2019/06/05 (Wednesday)
| Setup | DIB1/3/4 → 3.2 MHz filters 2/3/1 → cable 3/1/2 → gender changer ?/3/2 → HVConn 1/3/2 → HV 1/2 → DCRCX(4)/DCRC55(1) DCRC33(2) → 1kV power supply → HV input |
|---|---|
| 0000 | Tower = 46.7 mK |
| 0855 | Tower/MC/Coldplate = 46.4/31.0/178 mK, Still = 1.02 K / 0.095 mbar, G1/G2 = 188/182 mbar, 1K Pot = 1.64 K / 24 mbar. MB = 74.9% coldplate and still are reading in their “low” modes for whatever reason |
| Tower temperature continues to fall linearly, slowly | |
| 0900 | Fill LN2 Cold trap bath |
| 1315 | Let's take a look at that shorted test SQUID again. Stop DCRC programs and power down DCRCs 1 and 4. Plug DCRC1 into DIB3 directly. |
| 1320 | Power up DCRC1. Disable DCRC4 in the ODB and start up DCRC programs. Tower spikes by 0.5 mK, then cools again. |
| 1322 | With SQUID settings from HV setup, can see pulses on DCRC1 PC and noise on DCRC1 PB,PD as expected. Set DCRC1 QET bias = 0/0/0/0. Change DCRC1 phonon trace length to 52.4288 ms. Zap DCRC1 SQUIDs. |
| 1345 | Tower/Coldplate/Still = 46.5/178/1010 mK |
| 1350 | Do a baby IbIs with DCRC1 pB. Looks nice and linear with slope of -16.9 mV/uA As described on the squid_temperature Run 70 summary, this implies a wirebond resistance of 5.0 mΩ. This isn't much different from the resistance inferred at ~4 K on 2019/05/25 by the same method. |
| Ib (uA) | Vo (V) |
|---|---|
| -10 | 0.5704 |
| -5 | 0.4856 |
| 0 | 0.4005 |
| 5 | 0.3172 |
| 10 | 0.2317 |
| 1400 | Just looking at PSDs on the pulse viewer, it looks like DIB1 PD (SQUID only) has very low, flat noise at around 2.91e-7 @ 1kHz (in some units? Not sure how the SQUID tuning GUI PSD is normalized). DIB3 PB has a flat spectrum at 1.44e-6 with R/L roll off as expected for Johnson noise. |
|---|---|
| 1406 | Start 100s series with 20 Hz in run randoms and 1000 BOR randoms, DCRC1 PB/PD QET bias = 0, and phonon trace length = 52.4288 ms |
| 1415 | Stop DCRC programs and power down DCRC 1. Connect it back to the HV board and restore correct DCRC1 settings in ODB. |
| 1430 | Power DCRCs and start DCRC progams. Zap DCRC1,4 SQUIDs. |
| 1550 | Tower = 46.3 mK |
| 1557 | Use the python code to generate a sequence to do 5x 300 V bias drift series |
python HVscan.py -iDCRC 2 -HVs 300,300,300,300,300 -HVcalFile HV_calibrations/HV_DCRC2_control_cals_440V_190603.txt -HVdriftRate 0.065 -HVpreBias 5/1 -tSeries 25 -DCRCs2Flash 1,4 > ../../sequencer/HVdrift_5x300V_3p9VPM_prebias_5_1_dur_25min_flashDCRC_1_4.msl
| 1605 | Just noticed that HV is 520 V and has been since 1430!. Eventually realize that DCRC2 is set as RevC.1 in the ODB, though all other settings seem correct. This probably happened earlier today when a second instance of the SQUIDTuningGUI was opened on jiajin, in addition to the one which was running on cdms2. Those are both now stopped. |
|---|---|
| 1610 | Play with DCRC2 QI bias to bring HV down to ~0V, then unplug HV from HV board. Stop DCRC programs, power cycle DCRC2. Change DCRC2 to Rev3.0 in the ODB. Restart DCRC programs. Play with DCRC2 QI a little bit, it seems to be working properly now. |
| 1619 | Set HV=0 and plug HV cable back into HV board. Channels still look okay, adjust some driver offsets. |
| 1620 | Fill LN2 cold trap full. |
| 1623 | Flash detector for 30s. |
| Now that we have replaced DCRC3 with DCRC1, when we flash we get much greater heating. Tower rises to 160 mK. Before it only rose a few mK. | |
| 1700 | Start 5×25 min series at HV = 300→400V, rate +3.9V/min with 5% prebias for 1 minute, flash at end with 30 min wait. CDMS detector: Driver Gains = 1x, DCRC4/1 QET biases = (420/370/440)/(420/400/450) uA, Front-End Gains = (2/2/3)/(5/5/5) 1000 BOR and 1000 EOR randoms, , trigger on DCRC1pB/C/D threshold = 6/6/6. Use sequence HVdrift_5x300V_3p9VPM_prebias_5_1_dur_25min_flashDCRC_1_4.msl |
- It looks like the DAQ was very sluggish during these series. It took about 21 minutes to ramp the voltage up to 315 V, but it was set to increase at 1 V/sec - it actually took 4 V/s.
- similarly the ramp down took very close to 4 V/s, should have been 1 V/s
- Data-taking period took about 32 min but was set to 25 min. This is consistent with an extra 3-second overhead for every HV update, as above.
- The slower drift means we can expect that the detector bias and Luke gain would not stay flat, rather would fall over time, and we see this in the data
- Leakage was not the same from series to series: max leakage = 47/53/48/42/34 mV. The rise then gradual fall makes it seem like maybe it's tracking humidity changes. So let's start purging the Ebox bag with N2.