Spsc protodune-sp preliminary tdr review

2.4.3. Have the field cage profiles been finalized? What are the properties of the profile that has been selected for the FC? Given the profile and the distance to the nearest surface at ground, what is the breakdown voltage safety factor? Update this section of the TDR with description of the final design.
The field cage profile design has been finalized, following the recommendation of the DUNE Technical Board. The final choice is to use extruded Aluminium profiles with coated surfaces (SURTEC) to avoid oxidation. The choice was motivated by the the reduced weight and improved straightness compared to Steel. The breakdown voltage safety factor is 12kV/cm on the profile surface if there is no charge build up on the caps.

2.4.4. Has the CPA frame hanging scheme been finalized? Has the design for the CPA ground panels been finalized now? Update this section of the TDR with description of the final design.
The preferred top hinge is a metal one, which is being tested in the HV test setup installed the 35t cryostat. The ground plane tiles have been finalized. The latest design is essentially that described in the TRD; the only changes concern the holes for connecting the plates to the I-beams.
2.4.4. 1 kV/cm residual field in the gas is quite large! What is the acceptable maximum, and what determines this constraint? Re: “A FEA shows that the optimized overhang distance is 20 cm, provided that the surface of the LAr is 40mm above the GP. In this configuration, the maximal residual field is of the order of 13 kV/cm, with no greater than 1 kV/cm fields in the gas ullage at the top of the cryostat.”
The dielectric strength of argon gas is about 20% of that of air. The dielectric strength of air is about 30kV/cm at 1atm pressure. Therefore the dielectric strength of 1atm argon gas is about 6kV/cm. Considering that the argon density at boiling point is about 3 times denser than that of argon gas at 1atm and room temperature, the E-field requirement of 1kV/cm assumed for for cold argon gas seems to be safe value. It should be noted that double-phase LAr-TPC detectors (and protoDUNE-DP in particular) are designed to operate with electric fields in the gas phase derived from much higher voltages, e.g. the electric field required for electron extraction from liquid to gas exceeds 3 kv/cm.
2.4.5. What is the connection scheme for the last field cage profile to the cathode plane? Presumably this is near 180 kV. Has this been finalized?
The “last” field cage electrode, or profile #1 in our terms, is connected to terminals on the field shaping strip (-178500V) of the CPA through two flexible wires on the top and bottom modules. The connection between an end wall panel and the CPA is made through two flexible wires to the terminals on the HV bus at -180kV.
2.4.5. How is the integrity of the FC to APA ground verified during installation? How is this controlled under thermal cycling? (source of noise, QA)
The FC to APA connection is made by connecting pre-routed wires on the APA to the corresponding divider board terminals on the FC. The resistance of the last FC terminal to ground (APA frame) is well defined once connected to the wires. This will be verified by the person making the connection using a DVM, and verified by someone outside of the cryostat by the feedthrough. During cooldown, the resistance of each FC termination channel will be monitored by the slow control.
2.4.5. How is the beam plug secured to the field cage structure? How is thermal contraction mitigated in this support structure?
The beam plug has a mounting flange which is made from G10. It is bolted to a vertical FRP mounting plate connecting the two box beams holding the FC profiles. The G10 flange has very similar CTE to that of the FRP support structure of the field cage assembly.

2.4.5. What is the maximum power dumped in the event of discharge across the beam plug (165 kV!) How does this compare with the level at which damage to the readout electronics occurs (assuming capacitive coupling)?
The maximum power dumped in a full discharge is the stored energy of entire TPC, which is about 30J. The modular field cage units and resistive cathodes ensure the peak current injection to the readout electronics is at least two orders of magnitudes lower than that from a single conductive cathode.
2.4.5. A breakdown voltage of 30 kV/cm is given here, and compared with the maximum field around the beam plug of 15.7 kV/cm, but does this breakdown voltage account for impurities?
What if there is >ppb level of O2?
This breakdown limit is for ultra-pure TPC-quality liquid argon. In the literature it is demonstrated that the dielectric strength increases as one adds oxygen to the liquid argon. If the experiment has >ppb level O2, the field the experiment can withstand would be higher. However, in this case the corresponding electron lifetime of less than 300 µs would be insufficient for the operation of large LArTPC detectors (drift distance greater than a metre).
Also this is over quite a large area, while a factor of ~5 drop in breakdown voltage was observed over 1 cm2 area in FNAL tests for MicroBooNE.
The area relationship generally does not extrapolate; it was demonstrated only in a ball-plate geometry. A recent study using parallel plates (uniform field) and a larger areas gave results inconsistent with the area relationship in that paper; the field at breakdown was larger than that predicted by the relationship

Re 1.3.1. When is the beam plug test planned? If this has now happened, what was the outcome? Re: “To ensure that the displacement plug does not compromise the ProtoDUNE-SP operation, a dedicated HV test at Fermilab is planned that will test the final beam plug in the exact field configuration planned for ProtoDUNE-SP.”
The beam plug will be tested in the HV test setup in the 35 t cryostat in late spring of 2017.

2.5 TPC high-voltage (HV) components
2.5. The filtering scheme of the HV noise is mostly inherited from the studies performed by WA105. However, as noted by the authors, the filtering requirements are much more severe due to the sensitivity of the electronics (600 e- ENC). How will such filter be tested in realistic conditions prior to installation in ProtoDUNE? (In general: the schedule of the Q/A tests should be provided for this item and for all most critical ones.)
The HV filter system evolved from MicroBooNE, through to the 35-t TPC to the current protoDUNE-SP design. It was not influenced by the WP105 work. The filter system consists of two stages of simple low pass filter network. Assuming that the capacitance and resistance of its components do not dramatically change with voltage, the filtering efficiency can be easily measured by connecting a low voltage signal generator with adjustable frequency to the HV power supply plug and measuring the output of the filters. This test will be repeated after the TPC is assembled to verify that there is no unintended coupling into the cryostat. It is likely that the TPC will be powered up with a moderate high voltage in air prior to LAr fill, this is also an opportunity to evaluate the HV noise to the electronics.
2.5.3. For the HV monitoring, did you conduct COMSOL (or similar) simulations to understand the uniformity of the electric field and spark rates? Please update the TDR to provide detailed information about those simulations and the results obtained.
Electrostatic studies of the TPC HV components are summarized in dune-doc-1908. They are not included in the TDR in order the keep the document to a manageable size. We are not aware of any model that can predict spark rate so we are limit ourselves to a 30kV/cm electric field threshold.

2.6 TPC Front-end Electronics
General: this chapter should be updated to integrate the information from the DOE-DUNE internal review and from the Q/A tests at FNAL and BNL.
As stated in the reply to comment 5, we plan to update the TDR to incorporate new information, but the full review report goes beyond the scope of the TDR. The result from the FNAL/BNL integration tests will not be mature prior to completion of the TDR. We are happy to supply the SPSC with additional documentation

What is the requirement on the reliability of the electronics? Which fraction of channels must work in order to achieve the physics goals?
The DUNE requirement is for the cold electronics to be reliable for >25 years at cryogenic temperatures, far longer than ProtoDUNE operation. The cold front-end (FE) and ADC ASICs have been designed for low power in LAr and shown in prototype tests to meet the DUNE requirement. Tests of the other cold electronics components at LN2 temperature are ongoing, but all are expected to operate reliably in LAr for longer than ProtoDUNE operation, based on prior detectors such as MicroBooNE, LArIAT, and 35ton.

To achieve the precise event reconstruction necessary for the DUNE physics program the requirement is >98% of the active TPC volume is observed by all 3 active wire planes, which results in a requirement of <0.5% dead TPC channels per wire layer. For protoDUNE physics, the requirements are less well defined, but we expect a level of 1% dead channels will not impact the physics program (as stated in reply to comment 5ii).

The plan is to have all APA readout electronics channels fully tested and functioning during the installation with 0% dead channels. Any Front-End Motherboard (FEMB) with any malfunction or bad channels will be replaced during the installation. All FEMB will have been fully tested and functioning at both room temperature and LN2 temperatures multiple cycles before they are shipped to CERN. Further testing with all 20 FEMB on each APA at gaseous Nitrogen temperature will be done prior to installation.

Which are the most critical common failure modes of an FEMB and of a whole APA? Can the rest of the detector be operated normally if a whole group of wires is lost?
Two failure modes (start-up issue and ESD damage) of FEMB identified in MicroBooNE/35ton have been addressed in the new cold FE ASIC design. The layout of the FEMB has been updated to make critical signal paths such as the 2 MHz digitization clock more robust.

For the APA, the main risk of breakages is likely to be associated with “nicks” produced in the production/installation process (as stated in reply to comment 2.2.3), although the impact of a single broken wire on the surrounding APA is hard to predict. Each APA readout channel is relatively independent, if a whole group of wires is lost, the rest of the detector can be operated normally.

Would a short on any of the wires create a critical electrostatic force between wires?
The short on any of the wires from different planes will show extra current draw on the wire bias voltage line and higher noise on readout channels, it is not expected to have critical electrostatic force between wires.

2.6.2. What is the frequency domain of the wire and photon signals? Are they well separated from the clock frequency or other common noise frequencies in the signal? If the clock signal runs parallel to the power lines are you safely protected from pickup from the clock?
The wire readout digitizes at 2MHz with <500kHz bandwidth at 1us peaking time, while the photon readout is operating at 150MHz with >10MHz bandwidth. They are separated from the clock frequency (50 MHz) and common noise frequencies through the FE ASIC and cabling designs. All clock signals are transmitted differentially with individual shield to avoid the interference to power lines.

2.6.3. Amplifiers with 50 Ohm input impedance are used. Is this matched to the output impedance of the signals?
The FE ASIC has high input impedance; the sentence on page 2-51 is revised to “The amplifier circuit has a 22-nF coupling capacitor at input to avoid leakage current from protection diodes.”

2.6.3. The statement on p. 2-51 “Clamping diodes limit the input voltage received at the amplifier circuits fo between zero and 1.8 Volts” is inconsistent with the circuit diagram shown in Fig. 2.34. There the signal is clamped between 1.8V + U_D and -U_D, where U_D is the breakdown voltage of the diode (0.6V?). Please explain.
Good comment. The sentence on page 2-51 is revised to “Clamping diodes limit the input voltage received at the amplifier circuits to between 1.8V + U_D and -U_D, where U_D is the breakdown voltage of the diode around 0.7V.”

2.6.3. The statement is made that a very long time constant is chosen for the undershoot to make it small. This is possible, but it makes the undershoot very long as the integral is essentially fixed. Is there a risk that the undershoot piles up on the undershoot of previous events?
Spice simulation shows the 1ms time constant has <1% undershoot for a signal with 1us peaking time. So the risk of pileup is very low, even for above ground operation with a few tens kHz event rate.

2.6.4. What is the reason to separate the amplifier and the ADC into two independent ASICs?
The FE ASIC development was started before the ADC ASIC, and its design became mature early enough to be deployed in the MicroBooNE experiment in 2013. The original plan was to integrate the FE and ADC into one ASIC when both designs became well developed, however the ADC ASIC development took longer due to technical challenges.

2.6.5. Is a method to update the firmware foreseen? What protection is in place to prevent loading faulty firmware?
By default the cold FPGA will load firmware from cold flash onboard the FEMB. The remote update of FPGA firmware via JTAG chain is available through the WIB, with 4 differential cables to each FEMB from the WIB. In addition, the cold flash can be programmed remotely as well. All of these features will be validated in the QC testing at BNL prior to shipping to CERN. In the case that faulty firmware is identified in-situ by slow control and monitoring, an update of the firmware will be initiated.

2.6.5. Are the cables shielded individually?
The cold data cable is shielded individually for each pair. Currently the cold power cable is being tested in the test stand at BNL. If proven necessary, the cold cable bundle to each FEMB will be shielded individually.

Fig. 2.44, top right. Is this a continuous cable or is the feedthrough included in the measurement?
The connection through feed-through is emulated by proper connectors and test board in the measurement.

2.6.5. How large is the expected power loss on the LV cables?
The power loss on the 7 meter LV cables to each FEMB is ~0.1W at room temperature, or ~2W per APA, and will be further reduced when operating in LAr.

2.7 Photon Detection System (PDS)
General: a critical decision between the two systems should be taken in the view of the referees, in light of the challenging schedule.
There is no “critical decision” to be taken among the two versions of the PDS. DUNE has taken the decision to test both versions in protoDUNE-SP, each one covering half of the LAr active volume. Both are options for the far detector, but it is too early (and it is unnecessary) to make a definitive choice. The two PDS versions are identical in dimensions, readout, cabling and signal digitization. Also, the installation and test procedures are identical. The differences lie in some aspects of the detector design and in the materials used, which have different optical characteristics. ProtoDUNE-SP is a prototype and the goal is to compare the two options in real experimental conditions, before making a downselect.
2.7.2. The overall conversion/detection efficiency of the designs is not clear. What are they for the different designs? Why does ProtoDUNE-SP need two designs of the PDS?
There is not a “need” for two designs. The concept for the DUNE PDS has been developed over the course of a number of years and it assumes a solution with WLS and light-guide bars embedded in the APAs, coupled to SiPM photo-sensors. The two options to be installed in protoDUNE-SP provides different realisations of this concept. ProtoDUNE-SP provides the opportunity for the extensive comparison of the two solutions under real experimental conditions.

2.7.3. The detection efficiency (41%) is quoted at a rather large bias voltage of U_br + 5.0 V. At which bias voltage are the dark rates measured?
U_br at liquid argon temperatures is around 20.5 V. The characterization of SensL C-Series 6 mm2 SiPM response was performed through extensive tests. Setting the bias Voltage in the range 24.5 to 25.5 V was found to be adequate to reach the design single PE resolution and signal-to-noise ratio, and well within the over-voltage range suggested by the manufacturer. A dark rate around 10 Hz (average value) was found at ~4.8 V above U_br.
2.7.5. How does the alternative (third) photodetector fit in the test program?
ARAPUCA is a new PDS concept for a LArTPC, which is compatible with the concept of in-APA-frame bars. It is substantially different from the two reference PDS designs (light-guide bars). It can be adapted to fit all the space and geometrical constraints imposed by the integration in the APA frame (i.e. the “ARAPUCA bar” design). It can also use of the same existing PDS read-out electronics and cabling. The potential advantage is substantially increasing the light detection efficiency. ProtoDUNE-SP provides an opportunity to test this design under real experimental conditions and to compare its performance with the light-guide bar designs.
What kind of tests are foreseen to be run?
Two of the 60 slots for the PDS bars (light-guide reference designs) will be reserved for two “ARAPUCA bars”, identical in dimensions to the reference design(s) bars. The ARAPUCA bars will collect light signals during the test beam run. The data will be analysed to provide measurements of the detector performance (e.g. efficiency, light yield, single/multi-PE spectra), which will be compared to those from the light-guide bar design in exactly the same experimental setup.
What is the expected gain in efficiency?
An improvement of an order of magnitude in photon detection efficiency is possible.
Has the performance of the ARAPUCA arrays been demonstrated in the lab?
Yes, several tests have been performed at FNAL and at UNICAMP, the last being currently on going now at FNAL PAB. This tests provided the first proof of the concept, followed by more detailed characterization tests.


2.8 PDS electronics
2.8. Has the noise performance been demonstrated in the lab with a realistic setup (long cables, feed through, transition from cold to warm)?
The protoDUNE-SP PDS read-out has undergone extensive testing as the design has evolved (new cables, new connectors, front-end modifications). The most recent tests at ANL and CSU show the expected low noise level, 2-3 counts. The current ARAPUCA test at FNAL PAB (TallBo LAr test facility) also observed a similar low noise level, which is much lower than the single PE pulse height (16 ADC). This program of careful testing will continue, for example further tests will be performed when the new feed-through flange for protoDUNE-SP is fabricated.
The experience from the 35t prototype run at FNAL in 2016 has made us very sensitive of the potential issue. Here the noise level for the PDS was ~2-3 PE when the TPC cold electronics were off, and increased even more when the CE was operational. The current extensive test program is intended to mitigate the risks of a similar level of performance for protoDUNE-SP.
Table 2.7. I don’t understand how the requested resolution can be achieved with a 14-bit ADC.

In terms of amplitude: you want 0.25% resolution, i.e. 1 ADC count corresponds to 0.25% of a photoelectron signal and you need between 8 and 9 bit for the signal of 1 photoelectron. For a dynamic range of 1000:1 you need another 10 bits. Therefore 18 to 19 bits would be requested.

In terms of charge integration you save a few bits by measuring several samples of the signal pulse, but I doubt that 14 bits suffice. Please explain.
The 0.25% is the typo. Thank you for pointing this out, it will be corrected. It should be 0.25 PE (or 25% PE) as a requirement in Table 2.7. It is worth noting that for the 35 ton prototype readout we achieved better resolution: 16 counts (4 bits) for the single PE, and the rest of the dynamic range was 1000:1 with remaining 10 bits.

2.9 Data Acquisition (DAQ)
2.9. Why test two independent readout systems (RCE-based and FELIX-based)? What is the justification for this approach, particularly in light of the very challenging schedule?
It is too early to fix the final DAQ architecture and/or hardware for the DUNE far detector. This decision will be taken in 2019, allow us to take advantages of the most up-to-date technology developments. For this reason, the DUNE collaboration took the decision that the DAQ system for protoDUNE-SP would not a 100% prototype for the far detector. The RCE-based DAQ design provided a readily available solution, implemented and tested in previous/current experiments. Considering the very challenging schedule we adopted this solution. Even though the hardware implementation is unlikely to be used for the DUNE far detector, the architecture is one of the favoured solutions. The FELIX-based solution is an alternative DAQ scheme relying on commercial hardware. It adopts a different concept compared to the RCE-system. The DUNE collaboration agreed that, if ready, it would be implemented for the read out of one of the six APAs. The impact on the overall DAQ system is rather small as the existing downstream DAQ architecture and infrastructure for the RCE readout can also be used for the FELIX system.
The collaboration carefully considered the additional resources required to carry two systems and the potential schedule risks. Because the teams working on the RCE and FELIX systems (both of which have broader applications) are different, the additional burden on human resources is very limited. The collaboration is producing RCE-based readout for six APAs. If the schedule for the implementation of the FELIX system is delayed, the entire system will be read out using the RCEs. The collaboration believes there is a net positive impact of testing the FELIX based system in the real experimental environment of protoDUNE-SP.
2.9.1. It is stated here that cosmic data are also acquired at an appropriate rate. What is this rate, and what kind of trigger is used? This should be quantified in the TDR.
A cosmic muon tagger (CRT), composed by arrays of large area scintillator paddles (with X-Y strip read-out), will be implemented covering a large fraction of the front and back side of the Cryostat. The CRT will provide with a fast trigger for large zenith angle (nearly horizontal) cosmic muon crossing the TPC volume. The muon trigger rate can be adjustable, in the ~1 Hz to few Hz range, depending on the selected angular acceptance. In addition, the trigger can be prescaled.