Choose timezone
Your profile timezone:
Pulse Compressor Prototype Review
→
Europe/Zurich
18/3-008 - CLIC Meeting room (CERN)
Pulse Compressor Prototype Review (10 Apr 2013)
Present: Said Atieh; Thierry Bogey; Thierry Callamand; Nuria Catalan Lasheras; Luca Dassa; Miguel Gil Costa; Raphael Leuxe; Gerard Mcmonagle; Emilien Rigutto; Igor Syratchev; Thierry Tardy; Agostino Vacca; Ben Woolley
Chaired by: Dr. Catalan Lasheras, Nuria
Igor presented (in French!) a small description of the functioning of the pulse compressor, the advantages against the SLED and BOC concepts already in use at CERN in terms of stability, tolerances and price. He also mentioned the cooling problems that were faced by the current Pulse compressor installed in Xbox1 (X-band test stand in the CTF gallery). The bad thermalization of the pistons in this model creates a low frequency oscillation between on/off powering after a breakdown or a stop.
The new pulse compressor relies on an adiabatic shaping of the cavity that allows reducing the machining tolerances to 20 microns. The machineable pistons allow to tune very precisely both cavities to match the frequency of both of them. Indeed, the first measurements of the cavities showed a discrepancy of more than 150 kHz from the target frequency which was recover by machining both pistons. After a second machining of the piston by the AP and a manual one on the tuning chamfer, both cavities were tuned to the ideal frequency. Taking the RF measurements data, Igor has recalculated the compression factor that should be larger than 3. The compressor also shows a quality factor Q larger than predicted! The origin of the original frequency shift is not understood yet but indicates an error of 1.5mm. Raphael will contact Ahmed to make some kind of non-contact metrology on the piece. Metrology after every step should be considered in the future.
The working of the Xbox1 SLED2 pulse compressor was showed by Ben as well as the RF measurements of the new pulse compressor in the lab and the actual pulse compressor that is slightly smaller than the factor 3 foreseen by the simulations. A calibration problem may be at the origin of this discrepancy.
Nuria estimated the final cost of the pulse compressor. The final cost close to 75k is mainly dominated by the main workshop costs (machining and brazing) and by the production and modification of drawings which can be, at least partially considered non-recurrent engineering. Materials represent a fix cost of less than 15kCHF. As an educated guess, a new compressor could cost now between 50 and 60kCHF.
The total time invested on the manufacture was 15 months. Three months were invested on the preliminary drawings. The coating and modification of the flanges generated a three months delay and problems with the vacuum and water circuit leaks set us back another two months. Some delays could have been easily avoidable. In summary, it should be possible to do another unit in six months.
Machining problems were mentioned by Emilien. The final time for machining the copper cavities was underestimated us new tooling had to be designed and fabricated to adapt to the small opening. Now that the tooling exists, the initial estimate of 130 h should be sufficient. There was a discussion on the cavity wall thickness and the advantages to reduce it. It was concluded that there was no obvious reasons to do so and having thick walls made machining easier.
For the stainless steel water jacket machining Emilien proposes a change of strategy. Forming the cylinder from a thinner sheet will easy manufacturing and have similar results. It is accepted to use a sheet of stainless steel tor the future compressors. The cooling cap fixture could be extruded from the sheet.
The problem of brazing copper and stainless steel was discussed lengthy. Luca presented measurements and models of the temperature gradient that each part of the compressor suffers during the brazing process. There is no clear what caused about 3mm of displaced brazing between the copper body and the SS jacket. However, it is clear now that the two parts had ad difference in temperature of around 400 degrees and thus a difference between the top corner of both pieces equivalent to 3.2mm. As a first measure the temperature can be increased slower (50 degrees/hour) which will reduce the gap in half.
Different joining methods were discussed like EB welding or the use of a copper jacket. The copper jacket will not improve the situation as the heat transfer is done mainly bt radiation. EB welding will be very expensive when going to industrialization. A solution based in brazing two rings of stainless steel on the cavity copper and then tig welding the jacket to these rings is the preferred solution. There are still open questions about the optimum sequence for brazing to avoid stress in the jacket during the different passages in the oven. The model produced by Luca shall answer some of these questions.
Raphael added some other modifications that need to be undertaken. Aside from the ones already mentioned, the only remaining point is the mechanic support which needs to be re-though also to take into account a reduction of friction during assembly.
Last for the near future, we need to fabricate five more units (four units for Xbox3 plus a spare) which need to be ready by the end of 2014. As some major modifications are going to be introduced, the decision is to proceed with a second unit produced "in house". Gerry will order the cooper cylinders and Raphael will modify the drawings as soon as possible. The machining in the AP being fully booked until September, Emilien proposed to subcontract the machining of the two copper cavities. This will also serve to qualify an external company with no mayor over cost. The production of small pieces can still be done by the AP. Brazing should be done by MME but the capacity and scheduling need to be discussed with the group management. Said will discuss inside the group the possibilities.
For the hybrid, drawings already exist and the pieces were manufactured in France by Frank Peauger. Nuria will contact him to know the price, delay and company who took care of it in the past.
The goal thus is to have a second unit produced at CERN by the end of the year. The rest of the production (four more units) could be done in house depending on MME availability or subcontracted to the industry. Some assistance from MME will be in any case needed in the second scenario for technical specifications and contract follow up.
12.04.2013
reported by
Nuria Catalan Lasheras
Present: Said Atieh; Thierry Bogey; Thierry Callamand; Nuria Catalan Lasheras; Luca Dassa; Miguel Gil Costa; Raphael Leuxe; Gerard Mcmonagle; Emilien Rigutto; Igor Syratchev; Thierry Tardy; Agostino Vacca; Ben Woolley
Chaired by: Dr. Catalan Lasheras, Nuria
Igor presented (in French!) a small description of the functioning of the pulse compressor, the advantages against the SLED and BOC concepts already in use at CERN in terms of stability, tolerances and price. He also mentioned the cooling problems that were faced by the current Pulse compressor installed in Xbox1 (X-band test stand in the CTF gallery). The bad thermalization of the pistons in this model creates a low frequency oscillation between on/off powering after a breakdown or a stop.
The new pulse compressor relies on an adiabatic shaping of the cavity that allows reducing the machining tolerances to 20 microns. The machineable pistons allow to tune very precisely both cavities to match the frequency of both of them. Indeed, the first measurements of the cavities showed a discrepancy of more than 150 kHz from the target frequency which was recover by machining both pistons. After a second machining of the piston by the AP and a manual one on the tuning chamfer, both cavities were tuned to the ideal frequency. Taking the RF measurements data, Igor has recalculated the compression factor that should be larger than 3. The compressor also shows a quality factor Q larger than predicted! The origin of the original frequency shift is not understood yet but indicates an error of 1.5mm. Raphael will contact Ahmed to make some kind of non-contact metrology on the piece. Metrology after every step should be considered in the future.
The working of the Xbox1 SLED2 pulse compressor was showed by Ben as well as the RF measurements of the new pulse compressor in the lab and the actual pulse compressor that is slightly smaller than the factor 3 foreseen by the simulations. A calibration problem may be at the origin of this discrepancy.
Nuria estimated the final cost of the pulse compressor. The final cost close to 75k is mainly dominated by the main workshop costs (machining and brazing) and by the production and modification of drawings which can be, at least partially considered non-recurrent engineering. Materials represent a fix cost of less than 15kCHF. As an educated guess, a new compressor could cost now between 50 and 60kCHF.
The total time invested on the manufacture was 15 months. Three months were invested on the preliminary drawings. The coating and modification of the flanges generated a three months delay and problems with the vacuum and water circuit leaks set us back another two months. Some delays could have been easily avoidable. In summary, it should be possible to do another unit in six months.
Machining problems were mentioned by Emilien. The final time for machining the copper cavities was underestimated us new tooling had to be designed and fabricated to adapt to the small opening. Now that the tooling exists, the initial estimate of 130 h should be sufficient. There was a discussion on the cavity wall thickness and the advantages to reduce it. It was concluded that there was no obvious reasons to do so and having thick walls made machining easier.
For the stainless steel water jacket machining Emilien proposes a change of strategy. Forming the cylinder from a thinner sheet will easy manufacturing and have similar results. It is accepted to use a sheet of stainless steel tor the future compressors. The cooling cap fixture could be extruded from the sheet.
The problem of brazing copper and stainless steel was discussed lengthy. Luca presented measurements and models of the temperature gradient that each part of the compressor suffers during the brazing process. There is no clear what caused about 3mm of displaced brazing between the copper body and the SS jacket. However, it is clear now that the two parts had ad difference in temperature of around 400 degrees and thus a difference between the top corner of both pieces equivalent to 3.2mm. As a first measure the temperature can be increased slower (50 degrees/hour) which will reduce the gap in half.
Different joining methods were discussed like EB welding or the use of a copper jacket. The copper jacket will not improve the situation as the heat transfer is done mainly bt radiation. EB welding will be very expensive when going to industrialization. A solution based in brazing two rings of stainless steel on the cavity copper and then tig welding the jacket to these rings is the preferred solution. There are still open questions about the optimum sequence for brazing to avoid stress in the jacket during the different passages in the oven. The model produced by Luca shall answer some of these questions.
Raphael added some other modifications that need to be undertaken. Aside from the ones already mentioned, the only remaining point is the mechanic support which needs to be re-though also to take into account a reduction of friction during assembly.
Last for the near future, we need to fabricate five more units (four units for Xbox3 plus a spare) which need to be ready by the end of 2014. As some major modifications are going to be introduced, the decision is to proceed with a second unit produced "in house". Gerry will order the cooper cylinders and Raphael will modify the drawings as soon as possible. The machining in the AP being fully booked until September, Emilien proposed to subcontract the machining of the two copper cavities. This will also serve to qualify an external company with no mayor over cost. The production of small pieces can still be done by the AP. Brazing should be done by MME but the capacity and scheduling need to be discussed with the group management. Said will discuss inside the group the possibilities.
For the hybrid, drawings already exist and the pieces were manufactured in France by Frank Peauger. Nuria will contact him to know the price, delay and company who took care of it in the past.
The goal thus is to have a second unit produced at CERN by the end of the year. The rest of the production (four more units) could be done in house depending on MME availability or subcontracted to the industry. Some assistance from MME will be in any case needed in the second scenario for technical specifications and contract follow up.
12.04.2013
reported by
Nuria Catalan Lasheras
There are minutes attached to this event.
Show them.
