lecture 3: magnet

1

Norm

al‐conducting accelerator magnets

© Thomas Zickler, CER

NJUAS 2015

Archam

ps, 23.‐27. February 2015

Lecture 3: Magnet production

Magnetic materials

Manufacturing techniques

QA & Acceptance tests

Recurrent issues

Magnetic measurements

Cost estimates and optimization

Norm



NJUAS 2015

Archam


Magnet manufacturing

Mechanical design

Procurement

• Raw materials

• Tooling

Yoke production

Coil productionMagnet assembly

Tests & measurements

Magnetic materials – Manufacturing – QA – Magnetic measurements – Costs – Summary

2

Norm



NJUAS 2015

Archam


Massive or laminated yokes

Historically, the primary choice was whether the magnet is operated in persistent mode or cycled (eddy currents)

+ no stamping, no stacking

+ less expensive for prototypes and small series

‐ time consuming machining, in particular for complicated pole shapes

‐ difficult to reach similar magnetic performance between magnets

+ steel sheets less expensive than massive blocks (cast ingot)

+ less expensive for larger series

+ steel properties can be easily tailored

+ uniform magnetic properties over large series

‐ expensive tooling

Magnetic materials – Manufacturing – QA – Magnetic measurements – Costs – SummaryNorm



NJUAS 2015

Archam


Magnetic steel

Today’s standard: cold rolled, non‐oriented electro‐steel sheets (EN 10106)

– Magnetic and mechanical properties can be adjusted by final annealing

– Reproducible steel quality even over large productions

– Magnetic properties (permeability, coercivity) within small tolerances

– Homogeneity and reproducibility among the magnets of a series can be enhanced by selection, sorting or shuffling

– Material is usually cheaper, but laminated yokes are labour intensive and require more expensive tooling (fine blanking, stacking)

Courtesy of ThyssenKrupp


3

Norm



NJUAS 2015

Archam


The rolling process produces a thickness variation perpendicular to the rolling direction:

Profile of steel strips






NJUAS 2015

Archam


NGO steel properties

Data source: VOESTalpine Austria

Data source: VOESTalpine Austria

ISOVAC 1300‐100A: Hc = 65 A/m

ISOVAC 250‐35HP: Hc = 30 A/m

Sheet thickness: 0.3 ≤ t ≤ 1.5 mm

Specific weight: 7.60 ≤ δ ≤ 7.85 g/cm3

Electr. resistivity @20°C: 0.16 (low Si) ≤ ρ≤ 0.61 μΩm (high Si)


4

Norm



NJUAS 2015

Archam


Sheet insulation

Surface coating:– electrical insulation of several μm thickness

– one or both sides

– oxid layer, phosphate layer, organic or inorganic coating

-active5.0 - 8.0

passive max.1.5

both sidescolorlessorganic bonding lacquer with one side heat treatment (passive)

Combined insulation

-5.0 - 8.0one or both sides

colorlessorganic bonding lacquer (active)STABOLIT 70

> 5> 15> 50

0.3 - 1.01.0 - 2.02.0 - 3.5

both sidesgreyinorganic with organic components pigmentedSTABOLIT 60EC-5

> 903.0 - 5.04.0 - 7.06.0 - 9.0

one or both sides

greyorganic pigmentedSTABOLIT 40EC-6

> 50.5 - 1.5both sideslight grey

inorganic with organic componentsSTABOLIT 30EC-5-P

> 50.5 – 1.5both sidesgrey-green


> 15max. 1.5both sidesyellow-green

organicSTABOLIT 10EC-3 by prior arrangement only

Insulation resistanceat room temperatureto ASTM A717/A717M-95Ωm²/Lamelle

Coating thickness each sidein µm

CoatingColor1)Insulation typeInsulation designationIEC 60404-1-1

-active5.0 - 8.0

passive max.1.5

both sidescolorlessorganic bonding lacquer with one side heat treatment (passive)

Combined insulation

-5.0 - 8.0one or both sides

colorlessorganic bonding lacquer (active)STABOLIT 70

> 5> 15> 50

0.3 - 1.01.0 - 2.02.0 - 3.5

both sidesgreyinorganic with organic components pigmentedSTABOLIT 60EC-5

> 903.0 - 5.04.0 - 7.06.0 - 9.0

one or both sides

greyorganic pigmentedSTABOLIT 40EC-6

> 50.5 - 1.5both sideslight grey


> 50.5 – 1.5both sidesgrey-green


> 15max. 1.5both sidesyellow-green

organicSTABOLIT 10EC-3 by prior arrangement only

Insulation resistanceat room temperatureto ASTM A717/A717M-95Ωm²/Lamelle

Coating thickness each sidein µm

CoatingColor1)Insulation typeInsulation designationIEC 60404-1-1

Source: ThyssenKrupp




NJUAS 2015

Archam


Other magnetic materials

1. High purity irons– Iron referred to as "high purity" when total concentration of impurities (mainly C, N, O, P, S ,Si and Al)

does not exceed a few hundred ppm– Otherwise Low Carbon Steel or Non‐alloyed Steel

– Very pure F: high electrical conductivity not suitable for AC applications – For high permeability at B > 1.2 T it is advisable to anneal at max. 800 °C and cool slowly

2. Low‐Carbon Steels – e.g. type 1010– Disadvantage: Magnetic ageing (increase of coercivity with time)

3. Non‐grain oriented Silicon Steels (NGO)Advantages:– Increase in permeability– Decrease in hysteresis loss– Eddy current loss decrease due to higher resistivity (Al and Mn added as well)– No ageing

4. Grain‐oriented Silicon Steels5. Iron alloys

a. Iron‐Nickelb. Iron‐Cobalt alloys with high magnetic saturation

6. Compressed powdered Iron and Iron alloys7. Ferrites8. Innovative materials and rare earths

Reference: S. Sgobba: Physics & Measurements of Magnetic Materials, CAS 2009, Brugges


5

Norm



NJUAS 2015

Archam


Yoke manufacturing

Stamping laminations

Stacking laminations into yokes

Gluing and/or welding

Machining

Assembly (preliminary)




NJUAS 2015

Archam


Lamination punching• Punching or fine blanking

• Fine blanking requires more expensive tooling

• Tolerances less than +/‐ 8 μm achievable (depending on thickness, material and layout)

• Material can be delivered in sheets or strips (coils)


6

Norm



NJUAS 2015

Archam


Alternatives to punching

...or a combination of different techniques

TechniqueAccuracy [mm]

Repeatability [mm]

Drawbacks

Water jet cutting > 0.13 >0.025rough cutting edge, relatively slow

Laser cutting >0.01 >0.005cutting edge ‚burnt‘, relatively slow

CNC machining 0.01‐0.001 0.01‐0.001 stacks only

Wire‐cut EDM > 0.002 >0.001 very slow, limited size




NJUAS 2015

Archam


Tooling for:• stacking

• baking

• welding

Yoke stacking


7

Norm



NJUAS 2015

Archam


Bonding & insulation

Special coatings have been developed for the adhesive bonding of laminations:

• provide electrical insulation and mechanical bonding

• based on epoxy resins

• available in B‐stage (partly cured) and C‐stage (fully cured)

• Referred to as STABOLIT 70 by ThyssenKrupp

Courtesy of Rembrandtin




NJUAS 2015

Archam


Glueing vs. Welding

Welding

+ mechanically more ridgig

+ no aging

‐ massive end plates/tension straps needed

‐ continous welding introduces stress and deformation

‐ sophisticated welding procedure

/ requires stacking fixture

Glueing

+ no stress, no distortions

+ no tension straps, no end plates( no eddy currents)

‐ glue sensitive to radiation and aging

‐ requires clean laminations and conditions

‐ requires baking oven

/ requires stacking fixture

Recommendation: combine gluing, welding & bolting


8

Norm



NJUAS 2015

Archam


Coil manufacturing

Define conductor type and material

Conductor insulation

Winding

Ground insulation

Epoxy impregnation




NJUAS 2015

Archam


Conductor materials

Key‐stoning: risk of insulation damage & decrease of cooling duct cross‐section

%6.33

A

AAR

Al Cu (OF)

Purity 99.7 % 99.95 %

Resistivity @ 20°C 2.83 μΩ cm 1.72 μΩ cm

Thermal resistivity coeff. 0.004 K‐1 0.004 K‐1

Specific weight 2.70 g/cm3 8.94 g/cm3

Thermal conductivity 2.37 W/cm K 3.91 W/cm K


9

Norm



NJUAS 2015

Archam


In a magnet coil, the electrical insulation ensures that current flows only along the conductors and not between individual conductors or between the conductors and other parts of the magnet

Dielectric materials can be distinguished in three main classes: – inorganic materials: ceramics, glass, quartz, cements and minerals (e.g. mica) – organic materials: thermoplastic (Rubber, PA (Nylon), PP, PS, PVC, PC, PTFE) or

thermosetting: Polyethylene, PI, PEEK, Epoxy, phenolic, silicon, polyester resins– composites: fully organic (aramidic fibres‐epoxy tapes) or mixed (epoxy‐mica tapes)

The electrical insulation is stressed by several factors: – electric– thermal– mechanical– chemical (including oxidation)– radiation

A weak electrical insulation may produce: – current leaks with local heating up to melting and possible fire – progressive damage of the leakage path up to a short circuit – unbalanced circulating currents (magnetic field distortion) – incorrect functioning of protections

Montsinger’s rule / Arrhenius equation:

A temperature rise of 10 K halves the expected live time of an insulation system

Reference: D. Tommasini: Dielectric insulation and high‐voltage issues, CAS 2009, Brugges

Coil insulation

)(5.0)10( TtKTL




NJUAS 2015

Archam


Radiation hardness

Radiation hardness is an important criterion for insulation materials used for accelerator applications

Source: CERN Yellow Report CERN 98‐01

Above 108 Gy special insulation techniques are required!


10

Norm



NJUAS 2015

Archam


Radiation hardness

Source: CERN Yellow Report CERN 98‐01




NJUAS 2015

Archam


Coil insulation

Conductors with small cross‐section:straigthening cleaning conductor insulation winding ground insulation


11

Norm



NJUAS 2015

Archam


Coil insulation

Conductors with large cross‐section:straigthening winding sand blasting cleaning conductor insulation ground insulation




NJUAS 2015

Archam


heating and evacuating mold and coil (auto‐clave or vacuum mold) mixing resing heating and degassing resin injecting resin curing cycle cooling

Coil impregnation


12

Norm



NJUAS 2015

Archam


Magnet assembly

By hand....

... or with the help of tooling




NJUAS 2015

Archam


Auxiliary components

• Electrical connections

• Hydraulic connections

• Interlock sytem (temperature, pressure, water flow)

• Magnetic measurement devices (pick‐up coils, hall probes)

• Alignment tragets, adjustment tables and support jacks


13

Norm



NJUAS 2015

Archam


Hydraulic circuits

• Water cicuits are most critical items• 95% of all magnet failures due to water leaks:

– Corrosion– Errosion– Poor brazing quality– Poor welding quality– Failure or aging of joints– Inadequate materials– Incorrect assembly– Radiation damage– Inadequate design

• Leaks can be detected and repaired during magnet acceptance tests and commissioning...

• ... but, many leaks occur only after years in operation• Often not monitored magnet damage (short cicuits, corrosion of

iron yoke) and collateral damages on other equipment possible




NJUAS 2015

Archam


Interlock SensorsThermo‐switch:

Flow‐switch:


14

Norm



NJUAS 2015

Archam


QA & Acceptance tests

QA is important at each production stage• Constant monitoring of critical items from the raw material, to semi‐finished parts, to sub‐

components to the final product

• Sample testing (destructive or non‐destructive) to qualify materials, manufacturing techniques and processes

• Acceptance test can include electrical, hydraulic, mechanical, thermal, and magnetic measurements

• Tests/measurements can be systematically (entire series) or on specific/random samples

• Complete recording and documentation indispensible (back‐tracing in case of doubts or failures)




NJUAS 2015

Archam


Sample testing

Typically the following samples are tested to validate material performance and production processes:

• Magnetic steel

• Laminations

• Bond strenght (laminations)

• Brazing

• Welding

• Bond strength (coil)

• Impregnation


15

Norm



NJUAS 2015

Archam


Sample testing


• Magnetic steel

• Laminations


• Brazing

• Welding


• Impregnation




NJUAS 2015

Archam


Sample testing


• Magnetic steel

• Laminations


• Brazing

• Welding


• Impregnation


16

Norm



NJUAS 2015

Archam


Sample testing


• Magnetic steel

• Laminations


• Brazing

• Welding


• Impregnation




NJUAS 2015

Archam


Sample testing


• Magnetic steel

• Laminations


• Brazing

• Welding


• Impregnation


17

Norm



NJUAS 2015

Archam


Sample testing


• Magnetic steel

• Laminations


• Brazing

• Welding


• Impregnation




NJUAS 2015

Archam


Mechanical measurements

Yoke shape and dimensional accuracy is essential for magnetic field quality in iron‐dominated magnets

Mechanical errors can significantly influence the magnetic performance

We distinguish between:• Systematic errors (lamination contour) ,

can be relativley easy determined and corrected

• Random errors (stacking, machining, assembly), difficult to predicted, in general known only after series production, can be partly minimized by imposing low mechanical tolerances (expensive), enlarged design margins (expensive), strict quality assurance and close follow‐up of production processes

Mechanical measurements are required after each manufacturing step:• Lamination punching

• Yoke pieces (half‐yoke, quadrants) manufacture

• Yoke assembly

• Reproducability: after disassembling and re‐assembling (dowel pins)

• Stability: no deformation due to handling, lifting, transport (lifting test)


18

Norm



NJUAS 2015

Archam


Various optical and mechanical techniques with different accuracy, precision and resolution are available for mechanical measurements

Mechanical measurements




NJUAS 2015

Archam


Insulation testing

• Insulation tests are an integral part of the electrical testing in addition to the measurements of the electrical coil resistance and inductance

• An electrical insulation shall be tested to ensure its ability to provide the required insulation levels for specified operation and faulty conditions and during a specified time tests shall reveal manufacturing deficiencies

• In most cases, this means specifying test levels much beyond the real operational conditions

• For acceptance testing we typically perform two types of tests:– High voltage test of a conductor to ground

– Capacitor discharge test for inter‐turn insulation


19

Norm



NJUAS 2015

Archam


Power and heating tests

Power test are performed to verify:

– correct interlock functioning

– no accidental over‐heating

– correct cooling performance

– absence of poor electrical contacts

– absence of moving or loose parts (pulse test)




NJUAS 2015

Archam


Recurrent quality issues

Despite a severe quality control by the manufacturer, we often find quality deficiencies during the acceptance tests and certification at CERN

Amongst several other recurrent issues, the following are the most frequent and most serious:– Poor brazing quality

– Poor bonding stength

– Poor coil insulation/impregantion

– Insufficient rust protection

– Loose or moving parts

– Covers not respecting IP2X

– Insufficient cable cross‐section

– Obstructed cooling circuits

– Transport damages due to inadequate packaging


20

Norm



NJUAS 2015

Archam



Lack/excess of brazing filler




NJUAS 2015

Archam



Lack of resin: bubbles, voids, fissures, cracks, poor penetration, poor wetting

Excess of resin: volumes of pure resin


21

Norm



NJUAS 2015

Archam



Poor lamination bonding stength




NJUAS 2015

Archam


Magnetic measurements

• Magnetic measurements as final proof if design and manufacturing is correct are an

essential part of the magnet qualification process

• Magnetic measurements are complementary to computer simulation and beam based

measurements

• No single instrument or method can cover all requirements

– Multiple instruments are complementary

– Overlaps provide estimation of absolute uncertainty and error correction

• Commercial availability is very limited requires R&D, time and experience

– Precise mechanics, stable materials, heavy benches are the foundation of good instruments

• Different strategies for error correction:

– random errors are reduced by repetition

– systematic errors are reduced by cross‐checks and by the use of symmetries


22

Norm



NJUAS 2015

Archam


Overview of measurement techniquesMethod

Type1 Range

[T]u2

[‐]BW3

[Hz]Cost4

[€]Size5

[mm]Length6

[mm]Notes

SQUID V 10‐1410‐4 10‐5 DC100 103 < 1 < 1 Highest sensitivity, needs cryogenics

Fluxgate V 10‐710‐2 10‐2 DC103 102 3 10 Main application: geomagnetic fields

Fiber optic V 10‐1210‐3 10‐6 DC102 104 30 100Magnetostrictive sample deformation detected by Mach‐Zender interferometer

M‐optical V >10‐4 10‐2 DC109 103 >100 >100Faraday effect (rotation of polarization). Extreme BW, compatible with extremely high fields

M‐induction V 10‐610‐3 10‐2 DC103 100 5 5Used for low‐power compass devices in GPS, smartphones …Digital out, T‐insensitive

M‐resistance V 10‐910‐3 10‐2 DC109 101 < 1 < 1GMR sensor vital for magnetic recording industry; non‐linear

M‐diode V 10‐5101 10‐1 DC104 101 < 1 < 1 High sensitivity = 10x Hall, non‐linear

M‐transistor V 10‐6100 10‐1 DC104 101 < 1 < 1 Very high sensitivity = 100x Hall, non‐linear

Hall effect V 10‐5102 10‐3 DC104 103 <1 <1Higher low‐prec BWsimple, cheap, commercially availableaccuracy requires laborious calibration

Fixed coil V >10‐12 10‐4 10‐2106 104 5500 1 mm 10 m Linear, sensitivity increases wth BW

Rotating coil V >10‐12 10‐4 DC101 104 8‐400 1 mm 2 m Best all‐round performer for accelerator magnets

Translating wire V >10‐3 10‐4 DC 104 0.1 < 20 m Classical system

Oscillating wire V >10‐3 10‐3 DC 104 0.1 < 20 m Ongoing R&D

NMR S 10‐5102 10‐6 DC100 104 10 10Low field range only with custom flowing/water method

FMR/EPR S 10‐810‐1 10‐4 DC103 104 40 40 Experimental

Optical pump S 10‐510‐4 10‐6 DC101 104 100 1 mBased on Zeeman effect in Rb, CeBest resolution: 10 pT

(1) V=vector, S=scalar (2) u=typical uncertainty (3) bandwidth (4) system cost (5) transversal sensor size (6) sensor size along the beam direction




NJUAS 2015

Archam


Field range and uncertainty1.E‐07

1.E‐06

1.E‐05

1.E‐04

1.E‐03

1.E‐02

1.E‐01

1.E‐12 1.E‐11 1.E‐10 1.E‐09 1.E‐08 1.E‐07 1.E‐06 1.E‐05 1.E‐04 1.E‐03 1.E‐02 1.E‐01 1.E+00 1.E+01 1.E+02

Fixed/rotating coils

Hall sensor

FMR

Stretched wires

NMR


23

Norm



NJUAS 2015

Archam


Hall based instruments

b3/b5 probe: based on rings of fixed Hall plates

3D/3D hall probe mapper

Hall head

rotating hall probe polarity checker

3D Teslameter based on 3 Hall probe sensors(typically 1% absolute uncertainty)




NJUAS 2015

Archam


Fixed coils (Fluxmeter)

• Trivial solution to measure AC fields over large areas• Arrays of coils measure field quality in one go• PCB technology for inexpensive, precise large series runs• Absolute calibration requires adequately sized reference magnet often not possible• Used for relative Bdmeasurements (tracking)


24

Norm



NJUAS 2015

Archam


Rotating coils

Provide simultanously information about field strength, quality (harmonics), direction and center

n=2, B20n=1, A10n=1, B10

Discrete sampling of flux

→ Fourier components

→ Field harmonics




NJUAS 2015

Archam


Stretched Wire


25

Norm



NJUAS 2015

Archam


NMR technology

Metrological reference: • NMR devices are used as calibration standard for uniform fields ( = 0.1 ppm, u=5 ppm)

Limitations:• lowest field level 430 G • tracking is slow (Hz): maximum field variation tolerated for latching δB/B < 1 % /s (4.5 G/s @ 450 G)• field gradients blur the signal: max. field gradient for latching B/B < 10 …100 ppm/mm

Courtesy www.metrolab.com

10 T

demodulated RF output in “marker mode”




NJUAS 2015

Archam


Cost estimate

Production specific tooling:

5 to 15 k€/tooling

Material:

Steel sheets: 1.0 ‐ 1.5 € /kg

Copper conductor: 10 to 20 € /kg

Yoke manufacturing:

Dipoles: 6 to 10 € /kg (> 1000 kg)

Quads/Sextupoles: 50 to 80 € /kg (> 200 kg)

Small magnets: up to 300 € /kg

Coil manufacturing:

Dipoles: 30 to 50 € /kg (> 200 kg)

Quads/Sextupoles: 65 to 80 € /kg (> 30 kg)

Small magnets: up to 300 € /kg

Contingency:

10 to 20 %

Magnet type Dipole

Number of magnets (incl. spares) 18

Total mass/magnet 8330 kg

Design 14 kEuros

Punching die 12 kEuros

Stacking tool 15 kEuros

Winding/molding tool 30 kEuros

Yoke mass/magnet 7600 kg

Used steel (incl. blends)/magnet 10000 kg

Yoke manufacturing costs 8 Euros/kg

Steel costs 1.5 Euros/kg

Coil mass/magnet 730 kg

Coil manufacturing costs 50 Euros/kg

Cooper costs (incl. insulation) 12 Euros/kg

Total order mass 150 Tonnes

Total fixed costs 71 kEuros

Total Material costs 428 kEuros

Total manufacturing costs 1751 kEuros

Total magnet costs 2250 kEuros

Contingency 20 %

Total overall costs 2700 kEuros

Total costs

Magnet

Fixed costs

Yoke

Coil

NOT included: magnetic design, supports, cables, water connections, alignment equipment, magnetic measurements, transport, installationPrices for 2011


26

Norm



NJUAS 2015

Archam


Cost optimization

Focus on economic design!

Design goal: Minimum total costs over projected magnet life time by optimization of capital (investment) costs against running costs (power consumption)

Total costs include: capital costs of

magnets

capital costs of power converters

capital costs of power

distribution

capital costs of cooling system

estimated operation costs of these items




NJUAS 2015

Archam


Cost optimization

2 3 4 5 6 7 8 9 10

Norm

alized costs

Current density j [A/mm2]

Investment vs running costs

2 3 4 5 6 7 8 9 10

Norm

alized costs



2 3 4 5 6 7 8 9 10

Norm

alized costs



2 3 4 5 6 7 8 9 10

Norm

alized costs



2 3 4 5 6 7 8 9 10

Norm

alized costs



2 3 4 5 6 7 8 9 10

Norm

alized costs



Magnet capital

Power equipm. capital

Total capital

Running

Total


27

Norm



NJUAS 2015

Archam


Cost optimization

2 3 4 5 6 7 8 9 10

Norm

alized costs


Total cost (j, energy costs)

n euros/kWh

2n euros/kWh




NJUAS 2015

Archam


Summary

• The manufacturing techniques shall be adapted to meet the specified requirements of the finial product in all respects

• Tight QA is the key for success and important at each production stage

• Sample testing to qualify materials, manufacturing techniques and

processes

• Acceptance tests to verify the correct performance of the final product

• The yoke shape and dimensional accuracy is essential for magnetic field

quality in iron‐dominated magnets

• Magnetic measurements are an essential part of the magnet qualification

process and complementary to computer simulations and beam

measurements

• Cost optimization is an important design aspect, in particular in view of future energy costs


28

Norm



NJUAS 2015

Archam


Thanks for your attention…

… and many thanks to all my colleagues who contributed to this lecture, in particular L.Bottura, M.Buzio, B.Langenbeck, N.Marks, S.Russenschuck, D.Schoerling, C.Siedler, S.Sgobba,

D.Tommasini, A.Vorozhtsov

Norm



NJUAS 2015

Archam


Literature

• Fifth General Accelerator Physics Course, CAS proceedings, University of Jyväskylä, Finland, September 1992, CERN Yellow Report 94‐01

• International Conference on Magnet Technology, Conference proceedings

• Iron Dominated Electromagnets, J. T. Tanabe, World Scientific Publishing, 2005

• Magnetic Field for Transporting Charged Beams, G. Parzen, BNL publication, 1976

• Magnete, G. Schnell, Thiemig Verlag, 1973 (German)

• Field Computation for Accelerator Magnets: Analytical and Numerical Methods for Electromagnetic Design and Optimization, S. Russenschuck, Wiley‐VCH, 2010

• CAS proceedings, Magnetic measurements and alignment, Montreux, Switzerland, March 1992, CERN Yellow Report 92‐05

• CAS proceedings, Measurement and alignment of accelerator and detector magnets, Anacapri, Italy, April 1997, CERN Yellow Report 98‐05

• Physik der Teilchenbeschleuniger und Synchrotronstrahlungsquellen, K. Wille, Teubner Verlag, 1996

• CAS proceedings, Magnets, Bruges, Belgium, June 2009, CERN Yellow Report 2010‐004

lecture 3: magnet

Documents