ILC EMI and bunch length measurements Gary Bower
ILC EMI and bunch length measurements Gary Bower, SLAC Nick Sinev, U. Oregon, speaker Sean Walston, LLNL July 20, 2006 ILC EMI & bunch length studies
Important Contributors • • • Ray Arnold, SLAC Karl Bane, SLAC Eric Colby, SLAC Joe Frisch, SLAC Doug Mc. Cormick, SLAC Marc Ross, SLAC Yasuhiro Sugimoto, KEK Mike Woods, SLAC Hitoshi Yamamoto, Tohoku University Japan-US Cooperative Program in High Energy Physics by JSPS July 20, 2006 ILC EMI & bunch length studies 2
Overview • • • EMI issues EMI measurements EMI and electronics Bunch length issues Bunch length measurements July 20, 2006 ILC EMI & bunch length studies 3
EMI - Introduction • Going back to the 70 s there has been concern about beam generated EMI affecting detector electronics. • SLD Vertex Detector electronics. • As part of the SLAC ILC test beam studies: – Make EMI measurements with antennas. – Expose the SLD VXD electronics to beam EMI. July 20, 2006 ILC EMI & bunch length studies 4
Beam line sources of EMI • Accelerator beam is usually enclosed in evacuated conducting beam pipe. – The beam pipe is thick enough to contain all wakefield radiation. • However, to monitor beam properties (location, emittance, current, etc) “gaps” in the conducting beam pipe are needed. – The instrumentation gaps may allow leakage of wakefield radiation into the ambient environment. July 20, 2006 ILC EMI & bunch length studies 5
Beam line & EMI antennas July 20, 2006 ILC EMI & bunch length studies 6
“old” ceramic gap & 100 GHz horn July 20, 2006 ILC EMI & bunch length studies 7
New ceramic gap & VXD July 20, 2006 ILC EMI & bunch length studies 8
Antennas • Used two AHSystems EMI antennas: – Biconical: Precision calibrated for 30 -330 MHz. – Log-periodic (“yagi”): Precision calibrated for 650 -4000 MHz. • However, both antennas are sensitive to roughly the same, much larger range. July 20, 2006 ILC EMI & bunch length studies 9
Antenna measurement technique • Antennas were connected via 50 ohm coax signal cable to a 2. 5 GHz resolution digital scope – Amplitude distorted signals observable up to ~10 GHz (scope can do 20 Gsamples/sec) • This enabled measuring E field signal shape and strength. • By orienting antennas polarization could be measured. July 20, 2006 ILC EMI & bunch length studies 10
Far field regime • Radiation from moving charges has a complex near field structure and a simple transverse EM wave far field structure. • Near field( E~1/r 2 ); Far field (E~1/r). • A signal 1 meter from a source is dominated by the far field regime for a 1 GHz signal (λ=0. 3 m). • In the far field regime, measure only E field strength and infer B field strength using the wave equation. July 20, 2006 ILC EMI & bunch length studies 11
Measurement results • The following slides present the results and interpretations of measurements that were made to characterize the beam induced EMI radiation. • All this analysis is PRELIMINARY in nature and subject to change with additional data and further analysis. July 20, 2006 ILC EMI & bunch length studies 12
Using time delays to find sources • Trigger on an external accelerator beam crossing signal. • Know the length (time delay) of cables. • Know antenna locations and compare relative signal strength at different locations. • Can determine the location of EMI sources along the beam line. July 20, 2006 ILC EMI & bunch length studies 13
Data: Typical waveform full scale: 50 ns – apparent frequency ~800 MHz July 20, 2006 ILC EMI & bunch length studies 14
Waveform shape implications • The waveform was very stable under a wide range of machine conditions (current, emittance, bunch length, etc). – This suggests the shape is determined primarily by the beam pipe geometry. • The waveform was stable against moving the antenna large distances from the source. • The amplitude ~1/r when antenna is moved. – These observations support making the far field assumption above for the signal frequencies observable with these antennas and this scope. July 20, 2006 ILC EMI & bunch length studies 15
Data: Absolute field strength • Signal seen on scope is attenuated by – antenna factor and cable attenuation. • Accounting for these we find an absolute E peak to peak field strength of ~20 volts/m at ~1 meter from the gap for a current of ~1. 5 x 10^10. • This result is approximate since it is based on a linear extrapolation for cable attenuation that needs further checking. July 20, 2006 ILC EMI & bunch length studies 16
FFT of sample waveform scale in GHz July 20, 2006 ILC EMI & bunch length studies 17
Power spectrum analysis • FFT analysis assumes infinitely long waveform. • Problems with finite length waveform FFTs. – Example: consider a fixed frequency signal, f, modulated by a Gaussian. Then overlap several such pulses all with the same frequency. An FFT will not recover f. • Try FROG or Wavelet analysis. (To do. ) July 20, 2006 ILC EMI & bunch length studies 18
Wakefield theory predicts • E field strength ~ bunch charge. • E field strength ~ (1/bunch length)1/2. • Radiation is very forward directed, however, gap diffraction has different effects on different wavelength components. July 20, 2006 ILC EMI & bunch length studies 19
Data: E vs charge • Plot shows linear relation as predicted. July 20, 2006 ILC EMI & bunch length studies 20
Data: E vs bunch length • No effect is seen in the ~ 1 GHz range. • There is an effect in the 100 GHz range (see later slides on bunch length measurements. ) July 20, 2006 ILC EMI & bunch length studies 21
Diffraction effects • Wakefield radiation at gap is very forward. • However, diffraction occurs at the gap. • Coherent radiation from a source obeys (gap width*divergence) ΔxΔθ ≈ λ/2. • For gap width ~5 cm: – λ~0. 3 cm (100 GHz) Δθ~1. 7 o (radiation remains forward). – λ~30 cm (1 GHz) Δθ ~170 o (radiation is no longer forward). July 20, 2006 ILC EMI & bunch length studies 22
Data: E vs θ • Scope limits frequency resolution < 10 GHz • Observed signals ~< 1 GHz • No polar angle dependence observed due to diffraction effects. July 20, 2006 ILC EMI & bunch length studies 23
EMI measurements summary • Typical waveform rings @ ~1 GHz for about 50 ns. • Absolute peak to peak E field strength ~20 V/m at 1 m at ~1. 5 x 10^10 current. • E is linear with current. • E shows no dependence on bunch length or polar angle in the ~1 GHz range. July 20, 2006 ILC EMI & bunch length studies 24
EMI and VXD electronics • An SLD VXD electronics module was placed near the new ceramic gap. • EMI antennas were placed at the same location. • The phase-lock loop signal circuit was monitored. – When working properly it asserts a DC voltage. When it fails it asserts 0 voltage. July 20, 2006 ILC EMI & bunch length studies 25
VXD phase lock loop drops Top trace: VXD board phase-lock loop signal Other traces: the two EMI antennas. Time offsets are due to cable length differences. July 20, 2006 ILC EMI & bunch length studies 26
VXD Observations • The PLL signal displays an EMI-like ringing signal at beam crossing. • The PLL signal sometimes drops to 0. • 20 -40 ns after the EMI waveform appears the DC signal drops to 0 in < few ns. • It always drops at the bottom of a wave cycle in the waveform. July 20, 2006 ILC EMI & bunch length studies 27
Cause of VXD failure • By various combinations of shielding cables and the VXD module, it is determined that: – The failure is not due to ground currents induced by beam image charges. – The failure is not due to amplifier overload. – The failure is not due to EMI radiation on the cables. – The failure is due to EMI radiation on VXD module. July 20, 2006 ILC EMI & bunch length studies 28
VXD failure rate vs EMI strength • The VXD module phase lock loop lost lock on about 85% of beam crossing when the module was exposed to ~20 V/m of EMI. • The VXD module lost lock about 5% when exposed to ~1 V/m of EMI. July 20, 2006 ILC EMI & bunch length studies 29
Bunch length measurement • The power spectrum radiated by a bunch is related in a non-trivial way to the length of the bunch: – Shorter bunch = more power, especially at shorter wavelengths – Longer bunch = less power overall, and what’s there resides at longer wavelengths • A ceramic gap is used to get signals out of the beam pipe • With the 300 micron bunch at ESA, we ideally want to look at millimeter and sub-millimeter wavelengths • By begging, borrowing, and stealing, parts were scrounged from around SLAC and an RF bunch length monitor was built at a ceramic gap in End Station A – Many thanks to Doug Mc. Cormick, Eric Colby, Joe Frisch, and Marc Ross for parts and technical assistance July 20, 2006 ILC EMI & bunch length studies 30
Three Frequency Ranges So Far… • X band: – WR 90: Cutoff Frequency = 6. 6 GHz – Low Pass Filter: 16 GHz • Ku band: – WR 75: Cutoff Frequency = 7. 9 GHz – Low Pass Filter: 23 GHz • W band: – WR 10: Cutoff Frequency = 59. 1 GHz 10 20 July 20, 2006 30 40 50 60 70 GHzlength studies ILC EMI & bunch 80 90 100 120 31
RF Bunch Length Monitor for ESA To 16 GHz and 23 GHz Diodes WR 90 Waveguide (0. 9 x 0. 4 inches) WR 90 Waveguide To 100 GHz Diode Ceramic Gap Beam Pipe Ceramic Gap ~8 cm WR 10 Waveguide (0. 1 x 0. 05 inches) To 100 GHz Diode Initially, there was too much WR 10 Waveguide signal in the 100 GHz diodes, so the horns were removed and the July 20, 2006 ILC EMI & bunch length studies waveguides retracted ~8 cm. 32
23 GHz Low Pass Filter 16 GHz Low Pass Filter Diode WR 90 -WR 75 Taper WR 90 Waveguide Diode Horn WR 10 Waveguide July 20, 2006 ILC EMI & bunch length studies 33
Measurement Electronics Gated integrator installed to allow ~2 ns gates for expected signals from diodes (Actual signals ~20 ns, so standard GADC being used for July run) DC output from gated integrators read by SLC control system SAM July 20, 2006 ILC EMI & bunch length studies 34
Slow Diodes 16 GHz 80 ns/div 5 m. V 23 GHz 80 ns/div 5 m. V 100 GHz Diodes Raw Diode Signals from 5 GS/s Scope 100 GHz Left Diode 20 ns/div 10 m. V 100 GHz Right Diode 20 ns/div 20 m. V July 20, 2006 ILC EMI & bunch length studies 35
A Few Observations • Slow diodes correlate well with toroid signal (bunch charge) • Left and right fast diodes highly correlated with each other • Fast diodes vary with damping ring phase and bunch compressor voltage suggesting they may actually be measuring something proportional to the bunch length July 20, 2006 ILC EMI & bunch length studies 36
Strong Correlation Between Left and Right 100 GHz Diodes July 20, 2006 ILC EMI & bunch length studies 37
100 GHz Diode Signal Bunch Compressor Voltage July 20, 2006 ILC EMI & bunch length studies 38
100 GHz Diode Signal Damping Ring Phase Ramp July 20, 2006 ILC EMI & bunch length studies 39
- Slides: 39