|May 4, 1998||All That Fits is News to Print||Vol.11,No.2|
Page contact and owner at end of this issue.
February 20, 1998
|Author: Zelazny, Gromme, Hendrickson, Himel||Subsystem: PEP||User Impact: Small|
|Panel Changes: Few||Documentation: Yes||Help File: Yes|
At LEP a very useful and accurate method to measure the betatron phase advance using BPM 1000 turn data was developed by Pedro Castro. We have implemented this technique for PEP.
The basic idea is to get the beam excited at the tune frequency. For large enough beam currents this happens naturally. For smaller currents, it is done with the tune measurement system. Synchronized data for 1024 turns are collected. An oscillation with a frequency given by the fractional part of the tune should be evident (an amplitude of 1 to 2 mm gives a good measurement). The DSPs in the BPM processors fit sine functions of the form x=Asin(2*pi*i*q+phi) where i is the turn number and q is the fractional tune. The fitted amplitudes and phases (A and phi) are returned to the SCP for display. These fitted phases are the desired phase advances except for an arbitrary phase offset (the same for all BPMs, but different for different data sets). The measured phase advances on three nearby BPMs can then be used to calculate the beta at a BPM.
Typical measurement errors on the phase advances are about a degree. Betas are determined to a few percent. (If the phase advance between BPMs is 90 degrees then the error on beta gets very large.)
The technique can be (in fact it has been) used to measure beta beats so they can be minimized by tweaking a quadrupole. By measuring how the phase advances change when the RF frequency is changed, one can see how the chromaticity is distributed around the ring. By measuring how the phase advances change when the beam current is changed, one can measure the distribution of the reactive transverse impedance.
The BPM phase advance and beta function package gives one the ability to make all of these measurements online. It also allows data to be saved for further off-line analysis.
Note that measurements can be taken for x or y, but not both at the same time.
First setup the BPMs to read accurately. Make sure there is a recent calibration done with the correct bunch intensity. Make sure that if you have a single bunch you are reading in single bunch mode and that if you have a bunch train (at least 20 bunches spaced 4.1 ns apart) that you are reading in bunch train mode. If you are using a single bunch use as large a bunch current as allowed (presently 0.5 mAmps). This setup will get the best BPM resolution and hence the best phase advance resolution.
Next use buffered BPM data acquisition to acquire data for a BPM or two. I usually use BPMS PR10 8022 for the HER because it has both x and y readout. You want to get a fairly stable 1-2 mm peak-to-peak oscillation at the tune frequency. At low currents (roughly less than 100-200 mA) you will have to excite the beam as explained in the next section. At high currents the beam may be too unstable with an oscillation whose amplitude is varying significantly with time. At medium currents you may get a good oscillation without excitation. If so, use the tune spectrum analyzer to measure the tune frequency (in Hertz).
If you need to excite the beam, it is done with the source of the tune spectrum analyzer which is hooked into the transverse feedback's amplifier and kicker. First measure the tune. To see the tune peak it may be necessary to turn on the source in chirp mode. Make sure the source is switched into the amplifier. This is controlled from the SCP Tune Monitor Panel. Measure the tune frequency. Set this frequency as the source frequency of the analyzer. Change the source from chirp mode to CW mode. Increase the amplitude of the source (typically 10 or 19 dB works) to get a 1-2 mm peak-to-peak oscillation on the buffered BPM data acquisition. Note that smaller amplitudes may work; the measurement errors will be larger.
Now go from the PEP-II Index to the TUNING APPLIC INDEX to the PHASE ADV & BETA panel. This is the completely new panel that is used to measure the phase advance and beta function.
The 2nd and 3rd rows of buttons are used for setup. Two buttons allow you to select the BPM definition to use: e- or e+ in single bunch mode. If you have already setup the BPMs as described above there is no need to use these. If you start from this panel (as experienced users may), then these provide a short-cut for the BPM setup if you are using single bunches.
Next select whether you are measuring in x or y by using the SELECT X OR Y toggle button.
A button allows you to enter the BPM resolution for a single (unaveraged) BPM reading. This resolution is used to calculate the measurement errors on the phases and betas. It defaults to 100 microns. Apparent oscillations due to the ``21 Hz" problem do not affect the phase measurement and can be ignored when estimating the BPM resolution.
Next enter the tune frequency. This should be exactly the frequency (in Hertz) that you measured on the spectrum analyzer and may be using to excite the beam. Don't worry about whether the tune is more or less than one half or whether it was measured at some harmonic of the turn frequency. The software takes care of all that. Note that tune changes of up to 200 Hz have very little effect on the measurement; never-the-less, go ahead and type in all the digits of the frequency.
To acquire data, hit ONE SHOT DATA. This causes all the BPM processors to synchronously record 1024 turns of data. The DSPs in the processors then fit the data to a sine function and return the phase and amplitudes from the fits to the micros and from there to the Alpha and thence the SCP. Note that the actual data from the 1024 turns is never even read-out by the micro and hence this measurement is very fast (a few seconds).
Immediately a plot of the measured phases and amplitudes is displayed. The arbitrary phase offset is removed by forcing the first BPM in the line to have a zero phase. All phases are forced to be between -pi and pi. There is a toggle button to change displays from radians to degrees.
Many other plots are available. They each have HELP displays which give details about them. One of the most useful is PHImeasured minus PHImodel. If everything is perfect, this will be flat to within the measurement error. A beta beat shows as every other point being higher than the others.
The plot of the measured amplitude over the sqrt of BETAmodel provides a data quality check independent of the actual phase measurement. If all the BPM gains are correct and the model is correct, this plot will be flat. Once the accelerator agrees with the model, this can be used to measure the errors in the BPM gains.
There is a plot of BETAmeasured overlaid with BETAmodel. This is another way to see the size of beta beats and to see how well the accelerator agrees with the model. Note that only the betas at BPMs are plotted. Hence this display looks different than the one on the model optics panel. In particular, the maximum betas at IR2 are smaller. BETAmeasured at a BPM is calculated using its measured phase advance and those of two other nearby BPMs. The model phase advances at those BPMs are also used, so this calculation of beta is not completely model independent. Note that the measured amplitudes are not used so any errors in BPM gain calibration are irrelevant. The software selects which two nearby BPMs to use in the calculation by trying many different combinations and propagating the measurement errors. The pair that has the best combination of being nearby and having small measurement error is used. You can print the FITTED VALUES DISPLAY to see which BPMs were used in the calculation. (It is not visible on the screen.)
While the above plot usually looks very good, it is difficult to see small disagreements as the scale is large enough to show the large betas at IR2. The plot of BETAmeasured over BETAmodel overcomes this problem and makes it easy to see small errors.
Buttons in the bottom two rows allow you to save measurements to configuration files. These can be used for offline analysis (the HELP for SAVE CONFIG gives the meanings of the parameters that are saved) or loaded back in for more online analysis.
Having saved a configuration, one can load it back in with the LOAD-D button and look at all the above displays as though that measurement had just been made.
Using the LOAD-1 and/or LOAD-2 buttons allows one to look at changes in the phase advance. The PHImeasured minus PHIcnf1 and PHIcnf1 minus PHIcnf2 buttons provide these displays of phase advance differences. By doing two consecutive measurements and looking at the difference, you can estimate the measurement error in the phase advance. Note that the display shows the RMS of the difference to aid in this. If that disagrees with the error bars, perhaps the BPM resolution is different than what you entered. By doing measurements with different conditions (e.g. RF frequency or beam current) and looking at the difference, you can measure the distribution of machine properties (e.g. chromaticity or impedance).
February 10, 1998
|Author: Grossberg,Hendrickson,Shoaee||Subsystem: FBCK||User Impact: Small|
|Panel Changes: Few||Documentation: Yes||Help File: No|
In order to damp high frequency noise near 60 Hz, the LI29 fast feedback has been upgraded to include a timeslot control algorithm. Linac pump vibrations cause 59 Hz oscillations, and the timeslot algorithm is able to damp at that frequency. This feedback algorithm has been used in FB69 since 1996. The goal of the timeslot design is to damp at low frequencies and at frequencies near 60 Hz, without causing excess amplification of beam jitter at other frequencies. In preliminary measurements, the 59 Hz oscillation is damped by a factor of about 0.29. Additional study is needed to ensure the feedback design is optimal for all frequencies.
The label for the LI29 feedback loop has been changed from ``LI29 FBCK LOOP'' to ``LI29 TMSLOT LOOP''. The feedback now acquires and buffers data at 120 Hz, but feedback control is only performed at 60 Hz. The feedback calculations include the average and differences between measured values for timeslots 1 and 4 (i.e., the alternating 120 Hz pulses). The feedback includes ``pseudoactuators'' for timeslot averages and differences, before implementing separate physical control correctors for the 2 timeslots separately. Pulsed Amplitude Units are used to implement separate hardware controls for the alternating pulses. In order to damp the 59 Hz oscillations, the calculated timeslot difference pseudoactuator produces a 1 Hz wave as it beats against the 60 Hz calculation.
January 28, 1998
|Author: Phyllis Grossberg||Subsystem: Fast Feedback||User Impact: Small|
|Panel Changes: One||Documentation: No||Help File: Yes|
To combat power supply problems with the fast correctors in LI29, a new functionality has been added to Fast Feedback to enable users to limit actuator movement per feedback iteration. This ability is provided through a new button MAX ACT MOVMNT on the Fast Feedback CALB/DIAG panel.
A limiting value has been set for the LI29 feedback loop and it should be changed by knowledgeable users only. Currently there are no plans to use this functionality for loops other than LI29 but it can be and if it is, see the HELP behind the button for details on its use.
Do not use this functionality for loops with non-linear actuators.
April 16, 1998
|Author: G. White, M. Zelazny||Subsystem: Correlation Plots||User Impact: Small|
|Panel Changes: Few||Documentation: Yes||Help File: Yes|
There is a new release of Correlation Plots which is able to read BPMs and BPM-like devices from 2 measurement definitions in the same acquisition cycle.
The two measurement definitions are specified using the new features of the BPM Measurement panel. To do so, first press a region button and create a measurement definition for it in one of the five available buttons on the bottom of the panel. Then select another region button so that it is highlighted and press
also on the BPM Measurement panel. Select ``2ND-BPMD" from the menu (you may have to scroll down to see it in the list). You'll see then that the first active measurement definition button reads ``1st Mdef" in its 4th line. You will also see that the second active measurement definition button reads ``2nd Mdef" in its 4th line.
Now, in Correlation plots you can read devices in either region. If a device can be read in either measurement definition it will be read in the first active measurement definition. Note that, where previously a Correlation Plots variable required that a particular single measurement definition be set up, such as the PHYS variable for BPM orbit fitting and many others, the measurement definition to be used will be assumed to be the first active measurement definition.
Back to top of this issue
|April 16, 1998||Index Panel||indx112|
Translated from original PlainTeX by index2html.pl.
*Links followed by an asterisk are limited to SLAC clients only.