Download presentation
Presentation is loading. Please wait.
1
Michele Martino TE-EPC
Status Update of Power Converter Precision & Accuracy Performance: Noise Michele Martino TE-EPC 88th WP2 Meeting β CERN 6/R-012 β 21/03/2017
2
? βπ΅ π΅ = βπΌ πΌ Final goal of the study
Beam physics constraints are usually specified in terms of βπ΅ π΅ Power Converters conversely are specified in terms of electrical quantities Usually/mostly in terms of current βπ° π° Assuming βπ΅ π΅ = βπ° π° βseemsβ safe but it can be costly as it might lead to over-specification!! M. Martino TE-EPC - WP2 Meeting - 21/03/2017
3
Model of PC output to magnetic field
two regimes: current control (<0.1 Hz): voltage control (>0.1 Hz): with current ripple (Power converter specifications) voltage ripple (Power converter specifications) transfer function of the load (circuit) seen by the power converter transfer function from the input current of the magnet to the magnetic field (assumed constant) transfer function of the cold bore, absorber, beam screen etc., TVacuum β€1 (not taken into account) M. Giovannozzi - HL-LHC Circuit Review March 2016
4
Model of PC output to magnetic field
two regimes: current control ( <π 0 ): voltage control ( >π 0 ): with current ripple (Power converter specifications) voltage ripple (Power converter specifications) transfer function of the load (circuit) seen by the power converter πΌ π = π(π) π
πππππ’ππ‘ +2ππβ πΏ πππππ’ππ‘ (π)βπ transfer function from the input current of the magnet to the magnetic field (assumed constant) transfer function of the cold bore, absorber, beam screen etc., TVacuum β€1 (not taken into account) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
5
Model of PC output to magnetic field B
Current control ( π<π 0 ) π 0 is a design parameter and can be chosen conveniently (with limitations) For SC magnets (large L, small R) π 0 is not going to be chosen above a few Hertz In this regime both π π£πππ’π’π (π)β
1 and π πΌπ‘ππ΅ (π)β
ππππ π‘ are deemed to hold true! βπ΅ π΅ = βπ΅ ππππ π΅ πππ ππππ = π© ππππ (π) π© πππ ππππ β
π° πππππππ π π° πππ πππππππ = βπΌ πππππ’ππ‘ πΌ πππ πππππ’ππ‘ = βπΌ πΌ M. Martino TE-EPC - WP2 Meeting - 21/03/2017
6
Model of PC output to magnetic field B
Voltage control ( π>π 0 ) this regime requires a more accurate modelling and experimental validation as neither assumption might be justified! π π£πππ’π’π (π) depends entirely on material properties and geometry of cold bore, absorber, beam screen and temperature! Many phenomena might be at play for the π πΌπ‘ππ΅ π : Eddy currents (and other βlossesβ) Superconductive material AC behaviour (type II) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
7
πΏ=πΏ(π): existing results for LHC 1085 Quadrupole
No DC bias Dramatic effect due to the Beam Screen Small signal excitation π π£πππ’π’π π ? Courtesy TE-MSC C. Giloux M. Martino TE-EPC - WP2 Meeting - 21/03/2017
8
Modelling the phenomenon : assumptions
π π·πΆπΆπ π π΅ ππππππ‘ π π΅ ππππππ‘ = π // π // +π πΏ π·πΆ π π·πΆπΆπ π π // πΏ ππ πΏ ππ = πΏ π·πΆ 1+π πΏ π·πΆ π // π π΅ ππππππ‘ = π
// π
// +π πΏ π·πΆ π π·πΆπΆπ = 1 1+π πΏ π·πΆ π
// π π·πΆπΆπ If π // is a resistance π
// then (In this case active power losses would have a component β π 2 ) In any reasonable case π π΅ ππππππ‘ is a βlow passβ version of π π·πΆπΆπ M. Martino TE-EPC - WP2 Meeting - 21/03/2017
9
πΏ=πΏ(π): existing results for LHC 1085 Quadrupole
With Beam Screen With Beam Screen M. Martino TE-EPC - WP2 Meeting - 21/03/2017
10
πΏ=πΏ(π): existing results for LHC 1085 Quadrupole
With Beam Screen With Beam Screen Good fit up to 1kHz ο M. Martino TE-EPC - WP2 Meeting - 21/03/2017
11
Results for MQXFS3: No DC @ WARM
150Hz Good fit up to 3kHz ο M. Martino TE-EPC - WP2 Meeting - 21/03/2017
12
Results for MQXFS3: No DC @ COLD
50Hz Very good fit up to 3kHz ο M. Martino TE-EPC - WP2 Meeting - 21/03/2017
13
Beam Screen Simulations : Q1
Approximation: β₯ B-field infinitely long thin cylinder ππΈβ
π π β π π π 0 β+ π 0 2 π π β πΎ 2 Shielding Efficiency β thickness πΎβ
1+π πΏ π conductivity π perimeter ππΈβ
1+π π 2 π 0 βππ Single pole approximation: 1st pole accurate to Β±10% M. Martino TE-EPC - WP2 Meeting - 21/03/2017
14
Beam Screen Simulations HL-LHC: Q1 @ 80K
<150 Hz M. Martino TE-EPC - WP2 Meeting - 21/03/2017
15
Beam Screen Simulations HL-LHC: Q2 @ 80K
<300 Hz M. Martino TE-EPC - WP2 Meeting - 21/03/2017
16
Beam Screen Simulations HL-LHC: D1 @ 80K
<350 Hz M. Martino TE-EPC - WP2 Meeting - 21/03/2017
17
Beam Screen Simulations HL-LHC: D2 @ 20K
<120 Hz M. Martino TE-EPC - WP2 Meeting - 21/03/2017
18
Beam Screen Simulations LHC: Dipole @ 20K
200 Hz M. Martino TE-EPC - WP2 Meeting - 21/03/2017
19
Beam Screen Simulations LHC: Quadrupole @ 20K
<500 Hz BS acts like as a 1st order (at least) low pass 50Hz bandwidth! M. Martino TE-EPC - WP2 Meeting - 21/03/2017
20
Estimation of Noise LHC MQFA.A12
π π·πΆπΆπ sampled at 1kS/s (including measurement noise) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
21
Estimation of Noise LHC MQFA.A12
π π·πΆπΆπ filtered by a 1st order low pass with 50Hz bandwidth M. Martino TE-EPC - WP2 Meeting - 21/03/2017
22
Conclusions Beam Screen represents the major contribution of power losses due to eddy currents but the magnet itself dissipates some! These effects can be electrically modelled as parallel impedances to the βidealβ superconducting magnet modelled as πΏ π·πΆ ! Both effects imply that the current that produces the useful magnetic induction field π π΅ ππππππ‘ is a low pass version of π πππππ’ππ‘ Measurements, magnetic and electrical, are needed to confirm the model and assess the final impact on useful magnetic field hopefully in April 2017 for MQXFS5 and summer with Beam Screen HL-LHC Beam Screens will introduce more than 20dB attenuation at 300Hz - for LHC Quadrupoles this is just a bit less (simulations) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
23
Thank you for your attention
Credits to: Miguel Cerqueira Bastos WP6B, Riccardo de Maria WP2, Lorenzo Bortot WP7, Marco Morrone WP12 M. Martino TE-EPC - WP2 Meeting - 21/03/2017
24
Back-up slides M. Martino TE-EPC - WP2 Meeting - 21/03/2017
25
Time domain behaviour π(π )=π π π + π πΏ π·πΆ π·πΆ =π π (π )
π(π )=π π π + π πΏ π·πΆ π·πΆ =π π (π ) π΄(π )= 1 π πΏ π·πΆ + 1 π π β π΄ (π‘)= 1 πΏ π·πΆ 1 π‘ + 1 π ππ‘ 1(π‘) β π§ (π‘)=π 1 ππ‘ β π πΏ π·πΆ erfcx π πΏ π·πΆ π‘ β‘ 1(π‘) π(π‘)= β π΄ π‘ βπ£(π‘) π£(π‘)= β π§ (π‘)β π ππ‘ π(π‘) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
26
Magnetic flux βseenβ by the circuit
πΏ ππ is the βapparentβ or βsecantβ inductance π πππ πππππ’ππ‘ = πΏ ππ ( π πππ )π πππ π πππ πππππ’ππ‘ is assumed directly due to π΅ πππ ππππ (DC operation at nominal current) For a simplified circuit with resistive cables we can βalwaysβ write : π£ πππππ’ππ‘ π‘ =π
π πππππ’ππ‘ π‘ + π π πππππ’ππ‘ ππ‘ π£ πππππ’ππ‘ π =π
π πππππ’ππ‘ π +π2ππ π πππππ’ππ‘ (π) π πππππ’ππ‘ (π)= πΏ ππ (π) π πππππ’ππ‘ (π) πΏ ππ (π) is such that πΏ ππ 0 = πΏ ππ (at π πππ ) π πππππ’ππ‘ π = π£ πππππ’ππ‘ π π
πΏ ππ (π) +π2ππ π πππππ’ππ‘ π = π£ πππππ’ππ‘ π π
πΏ ππ (π)+π2ππ β€ π£ πππππ’ππ‘ π 2ππ βπ π = π πππππ’ππ‘ π π πππ πππππ’ππ‘ = π πππππ’ππ‘ π πΏ ππ π πππ β€ π£ πππππ’ππ‘ π 2ππ πΏ ππ π πππ M. Martino TE-EPC - WP2 Meeting - 21/03/2017
27
From the circuit to the beam
From π£ πππππ’ππ‘ π , πΏ ππ and π πππ we can easily estimate maximum βπ πππππ’ππ‘ π πππππ’ππ‘ βπ΅ ππππ π΅ ππππ = βπ πππππ’ππ‘ π πππππ’ππ‘ ? βπ΅ π΅ = π΅ ππππ (π) π΅ πππ ππππ β
π πππππ’ππ‘ π π πππ π< π 0 π π£πππ’π’π π Γ π π£β π΅ ππππππ‘ β² π Γ π£ πππππ’ππ‘ π πΏ ππ πΌ πππ π> π 0 π π£πππ’π’π π β€π π£πππ’π’π 0 =1 π π£ β π΅ =π π β2 π π£ β π΅ β² =π M. Martino TE-EPC - WP2 Meeting - 21/03/2017
28
From the circuit to the beam
π π£β π΅ ππππππ‘ β² π = π π πππππ’ππ‘ β π΅ ππππππ‘ (π) π π πππππ’ππ‘ β π΅ ππππππ‘ (0) 1 π
πΏ ππ π +π2ππ β€ π π£β π΅ ππππππ‘ β²β² π 2ππ βπ΅ π΅ = π΅ ππππ (π) π΅ πππ ππππ β
π πππππ’ππ‘ π π πππ π< π 0 π π£πππ’π’π π Γ πβ² π£β π΅ ππππππ‘ β² π Γ π£ πππππ’ππ‘ π 2πππΏ ππ π πππ π> π 0 EPC + MSC will measure π£ πππππ’ππ‘ π , π πππππ’ππ‘ π , π΅ ππππππ‘ (π) and ideally π΅ ππππ (π) We will then estimate πβ² π£β π΅ ππππππ‘ β² π and ideally π π£πππ’π’π π M. Martino TE-EPC - WP2 Meeting - 21/03/2017
29
Existing results NC: Integrated Gradient / Current
Vacuum chamber βπ΅ π΅ βͺ βπΌ πΌ No DC bias Very small excitation current Courtesy TE-MSC-MM Buzio et al. 2014 M. Martino TE-EPC - WP2 Meeting - 21/03/2017
30
Minutes of the last meeting
Michele showed the expected stability of power converters of Class 1 and Class 2. Measured data, although limited, shows that Class 1 power converters are stable below 2 ppm p-p at 6 Hz and below 1 ppm at 0.1 Hz. The regulation frequency has been set to 0.5 Hz, however it could be increased up to 5 Hz (with some limitations) if needed since the sampling frequency is done at 10 Hz. The yearly stability has been measured and quantified in offset (about +-2 ppm) and gain (-7 to -2 ppm). Gain and offset are calibrated through dedicated measurements once or twice a year. Continuous online monitoring is not currently possible. At frequency above 10 Hz (e.g. 1 kHz), measurements are dominated by noise of the equipment and it would be complicated to use AC probes. For class 2 PC the stability is < 2 ppm (in 1h) and the noise is < 2 ppm (p-p 0.1 Hz BW) and 6 ppm (p-p 6 Hz BW). Gianluigi commented that for high frequency we could concentrate on 600 Hz and 300 Hz frequencies since they are the ones at which the noise impact the most the simulations. Massimo reminded that J-P. Burnet offered the possibility to bring the noise of 50Hz lines and multipoles below any specified value. Action: M. Cerqueira, M. Giovannozzi and R. Tomas will follow-up on the subject in separate ad-hoc meeting. M. Martino TE-EPC - WP2 Meeting - 21/03/2017
31
Minutes of last meeting
A discussion touching AC magnetic modelling and simulation followed. Miguel asked whether there are important changes in the differential inductance with the frequency. Stephan replied that about 4% has been reported in the literature. Complete simulations of AC magnet and beam screen response would be too complicated, however separate simulations of a magnet and the cold bore and beam could be performed (already quench stress simulation have been performed). It would be relatively easy to measure AC response without beam screen with an anti-cryostat in SM18 on a single dipole (a first opportunity may come in about 3 months). A measurement with beam screen would imply building the appropriate cryogenic infrastructure. In order to make AC measurement a special coil has to be built, but this is relatively straightforward. Lucio and Stephan will see what the possibilities are. The measurements could be used to benchmark simulation that could be then extended to new magnets. Gianluigi asked whether Thyristor or switching power converter will be used. Miguel replied that this has not been decided yet. The EPC expert could provide a more realistic noise spectrum. Riccardo reminded the present discontinuity from current regulation regime and voltage noise dominated regime. Gianluigi asked to Stephan whether the field quality degradation due to the beam screen could be simulated, Stephan replied he will enquire with Susana and Ezio. M. Martino TE-EPC - WP2 Meeting - 21/03/2017
32
Proposed Measurement in SM18
M. Martino TE-EPC - WP2 Meeting - 21/03/2017
33
Measurement Design RPHFA.SM18.RM.H converter seems the best fit in terms of both DC range and BW RPHFA.SM18.RM.H to run in Voltage mode at different DC levels + frequency sweep Additional (DC)CT might be needed since RPHFA.SM18.RM.Hβs DCCT is rated 13kA π£ πππππ’ππ‘ π‘ , π πππππ’ππ‘ π‘ , π΅ ππππππ‘ (π‘) (and/or π΅ ππππ (π‘) ) synchronously acquired Acquisition with NI PXI bit programmable gain setting 4-ch (up to 200 kS/s) Sine-fit post-processing to extract amplitude and phase of: π£ πππππ’ππ‘ π , π πππππ’ππ‘ π , π΅ ππππππ‘ (π) (and/or π΅ ππππ (π) ) M. Martino TE-EPC - WP2 Meeting - 21/03/2017
34
Q1-Q2a-Q2b-Q3 HL-LHC example
βπ π β€ π£ πππππ’ππ‘ π 2ππ πΏ ππ π πππ βπ π β€ π π πππ 2π 300 π»π§ 255 ππ» 16.5ππ΄ =1.78Γ 10 β9 =1.78 πππ M. Martino TE-EPC - WP2 Meeting - 21/03/2017
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.