shielding and grounding in large detectors radeka bnl65912

7/29/2019 Shielding and Grounding in Large Detectors Radeka BNL65912

1/5

Abstract

Shielding effectiveness as a function of shield thick-ness and conductivity vs the type and frequency of theinterference field is described. Noise induced in trans-mission lines by ground loop driven currents in theshield is evaluated and the importance of low shieldresistance is emphasized. Some measures for preven-tion of ground loops and isolation of detector-readoutsystems are discussed.

1. INTRODUCTION

Prevention of electromagnetic interference (EMI),or noise pickup, is an important design aspect in large

detectors in accelerator environments. It is of particu-lar concern in detector subsystems where signals havea large dynamic range or where high accuracy positioninterpolation is performed. Calorimeters are very sen-sitive to coherent noise induced in groups of readoutchannels where energy sums are formed, covering alarge dynamic range. There are several potential noisesources and means of transmission:1) Noise from digital circuits generated locally on a

single front end read-out board on the detector;2) electromagnetic radiation in the space around the

detector generated by other detector subsystems,power supplies, silicon-controlled rectifiers, ma-

chinery, etc.;3) noise induced by penetrations into the detector en-

closure (e.g., cryostat) and the front end readoutelectronics located on the detector;

4) currents coming through ground loops, of which thedetector enclosure and the front end electronics area part, caused by any apparatus and machinery out-side the detector.Problem 1) of internally generated noise is being

addressed by a careful layout and filtering on theboard(s), shielding of preamplifiers, and by minimiz-ing digital operations on the board.

The effects of externally generated noise in the formof EM radiation are best reduced by a well designedFaraday shield (cage). The effects of noise currentsflowing through the shields, due to ground loops, arealso reduced by the Faraday cage.

Ideally, ground loops should be avoided entirely.In practice, preventing formation of ground loops meansto increase a ground loop impedance over most of thefrequency range as much as possible.

Figure 1 illustrates some of the basics of shieldingand ground loop control. No leads of any kind shouldenter a Faraday cage without their shield being con-nected to the detector enclosure at the penetration, oth-erwise a lead connected to the enclosure would injectnoise (it presents a coupler) into detector electrodes andfront end electronics. If the detector enclosure (i.e., itsfront end electronics) and the readout electronics in thecounting area have to be connected electrically and notonly by optical links, then the cable shield should havea low resistance and a high inductance to minimize thenoise at the receiving end due to the ground loop volt-

age. Ground loop currents in the enclosure walls andthe transmission lines will be minimized by isolatingthe detector enclosure from the surrounding.

In Section 2, some basics of shielding against exter-nal EM fields are reviewed. In Section 3, noise pickup intransmission lines from ground loop voltages and currentsis discussed. In Section 4, potential ground loops in a largedetector subsystem are illustrated in the example of theATLAS liquid argon calorimeter. Some practical steps forisolation are described in Section 5, and in Section 6, thequestion of the safety ground is addressed.

SHIELDING AND GROUNDING IN LARGE DETECTORS*

Veljko RadekaBrookhaven National Laboratory, Upton, NY 11973-5000 ([email protected])

*This work is supported by the U.S. Department of Energy:Contract No. DE-AC02-98CH10886.

Coupler:

Detector Counting Area

Shield (Faraday cage)

Cable shield connected to detectorshield at the penetration:

Solid ground ( earth )

Signal, Low V,High V, control lines

insulateelectrically!

ferritecore

vacuum pumps, cooling watercryo lines, mechanical supports

acpower 1

ac power 2major ground loops (global)(low impedance)

~ 5 50 m

Figure 1. An illustration of shielding and ground loopcontrol concepts.


2/5

2. SHIELDING EFFECTIVENESS

AGAINST EM RADIATION

2.1 Thick Shields (t)

Shielding properties of enclosures are analyzed indetail in Ref. 1. Only some main points are emphasizedhere. A hermetic detector or an electronics enclosure

of highly conductive material, such as copper or alumi-num, provides very high attenuation against external EMfields in the frequency range from a few kHz up to veryhigh frequencies. The shielding effect is obtained byreflection and absorption of the EM wave. Attenuationby reflection of an external plane EM wave at normalincidence to the shield, with wave impedance Z

Wis:

Ar= 20 log [dB]Zw4Zs

(1)

where ZS is characteristic impedance of the shield

material,

ZS = (/)

= 3.7 10-7

f

[] (2) (for copper)

where = 2f is the frequency, is the permeabil-ity, and is the conductivity of the shielding material.Z

Scan be expressed in terms of the skin depth = 2/

( ) as,

Zs=2 . (3)

1/ is simply dc sheet resistivity per square of theshield layer, one skin depth thick. Characterist ic im-pedance of the shield is very low, Z

S1m at 10 MHz;

0.1m at 100 kHz.

The wave impedanceZW (the ratio of the electric fieldand the magnetic field) depends on the nature of the source(electric or magnetic antenna) and the distance from thesource. In the far field, i.e., distance greater than /2, itapproaches the impedance of the free space (and air),377. Atf= 10 MHz, = 30 m, so that for frequencies lessthan 10 MHz, most detectors will be in near field condi-tions. For an electric antenna (high voltage and low cur-rent), the wave impedance varies as 1/r, and for a mag-netic antenna (high current and low voltage), it varies asrin the near field r< /2. Taking the above values forZ

S

and for ZW

= 377, the shielding effectiveness in the farfield, due to reflection, is:

Ar = 9.4 104 = 99.5 dB at 10 MHz= 9.4 105 = 119.4 dB at 100 kHz.

The attenuation is higher in the near field for anelectric source, and much lower for a magnetic sourceat low frequencies[3].

Shielding attenuation by absorption due to skin ef-fect is,

Aa = 6.2 t() = 8.7 t/ [dB] (4)(for copper)

where t is thickness of the shield.

A copper shield 0.5 mm thick provides a far fieldattenuation at 100 kHz of only about 21 dB by absorp-tion, and nearly 120 dB by reflection. In this case, ab-sorption becomes dominant only above ~ 5 MHz. Thereflection attenuation increases with the angle of inci-dence.

2.2 Very Thin Shields (t


3/5

3. NOISE INDUCED IN SHIELDED

CONDUCTORS BY GROUND LOOP

CURRENTS

Derivation of noise current into the receiver at theend of a coaxial transmission line is outlined in Fig. 2.It is based on the magnetic coupling between the centerlead and the shield. It can be shown that the mutualinductance between the two is equal to the(self)inductance of the shield[1]. The potential differ-ence between the two ends of the shield is determinedby the ground loop current and the resistance and in-ductance of the shield. The induced emf in the centerlead is equal to the voltage across the shield inductanceonly. The noise current into the receiver is the result of

the potential difference at the receiver end of the line.The shield resistance in relation to the characteristic im-pedance of the transmission line determines the magni-tude of the noise current into the receiver. In many casesthe ground loop voltage, vext, between the sending endand the receiving end is generated with a very low im-pedance. The ratio of the shield resistance to the shieldinductance is then a determining parameter for the re-ceiver noise current.

For shielded, balanced transmission lines, noise re-jection is improved by the common mode rejection of the

receiver (i.e., cmr/4). A principal role of double shieldingfor terminated transmission lines is to reduce further theshield resistance, rS. Figure 3 illustrates a transmissionline connection for analog signals with a very high dy-namic range (~ 5104), which has been proven in practice.Inductance of the shield can be artificially increased by

several turns on a ferrite core. The noise current in Fig. 3is given for a direct connection in place of Cb. A capaci-

tance, Cb, of 100-300 pF reduces further the shield cur-

rents at lower frequencies, and prevents unbalancing thetransformer due to the stray capacitance, C12, at high fre-quencies. Differential amplifiers are also commonly usedinstead of transformers, with somewhat lower rejection ofthe noise and crosstalk.

The transmission line case illustrates the importanceof a low shield resistance. The same conclusion can bereached, albeit in more complex geometry, for any Fara-day cage and, in particular, for any configuration wherefront end electronics is located in a shielded enclosure

attached to the detector. This is the case for almost allsubsystems in LHC experiments. Any gaps in the en-closures are particularly important. This is where thewell developed technology[2] of rf gaskets may haveto be applied. Special attention has to be paid to gal-vanic compatibility of the metals used, to ensure lowcontact resistances over the lifetime of the experiment.In particular, contacts with bare aluminum have to beavoided. Aluminum has to be chromate or tin-plated or,if that is not practical, a brush-on coating has to be ap-plied to contact surfaces.

Prevention of noise injection by ground loop cur-rents is usually more difficult than shielding against EMradiation.

4. POTENTIAL GROUND LOOPS IN A

LARGE DETECTOR SUBSYSTEM

Large detector subsystems have a large number ofconnections to the surrounding world for signals, moni-toring, cooling, power, etc., that if left to chance, a be-wildering network of ground loops will arise. Even incases where all signal transmission to and from the de-tector is digital, and via optical links, power and ser-

a) Single ended PA

b) "Common Mode"

c) Double Shield

* v in the shield = iext ( jLs + rs) for rs C12

Cb

C12Z0Z0 inZ0, Ls, rs

figure 2. Noise pickup in cables due to currents causedby ground loops.

Figure 3. Balanced transmission line with high rejectionof ground loop noise.


4/5

vices connections may create loops which will injectnoise currents into the critical contact areas of the de-tector and front end electronics enclosures. The solu-tion that offers some degree of control is to isolate, elec-trically, the detector. This is illustrated for the case ofthe ATLAS Liquid Argon Calorimeter in Fig. 4.

An objection is sometimes made that a large object,such as a cryostat, has a capacitance of severalnanofarads to the support structure and other sub-systems. However, at some intermediate frequencies,say 100 kHz, this presents several orders of magnitudehigher impedance than the resistance of a direct con-nection. Noise at frequencies lower than the center fre-quency of the signal processing chain is important, sinceit can be induced by various paths of currents throughnonhermetic shields, into the wide band stages of frontend electronics, such as analog memories and ADCs.

5. AN OUTLINE OF SOME ISOLATION

MEASURES

Figure 5 illustrates some of the practical configu-rations for communicating signals, power and varioussensor lines with the interior of a Faraday cage. Theintent of all of them is to divert any ground loop cur-rents into the shield (enclosure). In examples 2 and 3,the impedance of the connecting lines is increased by abalun transformer, or by resistors in each line where thecurrent in the leads is very low.

Figure 6 illustrates floating dc supplies for low volt-age (high power) and for high voltage (very low cur-rents).

Figure 7 shows the connection for multiple remotelysensed power supplies. The common can only be atthe location of front end electronics (to avoid makinginterdependent feedback loops).

6. THE QUESTION OF SAFETY GROUND

Prevention of ground loop currents, by increasingthe impedance of any loop as much as possible, leads tothe following guidelines:

All detector subsystems will be electrically isolated; There will be no connection to ground other than

Safety Network;

There will be no connection between different de-tector subsystems.(These have been adopted as the primary guidelines

in the ATLAS Policy on Grounding.)The goal of the isolation is to prevent numerous pos-

sible ground loops (illustrated in Figs. 1 and 4), to al-low checking for inadvertent connections to variousgrounds (i.e., objects which appear to be near zeropotential, such as the experiment support structure), andto allow for a safety connection to a single point with-out creating a ground loop.

cold

Accordion

warm

FEE

1. Lar cryo lines

2. HV supplies

3. Data (opt. fibers)

4. Cooling circuit (water)

5. LV supplies

6. Level 1 sums

~ 5103

channels

7. clock

8. slow control

(parameter input)

9. sensors

10. solenoid cryo & supply

11. mechanical supports

12. feedthrough heater (dc)1., 4., 11. insulators

2., 5. floating supplies, as in Fig.6

6. differential transmission transformers (balun or signal)

7., 8. opto-couplers (or transformers)

9. insulate sensors; different techniques at various receivers

10. solenoid line to be insulated, power supply floating

12. heater insulated, capacitors to pedestal, floating supply

~20m

~70m

1. Coaxial Cables

2. Power Supplies

3. Probes, HV

Shield connected to cryostat beforepenetrating Faraday cage

Short connection, low inductance

Performed on standard feedthroughs

Capacitors with short leads, close to cryostat lowinductance connection

Capacitors with short leads, close to cryostat,low inductance connection

No net DC Current in Balun to avoid saturatingFerrite (pass power and return). In magnetic fieldup to ~ 300400 gauss, use 3D3 type ferrite

R > 1 k can be replaced by L > 1mH whenno current flows

LAr

LAr

Cryostat

Cryostat

LAr

Balun

100 n

100 n

100 n

100 n

100 n

r

R

5 - 10 turns on

Ferrite core

10+

0

_

+

0

_

to floating

power supply

i = 0

B

1.27 cm

to electrode

Figure 4. Vital lines in the ATLAS Liquid ArgonCalorimeter (i.e., potential ground loops).

Figure 5. Rules for Entering a Shielded DetectorEnclosure (e.g., cryostat).


5/5

Once the subsystem isolation has been ensured, awell-defined safety ground must be established. To whatpoint?

In the case of the LAr Calorimeter, a dominant con-sideration is to preserve from EMI the Level 1 trigger

1. LV Supplies 100 H

Detector

Cs

Figure 8. Safety ground for an isolated detectorsubsystem with any copper connections to the level 1trigger electronics (and/or DAQ).

shielding and grounding in large detectors radeka bnl65912

Documents