• This morning, before starting to work on the BPL loop, we unblocked the beam transmitted from the PMC and we took the references of its position on the BPL QPDs (see fig 1).
• Then we plugged the PZT to the DAC in Eeroom. The PZT driver accepts -2V +12V, while the DAC dynamics is +-10V but due to the LEMO 3 PIN - BNC adaptation it is restricted to +-5V. On the PZT driver the DAC input is summed with a offset (from 0 to 10 V) that can be manually adjusted, and then amplified by slightly more than a factor 10, to -30 +130 V.  We manually set the offset to mid range (5V), so that we can correct around this value with the DAC.

lines 127-130 in  /virgoData/VirgoOnline/ISYS_EER_dac.cfg

ACL_DAC_CH    dac1955_EER_DAC2_ch03    1    DAC1955_FREQ    BPL_PZT_EIB_M1BH_CORR            0            1            "None"                        "ad1955-10v"    ""
ACL_DAC_CH    dac1955_EER_DAC2_ch04    1    DAC1955_FREQ    BPL_PZT_EIB_M1BV_CORR               0            1            "None"                        "ad1955-10v"    ""
ACL_DAC_CH    dac1955_EER_DAC2_ch05    1    DAC1955_FREQ    BPL_PZT_LB_M14H_CORR            0            1            "None"                        "ad1955-10v"    ""
ACL_DAC_CH    dac1955_EER_DAC2_ch06    1    DAC1955_FREQ    BPL_PZT_LB_M14V_CORR            0            1            "None"                        "ad1955-10v"

• We injected a 2Hz line on the PZTs during 90s. Here the gps for each actuator

gps = [1462194492 1462194603 1462194722 1462194871]; actuators = {'INJ_BPL_PZT_EIB_M1BH_CORR','INJ_BPL_PZT_EIB_M1BV_CORR', 'INJ_BPL_PZT_LB_M14H_CORR', 'INJ_BPL_PZT_LB_M14V_CORR'};

• The sensors are: {'INJ_BPL_QF_h_norm', 'INJ_BPL_QF_v_norm','INJ_BPL_QN_h_norm', 'INJ_BPL_QN_v_norm' };
  • QF mainly see the tilt (done with EIB M1B and also a bit with LB M14)
  • QN mainly see the shift (done with LB M14)
• We computed the driving matrix with the attached script and closed the loop. We started to try different filters (see 2nd plot). As a final test we removed the resistences at the output of the PZT driver, from around 15:58 UTC to 16:20 UTC (but there is a lot of noise on the signals, introduced by the amplifier itself, so we put it back).

To do next:

• to perform noise injections and design a better filter which does not introduce noise on QF h between 20 and 100 Hz
• check availability for a DAC +-20V
• add in the ISYS_BPL code:
  • a latch to keep the setpoint when the loop is open 
  • clip on the corrections to avoid too large beam excursion
  • trigger to open the loop if there is no signal

%%% noise injections on BPL %%%

We injected this noise BPL_noise_flt

ACL_FILTER_SET     "BPL_noise_flt"    1     1    20    20
ACL_FILTER_ZEROS   "BPL_noise_flt"      0.01      0
ACL_FILTER_POLES   "BPL_noise_flt"      0.5    0
ACL_FILTER_POLES   "BPL_noise_flt"      500    0.5

on the POST error signals. The error signals are filtered with the filter named BPL_PAOLO, which is taken from the BPC loop

ACL_FILTER_SET        "BPL_PAOLO"    1         -3.3     10        20    
ACL_FILTER_POLES    "BPL_PAOLO"    0        0
ACL_FILTER_POLES    "BPL_PAOLO"    200        0.7
ACL_FILTER_ZEROS    "BPL_PAOLO"    10        0

The problem with this filter is that we are reintroducing noise at 10-100 Hz ( fig 2. in the entry made yesterday #69090), so the aim was to design a more adapted filter.

The GPS are the following (each duration is 3 min):

tilt x 08:14:00 UTC (fig. 1)

tilt y 08:20:40 UTC

shift x 08:35:00 UTC (fig. 2)

shift y 08:42:00 UTC

The last plot shows the loop open and closed with the new filters designed by Manuel (in BPL filters: BPL_tiltx_flt, BPL_tility_flt, BPL_shiftx_flt, BPL_shifty_flt).

%%% clipping on the corrections %%%

The corrections sent to the DAC are now clipped  -9.9 to +9.9 V (since yesterday we observed a misbehaviour of the PZT at 10V )

Profiting of the measurements made today, we extracted the plants of the four DoFs needed to design the control filters. The basic idea was to measure the several pole-zero structures in the mechanical plants, to compensate them in the control filters, in order to flatten the mechanical responses and have for all the four degrees of freedom the same open-loop with UGF at 40Hz, controlled by an integrator and a double pole to roll-off the transfer functions (see fig.1) (same as it was done for the BPC controls).

For future reference, the measured and fitted plants for tilt X/Y and shift X/Y are reported in fig.s 2,3,4 and 5, respectively.

The control filter to reduce the excess of noise visible on the TILT X dof spectra has been updated but will be tested next time.