tension. Measure each turn angle and longitudinal position. Consider making another 3D printed LE part(s) with a steeper angle and wind a single turn to evaluate how the cable responds (Block 1.b, see page 10 of Miao’s talk). In general, make 3D printed parts with various final edge angles (nose angle) and do single turn winding tests of problematic current blocks. It was mentioned that end parts were not inspected prior to coil winding. Recommend using QC inspected end parts prior to coil winding. Recommend always making end part spacer legs that transition to a wedge to be as short as possible. As was recommended recently in the 11 Tesla Dipole Readiness Review, do the necessary analysis and mechanical testing of laser sintered parts as part of the qualification process for the material and the laser sinter process. This could be shared/coordinated effort between the 2 projects. Using end parts that don’t need to filed or modified during coil winding is a must for coil mass production. Recommend testing the effectiveness of the flexibilize/accordion style end part design in a magnet as soon as possible. Investigate the use of a flexibilize end saddle as tooling for coil winding to help protect cable insulation and replace after curing with a monolithic end saddle. 21/10/2013 G. Ambrosio and P. Ferracin14

15. From Shlomo My opinion regarding end-parts criteria is that if the conductor looks good the spacers are good enough. Having said that we should ensure that: a) The first turn wraps around a spacer without a gap b) No gap between the last turn and the following spacer c) Following impregnation use an internal inspection to ensure turns linearity, no or little S curves. Recommendations: 1) Be reasonable with whatever computer program is used, it is based on a mathematical concept (zero strain) and does not take into account material properties (e.g. no stress analysis). Personally I would use BEND. 2) The experience with BEND has been to reduce the “nose” angle even if strain appears. Reduce that angle but stay above the point where errors or cable reversal appears. Vary the A-length (or shift or blunt) to try and not go much above 15 degrees. 3) The FAT and KEY variables are empirical. I have used them to “cover up” uncertainties especially when a new cable is used for the first time. 4) Use rapid prototype parts as a variable. For a given choice of A-length and angle, vary the angle by +- 3-5 degrees and try several such parts during winding. The techs should be able and provide useful feedback. 5) Fix each part as you become aware of it, we cannot afford the time or justify the performance by trying to reduce problems one at the time. 6) To be consistent maintain a cable tension that minimizes strands popping. In my experience it will have a small affect on the shape of the spacers. 7) Place the last turn next to the shoe of layer 2 on top of the inner layer shoe. Avoid the last turns of both layers from overlapping. 8) Flexing end-parts will make the winding process easier but may have a negative impact on the magnet performance exactly because they are flexible. As suggested they could be used during winding and later replaced after curing. Their use in the magnet itself should remain an R&D item. 9) The straight-section wedges should be staggered as they are terminated, e.g. the wedge length near the pole should be shorter than the one close to the mid-plane. Avoid including a wedge-like straight section in the spacer. 21/10/2013 G. Ambrosio and P. Ferracin15

16. From Paolo Establish a check list for the selection and evaluation criteria on Establish common procedures from the beginning. Make clear to the technicians that they are not alone but another guy on the other side of the ocean is doing the same thing of him and the results shall be the same. If during agreed test it looks like a different option/solution is better, before you complete the assigned test and then you do the other one in addition. This will allow everyone to move forward together. Establish a unique way to control and measure turn by turn cable position and angle. This shall be analyzed to see what is happening and feedback in next generation. If reasonably well done is powerful. In addition here we are facing what it will be a multi-site production chain. Having comparable quantities to match is key to understand what is happening on each production site and understand if agreed changes produce the same effect on the product. It is the start of a common QA and QC plan If it would be possible back wind the coil and rewind with 25 kg, change the angle more vertical has probably to be done immediately. Nevertheless take care not to change too many parameters at the same time or the picture will get foggy (I would change only one per time). Use the time effectiveness of rapid proto to do this. Some years ago we could not do it, now you can. In key position multiply the rapid proto pieces to test different options In order to have stable winding condition for a series of coils differences in +/- 0.1 0.15 mm on the curve part of the spacer in contact of the cable shall not affect the final results. On the other hand material thickness in prolongation to the wedges shall be in +/-0.01 0.02 range. This dimension can be easily measured by hand and will provide indication of the quality of the pieces. Personally I would do 3D checking to understand the production capability of the rapid proto, but in series this shall be a statistical check Stagger the transition wedge –end spacer Complete qualification of the metallic rapid proto material but define before the criteria and do not make them unnecessarily demanding Piece shall not be modified in production. This is for time effectiveness, reproducibility but also avoiding risks of electrical defects. Flexible part could be a tool, but here I have some concern on the possibility that they could harm the electrical insulation to be checked. Investigate the use of a flexible end saddle as tooling for coil winding to help protect cable insulation and replace after curing with a monolithic end saddle. Flexible end part shall be tested in HQ or LHQ before use Straight part could be included in the spacer wedge during the design and initial proto phase. Then preparing the drawings for real coil, shift them to the straight part wedges. This will also reduce costs Use the program that you like the most, avoid calling one the BEND design and the other the ROXIE design. The key is the person using the tool, her/his understanding of the problem, what she/he wants to do. For me they are equivalent. Results matter. Nevertheless (see below) When a design has been agreed the source file for the production and the drawings shall come only from one place and they should be stored on a common web site If possible avoid superposition of the end(or beginning) of the last outer layer block and the last inner layer block. But do not look only to the section. Look at the developed heads to see if they cross/ align somewhere else 21/10/2013 G. Ambrosio and P. Ferracin16