Transcript 1-shot
Background: Use of TKR Trigger One-Shots • individual TKR strip channels “true” when analog shaped pulse is above threshold. hit single strip analog pulse and threshold level • strips in a plane are OR’d together. binary strip signal (also resulting layer OR if this is the only strip in the layer above threshold) – The resulting layer OR’d signal is the subject of discussion today. A one-shot with programmable width can be turned on in the circuit. Currently in our testing it is off. – This signal is used in forming the “3-in-a-row”: when these signals from spatially consecutive xy*yx*xy are “true”, a TKR trigger request is sent to the global trigger. • the global trigger process is started by transition edges, not levels (this is good!). Consider a situation with a long-duration pulse (e.g., a heavy ion in the TKR): – – – • a trigger request level that persists longer than the single-event deadtime will NOT retrigger the instrument. the channels involved in a long-time level are effectively dead in the trigger when subsequent particles pass through them; that situation persists until the level goes away. (see next slides) in the readout, a ghost of the track persists in any subsequent triggers until the levels go below threshold. once a global trigger decision process (“window”) has started, the logic is levelsensitive, not edge-sensitive. (again, this is good) • Purpose of today’s discussion: what is the most likely flight configuration for the one-shots: on or off; and what is the stretch duration? We want to test in the flight configuration, so a timely decision is necessary. • [Many details glossed over here] 1 Why does this matter? • Because the opening of the trigger window is edge-sensitive, this issue is not very critical (LAT won’t continuously retrigger). However, we still must make a choice. • The first, best-guess length of time for the one-shot stretch is ~1 microsecond (see later slide). – no identified benefits, and some possible complications, making it longer • We have found three cases in which turning on/off the oneshots matters: – impact on handling of heavy ions – impact on noise triggers – possible impact on mip trigger efficiency due to jitter (see talk by Su Dong) • In all three cases, the impacts are fairly small but potentially important. – “choose your poison” – J. Russell 2 Heavy Ions • • • TKR signal from passage of a heavy ion will often persist (pileup) into subsequent events (~150 microsecond time over threshold, with 4-10 kHz of charged particles passing through the TKR that could make triggers). The heavy ion rate is ~50 Hz. The layers exposed to the heavy ion are dead for subsequent triggering until the layer-OR turns off, and the WHOLE TOWER is dead for subsequent retriggering until its trigger request turns off (no more 3-in-a-row condition). Thus, with the one-shots OFF, for ~6 ms every second (0.6% of the time) a fraction of the instrument (~20%) is not retriggerable. [The impact on the dead area in the readout is effectively nil due to the huge number of strips.] If the one-shot is OFF, the layer-OR signal will persist for future triggers. – this is potentially GOOD, because the trigger readout info in affected events [only those WITHOUT TKR TRIGGERS] will tell us this happened and we can then monitor the rate of event pileup. Also, we can tell in those events if at least some hits in a layer are ghosts from pileup. Also, the readout and trigger information will disagree less than if the one-shot is ON. • only useful for statistical monitoring at best – this is NOT GOOD, because use of the TKR information in the trigger logic in some cases will be incorrect. In particular, the geographic VETO (one of our two possible veto’s), when engaged, looks for TKR triggers coincident with lists of ACD tiles. If a subsequent event happens to have a hit ACD tile associated with the persisting TKR level from a previous heavy ion event, the subsequent event will be vetoed. The error rate will be small (VETO is only used for small CAL deposits so backsplash rate is low), and we would be able to monitor the rate of those incorrect vetoes using the pass-throughs. 3 Noise Triggers • Two categories: – without the one-shots, after a heavy ion, and AFTER the 3in-a-row condition is no longer true, some of the layers will still be in saturation for a brief period. During that window, noise hits in some layers will again make a 3-in-a-row, resulting in a trigger. – A noise pulse in the TKR has a very short time over threshold, typically, so the plane OR is nominally also short. However, turning on the one-shot will increase the chance coincidence of noise pulses. • Noise considerations: best choice is to turn ON the one-shots but to keep the pulse width short (~1 microsecond). 4 Summary Issue One-Shot ON NOISE TRIGGERS [clear winner if stretch One-Shot OFF short] [incremental TRIGGER/READOUT TKR INFO MISMATCH disagreement O(0.1)%] WRONG VETO DEAD TRIGGER AREA [<1% effect: layers INDIRECT PILEUP MONITORING [may also have TOT info?] TKR TRIGGER EFFICIENCY [w/stretch] dead instead of whole towers] [small] = better choice 5 Stretch Value • need slide from Su Dong • answer is 14 ticks? 6 Conclusion • Recommendation: We recommend turning ON the one-shots for nominal operations, with stretch of 14 ticks (0.7 microseconds). – still important to test the full LAT with the one-shots turned off to make sure it works as expected. Surprises on orbit may lead to the desire to turn off the one-shots, at least for systematic studies. 7