TPC Ion Backflow Study

PAs of the study: J. Thomas (LBNL), G. Van Buren (BNL), with significant contributions from T. Ljubicic (BNL)

Note, in this document, I use the acronym IBF for ion backflow, the release of ions from the high gain wire region of the TPC through the gating grid out into the primary drift volume of the TPC.

Primary topic:
Operation of the TPC in a mode where the gating grid is not tied to triggers. This would have two potential benefits:
  1. Acquire data at rates higher than the maximum that can be delivered by the existing gating grid driver
  2. Allow readout of TPC clusters in the endcap regions of the TPC which are currently obscured by the delayed opening (and settling) of the gating grid
Currently, the gating grid cannot be decoupled from triggers (this would be work we expect Tonko might be able to do with more time).

In order to run this way, IBF must be managed by closing the gating grid by the offset in time that it takes ions (from the first event after gating grid opening) to drift from the anode wires out to the gating grid wires (shorter than that is fine). And it must stay closed at least as long as it takes the ions from the last event before closing to reach the gating grid wires (longer than that is fine).

Jim has done a back-of-the-envelope calculation to estimate that the ions take ~0.1 ms to drift from the anode wires to the cathode (or shield, or ground) wires, and then another ~1.6 ms to drift out to the gating grid wires. There is some room for uncertainty in factors such as:
  1. How far the ions can drift beyond the gating grid wires and still be pulled back
  2. What transparency of the gating grid is really achieved (e.g. 1 in 104 follows a ridge of stability and gets through?)
  3. The true drift paths & times in the non-uniform field around the anode wires
These are topics which could be explored by simulation. However, it is clear that simulation efforts will take a while, whereas we presently have some opportunity to take real data with the STAR TPC to gain further understanding of the existing device in situ.

Proposed test:
Tonko suggested a proxy for decoupling the gating grid from trigger: if the trigger can fire and accept TPC data immediately after the TPC dead time ends, then this will mimic operating the gating grid with a fixed rate.

To that end, a test can be run using only minimal limitations on the trigger. We probably all agree that some form of usual "minimum bias" trigger is a sufficiently open trigger, i.e. that we need not go as far as zero bias. If rates are on the order of 10 kHz, this means that we should see another trigger on average ~0.05 ms after the dead time ends (though we probably care more about the tails of this distribution than its mean). Getting a trigger with even higher rates is even more beneficial.

  • Trigger:

    It is important to recognize that the ~fixed gating grid opened and closed times is more important to this study than the kind of event that triggers the acceptance of data. It is true that we care about which events we use to try to see the effect of IBF, but cuts regarding which events to use in the analysis portion of the study may be applied offline. For example, a VPD coincidence may be used in offline analysis, but isn't necessary to have in the trigger. Input from trigger experts on which triggers to use would be appreciated. Present p+Al operations seem to show VPD coincidence rates more than x10 higher than ZDC coincidence rates (about ~300 kHz VPDx near the end of a fill vs. ~30 kHz ZDCx), so maybe VPD coincidence with no prescaling is a reasonable choice.
  • Detectors to include:

    Detectors desired would be:
    • TPX
    • BTOW
    • TOF
  • Scan settings:

    We propose to execute the test in two phases. The second, and main, phase is a scan of several different gating grid open time windows, using a fixed fraction of open/total time to control TPC anode currents, to get a systematic picture of the ion drift times which do and do not produce measurable IBF. A prior first phase would only attemp a couple time windows, to help in both (1) understanding the time scales and datasets involved in achieving the second phase (and consequently finalizing its details), and (2) practicing the procedure and tools.

    Each dataset used to look for IBF should include at least ~10k events, which should be enough to see the IBF if it is there in sufficient amount. This does not take very long. We propose to acquire data for at least 2 minutes so that anode currents can be measured, and online QA plots can be seen. In the below table, one can see that all proposed tests will acquire ~10k events in under half a minute. The time for performing the tests then is not constrained by the number of events, but instead by the number of settings we test.

    A preliminary plan for the second phase scan involves the following:

    Second Phase
    Run GG open time
    [ms] / [ticks]
    TPX dead time
    [ms] / [ticks]
    Time to 10k
    events [sec]
    A 0.043 / 400 0.05 / 470 6 standard operation, limited by 1800 Hz maximum GG rate
    B 0.522 / 4900 2.61 / 24500 26 DAQ (and GG) rate will be below 400 Hz
    C 0.480 / 4500 2.40 / 22500 23 dead times longer than this should be incapable of IBF, since the closed time (dead minus open) is >1.7 ms
    D 0.437 / 4100 2.18 / 20500 21  
    E 0.394 / 3700 1.97 / 18500 19  
    F 0.362 / 3400 1.81 / 17000 18  
    G 0.341 / 3200 1.71 / 16000 17  
    H 0.320 / 3000 1.60 / 15000 15  
    I 0.298 / 2800 1.49 / 14000 14  
    J 0.266 / 2500 1.33 / 12500 13 ions with 1.7 ms drift time should arrive after the next GG opening, thus maximizing IBF
    K 0.043 / 400 0.05 / 470 6 standard operation, limited by maximum 1800 Hz GG rate
    L 0.245 / 2300 1.23 / 11500 12  
    M 0.224 / 2100 1.12 / 10500 11  
    N 0.202 / 1900 1.01 / 9500 10  
    O 0.181 / 1700 0.91 / 8500 9  
    P 0.171 / 1600 0.85 / 8000 8  
    Q 0.160 / 1500 0.80 / 7500 7 at ~two closed cycles per 1.7 ms, this should also block much of the IBF
    R 0.149 / 1400 0.75 / 7000 7  
    S 0.139 / 1300 0.69 / 6500 6 1450 Hz is the maximum non-standard GG (and DAQ) rate we should see for the test settings
    T 0.043 / 400 0.05 / 470 6 standard operation, limited by maximum 1800 Hz GG rate

    An estimate for the entire operation might be:
    (1 minute setup time per setting + 2 minute acquisition time per setting) * 20 setting runs = 60 minutes
    [NB: This plan and time estimate have been revised based upon what we learned and experienced in the first phase.]

    The first phase would simply run a couple of these settings:

    First Phase
    GG open time
    [ms] / [ticks]
    TPX dead time
    [ms] / [ticks]
    Time to 10k
    events [sec]
    0.522 / 4900 2.61 / 24500 26 dead times this long should be incapable of IBF, since the closed time (dead minus open) is >1.7 ms
    0.341 / 3200 1.71 / 16000 17 ions with 1.7 ms drift time should arrive after the next GG opening, thus maximizing IBF

    We expect more overhead in setting up each run for the first phase, and we will want to run a little longer than 2 minutes per run to get a better look at anode currents and QA plots. An estimate for the first phase would then be perhaps:
    (2 minutes setup time per setting + 4 minutes acquisition time per setting) * 2 setting runs = 12 minutes
  • Anode currents:

    According to Alexei, anode currents during pAl operations are typically ~700 nA at the beginning of a fill, and near ~400 nA at the end of a fill. The anode current limits are set at 2000 nA. This implies that we could allow up to a factor of about x3 to x5 more anode current at the end of a fill than we currently do. The present currents are achieved under the condition that the GG is opearted at under 1800 Hz, with open times of 0.04 ms. This means that for every one second of operation, the GG is open for less than 72 ms, or less than ~7% of the time.

    Our proposal is to perform our tests near the end of a fill, where there is more room to allow anode currents to be increased. If we allow the GG to be open 20% of the time during our tests (to which all of the above test settings are designed), this represents a factor of ~x3 increase in open time over normal operation, and should lead to anode currents of ~1200 nA near the end of a fill, with some margin for going even higher if we start some time before the end of the fill.

    Alexei will be present during the tests to monitor the anode currents with us. This implies that the tests will be conducted during typical working hours.

Proposed analysis:
Look for unexpected changes in signed DCA distributions as a function of the scanned runs.