Oli, Fil, Camilla, Jenne,
We followed on from Keita's Friday issue with ALS WFS 1.6Hz wobble 91647, found it was WFS DOF 1 PIT, increasing or decreasing gain, this did not help. This pushes on ETMX (as does tidal) so Oli in turn turned on offsets in ETMX TEST banks while we watched all the inputs. All M0 was fine. In L1, while pushing on LL nothing moved. See attached. Fil found it strange that the H1:SUS-ETMX_L1_VOLTMON_LL_OUT16 did behave as expected though.
The L1 LL readback didn't react when UL was moved. It did move 6urad when we turned the YAW slider off (our initial test) but the shape of the trend is different than the others, and it didn't move after that, attached.
Fil and Oli went to EX to investigate.
Fil, Oli
We did a few tests at EX to try and figure out what was wrong with the ETMX L1 LL osem. We couldn't find anything wrong, but were able to rule out basically(?) everything outside of chamber. Tomorrow Fil will be checking the resistances of the L1 osems.
The LL channel generally seems to be moving a lot less than it should be(ndscope6). It also sometimes doesn't really line up with the normal breathing frequency of the other L1 osems (I had seen it happening earlier but wasn't able to find an example), but other times it does seem to breathe at the same frequency, albiet still moving a much smaller amount.
COILOUTF OFFSET tests
If you turn on smaller offsets (5,000 counts) in L1 COILOUTF, the UL, UR, and LR OSEMINF OUT values would react to each other's offsets, but LL would not react to their offsets and they would not react to an LL offset(ndscope1). If you put a larger (~50,000 counts) offset into COILOUTF osems, we found that sometimes LL would only react a tiny bit (hundredths of a micron) to the other osems' offsets, and it would either react a tiny bit to its own offsets, or it would move about 8 um. One of the times that we put a large offset into LL, it barely moved for over 3.5s before suddenly moving the entire 8 um in 1 second(alltraces,zoomedin). It was very strange (almost like it was stuck???????).
Coil driver to satamp voltage
At the ETMX L1 coil driver (SUS-C1 U29), Fil put a breakout board in the Coil Out to Sat path. Connecting to each channel, we turned on an offset in the COILOUTF filter bank for that channel and monitored the voltage going out to the satamp. Everything made sense - when there was an offset in the measured sensor, we saw a voltage. When there was no offset in that sensor, the coil driver wasn't sending any voltage out. The voltages measured all matched. We did this test with offsets of 5,000 and then of 50,000, and we correctly saw a difference in a factor of 10 in the voltage.
|
UL
|
OFFSET in COILOUTF [cts] | Voltage [mV] |
| 5,000 | -10.5 | |
| 50,000 | -105 | |
|
LL
|
OFFSET in COILOUTF [cts] | Voltage [mV] |
| 5,000 | +10.4 | |
| 50,000 | +102 | |
|
UR
|
OFFSET in COILOUTF [cts] | Voltage [mV] |
| 5,000 | -10.7 | |
| 50,000 | -108 | |
|
LR
|
OFFSET in COILOUTF [cts] | Voltage [mV] |
| 5,000 | +10.5 | |
| 50,000 | +105 |
Aux channel check with Coil driver to satamp chain
We also tried turning on all the L1 COILOUTF offsets at once (all offsets of 50,000) without the plug plugged into the other side of the breakout board. All channels react the same with their VOLTMONs reading in the 1000s and we were reading ~9 V on the mulitmeter, then once the cable is plugged back in, we see the FAST IMONs and RMS IMONs jump to their normal values, and the voltage on the multimeter dropped to the values in the table as the current was able to flow (ndscope4).
Swapping PD to AA inputs
We also tried swapping the PD to AA cables on the L1 coil driver (SUS_ETM_022) with the PD to AA cable on the R0 F1F2F3 coil driver (SUS_ETMX_014), and saw no difference in the LL signal.
Swapping satamps
We also tried swapping the ETMX L1 satamp with the ETMX M0/R0 LF/RT satamp. LL still reacted the same way as it had been reacting earlier. This rules out the satamp.
Fake osem signal
Fil had the idea of sending fake osem signals into the satamp to more confidently rule out the satamp/AA/ADC path. We sent a signal into each channel and LL saw the signal exactly the same as the other osems, so that rules out that path as the issue(ndscope5).
Attaching green locking this morning, when we locked green in March, and in O4, the L1 LL behavoiur is different now than before.
Plots of ETMX BOSEMs at UIM, Top Mass and Opleves show that there is no sign or major change of BOSEM degradation in LL. It has remained at comparable counts for the last two years.
