This offset has been added in May 2017, see figure 1. It seems to have happened during work on comparing SNEB to SWEB that was done on May 9, see figure 2. The trend data for May 9 are missing, so we only have the information for the day before and after.

Looking at figure 1, the correction signal changed for BL_V and FR_V by about 4V. These are coils at opposite end of the bench, and checking the configuration they have the same sign of actuation. So this change in correction correspond to an offset on BL_V being compensated by the same force applied on FR_V to have a zero angular force, but some force pushing the bench up or down (I am not sure of the sign).

I expect this is an offset left on by mistake after testing the sign of the actuation for the 4 vertical coils, and it is better to keep the offset at 0 (and if confirmed use the balancing mass on SNEB to keep the correction closer to zero too).

I have checked the other suspended benches. And the only remaing non zero offsets are for SDB1, they are 4 equal horizontal offsets, which correspond to a TY force. Presumably this is a torsion on the three supension wires to keep the bench roughly aligned with the beam even when the controls are off. Presumably it would be better to move that feedback offset up the suspension chain, and have it applied on the marionette coils, especially because the marionette is suspneded only from 1 wire and from 3 wires like the bench.

Later on March 29 2018 an intervention to fix the SNEB doors was finished. Around thes days the correction on BL_V jumped from -4V to -8V, see figure 1.

This correspond to a rebalancing of the bench on March 27 and March 29 (see log file below). The balancing was probably done by looking at the control loop correction signals, instead of the signals actually send in the coils. Which means that the balancing put the signal sent into the BL_V coil roughly at the offset constant left by mistake a year before. The right course of action is to keep the BL_V offset at zero, and once the beam in the north arm is found again, redo the bench balancing using the motorized counterweight (which should bring it back to the middle of the range).

SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h53m52-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 100 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h54m08-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 400 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h54m32-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 2000 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h54m49-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 2500 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h55m24-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h56m42-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h56m59-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h57m18-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps forward
SNEB_Pico_2018-03-27-14h52m35-UTC.log:2018-03-27-14h58m27-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps forward
SNEB_Pico_2018-03-29-07h16m29-UTC.log:PMC_MOTOR SNEB_NS    7 1
SNEB_Pico_2018-03-29-07h16m29-UTC.log:2018-03-29-07h17m11-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps backward
SNEB_Pico_2018-03-29-07h16m29-UTC.log:2018-03-29-07h18m18-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps backward
SNEB_Pico_2018-03-29-07h16m29-UTC.log:2018-03-29-07h18m29-UTC>INFO...-PicoControl> Motor 'SNEB_NS' moved 10000 steps backward