Superfast BPM Processor Scheme and First Results

Stephen Molloy, QMUL2nd Mini-Workshop for Nano-Project at

ATF

FONT at ATF

• Micron-level stabilisation of the ATF extraction line beam.

• Measure position of start of train, and correct the end.– To be accomplished within 56ns train.– Latency must be kept low.

• BPM processor should work in <5ns.

Processor Design

• Obtain difference from hybrid and mix with 714MHz from ATF control system.

• Low-pass filter with 200MHz cutoff– 5-pole Chebyshev chosen due to low latency, and

strong out of band fall-off.

Issues• Non-zero hybrid isolation causes zero-offset.

– BPM centre can be moved with variable attenuation if necessary.

• Stripline -> hybrid cables must be matched in time to better than ~50ps.– This is possible, and has been achieved.

• 714MHz LO input to the mixer should be very phase stable with respect to the beam.

• A lot of power at the beam bunching frequency.– Low-pass filter must limit this to a very small value

before signal reaches feedback amplifier.

Phase Stability of 714MHz

Simulated Output - 100μm

Comments• Positive points

– Simulated output is linear with beam position.– Hybrid common-mode residual shifts BPM centre in a

predictable way.– Imperfect cable lengths (within achievable limits)

merely shift BPM centre.– Mixer leakage easily reduced by low-pass filter.

• Potential problems– Relatively large amount of power at beam bunching

frequency remains after low-pass filter.– Predicted latency of entire system is ~6ns.

• Remember latency should be <5ns for feedback experiment.

Alternative Filtering

• 3 or 4-pole Bessel band-pass filter before mixer.– Centred at 714MHz, bandwidth ~400MHz.– Reduces 357MHz entering mixer.– Reduces out of band power entering mixer, thus increasing dynamic

range of BPM.• Less 357MHz means low-pass filter requirements are relaxed.

– Less poles results in faster filter.– 3-pole Chebyshev with ~170MHz cut-off.– Total latency should be equal to or less than the previous scheme as the

band-pass poles have a larger bandwidth than the low-pass poles

Simulated Response of 3-pole Bessel BPF

Simulated Output - 100μm

Simulated Output - 1μm

Comments

• 357MHz beam bunching is not observed.• Power at 714MHz

– Due to mixer leakage.– Level equivalent to DC output when beam has

1μm displacement.• Predicted latency ~4.5ns

– Longer by ~0.5ns when 4-pole band-pass filter is used.

Recent Beam Tests• Single bunch tests

– Verified cable lengths were correct.– Stepped through each component in turn and

verified signal.– Found correct LO phase.– Verified output was correlated with beam

position.– Measured latency.

Recent Beam Tests

• Multi-bunch tests– Calibrated each of three processors using

corrector magnets.– Recorded many extraction pulses to measure

resolution.

Reminder!

Single Bunch - Raw Signals

Single Bunch - After hybrid

Single Bunch - After Band-Pass Filter

Single Bunch – After Mixer

Single Bunch – Final Difference Output

Single Bunch - Corrector Sweep

Latency Measurement

• Triggered the scope with a sum signal.• Recorded raw stripline signal, and final

output of processor.• Results

– 3-pole band pass scheme – 4.2+-0.2ns– 4-pole band pass scheme – 7.3+-0.2ns– No band pass, 5-pole low pass – 6.1+-0.2ns

Multibunch – 3-pole band-pass filter

Multibunch – 4-pole band-pass filter

Multibunch – No bandpass, 5-pole lowpass

Calibration Run – BPM13

Bunch charge during Dec 9th shift