- Before beginning all hardware should be setup as instructed by
the Hardware Setup Procedure.
The VXI crate should be booted and
the station's EPICS displays should be live. No modules or EPICS
panels should be purple. The
120 Watt drive amplifier should be ON. It's best (but not totally
necessary) to let the crate warm up 30 minutes with the shielded rack doors
closed to allow temperatures to stabilize.
- Bring up the main EPICS station panel
for the appropriate station by clicking the desired station button.
- Verify that the RF reference is present.
Bring up the Phase/Powers EPICS panel and verify that IQA module #1 channel
#6 is between 20 mw and 40 mW. If it is not, then measure the reference
power delivered to the rear of the shielded rack. Should be approximately
10 mW. If not check reference distribution system power levels.
If it is OK than the amplifier in the clock module or a bad cable is most
likely to be at fault. Repair as required before proceeding.
- Measure the 471.1 MHz LO power delivered
to each IQA module
(top SMA connector on each IQA) with a trusted power
meter. The power should be +13dBm +/- 3dB. If out of range
the clock module or SMA cables/connectors must be repaired or replaced.
Be sure to reinstall each LO cable to the same IQA module it came from.
The cables should be labeled. If labels are missing be sure to apply
new labels after sorting out to which module the cable should be connected.
- If the IQA modules have not been previously
calibrated or if a clock module has been changed, the IQA modules should
be recalibrated as instructed in the
IQA module calibration procedure.
- Verify that the bunch 0 turn clock is
present. The "fiducial missing" LED on the clock module should be
OFF. If the turn clock is missing the problem must be corrected.
The turn clock is generated by a PEPII PDU/STB module pair in a CAMAC crate
and cabled over to the shielded rack via a Heliax cable. The turn
clock is a 60 ns NIM level (must be terminated in 50 ohms) pulse repeating
at 136.3 kHz.
- Bring up the RF Processing (RFP) Module
panel from the Modu Diags panel on the main station EPICS display. On the
left side about 1/3 down from the top of the RFP Module panel
is a purple button with two possible titles, "DAC
Set Points" or "Module
Status". With the right mouse button
select "DAC Set Points"
to bring up the DAC panel. On the lower left of this panel is a button
labeled "Calibrate".
Activate the calibration procedure with a left mouse button click.
Note that RF must be OFF to allow the calibration procedure
to begin (prevents accidental calibrates during operation).
- The DAC calibration procedure will take
10-15 minutes. It first uses a binary search method to find the null
(0 gain) DAC offset setting for each of the analog multipliers present
in the RFP module by dithering the input to each multiplier and measuring
the output as the DAC setting is varied. This technique does not
work for the comb filters since each comb filter has a DC nulling circuit
on the input which removed the dither. The comb loop coefficient
offset is simply preset to a reasonable value (130 counts offset).
You can watch the progress of the calibration procedure on the DAC panel
and messages on the local ASCII terminal on top of the shielded rack.
Coefficient offsets should range from 110-140 counts. The procedure
also nulls out DC offsets in the RFP module analog electronics using a
Newton's method search technique. Typical DC offsets range are between
-1000 and +1000 counts, full scale is -2047-->+2048 (12 bits bipolar).
This calibration procedure is not bullet proof. If it fails try it
again after typing in 0 for the entries the program reports that it failed
to complete. Large offset values may indicate a hardware problem.
A successful calibration will report "Calibration
Complete". Unsuccessful calibration
will report "Calibration Completed With Errors".
- The next step is to prepare the station
for running RF power. On the Klys and Circ panel, energize the klystron
filament and solenoids using the ON/OFF button on the right side of the
panel. The solenoids will come on immediately but the filament will
ramp up slowly. There is a ~20 minute timeout on the filament which
prevents closing the station contactor until the filament has reached full
temperature. Never override the filament timeout without instruction
from a qualified klystron department expert. Applying high voltage
to a cold filament can ruin the klystron. After the filament has
timed out close the high voltage contactor using the button on the main
station panel or the HVPS panel.
- A non-configured station should be activated
with the feedback loops on the following state. You will find the
loop controls on the Feedback panel which is
reached from the main station panel. Set the loops as listed below:
Direct Loop - OFF
Comb Loop - OFF
Tuner Loop - OFF
HVPS Loop - OFF
Ripple Loop - ON
Gap FF Loop - ON
LFB Woofer - OFF
All other loop control buttons will not affect the
calibration procedure since they only activate under certain conditions.
More on these details later. Close the Feedback panel and return
to the main station panel.
- The ripple loop parameters must be checked
to insure they are properly setup. From the Modu Diags button
bring up the RFP Module panel.
Check that the following entries are set:
IMSK is 197 - Interrupt mask for the module.
DSPE is /dsp/rippleRfp - Current DSP ripple loop program.
(NOTE: Use the load/run button to the right to reboot the DSP.)
DDSF is /dat/ripple.dat - This file is written
to view DSP data. Not yet used.
On the left side about 1/3 down from the top of the
RFP Module panel is a purple button with
two possible titles,
"Ripple Loop Parameters"
or "Module Status".
With the right mouse button select "Ripple
Loop Parameters" to bring up the RF Processing
Module Ripple Loop Params panel. Check that the following entries
are set:
PHSG to 0.0050 - Gain
term for the DC phase across the klystron.
ACDF is /tbl/AmplCoefs.tbl - Amplitude
coefficients for the ripple loop DSP code.
PCDF is /tbl/AmplCoefs.tbl - Phase
coefficients for the ripple loop DSP code.
After setting any of these parameters, use the
"Load" button on the right to update the
hardware.
AMSP is the amplitude set point (gain)
for the baseband modulator used in the ripple loop. Its value is
set by the gain tracking loop and does not require manual configuration.
The gain tracking loop is configured when running the "ConfDirect" Matlab
task. Normally AMSP is configured to 0.9 volts with no beam and decreases
as klystron gain increases.
Close the RFP and the Ripple Loop panels and open
the main station panel.
- Open the HVPS panel from the main station
panel. If the station has not been run for a long period of time
or if the klystron is new, a low amount of high voltage should be initially
applied. Under these conditions the "Stn Turn-On Voltage" should
be set to 10 kV otherwise 45 kV is the nominal value. Return
to the main station panel.
- Verify that the drive power settings are reasonable following
instructions for the Main Station panel.
- Activate the station to the "ON_CW" mode.
Drive power should come up to the "Drive Pwr Setpoint" level and the HVPS
voltage should reach the "Stn Turn-On Voltage" in a few seconds.
If you are trying to high voltage process the klystron gun, let the station
run for 10 minutes at the 10 kV level and then raise the HVPS voltage in
10 kV steps until the 45 kV level is reached. Observe the klystron
ion pump current to insure the tube is not gassy. Klystron experts
should be involved in processing new tubes. Once the tube appears
to be processed move on to the next step.
- From the main station panel access the
More MATLAB panel. Activate the "Tune Cavs" Matlab task.
This will open a Matlab window and execute the tune_cavities.m Matlab script.
This task will take 10-25 minutes depending on which Sun workstation the
EPICS display was running on. The tuners will be configured and the
tuner loop will be activated. If the Matlab task hangs-up and appears
frozen a "Control-c" while the Matlab window is selected will kill
the task. Be sure to recheck the "Drive Pwr Setpoint" before restarting
the "Tune Cavs" task since it may have been modified during the tuning
procedure. Note that there is also a "Tune Cavities Directivity"
Matlab task on the More MATLAB panel. This task corrects for the
non-perfect cavity coupler directivity and has been demonstrated but should
be considered experimental at this point. Directivity correction
is only important when we get into very precise gap transient matching
and "Magic Detuning". This may never prove necessary.
- If the cavities are in need of processing
to improve vacuum level, the HVPS loop can be placed in "PROC" mode which
will slowly increase the klystron power as vacuum levels improve.
A klystron expert should set the limits of the processing loop following
guidance for the Main Station panel.
Cavity processing may be done in "TUNE", "ON_CW",
or "ON_FM" modes. The first two options perform processing with CW
RF power, while "ON_FM" plays modulation files through the built-in network
analyzer in the RFP module to scrub more areas of the system. If
necessary, processing with "ON_FM" can be done before the tune cavities
task to improve vacuum if faults prevent the tune cavities operation from
reaching completion.
- We are now ready to configure the direct
RF feedback loop. There should be no beam in the ring during this
configuration procedure. Make sure the station is operating in "ON_CW"
mode and the station gap voltage is set to the value you plan to run the
station at. This is important because the gain tracking loop, which
reduces the direct loop gain as the klystron gain increases, is initially
set to its full gain and the direct loop gain is then configured.
If the station gap voltage is altered then the klystron power will be different,
and the klystron gain will not match the configured value in the gain tracking
loop.
Once the station gap voltage and drive power is set, you can
start the Matlab task to configure the direct loop. From the main
station panel access the More MATLAB panel. Activate the "ConfDirect"
Matlab task. This will open a Matlab window and execute the configure_direct.m
Matlab script. This task will take just a few minutes to complete.
It will find the proper gain and phase to achieve the preset phase margin
which is currently 66 degrees (desired_margin in configure_direct.m Matlab
script). The gain tracking EPICS loop will also be configured.
Both lead and integral compensation will be activated. The EPICS
task which closes the direct loop will initially offset the gain by the
initial gain offset (nominally -10 dB) and increase the gain up by the
increment setting (nominally +1 dB) until at the desired gain. In
the end it will measure the closed loop response of the damped cavity and
plot on top of the open loop response.
After the task has finished
we have been manually reducing the direct loop gain slightly. On
the Feedback panel change the Loop Gain on the upper left to 15.0 dB.
Be sure to hit return to record your entry. Direct loop is now all
set.
- At this point we could make an equalizer
for the comb filters and the woofer. There is a Matlab task "Make
Equal" which calls the Matlab script, make_equalizer.m . It should
not be necessary to repeat this task since the equalizers are made with
cavity detuning corresponding to full beam loading. The equalizer
is determined by measuring the phase response through the comb filter path
while the comb filters are set for a "through" response (no comb filtering).
There are, however, some constants in the make_equalizer.m script which could
be modified (such as roll off frequency) if there was a need to optimize
performance in the future. The parameter "added_ticks" in the Matlab
script is used to center the filter response across the 32 taps available.
I suspect it will not be necessary to alter the existing equalizers but
wanted to mention the possibility.
- We are now ready to configure the comb
filter loop. The station should be running at the desired gap voltage
and drive power with the direct loop operating. From the main station
panel access the More MATLAB panel. Activate the "Conf Comb" Matlab
task. This will open a Matlab window and execute the configure_comb.m
Matlab script. This task will take just a few minutes to complete.
It will find the proper gain, phase and one turn delay setting (delay is
on the comb filter module panel) to achieve 17dB loop gain (desired_gain
in configure_comb.m Matlab script). In theory, this gain could be
raised to 20dB but so far we have found it unnecessary to do so.
Comb delay can be set in 25ns increments. The configure comb task
(unlike configure direct) does not close the loop when the configuration
is done (but it could).
- Close the comb filter loop by using
the left mouse button on the Feedback panel to flip the comb loop button
from Off to On. The EPICS task which closes the comb loop will initially
offset the gain by the initial gain offset (nominally -20 dB) and increase
the gain up by the increment setting (nominally +5 dB) until at the desired
gain. The comb loop is now operating.
- To check that the RF feedback loops
are configured and operating properly, one can measure the closed loop response
of the RF cavities. With the station operating (with or without beam),
the Matlab task measure_direct_closed.m can be used to verify the performance
of the feedback electronics. From the main station panel or the feedback
panel access the More MATLAB panel. Activate the "MeasDirCls" Matlab
task. A small 50 ms burst of band limited noise will be injected
into the RF drive to measure the cavity impedance. The result will be
plotted and can be compared to other stations. Basically, it should look
like a rounded hump with notches removed by the combs. If beam is
present, revolution harmonics will appear between comb filter notches.
- At this point the station RF loops are
configured with the exception of the woofer (see the
Woofer Configuration Procedure.)