30]anua1‘y 1996 OCR Output

Geneva, Switzerland

Radec 95, Arcachon, France

will be exposed to 2.5 10*3 n / cm2 (E >140 keV) during 10 year operation life time.superconducting magnet protection of LHC. They will be placed inside the tunnel and theyelectrical characteristics was measured versus doses. These amplifiers could be used for thefrom 1.2 10** n/cmz (E > 140 keV) to 1.34 10“n/cm2 (E >140 keV). The degradation of thedescribed. These components were irradiated to total integrated neutron fluences varyingThe behaviour of some isolation amplifiers irradiated at the CERN PSAIF facility is

V. Remondino

in a Superconducting AcceleratorRadiation Damage in Amplifiers Used for Quench Detection

j) ,5 Q3 Q O fl

LHC NOTE 351

cum AT/95-47 (MA)CERN·AT·95·47

lllllIllMlIllllIll\|1l1|lEl|l}ll1Jll?7lI1||\

CERN LIBRARIES, GENEVA

EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH

"false" quenches [1, 2]. Fig l. Doses to LHC dipoles per year OCR Outputunderstand the origin of the quench and to identifywave form is recorded. This helps to localise andmagnet voltages. When a fault occurs, the voltage V zsecond set of amplifiers will continuously monitor the

In addition to the quench detection amplifiers, a /well above the noise level.

’//l'CEIsI`protection system, these detection signals must beappears and can be detected [2]. In order to trigger the 7.0 x 1012 n/cm?

500 Gymagnet. If a quench occurs, then a differential voltageYokeof voltage between the ends of the two coils of each

_________C, 5.0 Gyii 2.5 x 1011 n/cm?The detection of quench is based on measurements

Amplifiersdesuoyed by the local overheating.

“ i Avoltages [1]. Without the heaters the magnet could be xa0°the maximum temperature rise and maximum internalout through a large volume of the winding, limiting 1.0x1012 n/cm? 2.Ox1O13

20 Gy 7000 Gy 7outer layer of the magnet coil. This spreads the energyCoilOutside Cryostatstrip heaters, which provoke quenching along the

soon as possible. A signal must then be sent to tire 7/,avoid this, the incipient quench must be detected asin the section of the coil where the quench starts. To

unc Mamet P22magnets could cause local overheating of the magnetThe extremely high energy density of the LHC

960superconductivity or "quench" under fault conditions.Superconducting magnets are liable to loose their

for the large majority of the magnets [4]. The neutronradiation environment.will run at the limit of existing technology in a high the tunnel walls for the nominal beam parameters and

at various radial positions, from the beam centre tocircular orbit. Such large superconducting magnetsThe dose and neutron fluences are given in Fig. lnecessary bending power to maintain protons in their

ducting magnets will be powered in series to give the for a year of operation (taken to be 107 seconds).alarm threshold of the beam loss monitoring system,In the LHC many hundreds of large supercon

of the existing Large Electron Positron (LEP) collider. magnets of lO° protons per second, just below thetions are made assuming continuous losses on themagnets. This machine will be installed in the tunnellosses of the primary proton beam [3]. These calculaLarge Hadron Collider - LHC) using superconductingbeam losses due to beam scattering and the accidentalCERN intends to build a particle accelerator (thehitting the collimators and the vacuum chamber, theaccount the continuous losses due to the beam haloI. INTRODUCTIONconsidering a proton beam of 8 TeV and taking into

The dose in the LHC tunnel was estimated10 year operation life time.to 2.s X 10¤/cm(E >140 kw) during

" 2Il. IRRADIATION AND DOSIMETRYinside the tunnel and they will be exposed

protection of LHC. They will be placed"radiation hard" to the levels given by LHC.used for the superconducting magnetsuitable for this requirements but they must beversus doses. These amplifiers could be

Electronic isolation amplifiers are clearlyelectrical characteristics was measuredmagnet power circuit.(E >140 keV). The degradation of thesignals but also isolate the detection system from the(E > 140 kev) to 1.s4x10me m

" 2the magnet. Amplifiers must not only amplify thefluences varying from 1.2x10n/cm

l3 2across the magnet must be situated in the tunnel nearwere irradiated to total integrated neutronamplifiers which measure and amplify the voltagesfacility is described. These componentsproportional to the length of the cable. Therefore,amplifiers irradiated at the CERN PSAIF

For transmitted signals the pick—up noise level isAbstract—The behaviour of some isolation

CERN, 1211 Geneva 23, Switzerland

Vittorio Rcmondinc

DETECTION IN A SUPERCONDUCTING ACCELERATORRADIATION DAMAGE IN AMPLIFIERS USED FOR QUENCH

converter. They provide high accuracy with complete OCR Output4|AD210 (G=20) [ 6000[ 25[ 1.34g) ports. This eliminates the need for an extemal DC/DC3[AD289 (G=30,90) [ 2000[ 25[ 4.50x10 generated and supplied to both the input and output2[AD289 (G=10) | 3400| 25[ 7.60xlO supply, and a t 15 V isolated power is internallyPWS727, DESY to the module. They are powered by a single + 15 V1[AD289,ISO106, [ 500| 5| 1.10x10 complete isolation via a transformer coupling intemal

surface-mounted components. They provide a[Gy] [[Gy/hour][E>14O KeV

meas.Nr. DEVICES. These circuits are manufactured with[n/cmz]cost isolation amplifiers made by ANALOGExp. [Components Nr [Dose [Dose rate [Fluence

Such amplifiers are a wide band, accurate, low

A. AD289.l and AD210JNLIST OF ALL EXPOSURESON TIE AMPLFERSTABLE1

for HERA (High Energy Ring Accelerator)measurements carried out with PN diodes in 1993 [7]. Elektronen—Synchrotron lab in Hamburg, Germany)a 1 MeV equivalent dose. This was confirmed by

Magnetic amplifier : made by DESY (Deutsches(E > 140 keV) is almost equivalent to the fluence at

dc / dc converters: PWS727to the shape of the spectrum, the neutron fiuence

Isolation amplifiers: AD289J, AD210IN, ISO1061.1 x 10 n/cm(E>140 keV) per pulse. According

2gammasg per pulse of 1.4 x l0" protons andbeen chosen:

tion of our circuits are of 50 mGy (total dose mainlyfrom line transients, the following components have

The values to be taken into account for the irradia With a view to a complete isolation and protectionenergy bin Flux/BN/Log.

thousand volts.in Fig, 2, is given in n/cm2- s or in logarithmic between inputs and outputs is of the order of oneshape of the energy—spectrum of the neutrons, shown

the gain between 10 and 50, the voltage isolationspectrum by means of the LOUHI program [6]. The

frequency response rates range from dc to a few kHz,flux measurements and for evaluating the neutron

quench detection level is between 0.1 and 0.5 V, theActivation detectors and TLDs were used for neutron required for our purpose are not specially severe: thedosimeters (PAD) were used for dose measurements.

The electrical characteristics of the amplifiersmoluminescence dosimeters (TLD), polymer/alanine

Radiophotoluminiscence (RPL) dosimeters, therTESTEDprotons on the target.

III. CHARACTERISTICS OF THE CIRCUITShas been measured as a function of the number of

As described in [5], the dose inside the containerFig. 2. Neutron Spectrum in the PSAIF container

irradiation.

carried out in a safe radiation environment during then"“")

variable dose rate. On-line measurements can be Er? 16) x6;;p[1o° id 14;* uf ufaccelerator beam. The distance from the beam yields a

< Dale from LOUHI (RAL) >

(4.51 cl ttm flux ¤bo·4• I knv)lowered along a tube to variable distance from theE i0° (1.6 El} app on Inrqlf /4.8 3)(18 cm in diameter and 20 cm in height) can be

Total flux : I.234 EIO n/cm2:high—energy particles and neutrons. A containerQ i0°then produced by the strong emission of secondary

diation place, a typical accelerator radiation field is`E 1o’beam from the Proton Synchrotron (PS). At the irra

antiprotons, this target is hit by a 26 GeV protonsantiproton production target. During the production of(PSAIF). The irradiation area is sited behind thetests at the CERN PS-Acol Irradiation Facilitycircuits have been submitted to preliminary radiationon the electrical characteristics of the amplifiers, the readings at the end of each exposure.

dose. Finally, the final dose was compared to the RPLIn order to investigate the influence of irradiationamplifiers was then measured as a function of thewall where the amplifiers will be placed.

2.5 x 10n/cm(E>140 keV) per year at the tunnel sures. The degradation of the performance of the12 2

tored to evaluate the absorbed dose all along the expodose and fluence are expected to be 50 Gy andten are typical for proton accelerators, therefore the ated. The number of protons on the target was moni

radiophotoluminiscence dosimeters (RPL) were irradimachines have shown that dose fluctuations of a factor18° and 25° C. With each set of components, someSystematic measurements at the existing CERN

energy. ature during the irradiation was in the range betweenlevels and the dose rate on the amplifiers. The temperdamage in semiconductors is negligible below that

Table 1 shows the list of all radiation exposureflucnccs have been cut at 140 keV since displacement

istics. Dose levels were then progressively increased OCR Outputcorresponding current will change the flux in the core.any possible initial degradation of electrical character(E,¢0), is then applied to the control circuit, theexperiment, the dose level was kept low to monitormagnetic core with Ec = 0. lf an input voltage,period of about one month. At the beginning of theadjusted so that its peak value just fails to saturate the1.1 x 10n/cm(E>140 keV) was obtained over aIn operational conditions, the supply voltage E, is

13 2

the first radiation test. The total fluence ofcurrent (I,). The circuit behaviour is then non linear.AD289] and one magnetic amplifier was subjected toresistance R with a consequent high value for the

A batch of 2 samples of ISO106, PWS277,inductance falls. Most of voltage appears across theISO106, AD289 and AD210.bring the circuit almost into the saturation zone thethe beginning and at the end of the exposures on thethe current ia. If the applied voltage (E,) is such as totest (1000 Vms between input and output) was made atthe resistance R is then small and the same is true forAD289, AD2l0) and the offset voltage. An isolationcurve, the inductance is high, the voltage drops across10 Khz (Vi: 3 Vpm for ISO106, Vi: 1 Vpm forcharge. During the linear part of the magnetizationAD289, AD210), ac gain at 100 Hz, 1 Khz andcircuit is shared between the resistive and the inductivedc gain (V, = 1 5V for ISO106, V,= 1 0.5 V forWhen Ec: 0, the total voltage E, applied to the ac

Parameters monitored were: power consumption,metal-glass core driven by a 200 Hz ac voltage [8].data without stopping the irradiation.magnetically coupled by means of a high permeabilityexposures, all measurements were made in situ to takeThe ac circuit and the control circuit (Fig. 3) arewere measured prior to the irradiation. During theproperties of ferromagnetic materials.

The electrical characteristics of each componentcurrent amplification is obtained by using non linearoutside the irradiation facility in a safe environment.This is a low impedance passive device in which aneeded to test and power the circuits was located

D. DESY Magnetic Ampliercontainer in the irradiation area. All instrumentationcircuits, mounted on a printed board and placed in the

ISOlO6. Tests were carried out on completed electronicfier without an integrated power supply, like thecircuit could be used for powering an isolation ampli

IV. TESTING PROCEDURESt 15 mA at voltage accuracy of 1 0.75 V. Thisisolation voltage of 1500 V rms, output currents of

Fig. 3. Magnetic amplifier test circuitof the same value as the input. It provides a minimumsingle input voltage (10/18 Vdc) to bipolar voltagesconverter designed by BURR BROWN. It converts a

ac CimmCon¤·ol CircuitThe PWS727 is an isolated, unregulated dc/dc

0.SW

2700 > R ( uq]C. PWS727

i 25 mV..;l€'L-t¤@S2i9¤22r¤miQarinitial input offset voltage is in the range of

70 kHz; the non linearity gain is 1 0.04% and thethe typical frequency response varies from dc to

The continuous isolation is rated for 3500 V rms; 6 1 L S Z iacross a differential ceramic capacitive barrier (3 pF).encoding the input voltage and uniquely couplingisolates t 10 V analogue signals by digitally

r ····•|·•’·•··········•·signal and translate it to a high level. The amplifierbe preceded by amplifiers that precondition a low level

100 Hz, and is thus suitable for our applications.input voltages on the order of 1V to 10 V. They maylonger than 1 ms and its frequency response limited toamplifier made by BURR BROWN intended for levelhas good voltage insolation. Its response in time isThis amplifier is a unity gain buffer isolationcheap, reliable (no maintenance required), robust and

B. ISO106 expected to be radiation resistant. Moreover, it is

threshold switch. It has been tested because it is(-3 dB) and isolation of 2500 V between any ports. monitoring amplifier (recording of data) but only as alinearity of i 0.025%, a bandwidth of 20 kHz therefore strongly non linear: it can not be used as a

produces a 10 V signal on R. The circuit behaviour isThe AD210JN model assures a maximum noncontrol circuit: the resulting current in the ac circuitoutput of i 2500 V.machine a quench signal of ~100 mV is applied to thefrom dc to 20 kHz and isolation between input anda test winding (output 1-2 in Fig. 3). At the Heralinearity error of i 0.05%, frequency response (-3 dB)device is usually completed by ancillary equipment asThe AD289J model offers a guaranteed gain nonacross the resistance R will vary proportionally. Theor fault voltages.Also, as a consequence, the current (I,) and the voltagegalvanic insulation and protection from linc transients

Fig. 4. C) AD289: ac gain at l kHz OCR Output(less than 10%), stable dc gain, and a stable gain at(E>l40 keV): small variations of the voltage offset

0 500 1000 1500 2000 2500 :000 2500 IGYI4000radiation up to a fluence of 3.3 x 10n/cm13 2

The AD289J show good resistance to the 2.23x10m 4.45:01013 \ 6-7¤10`° In/¤m”1and non linear gain.

;2.5x10'° `V"radiation zone. However it has low input impedanceradiation, making it a good level detector in a high min G-10

mlnamplifier is the best choice for its insensitivity todose level. According to these results, the magnetic

woobserved on the resistance R with an increase of the {j\_\_max

6-10tion doses (Fig. 3). Only a small dc drift was `\‘~`resulting voltage was observed on R for different radia M

max G-10of 0.5 V was applied to the control circuit and the

ac gain at 1 KHz’2°was found for an ac voltage of 10 Vm. Then, a signal

the voltage across the resistor R. The optimum valueadjusting the ac voltage (Ea) at 200 Hz to minimise

Pig. 4. B) AD289: dc gain vs. doseradiation resistant. The circuit was optimised by

As expected the magnetic amplifier is 0 500 1000 1500 2000 2500 0000 a50¤[Gy14000components failed.

‘ 2.23xi0’° 4i45x10u \ 6-7¤*¤`° In/¤m“Ifluence larger than 2.9 x 10n/cm(E>l40 keV), the12 2 §2.5x10`°

output voltages became noisy, unstable and forfor a fluence of 2.9 x 10n/cm(E>l40 keV); the min \ mn

12 2nents strongly increases as the dose increases (> 60%

(G-3G'tance to radiation. The input current of the compo

G¤1OThe dc/dc converters, PWS727, show a low resisE `it"~.m¤¤for further experiments.

ma!external power supply. Therefore it was not considered100 ·I·-—··—.L......._.. ---- - ’-r “AD289J, it needs additional amplification and an

ISO106 isolation amplifier behaves better than the120

1.1 x 10n/cm(E>l40 keV). Although the13 2

decreases by 10-20% at the fluence offluence of 2.9 x 10n/cm(E>l40 keV) and then

12 2 Fig.4. A) AD289: test circuitchange. The input current is almost stable up to theas the dose is increased. The dc and ac gains did notvery good: the offset output can be considered constant

Lo InThe behaviour of the ISO106 amplifiers looks

23.70‘ ? Vi¤

47KHV. ANALYSIS OF DATA1.1 KQ H] ln

1.34 x 10n/cm(E>140 keV) during 10 days.`4 z

umten AD210JN amplifiers was exposed to a fluence ofI I I 2 I I 3 I I ‘ I I 5Finally, during the fourth exposure, a batch of

(E>140 keV) in 7 days.and exposed to a fluence of 4.5 x 10n/cm

13 2printed circuit, in pairs, exactly as in the second test AD289Jof 30 and 90. Ten new amplifiers were assembled ontion test to examine their behaviour at different gains

The AD289J were tested again in a third irradia W I \¤ I \¤ I 1 " I I 6

were added to the test series.any selfannealing effect. Eight new components

1 $ Vshow any change in performance. This fact excludes Out 2 Knbefore the new exposure test. The components did not

Their use in LHC is therefore limited to gains of 10.characteristics of the two AD289Js were checkedteristics deteriorated quickly already for a gain of 30.7.6 x 10n/cm(E>14O keV) during 13 days. Theradiation sensitive at higher gain (Fig. 4): the characwere exgposed to an additional fluence of 1 2hand, the performance of the AD289J was morea gain of 10: the two samples previously irradiatedto affect the performance of the circuit. On the otherthe behaviour of AD289J components in circuits withof the radiation dose: but this variation does not seemThe second exposure was performed to examineconsumption decreased by about 40% with an increasenuance or 2.9 X 10n/cm(E>14o kev).

*2 ;100 Hz (gain of the circuit = 10). The powerand no appreciable changes were observed below a

[8] KH. Mess (Desy), Private communication. OCR Outputthe mean value for the different radiation doses.

931`H0616-3.drawn with the minimum/maximum deviations from Semi-conductors Dosimeters", Radecs 93, IEEE Cat No

Measurements in Mixed Gamma-Neutron Fields by means ofirradiation. The mean value of the samples has been[7] M. Tavlet, E. L. Florian, "Dose and Neutron-Fluencenormalised to 100 % for the original value prior to LBL-6413 (1977).[6] A. Rindi: Unfolding neutron spectra: LOUHI for pedestrians.

Fig. 5. C) AD210: dc gain (G=20) vs dose 1991.Facility at CERN", Radecs 91, La Grande-Motte, France.

[5]M. Tavlet, E. L. Florian, "PSAIF: The PS-ACOL Irradiation

soo 1000 2000 0000 4000 5000 {Gy] soog1~en1’] z.2s¤10" 4.4sx10" s.7¤10" a.sx10" 1.12x`10“1.34x1 "inside and around LHC Dipoles", CERN/AC/DI/FA/Note 93

[4] L. Bumod, J.B. Jeanneret, H. Schonbachcr, "Expected DosesTunnel due to Beam-Gas Scattering", CERN/TIS-RP/93-6.20 · Jl { E2.s¤1o‘°

[3] G.R. Stevenson, I.M. Zazula, "Estimates of Dose in the LHCMA/93-81.

\_m1h "LHC Magnet Quench Protection System", CERN/AT[2] L. Coull, D. Hagedom, V. Remondino, F. Rodriguez-Mateos,

Canada, 1993."LHC Magnet Quench Protection System", MT13,. Victoria,

[1] L. Coull, D. Hagedom, V. Remondino, F. Rodriguez-Mateos,

-·-------. - .... M3!100

REFERENCESA¤21oJNdc gain (G-20)

120

irradiation exposures and dose measurements.

Fig. 5. B)AD2l0: Power consumption vs. dose Tavlet for the support provided in organising theof the circuits to be tested. I am equally grateful to M.tion system and his kind suggestions about the choiceo 5°° woo zcoo :000 4000 sooo (GY} sooocussions conceming the LHC magnet quench protec

1¤/¤§‘1 2.zz¤1o" 4.4s¤10" 6_7xio" s.9¤10" 1.m10" 1.34x{¤"I wish to thank L. Coull for the many useful dis$0 T I Ez.sr10"

‘—--__ min

`ACKNOWLEDGEMENT

_

’""""‘···-·--`70+ I `~the dose received.decrease in power consumption is strictly related tomonitored all over this test, in order to verify that the

Power consumption will be continuouslyshould reflect more closely the real situation.

AD21OJN100 (a dose rate only I0 times stronger then in LHC)total dose level of 500 Gy over 1 year. Such condition

Fig. 5. A) AD210: test circuit the Super Proton Synchrotron (SPS) at CERN at a

Ten new ADZIOJN will be tested in the tunnel of

HI Ln

Vi FUTURE DEVELOPMENTS

SOK

be used to monitor the dose.5.6 x 10n/cm(E>l40 keV). This property could

13 2A1 K i 4 mA prior to irradiation, t 2 mA at

v]\./ \J V4sure, but the spread of values stay very narrow:

The current consumption decreases with the expo100 K replaced if necessary during shut-down periods.other

quence, amplifiers can be monitored and eventuallyl Z 1 4 15 16 17 1I 19

with the increase of the radiation dose. As a conseAD21 OJ N

30 29 Moreover, the degradation of the circuits is gradualdetermined).

1mnF levels between 0.2 to 0.5 V (still to be finallyremained less than 2 mV compared to quench detectionprotons per second on the magnets. The input offsettrical characteristics of every component have beenrunning time, assuming a continuous loss of 10In figures 4 and 5, the measurements of the elecequivalent to between 20 and 200 years of LHCvariation (Fig. 5).fluence of 5.6 x 10n/cm (E>l40 keV, which islevel, dc and ac gain and offset show no appreciable 13promising results. They all operated adequately up to aconsumption decreased (- 40%) with the radiation

The isolation amplifiers AD210J N gave very0f 5.6 x 10n/cm(E>14O keV): only the power13 z

The elecuical characteristics stay stable up to a fluenceVI. CONCLUSIONSThe AD210JN reacted the best under radiation.