Microphonics Discussion
For LLRF Design Review
Tom Powers
13 June 2016
Not for release outside of JLAB
There are several MSWord documents located at:
M:\asd\asddata\C100Microphonics2016
LLRF Implications due to Microphonics
• Background microphonics. Requires more RF power and makes the
cryomodule more susceptible to trips due to outside excitation.
• Phase/RF Power signals in excess of about 60° / 12 kW will cause the
cavity to loose regulation and trip off.
• Microphonic transients driven by outside sources, such as trucks and
construction cause cavities to trip.
• When in SEL mode, with no gradient regulation, microphonics causes
perturbations in cavity gradients which can lead to ponderomotive
instabilities or trips/quenches if they are operated near the prompt
quench field or other gradient driven limits.
• “Excessive” energy jitter due to microphonics that is correlated from
cavity to cavity as well as, occasionally, from cryomodule to cryomodule.
Microphonic Mitigation Activities
•
Completed and documented.
– Log and review extensive data for.
• Background tuner motion during down.
• 0L04 trips during operations
• Background microphonics for C100/R100/F100
• Energy stability on cavity by cavity basis (synchronous in each cryomodule) for the south
linac as well as for the machine in Arc 1 and Hall A line.
– Preliminary tests on tuner stack damper
– Waveguide mechanical mode transfer functions SL22
– Waveguide/beam pipe/ion pump to cavity transfer function F100
•
Near Term Future
– Design install, test and optimize 2 or 3 waveguide constraint/damper designs.
– Choose the “best” solution and install on all C100/R100/F100 cryomodules.
– Test and optimize the prototype tuner damper on the F100 and repeat on a C100 that has
waveguide constraints installed.
– Investigate the 100 Hz noise and try to mitigate.
– Investigate and optimize the setup of the extra CM feet for reduced microphonics.
– Monitor a cryomodule during a thunderstorm using accelerometers and low power cavity
resonance monitor.
Example of Microphonics Driven Trip
0L04 Trip recorded in Feb. 2016.
•
•
•
•
Normal 20 Hz dominated microphonics for
extended period.
40 Hz Burst on cavity 8 followed by trip on
cavity 5
Similar events recorded for trips initiated
by cavity 1
40 Hz is a typical waveguide mode
Energy Jitter in Hall A Line
•
The energy jitter between 8 and 40 Hz is from the C100 cryomodules.
•
It represents about 30% of the energy jitter in the CEBAF.
•
While it may be OK for most physics experiments it does lead to moving beam spots in the arcs and can
not help the beam aperture.
•
This could be reduced if more gain could be applied to the cavity feedback loops at low frequency.
•
Also there is a pulsed energy instability leading to two beam spots during 5-pass tune up operations.
Waveguide Transfer Function Examples
Upper
• Strike WG 8 perpendicular to
beam line lower transition
flange.
• Measure acceleration at
Waveguide transition lower
flange (EW is in direction of
Beam line)
Lower
• Strike WG 8 perpendicular to
beam line lower transition
flange.
• Measure WG just before it
enters the vacuum vessel
Example of Tuner Damper Improvements
(There is Hope)
• In both cases the excitation
was an impulse strike just
above the beam line entrance
to the cryomodule up stream
end.
• Red is microphonic frequency
transfer function.
• White is acceleration transfer
function of the top of the
tuner motor along the beam
line direction.
• Upper is without a tuner
damper
• Lower is without tuner
damper struts.
• 12 Hz is the string mode for
the F100
• 9 Hz is the tuner stack
vibrational mode.
Known and Suspected* Modes
NL22 Data Taken in GDR mode on 2 Feb. 2016, no sandbags
*Waveguides 5, 8 *Waveguides 2, 6
*Down Stream Ion Pump
*Waveguides 7
“20 Hz” Half String Modes
*Waveguide 5
Single Cavity modes plus
*Various Waveguide Modes
“10 Hz” String mode
Tuner Stack
LCW
Microphonics and controls Spectrums
Transient Effects as Compared to Steady State Microphonics
NL22 with polybead-bags on Tuner Stacks 4 and 5
Full Bandwidth Shown
Area of Interest
Note: According to all published data +/- 20 Hz should be OK
Transient Effects as Compared to Steady State Microphonics
NL22 with polybead-bags on Tuner Stacks 4 and 5, Full Bandwidth Shown
Step is probably cavity tuner operating
Transient Effects as Compared to Steady State Microphonics
NL22 with polybead-bags on Tuner Stacks 4 and 5, 70 Hz to 500 Hz content Shown
• Same Scale as previous plot.
• It is not clear if this is when the tuner is running or just when it is starting to run.
Transient Effects as Compared to Steady State Microphonics
NL22 with polybead-bags on Tuner Stacks 4 and 5
70 Hz to 500 Hz content Shown
• Same as last slide with decreased vertical full scale, also slowing all cavities
• Note that 5 Hz is not considered an excessive amount of microphonics.
Output RF Phase Signal Full Bandwidth
• Same data set as the previous slides only it is the RF phase for the klystron
output rather than the cavity frequency shift.
• Note: A 60° phase shift is about all that the FCC can do without going unstable.
Output RF Phase Signal
70 Hz to 500 Hz content Shown
• Content between 70 and 500 Hz.
• Note the peak excursions due to the just 100 Hz component are almost big
enough to cause a fault.
• Also remember that the 100 Hz microphonics excursion was less than +/- 5Hz.
Concept for Frequency Dependent Gain
• The detune frequency is related to phase as follows:
𝝎𝑫
𝝋 𝒕 =
𝒔𝒊𝒏 𝝎𝑴 𝒕
𝝎𝑴
• Where 𝝎𝑴 is the frequency of the microphonics and 𝝎𝑫 is the amount
of cavity frequency shift.
• This means that the effect of the microphonics on the phase error goes
as 𝟏 𝝎𝑴 . Thus, before feedback, the effect of the 100 Hz perturbations
are 10 times less than the effect of the 10 Hz perturbations.
• Therefore one needs 10 times less gain at 100 Hz to regulate
microphonics induced perturbations to the same level assuming that the
two frequencies start at the same level.
• The coupler works in an opposite direction, which with high gain, means
that proportional gain feedback loop will increase the signal in order to
regulate the phase.
• It is suggested that, in addition to trying to reduce the source of the
higher frequency content that we experiment with different
filtering/gain schemes to maintain proper regulation without saturating
the “phase” loop because of high frequency content.
Needs/Issues For Dealing With Microphonics
•
Work to understand the source of the 100 Hz pulsed driving terms. If it is the tuner motor
investigate the parameter space for ramp speed, velocity and micro stepping rates to
minimize it.
•
Work to understand and eliminate the microphonics that seem to be driven (i.e. 30 Hz, 40
Hz, etc.)
•
Continue the effort reduce the susceptibility of the CMs to external vibrations.
•
If possible use the waveguides and tuner ports as a means to damp microphonics that are
introduced into the system.
•
Flexible frequency parameters in the gain loops. Expend efforts to reduce the gain at
higher frequencies with a goal of reducing excessive 100 Hz driven phase excursions while
improving regulation at microphonics frequencies. This implies more than just turning up
the integral gain term.
•
Separate out phase and magnitude loops, either though an I/Q phase rotation or by using
phase and amplitude control loops so that different gain / frequency domain parameters
can be applied to each.
•
Provide amplitude and phase error signals to both the DAC ports and EPICS scope tool so
that the errors are out of the bit noise on both. Provide an appropriate scaling on both.
•
Provide triggering synchronous to tune up beam so that the gains can be set to optimize
both CW and pulsed operation. Determine if feed forward is necessary/desired for pulsed
operation.
Backup Slides
Microphonics Spectrum Examples
2L24 one C100 < 4
2L25 C100 > 3
Note: Modes at 40 Hz, and 30 Hz are not modes of cavity string.
Examples of Pulsed Detuning with 90 to 150 Hz Content
RF Phase Shown, Data filtered BPF 70 Hz to 500 Hz
C100 < 4
C100 >3