Damping Rings RF Users Guide
The RF system provides a synchronous accelerating voltage to make up for the energy individual particles lose due to synchrotron radiation and the energy a bunch can lose from higher order mode losses. The amplitude of the RF voltage and the magnitude of the losses determine an equilibrium synchronous phase angle. The particles will oscillate about this equilibrium phase angle according to initial conditions or perturbations to the system. The stability of these oscillations is a key concern.
In addition, the RF provides a longitudinal restoring force that keeps the particles bunched together inside the RF bucket. The bunch length will also oscillate about an equilibrium value depending on initial conditions and perturbations. At high intensities the stability of these oscillations is also a key concern.
The orbit circumference determines the revolution frequency of the bunch but since we keep the RF frequency fixed the equilibrium energy of the bunch adjusts itself to match this frequency. The bending magnet strength determines the bending radius for a given energy, so the configuration value of the dipoles will determine the equilibrium energy of the beam.
If the beam is injected away from this equilibrium energy it will oscillate at the synchrotron frequency (~100 kHz) around this energy. We can observe this as transverse position oscillations on a bpm in a dispersive region, or as accompanying phase oscillations when the bunch phase is compared to the 714 MHz drive frequency.
Similarly, the bunch will make synchrotron oscillations if it is injected away from the synchronous phase. Moving the phase of the injected bunch w.r.t. the 714 MHz determines this matching.
Phase and energy are analogous to position and angle in transverse phase space. Just as we need two correctors in a beam line to steer the orbit flat, it is necessary to optimize both launch energy and injection phase to remove an injection energy oscillation.
The bunch makes oscillations in both phase and energy. Since individual particles in the bunch have slightly different energy and phase they occupy a certain area in the energy-phase plane which we call emittance in the longitudinal phase space. The magnitude of the RF voltage determines the size of the RF potential well, or bucket, that should contain the bunch. Primarily, our concern is that the bucket be big enough to contain the bunch.
The orientation of the bunch within the bucket does determine whether there are bunch length oscillations (quadrupole mode at twice the synchrotron frequency). Typically, we are not concerned with this at injection, only later when we induce a bunch rotation just before extraction.
Instabilities manifest themselves as sudden beam loss in the ring or as jitter in the extracted beam. At high current the tolerance of downstream systems to jitter decreases.
Robinson Synchrotron Instability
The synchrotron oscillations of the bunch can grow or decay depending on the damping mechanisms present. One such mechanism is the tune of the main RF cavity. As the tuning angle is made more negative the damping increases. If the tuning angle is positive the oscillations are anti-damped.
Since we prefer to run the cavities at a zero tuning angle to maximize the gap voltage external damping is provided by a synchrotron feedback circuit..
Pi Mode Instability
When 2 bunches are present they can also make synchrotron oscillations in anti-phase with one another, in which case the centroid of the bunches does not oscillate and the above damping mechanisms are ineffective.
For this case special Robinson damping is supplied by a pi-mode cavity tuned tuned near to an odd revolution harmonic of the beam.
Bunch Length (Sawtooth) Instability
At high intensities not only the bunch position oscillates but also the bunch length. The bunch length oscillations can grow very rapidly but they cease once the bunch grows beyond a certain size. This gives rise to a characteristic sawtooth behavior as the bunch blows up and then damps down again.
This instability is avoided by ramping the gap voltage so that the bunch length never damps down below the instability threshold.
Beam Loading Instability
An effect of ramping the voltage down is an increase in the relative beam loading in the RF cavity. This can lead to a Robinson zero-mode instability where the beam begins to oscillate at an arbitrary frequency.
Beam loading compensation feedback (Flemming loop) reduces this effect.
In addition to the all-unit displays there are several summary displays:
Note - these displays only update every 2 minutes
Shows temperature and vacuum interlock status,
cavity gap voltages and forward and reflected power from each cavity.
klystron and power supply interlocks,
klystron voltages and power.
Status, timing and phases or gains of:
Gap voltage amplitude loop
Klystron phase COMPensation loop
Tuner loop status,
tuner positions and tuning angles,
status of bypass switch for beam presence signal.
RF Timing Display
Summarizes timing for all sample and holds
and gap voltage ramp.
For each ring there are two dedicated remote scopes with dedicated control system triggers for diagnostics of RF and beam derived signals
A remote switch allows a choice of signals to be displayed on either channel of each scope.
714 MHz Beam Phase
Shows the phase of the bunch compared to the 714 MHz reference frequency.
Features visible are:
Closure of the S-Band loop
Phase ramp and phase at extraction
Small oscillation just prior to extraction, accompanying the bunch rotation.
Instabilities when they occur
714 MHz Beam Phase Referenced to Cavity
Shows the phase of the bunch compared to the 714 MHz cavity frequency.
Cleaner signal for studying instabilities,
Does not show phase ramp.
Shows the phase of the bunch compared to the linac 2856 MHz reference frequency.
Synchrotron Sum Signal
Monitor signal from the obsolete synchrotron feedback module which shows the peak sampled BPM sum signal
Signal proportional to peak bunch current and therefore inversely proportional to bunch length.
Clearly shows sawtooth bunch length instability
Shows induced bunch length oscillation for bunch rotation prior to extraction
Cavity Gap Voltage
Output from an amplitude detector connected to the cavity pick-ups.
The gap voltage should track the signal from the controller and show features such as:
The voltage ramp
The bunch rotation "munches" just prior to extraction
Any saturation effects if they occur
Klystron Drive Control Voltage
Monitors the drive voltage to the klystron.
Signal is controlled by the amplitude feedback loop and so should show the same features as the gap voltage.
Deviations from this indicate saturation problems in the loop.
Peak Power Display
HP PK POWER ANALZR panel <- DRNG RSCOPE INDEX
Video output from the HP Peak Power Analyzer.
input can be switched to:
klystron forward and reflected power
waveguide to each cavity forward and reflected power
klystron drive power and vector sum of Flemming feedback signal
HP Spectrum Analyzer
<- DRNG RSCOPE INDEX
Video output from the remotely operable HP Spectrum Analyzer.
Typically connected to a BPM electrode to measure
tunes, sidebands etc.
Often specially configured for machine experiments such as
a high frequency receiver for the Tektronix DAs
Tektronix Digital Spectrum Analyzer
<- DRNG RSCOPE INDEX
Video output from the remotely operable Tektronix Spectrum Analyzer with real time digital signal processing.
Connected either directly to a BPM for tune measurements or
to a down mixer to analyze GHz signals from the HP spectrum analyzer.
Cavity Tuner System
The cavity tuners are controlled by a feedback loop, accessed from the COMP panel <- Feedback index <- RF index.
Sample and Hold
The cavity tune changes with beam loading and so must be sampled at the correct time in the store cycle:
The tuner control module has a beam presence input to prevent the tuner loop from responding when there is no beam in the ring.
A bypass to the beam presence circuit is available to allow the tuners to tune for special low intensity operation (e.g. FF wires)
Tuner position readback is calibrated in cm
The tuners can be put in manual and knobbed, but not trimmed, under special circumstances when a problem is suspected with the loop.
The tuning angle is read at two times during the beam store:
at injection, offset_angle_a -only this one should match the config
at mid-store, offset_angle_b
Gap Voltage Set Up
Refer also to the RF Cavity Display
Dual Sample and Hold System
The cavity voltage is sampled twice during the store to be compatible with the gap voltage ramp: once at injection when the ramp has its maximum voltage (A) and once midway in the store when the ramp is at its minimum (B).
On the summary display only A is shown.
TRIG 310 is set 0.5 ms after injection
TRIG 606 is an n-2 trigger and occurs 3.4 ms before extraction
Required gap voltage vs. readback gap voltage
The desired gap voltage is set by the gap voltage ramp controls. The gapv max and min are presently set only in units of DAC volts. When these devices are trimmed only the output of the controller is checked for the trim function. The gap voltages readback on the summary display are the actual cell voltages, but are No Control readbacks.
The gapv max is the default gap control voltage. Only when the gap voltage triggers are activated does the gap voltage ramp down to gapv min during the store cycle.
The klystron is capable of 60 kW peak in its present configuration with the ramp. The dual sample and hold reads power levels at the two times at the klystron and at the cavity waveguides.
Phase Set Up
see also memo by J.Judkins in DR Operating Manual.
Station Phase 1
Sets the phase of the damping ring RF w.r.t. S1
Minimize injection oscillations.
If grossly misset can result in beam loss, skipping buckets
S-Band Phase 7
Reference phase when the S-band loop closes
Look at 714 phase signal and minimize jump at 1.7 ms when loop closes
Reference phase at the end of the phase ramp prior to extraction, relative to the linac.
This phase is pulsed to allow the two time slots to be tuned independently.
Set according to beam energy/energy spread in BSY
Phase of RTL compressor w.r.t. to phase ramp
At the zero crossing there should be no energy deviation visible in the RTL orbit in the high dispersion regions.
In principle moves all phases together in the correct relationship
Set Up for High Current Stability
The stability criterion is the jitter tolerance in the linac and the occurrence of flier pulses. The RF setup must control the bunch length and beam loading
Gap Voltage Ramp
The gap voltage is set to maximum at injection for best capture of the beam. It is then ramped downward to prevent the onset of sawtooth instability. Just before extraction the voltage is ramped back up to reduce the bunch length, but done fast enough that the sawtooth does not have time to reoccur.
The procedure for setting Gapv Min is to observe the sawtooth on the RF bunch length, or SYNC SUM signal, and gradually lower Gapv Min until the sawtooth just disappears
Beam Loading Compensation
Beam loading compensation is provided by the Flemming loop. The loop stability is controlled by the feedback phase and the feedback gain.
COMP loop panel <- FDBK panel
The procedure for setting the Flemming phase AMPL 37 (Flemming Fdbk ON) is to range it and find the two values at which the beam goes unstable. At one limit the beam will appear unstable at injection and at the other limit the beam will appear unstable at the bottom of the voltage ramp. Leave the phase midway between these two values.
The feedback gain is controlled by AMPL 39. Changing it will also change the gap voltage so the drive gain must be compensated with by AMPL 40. These two can be changed in unison with the multiknob FLEM FDBK GAIN. Adjusting the feedback gain is NOT a standard tuning procedure please.
Intermediate Gap Voltage during Rate Limiting
This module provides an intermediate voltage to the cavities for the period of time when there is no voltage ramp during rate limiting. This is to keep the average RF power in the cavity the same and hence its temperature constant.
Procedure for setting AMPL 110 GAPV INTER
Check that thetiming for both GINSTA TRIG0414, GINSTP TRIG0415 and GV_STA TRIG0503, GV_STP TRIG0504 are set up on the correct beam codes on the RF TIMING DISPLAY
Run correlation plot vs. time of DR RF Temperatures (restore button file user_disk_slc:[pkr.btn]rftemps.btn) while the machine is toggled between rate limit states
Adjust GAPV INTER until the cavity temperature does not change by more than a degree or two a minute after the machine has changed rate.
Bunch rotation minimizes the bunch length just prior to extraction in order to reduce chromatic effects in the RTL compressor region.
Immediate consequence of correct bunch rotation is a reduction in RTL PLIC
The bunch rotation is performed by making two, short, downward pulses in the RF gap voltage (munches).
Procedure for setting up the bunch rotation.
The scope signal of the cell gap voltage should show 2 clean downward spikes about 6ms apart. If the 2 spikes merge into one this indicates a problem with klystron saturation.
The SYNC SUM scope signal should show a sinusoidal bunch length oscillation with 6ms period.
The 714 PHASE scope signal should show part of an oscillation whose period is twice that of the SYNC SUM signal.
Both the munch timing and the munch amplitude need to be optimized
Each munch has an on and an off trig and the two munches together are timed together with the ALL MUNCH MULTI timing knob (MNCHR CNTRL PNL <- RF FDBK PNL)
The ALL MUNCH MULTI should be adjusted until the crest of the sinusoid on the SYNC SUM signal coincides with extraction.
The munch amplitude, MUNCH Volts PAU AMPL DR13 21, is maximized looking at the pulses on the cell voltage and at the SYNC SUM signal, making sure to avoid klystron saturation and instabilities.
The individual timing of the munches can also be adjusted (experts only), where the width of much 2 and 3 can be adjusted as well as their separation. The procedure here is to give the beam 2 equal kicks a half synchrotron period apart so that the oscillation on the 714 phase is just canceled.
The status of the munch timing is seen on a summary display:
Klystron Power and Saturation
The klystron can deliver around 60 kW peak power, beyond which it saturates so that the output power and hence the gap voltage is no longer linear w.r.t. the drive power.
There are a variety of conditions which can cause the klystron to saturate:
the requested gap voltage is too high, GAPV MAX
the ramp is incorrectly timed so that the GAPV MAX is held for too long a period in the cycle.
the munch amplitude is too large
The presence of saturation during the store cycle can be diagnosed from a number of signals:
scope signal of the cavity voltages will show a distorted pulse shape, not reaching the desired maximum
scope signal of the drive control volts will be maxing out as the feedback loop tries to drive the klystron even harder.
HP Peak Power Analyzer of the forward power signals will show the klystron power at its maximum
Several different feedback loops interact with varying response times within the RF control electronics. Refer to the schematic diagram. Stable operation require that they all function correctly.
The slowest loop with a response time of the order of seconds relying on mechanical plungers.
Ensures that the cavity is tuned to the correct frequency in the presence of beam. A bypass switch allows the beam presence to be over ridden for low current operation.
The loop gain is set from the front panel of the Tuner Controller.
Its control signal is derived from a phase comparison of the 714 MHz waveguide pickup signal and the cavity pickup signal (loading angle).
Beam Loading (Flemming) Loop
Fastest loop operating at MHz around the klystron and cavity.
Feeds the cavity pickup signal back to the klystron drive to compensate for the change in loading on the klystron as the bunch goes through the cavity.
Stability is dependent on the remotely adjusted phase and gain of the fedback signal.
Operates with a response of around 10 kHz.
Ensures that the gap voltage follows the desired voltage function during the store cycle, but is not fast enough to respond to beam loading transients as the Flemming loop does.
The amplitude detected signal from the cavity pickup is fed into the Gap Voltage Controller module in the DR Alcove racks.
The gain of the loop is set on the front panel of the controller module
The control signal from the module can be remotely monitored. If the signal exceeds 10 V at any time during the store cycle this is an indication that something within the loop is saturating (the klystron).
S-Band Phase Loop
Necessary to launch the beam into the linac.
Phase locks the damping ring 714 MHz RF to the 2856 MHz linac S-band RF so that the beam can be put into the linac at a predefined phase.
The phase lock loop only closes 1.7 ms after injection at which point the phase ramp can then move the bunch over to the desired linac phase.
Klystron Phase COMPensation Loop
Adjusts the phase of the 714 MHz reference to compensate for phase changes that may occur across attenuators in the loop that are used for amplitude control.
When the loop is closed the KLY_COMP_MON phase (PHAS 24) should be zero indicating the loop is locked.
In the rare event that the loop loses lock because of radical tweaking, opening and closing the COMP Loop is usually sufficient to restore lock. PHAS 23 may need to be moved to bring the loop in range, but this is unlikely unless hardware has failed.
Provides external damping of the synchrotron 0-mode bunch oscillations
The beam phase measured from a BPM electrode is compared to the 714 MHz reference and fedback to a fast 714 MHz phase shifter.
The feedback electronics can distinguish between two bunches in the ring and hence the 0 and the pi-modes of oscillation. Only the 0-mode is damped in this loop.
Pi Mode Cavity System
Principle of Operation
The pi mode cavity is an additional passive cavity added to the rings to provide Robinson-like damping to the pi-mode (coupled) oscillation mode of the two bunches.
Pi mode is when the two bunches make synchrotron oscillations in the ring in anti phase with one another. Without the pi-mode cavities this oscillation can grow unchecked.
When two equal bunches are in the ring only even revolution harmonics are generated in the bunch spectrum. If pi-mode oscillations occur the symmetry is broken and odd revolution harmonics also appear.
The pi-mode cavity is designed to absorb energy from this unwanted mode by tuning the cavity to the lower synchrotron sideband of an odd revolution harmonic (tuning it to an upper sideband would make the oscillations grow).
The cavity is tuned in the same way as the main RF cavities with mechanical tuners, but there is no tuner loop for the pi-mode cavities.
Two Bunch Operation
The cavities must be tuned to the correct frequency:
8.5 MHz x 125 - 100 kHz = 1062.4 MHz
It is usually sufficient to restore the tuners to their configuration values, but the tune of the cavities can be measured as follows:
no beam in the ring
connect either the HP Spectrum Analyzer with growler, or a Network Analyzer, to the two signal cables connected to the two identical ports on the cavity
Set up the instrument so that it sweeps the frequency with a span of about 500 kHz centered around the desired frequency above.
Adjust the output level of the instrument and the attenuation on the input signal until a peak can be seen in the amplitude response.
Adjust the tuner position AMPL 25 PI_TUNE until the peak of the signal is at the desired frequency ± 10 kHz.
Disconnect the instrument before restoring beam to prevent damage.
Single Bunch Operation
For single bunch operation the tune of the cavity may be too close to the revolution harmonic of the beam so that it absorbs a lot of RF power, outgasses and trips on vacuum.
The solution is to detune the cavity to a lower frequency:
Move AMPL 25 PI_TUNE to a value about 0.5 less than nominal.