Attached is the signal mapping that will be used for the BEAVERTAIL test.

Note we plan to use 5 FEBEX cards to run 8 Si detectors because we will only use one pocket setup this time. In future, a solution using only 4 FEBEX cards will need to be found.

Bias Select: In general, higher bias resistors -> lower electronic noise. However, higher resistance = larger voltage drop due to leakage current.
Our natural leakage currents at operating voltage of 60V range from 18nA (MSPad 15A) to 320nA (Design I 13). At maximum expected rate of 100Hz, our signal induced leakage current is 
8.9nA. Hence our voltage drop range is 0.54V to 6.6V with 20MOhm resistors.
Since we are not electronically noise limited with GeV signals, we decided to choose the lower resistance in case any detectors have bad leakage currents to ensure a good comparison.

For Silicon detectors: p-side collects positive signals, n-side collects negative signals.

Finally, the expected signal energy for each detector is:
 - 1st Si pad -> 2.2GeV
 - 7th Si pad -> 2.5GeV
 - CsI scint  -> 1.3GeV
 - GAGG scint -> 0.9GeV
 - BGO scint  -> 129MeV
As a result, all Si preamps were set to 5GeV max range (MSI#1 Ch1-8 & MSI#2 Ch1-4). CsI & GAGG will use 4GeV range preamps (MSI#2 Ch7-8). BGO will use 330MeV preamps (MSI#2 Ch5-6).

Attached is the final MSI-8 preamp setup using the above settings.

Three terminals/processes are required to record data on a local machine.
1. log in x86l-131. 
	cd to '2021-beavertail/febex3-x86l-131'
	'resl' to clear, then run 'mbs'
	'@startup' -> initialises mbs by running startup script,
	'show acq' gets status, check if there are triggers incoming

2. new terminal on lab machine, cd to /data.local2/2022_beavertail
	to record data, run ./rawDispRFIO64-wheezy, then data record is listening
	bacl to mbs terminal, to send data, '@connect.scom' -> this a a connection script to connect mbs to RFIO
	@openfile.scom -> opens data file to record and automatically names them run####.lmd. 
	@closefile.scom -> closes file and stops file writing
	'show file' show connection to file, and whether its open/closed
	'type event' shows header, 1 words = related to data readout. -v shows full event output.

3. monitor with go4
	cd ~/beavertail/go4/CsISiPHOS_febex/
	source /u/litv-exp/go4login
	go4 x86l-131.hotstart

Our DAQ machines are x86l-131 (DAQ-1), x86l-124 (DAQ-2), x86l-128 (Time-sorter)

litv-exp has /beavertail folder for our testing.
x86l- DAQ computers have remote folder kept somewhere. These will run.

To login remotely: Log into lx-pool.gsi.de with litv-exp user:
ssh -XY litv-exp@lx-pool.gsi.de
From lx-pool you need another tunnel: ssh lxg1050
From there (or any other lxg-machine) you can do: ssh x86l-131

Beam: 208Pb(82+), 269.0 MeV/u, ~2*10^7 ions in ESR

Position pocket outside: GE01DD2AG (A= aussen, outside)
Position -126mm outside (very left), beam at ~-15mm

- BGO runs
electron cooler current: 20 mA (at ~2:00 am) (SC06): Moved pocket to -20mm
run005-033 with traces on. runs034-037 no traces rate is 1.2kHz
At ~2:35am (run038) changed e current to 100 mA, rate increase to 3kHz.: 
run038-080(?) no traces. run080-117 incl traces

- GAGG runs ~6*10^6 ions
Pocket at -20 mm, e current at 100 mA
Moved pocket to -50mm, 200 mA e current
run0114 - run0143 incl. traces
run0149 - run0157 no traces -> moving detector in/out
run0158 - run01?? cooler voltage changed from 100mA to 200mA

- large CsI
~2*10^7 particles in ring
100 mA e current
run0167 - run0197 incl. traces. Moved detector position from -70mm to -20mm in steps of 10mm
run0??? - run0210 no traces. Status detector pocket.

Attached is the detector arrangement for the 3 setups used during the tests.

Here were the f_user.c choices we implemented for this experiment.

Ln029: No white rabbit time stamp available so commented out. This needs to be done in the Go4 analysis script too.
Ln087: We used SFP 1 (random port choice) with 5 FEBEX cards (see signal map)
Ln104: We controlled traces.
Ln163: Channel control: we used 0x91000000 (all enabled,etc) for positive signals, 0x81000000 for negative signals.
Ln191: Sparsifying turned on for all channels.
Ln217: Trigger enabled as OR of pads 2, 4, and 6 + PD front and back: 0xef00, 0x0000, 0x00ef, 0x00ef, 0xc02a.
Ln257-261: Thresholds set to 0x85f as value that partially suppressed our pulser. Tbh, thresholds didn't seem to make an impact to our signal so we just left it at that.

In the previous signal map, I incorrectly mapped the DSSD channels. To correctly read out the DSSD for the new test, we will swap the Lemo cables into the MSI-8, rather then messing with the flange. Hence, plug the old labels into the new channels as outlined 
in the attached table.

Uwe and I were able to get some noise testing done before the ESR was closed with asbestos exposure. We used the new flange so didn't address any issues with the old flange. Our main findings were:
   1.	majority of noise comes from the connection to the ESR. Compare the two scope pics attached with a) the detector isolated from the ESR, and b) the detector in the pocket. It is clear that the large high-frequency spikes of amplitude 40mV plus some low 
frequency modulation is coming from the ESR.
   2.	The signal loop for the MS Pads is through both the MPR-32s and the MSI-8s, so ensuring those preamps are collectively grounded is paramount. Having a short grounding cable that connects all of them to the NIM crate containing the preamp power supply 
is ideal. Wrapping in aluminium foil also seemed to help.
           a.	Perhaps creating a Lemo to preamp ground signal would also help directly connect the flange to the preamps.
   3.	Shielding of the preamp cables seemed to help a bit, but not a huge amount. It may be worth investigating more the effects of shielding, but my feeling is that the above two sources are more important so shielding may not be worth the hassle.

Remaining tasks to investigate:
   1.	Try and quantify impact of isolation from ESR by wrapping mounting system in Kapton tape. If this makes a big difference, perhaps we can create something more purpose built.
   2.	Quantify impact of noise on our signal from data analysis to see if its worth all this effort.

MSI-8 module details.

MMPR1 card configurations:
   - MMPR1_2000_50_V62hf_HV: 2GeV range, 20/60 MOhm bias resistors. 50us decay time. Batch # 37_20: 1/18 - 16/18.                 Dillmann PLEIADES CFI.
   - MMPR1_1G_25_V62_HV:     5GeV range, 20/60 MOhm bias resistors. 25us decay time. Batch # 42_13: 13/54 - 17/54, 21/54 - 28/54. Old LISA modules.
   - MMPR1_20G_25_V62_HV:    20GeV range, 2/11 MOhm bias resistors. 25us decay time. Batch # 29_13: 33/36 - 35/36.                Old LISA modules.

MSI-8 1+2: All 2GeV cards.

MSI-8 3:   Positions 1-5 are 5GeV cards, 6-8 are 20GeV cards

MSI-8 4:   All 5GeV cards

Attached are photos of leakage currents for the current GAGG and BGO setups.
This setup uses the new flange (blue gas inlet), all pads are biased to +60V, the DSSD to +40V and the SiPD to +100V.
Dry nitrogen from and LN2 dewar was flowing over the detectors.

Attachment 1, GAGG housing. Contains:
   MSPad 17A       - 310 nA
   DSSD 161054     - 249 nA
   MSPad 17B       - 316 nA
   MSPad 17C       - 334 nA
   MSPad 16A       - 339 nA
   MSPad 16B       - 361 nA
   MSPad 16C       - 361 nA
   GAGG SiPD       - 293 nA

Attachment 2, BGO housing. Contains:
   Design-I 14     - 135 nA
   DSSD 121813     - 30 nA
   Design-I 13     - 583 nA
   MSPad 15B       - 78 nA
   Empty           - 
   Design-I 6      - 586 nA
   Design-I 8      - 518 nA
   BGO SiPD        - 162 nA

Last week we tested both the new and old flanges with the GAGG housing set up described in the previous ELog entry.

Performance of the new flange was excellent, with clear signal and low noise contribution. We didn't take any photos of this as it was so textbook.

The old flange we identified as performing poorly in June was installed and we carried out the same pulser tests.
This flange clearly creates a different noise environment and initially showed noise comparable/exceeding the 500mV pulser we were inputting.
A photo of the scope signal from one of the pad detectors is shown in Attachment 1.

We thought this could be due to the separate ground planes the MPR and MSI modules are on on the flange PCB so connected these ground planes together. 
This did not really improve the very noisy signal, but made the MSI-8 slightly worse. All-in-all, no significant improvement.

We noted an electrical connection between the D-Sub connector housings and the L-brackets on the flange, connecting the D-Subs, their signal and grounds directly to the ring.
Isolating them from the flange was as simple as applying some tape to the inside of the bracket.
I installed the old flange in the ESR with the GAGG housing and the noise is dramatically improved. There remains some significant wobble in the signal baseline

I have changed the configuration of Housing 2 (BGO) to account for the detector being fixed (bond wires) and provide a "plug-and-play" housing ready to go for any future experiments.

The detector ordering is as follows:
Slot 1 - Design-I #14 
Slot 2 - Canberra DSSD 121813
Slot 3 - MSPad 15B
Slot 4 - Design-I #13
Slot 5 - Design-I #8
Slot 6 - Design-I #6
Slot 7 - EMPTY
Slot 8 - Design-I #3

Design-I #14 previously fell out of its frame during tests at TRIUMF. Its leakage current is comparable with other Design-I detectors but if real-world performance is different in future this may be the 
cause and the detector should be swapped out.

The housing is also a little warped around the slots for the screws to attach the detector to the end of the flange arm due to overtightening of those screws. This makes inserting a detector into slot 5 very 
difficult for the first ~1cm and should be done with considerable care.

Whilst constructing a new flange to replace the old one, we did some tests again to compare the performance of new vs old flange.
See attached images:
1. Old flange, in the lab, no grounding attempts made, biased, n-sides.
2. Old flange, lab, ground plate connected to flange, biased n-sides.
This comparison shows that for the old flange, connecting the ground plane to the flange improved the 50Hz noise. This was also true in the ESR, so all following screenshots had this 
ground connection.

3. Old flange, in the ESR, grounded, biased, n-sides.
4. Old flange, in the ESR, grounded, biased, p-sides.
5. New flange, in the ESR, biased, n-sides.
Attachment 3 v 5 is an apples to apples comparison. This confirmed that even for the best case scenario, the noise in the old flange is an order of magnitude worse. Note that the n-
sides have the worse profile, and for the old flange, the high frequency noise is almost the same size as the pulser. For the new flange, that same noise can be seen, but much smaller 
magnitude.
Grounding on the new flange made no appreciable difference. But in general, we think connecting the electronic ground to the ESR is the best decision. 

We tested the new new flange using the same setup as in ELOG entry 14. The noise level seemed to be somewhere between the middle flange and the old flange.
See attachments:
1. Best noise conditions (grounding two planes and connection to ESR) on worst n-side pad.
2. Worse noise conditions (no grounding) on worst n-side.
3. Best noise conditions on a better n-side. We were unsure why some n-sides were different though.

Thus, for a final comparison of the flanges (best conditions on n-sides):
 - 2016 flange - noise scale = ~5V
 - 2022 flange - noise scale = ~0.4V
 - 2023 flange - noise scale = ~0.8V

MSI-8 and MPR-16 preamps have nominal differential output ranges of +/-1.8V and +/-1.2V respectively at maximum input (ie 5GeV).
All our FEBEX cards are now modified to accept +/-2V range. If serious pileup occurs, this could make it difficult to disentangle events for the n-sides.

To test whether out preamps satisfied this output range, we attempted to calculate the expected height from a 500mV pulser. This should have been:
 input height * (pulser coupling range / preamp energy range) * nominal output range

Mesytec noted that all coupling components are only tested to be within +/-10%, including the pulser coupling capacitors. Thus exact testing is challenging.
The pulser inputs of our MSI-8 cards are:
MMPR-1: Pulser inputs:
Type: 2GeV Pulser : 330MeV/V
Type: 5GeV Pulser : 330MeV/V
Type: 20GeV Pulser: 1GeV/V
Thus in the attached table, we can see the 2GeV and 5GeV cards are well within tolerance, but the 20GeV card is a bit outside. Thus we should probably expect a large output range from those cards.

We also tested the MPR-16 cards. However these cards do not have a terminated pulser input, and it is not clear to us what impact that would have. In the attached table, it seems to have a pretty consistent 
factor of 4 difference though.  

We took a few lmd-files with different settings of the febex DAQ to provide data for go4 development, when the DAQ is not running.
The files are located at lxg1048:/data.local2/2023_pleiades

timesorter_sparcifying.lmd - both nodes delivering data to the timesorter (sparcifying on)
timesorter_traces.lmd      - both nodes delivering data to the timesorter (sparcifying off)
x86l-124_sparcifying.lmd   - single node data (sparcifying on)
x86l-124_traces.lmd        - single node data (sparcifying off)
x86l-131_sparcifying.lmd   - single node data (sparcifying on)
x86l-131_traces.lmd        - single node data (sparcifying off)

comment: 
Both systems run on a 70 Hz pulser via 4 MesyTec preamps into 3 febex cards.
For unknown reasons x86l-124 delivers more data than x86l-131 when sparcifying is on. However, the threshold settings were not checked.