- Load in lock-in processing filters
    - H1:SUS-MC2_LKIN_P_DEMOD_SIG: BP4.0Hz
    - H1:SUS-MC2_LKIN_Y_DEMOD_SIG: BP4.0Hz
    - H1:SUS-MC2_LKIN_P_DEMOD_I: CLP50mHz
    - H1:SUS-MC2_LKIN_P_DEMOD_Q: CLP50mHz
    - H1:SUS-MC2_LKIN_Y_DEMOD_I: CLP50mHz
    - H1:SUS-MC2_LKIN_Y_DEMOD_Q: CLP50mHz
  (Turn them all on)

    [[These were already installed for M3 balance, so didn't need to do anything]]

H1:SUS-MC2_LKIN_P_DEMOD_PHASE- Open StripTool of lock-in output channels
    H1:SUS-MC2_LKIN_P_DEMOD_I_OUTPUT
    H1:SUS-MC2_LKIN_P_DEMOD_Q_OUTPUT
    H1:SUS-MC2_LKIN_Y_DEMOD_I_OUTPUT
    H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUTPUT
Make sure they're on the same range

   [[already open]]

- Set coil drivers to maximum range
    - Set COIL BIO switch to -2 (for "ACQ ON, LP OFF," in manual mode)
    - Turn off all compensation filters in COILOUTF filterbank

- Turn on *both* oscillators for demodulation purposes (even though only one will actually drive the DAC)
   OSC     Frequency [Hz]     Amplitude [ct]     Sin [ct]      Cos [ct]
    P         4.0              120000 (1.2e5)       1000           1000
    Y         4.0              120000 (1.2e5)       1000           1000

   [[already on]]

- Turn on LKIN EXC switch at given stage to be balanced (say M2)         
         P   Y
    UL  +1   0
    LL  -1   0
    UR  -1   0
    LR  +1   0
(drive should now be going out to the DAC)

- By hand/eye, tune demod phases in LKIN by minimizing Q phase (bring abs(Q) = 0) in P and Y oscillators
   H1:SUS-MC2_LKIN_P_DEMOD_PHASE = 104.5
   H1:SUS-MC2_LKIN_Y_DEMOD_PHASE = 82

- Quantitatively fine-tune demod phase
    - Measure average demod output after 50 mHz impulse response has settled
       cdsutils avg -s 100 H1:SUS-MC2_LKIN_P_DEMOD_I_OUT H1:SUS-MC2_LKIN_P_DEMOD_Q_OUT H1:SUS-MC2_LKIN_Y_DEMOD_I_OUT H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUT
                   Channel                   Mean              Std
H1:SUS-MC2_LKIN_P_DEMOD_I_OUT    -765.277312518    554.657713265
H1:SUS-MC2_LKIN_P_DEMOD_Q_OUT    -0.462407044597    2.05083625228
H1:SUS-MC2_LKIN_Y_DEMOD_I_OUT    156.427810545    112.812236568
H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUT    -0.15273894759    2.64449939029

- Perturb coil balance by 5% in P, measure demod outputs. Make sure DAC is not saturating. 
      /ligo/svncommon/SusSVN/sus/trunk/Common/PythonTools/perturbcoilbalance_fourosem.py H1 MC2 M2 PIT 0.05
	BIG PITCH: (note for future balance fine tuning: +P is the bad direction, and need much smaller than 5%)
                   Channel                   Mean              Std
H1:SUS-MC2_LKIN_P_DEMOD_I_OUT    -812.812304845    25.0834467432
H1:SUS-MC2_LKIN_P_DEMOD_Q_OUT    -17.1365320439    11.8038789533
H1:SUS-MC2_LKIN_Y_DEMOD_I_OUT    2443.56064935    1630.24135426
H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUT    1186.36074766    848.662710832

    - Calulate new YAW phase with big pitch using YAW demod outputs
          dPhi = 180/pi * atan( (Q_DC^{before} - Q_DC^{after}) / (I_DC^{before} - I_DC^{after}) 
               = 180/pi * atan((-0.15273894759 - 1186.36074766)/(156.427810545- 2443.56064935)) =  27.4193
       New Yaw Phase: 109.4

	Though this caused a huge change in *both* YAW demod outputs, and didn't bring the Q phase to zero, I removed the imbalance, and the Q didn't change. So this is the case where there's some leftover noise that's unrelated to the two knobs I'm turning.

    - Restore COILOUTF gains to +/-1

    - perturb coil balance by 5% in Y, measure demod outputs. Make sure DAC is not saturating. 
      /ligo/svncommon/SusSVN/sus/trunk/Common/PythonTools/perturbcoilbalance_fourosem.py H1 MC2 M2 YAW 0.05
        BIG YAW: (note for future balance fine tuning: +Y is the right direction)
                   Channel                   Mean              Std
H1:SUS-MC2_LKIN_P_DEMOD_I_OUT    -4694.02541529    2769.4692848
H1:SUS-MC2_LKIN_P_DEMOD_Q_OUT    -189.01067855    123.05200929
H1:SUS-MC2_LKIN_Y_DEMOD_I_OUT    262.151066908    1575.91656211
H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUT    -55.5393103903    894.353744908

    - Calulate new PITCH phase with big yaw using PITCH demod outputs
       Delta Pitch phase:
          dPhi = 180/pi * atan( (Q_DC^{before} - Q_DC^{after}) / (I_DC^{before} - I_DC^{after}) 
	         180/pi * atan((-0.462407044597 -189.01067855) / (-765.277312518 - -4694.02541529)) = -2.7611
          New Pitch Phase: 107.3

    - Restore COILOUTF gains to +/-1
    - Capture final numbers after demod phase tuning:
                   Channel                   Mean              Std
H1:SUS-MC2_LKIN_P_DEMOD_I_OUT    -815.784070443    2794.00515103
H1:SUS-MC2_LKIN_P_DEMOD_Q_OUT    41.9716489425    168.027560173
H1:SUS-MC2_LKIN_Y_DEMOD_I_OUT    142.161832216    68.0314577658
H1:SUS-MC2_LKIN_Y_DEMOD_Q_OUT    -75.0713564468    6.77278914281

- With tuned demod phases, I phases of P and Y should be measuring P and Y perturbations independently, and Q phase should do nothing with a perturbation. Now perturb coil balance in small increments to bring I's to zero, and that's your balance.
    caget  H1:SUS-MC2_M2_COILOUTF_UL_GAIN; caget  H1:SUS-MC2_M2_COILOUTF_LL_GAIN; caget  H1:SUS-MC2_M2_COILOUTF_UR_GAIN; caget  H1:SUS-MC2_M2_COILOUTF_LR_GAIN
H1:SUS-MC2_M2_COILOUTF_UL_GAIN 0.987841
H1:SUS-MC2_M2_COILOUTF_LL_GAIN -0.992495
H1:SUS-MC2_M2_COILOUTF_UR_GAIN -1.00737
H1:SUS-MC2_M2_COILOUTF_LR_GAIN 1.01212


- Quantify the goodness of the balance
   Started with
   - /ligo/svncommon/SusSVN/sus/trunk/HSTS/H1/MC2/Common/Data/2014-08-04_H1SUSMC2_M3_CoilBalancing.xml
   [[no need to change names]]

   - Measure with above balanced gains in place
   - Collect reference

   - Restore gains to +/-1, measure again

   - Compare the reduction in 4.0 [Hz] signal. (just use cursor to determine amplitude)
                P         Y
Not Balanced   2.366      0.422
With Balance   0.120      0.160
ratio          19.71     2.6375  

remeasured later, with damping turned OFF. Whoops. Same results though.