Phase jumps in FLASH

A. Winter, B. Steffen

Overview

• Presentation of observed phase jumps

• Conclusions?

•Direct measurement on 1.3 GHz

•On downconverted synchronization signal of various laser systems

Where were measurements taken?

• EOS-hutch– Separating transformer = separated ground – DC-blocks for RF– Star-topology grounding concept– General care was taken to provide low-noise

environment

1.3 GHz transients• Measured in EOS hutch with Agilent SSA

•Frequency jumps of 20 kHz pp = phase jump of 6 deg pp (or 12 ps pp) can be observed 3 ms before the bunch and 2 ms after the bunch

A little more zoomed in…

• 1.3 GHz slope was subtracted from measurements

• Where does this come from?

• A shoot at an explanation later….

This translates into phase noise…•Red: how it should look

•Black: how it is during a disturbance

= PROBLEM !

How does this transform onto the laser synchronization signal

•1.3 GHz from machine is mixed with 1.3 GHz harmonic of laser repetition rate

•Mixer output signal is used to lock repetition rate of laser and observed on a scope

•Very sensitive measurement for transients (6 mV/deg phase)

measurements

bunchPeak at –3 msPeak at +2 ms

+800 us

125 us

-800 us

-1.5 ms•Modulators on•No RF

•Machine running

•Peak 3 ms before bunchis synchronousTo timing systemPre-trigger

Zoom in on the bunch•Things get around a factor of two worse, and that in the most critical time period!!

Conclusions• We see essentially everything on the 1.3 GHz RF with

significant degradation of synchronization performance for the various laser systems

• DC blocks are a somewhat double-edged sword:– They prevent the influence of ground loops,– But they also prevent common-mode reduction when systems

follow each other• Some noise sources could be identified, but how to get rid

of them ???

• Suggestions always welcome!!!

Some Remarks on Shielding

Herbert Kapitza (FLA)(using slides from a talk by Mike Thuot)

DESY, 09.10.2006

Shields are either used to confine the radiated field from a noise source or to exclude radiated noise by reflecting and/or absorbing the energy. The latter is the interesting situation in the FLASH injector rack area.I was asked: “How good a shield is 0.4 mm Cu sheet?”Well, that depends ...

• Shields are metallic partitions used to control the propagation of electric and magnetic fields.

• Basic shield properties:- material (σ, ε, μ; index r means w.r.t. copper)- geometry (thickness, shape, orientation, holes ...)• Derived shield properties:- skin depth δ = sqrt 2 / ωμσ - impedance |Zs| = sqrt ωμ / σ = 3.68e-7•sqrt f μr/σr

Shield Properties

• Basic radiation properties:- source type (e.g. electric / magnetic dipole radiation)- distance from the source (i.e. near / far field)- frequency- intensity• For shielding calculations it is convenient to use a

derived property, the wave impedance, Zw = E/H.

Radiation Properties

As a wave propagates through a material, the impedance of the wave, Zw = E/H, approaches the intrinsic impedance of the material. In vacuum Z0 = sqrtμ0/ε0 = 377 Ω.

High impedance sources

Low impedance sources

field impedance determined by source characteristics

field impedance determined by characteristics of the medium

In the near field, the electric and magnetic fields must be considered separately since the ratio E/H is not constant.

corresponds to 5 MHz500 kHz50 kHz5 kHz at r = 10 m

Shielding effectiveness is a measure of the reduction in magnetic and/or electric field strength caused by a shield.

The incident wave is partially reflected from a metal barrier at each interface, with a reflection coefficient that depends on a ratio of wave impedance to metal impedance. Inside the metal, the wave is attenuated at a rate of ~9 dB per skin depth.

R = 20 log |Zw| / 4|Zs|

A = 20 t/δ log e= 8.69 t/δ

R and A in dB:

The reflection loss is dependent on the type of field,frequency and the wave impedance

In copper, reflection loss for E fields is >> than for H fields in the near field

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅+=

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅+=

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅+=

rfR

fR

rfR

r

rm

r

rpw

r

re

2

23

log106.14

1log10168

1log10322

μσ

μσ

μσ

Far field (plane wave) reflection loss is greatest at low frequencies and for high conductivity materials.

Shield impedance is minimized by using materials with high conductivity and low permeability, so steel has much less reflection loss than copper

An EM wave passing through an absorbing medium is attenuated exponentially.

The decay is the result of ohmic losses by induced currents.

Skin depth in a material depends on the frequency, the conductivity and the permeability.

Skin depth is the surface thickness of a metal at any frequency for which 1-1/e or 63.2% of the current is flowing. Two skin depths = 86.5% and three skin depths = 95% of the total current flow. 99% of a current flows on a conductor’s surface within 4.6 skin depths

Absorption loss increases with frequency and shield thickness.

Steel offers more absorption loss than copper of the same thicknessA thin sheet of copper provides ~0 absorption loss below 1 kHz

Total shielding effectiveness (total loss) is a combination of absorption and reflection losses.

Total shielding effectiveness for 0.5 mm copper sheet

The reflection loss decreases with frequency since shield impedance increases with frequency. The absorption loss increases with frequency since the skin depth decreases with frequency.

Qualitative summary of solid shielding materials(no holes; no seams)

f [Hz] A [dB] Rwave [dB] Re [dB] Rm [dB]5,0E+01 0,4 151,0 251,0 51,61,0E+02 0,5 148,0 242,0 54,61,0E+03 1,7 138,0 212,0 64,61,0E+04 5,3 128,0 182,0 74,61,0E+05 16,6 118,0 152,0 84,61,0E+06 52,6 108,0 122,0 94,61,0E+07 166,3 98,01,0E+08 526,0 88,01,0E+09 1663,3 78,0

t 0,4 mmmur 1 Cu:1 steel: 1000

sigmar 1 Cu:1 steel: 0,1r 10 m

Back to the initial question: How good are 0.4 mm Cu sheet? If the nearby modulator is the main noise source for theinjector rack area, we are about 10 m away. This is in thenear field for f < 5 MHz. Even for 50 Hz magnetic fields theshielding effectiveness is still „average“.

Summary of Shielding• Before designing a shield, learn as much as possible about your noise!• Magnetic fields are harder to shield against than electric fields.• Reflection loss is very large for electric fields and plane waves but can be small for low frequency magnetic fields.• Use a good conductor to shield against electric fields, plane waves, and high frequency magnetic fields. Use a magnetic material to shield against low frequency magnetic fields.• A shield one skin depth thick provides approximately 9 dB of absorption loss.

Many illustrations for this talk were taken from:

• Ott, Henry W., Noise Reduction Techniques in Electronic systems, Wiley 1976

• Paul, Clayton R., Introduction to Electromagnetic Compatibility, Wiley 1992

• White, Donald R., Electromagnetic Shielding Materials and Performance, D.W.Consultants 1975