Beams-doc-2461-v2 September 20, 2006 Rob Kutschke The data shown in this note compare 2.5 Mhz data at MI60N for states 20 (antiprotons from ACC to RR) and 11 (antiprotons from RR to TeV), before and after adding a high pass filter to the transition boards. Both raw data an injection flash data are shown. For reference the corresponding data for MI60S, which have unmodified transition boards at all times, are shown. The files are: tbt60n.pdf - State 20 Injection flash turn by turn data for 60N tbt60S.pdf - State 20 Injection flash turn by turn data for 60S raw60n.pdf - State 20 Raw data for 60N raw60s.pdf - State 20 Raw data for 60S state11_tbt_60n.pdf - State 11 injection flash turn by turn data for 60N state11_tbt_60s.pdf - State 11 injection flash turn by turn data for 60S ( In v1 of this document the files 60n.pdf and 60s.pdf corresond to tbt60n.ps and tbt60s.pdf; the only change is the addition of the note in the bottom left corner of each page. ) All files contain both before and after data. This note first discusses the state 20 data then the state 11 data. In the file raw60n.pdf the left hand column shows data after the modification to the transition board, taken on September 19, 2006 in the mid afternoon, and the right hand column shows data before the modifcation to the transition board, taken on August 23, 2006. The newer data set was taken by Steve Foulkes and can be found at: http://home.fnal.gov/~sfoulkes/acc.shot1.zip The older data set was taken by Bob Webber and is in his directory 8_23_06_pbars_to_RR_shot2. The top row of plots shows the raw A signal and the second row of plots shows the raw B signal. Note the timing differences between the two data sets; this is not important for noise studies. The newer data set has a timing chosen to always show at least one empty turn before the first turn with beam. Because of seam issues there are some BPMs that show only a single turn with beam. The older data set always shows 3 turns with beam. I will explain below, in a discussion about hp522, that the first turn is missing and that the older data shows turns 2, 3, 4. This is not ideal but it is the best data set that I have for making the before and after comparison. The before and after plots are drawn using the same vertical scale; but the vertical scale does change from A to B. The third row of plots shows a detail, between 0 and 5 MHz, of the Fourier transform of the raw A signal. The bottom row of plots shows the same detail of the Fourier transform of the raw B signal. All four Fourier transform plots are shown using the same vertical scale. The transition boards in house 60S were not modified between these two data sets. So inspection of the plots in raw60s.pdf allows one to compare beam conditions between the old and new data sets. This shows that the signal level in the new data is a lower than that in the old data. It also shows that the background levels are, with a few exceptions, about the same. The most notable exceptions are hp520, which has much more noise in the new data, and hp524, which has much more noise in the old data. The bpm hp522 is an interesting case that is discussed below. If we assume that 60S is representative then, for most BPM locations in 60N, the noise levels incident on the transition boards should be about the same in the before and after data sets. Inspection of raw60n.pdf shows that, at frequencies below about 1 MHz, the noise is much less in the new data set than in the data set. In a few cases some signficiant noise does survive in the new data set, although strongly attenuated relative to the old data set; look, for example, at locations hp616 through vp623. From this we conclude that the high pass filters are working as intended. In the file tbt60n.pdf, the top row of plots shows the sum signal, the position and the fft of position, for 3 pbar transfers from BPMs in MI60N. These data were taken on Sept 19, after the addition of a high pass filter to the inputs of the transition boards in MI 60N. The bottom row of plots shows the corresponding plots for 3 pbar transfers taken on Sept 11, before the modifications to the transition boards. The file tbt60s.pdf shows the correspoding plots for BPMs in the 60S house, for which no changes were made to the transition boards. The data for both sets of plots was taken by Steve Foulkes. The new data sets are located at: http://home.fnal.gov/~sfoulkes/acc.shot4.zip http://home.fnal.gov/~sfoulkes/acc.shot5.zip http://home.fnal.gov/~sfoulkes/acc.shot6.zip and the old data are located at: http://home.fnal.gov/~sfoulkes/acc.shot1.tbt.tweaked-timing.zip http://home.fnal.gov/~sfoulkes/acc.shot2.tbt.tweaked-timing.zip http://home.fnal.gov/~sfoulkes/acc.shot3.tbt.tweaked-timing.zip In both cases the timing on 601 and 602 is set to the "wrong" value so that the seam is in the wrong place but the data is valid. Inspection of tbt60s.pdf shows that the noise was slightly worse on Sept 19 compared to Sept 11. This is roughly consistent with the observations in the raw data; the noise in these plots comes from in-band noise, which can be hard to see in the raw data plots. Again, we can assume, based on 60S, that a comparison of the two datasets in 60N will tell us something about the effects of the high pass filter. Inspection of tbt60n.pdf shows that the addition of the high pass filter has not changed much; on face value the noise is slightly worse in the new data set but we know from tbt60s.pdf that this is due to the origin of the noise, not due to signal processing. So we conclude that there is no evidence that the high pass filters either help or hinder the response of the BPMs in this mode. There remains significant noise that affects the position measurements. The case of hp522 in 60S is particularly interesting. First look at the injection flash on page 3 of tbt60s.pdf. The measured first turn position is far from the mean position of the rest of the data; this single outlier also makes the fft look odd. The effect is present in both the old and new data sets. At first I presumed that this was some sort of noise. In fact this is the first bpm that sees beam from the injection line and I guess that, on the first turn, the beam has not yet been kicked onto the correct orbit. This bpm is one of the extra wide aperture bpms and its orthogonal view, vp522, is shown on the next page; that view looks perfectly normal. Now look at page 3 of raw60s.pdf, which shows the corresponding raw data. The left hand side shows that the beam really is present at the first turn, around tick 1000; the signal in the B channel is strong but that in the A channel is barely visible. Inspection of vp522 on the next page confirms this. Inspection of vp523 on page 5 confirms that the beam entered between vp523 and vp522, going in the direction of vp522. In an meeting last week I worried about the low signal in the first turn of the A channel of hp522; I understand it now. Now look at the phase of the low frequency noise relative to the last bunch in the left column and the first bunch in the right column. This suggests that the the old data is an injection measurement timed to see turns 2, 3 and 4. State 11: The state 11 turn by turn files 11 have the same layout as the state 20 turn by turn files. The upper plots show data from a single transfer after the change and the lower plots show data from two transfers before the change. The data show the same general features as that for state 20: many bpms exhibit single turn outliers, in particular bpms hp610 through vp617, inclusive. The effect is often seen most clearly in the vertical bpms for which the betatron oscillations are not as large.