Measurement technique:
- Using DTT, grab amplitude spectral densities of individual OSEMs on M2 and M3 stages (or only M2, or including Optical Levers, depending on which sensors the suspension has), e.g. 
     H1:SUS-${OPTIC}_${M2/M3}_OSEMINF_${OSEMDOF}_OUT_DQ
     H1:SUS-${OPTIC}_M3_OPLEV_${PIT/YAW}_OUT_DQ
- It is good practice and helpful to also grab an ASD of the excitation channel used below, treat it as an "A" channel, and plot the coherence as well.
While exploring for the given mode, use 5 exponential (aka "running") averages, with a 0.05 Hz frequency resolution. Once you've found the mode, increase the resolution to 0.01 or 0.005 Hz resolution, and measure for 5+ averages, depending on your patience and/or the amount of time you have.

- Using the COIL DRIVER Binary IO, switch the coil driver state to the highest possible range (State 2 -- ACQ ON, LP OFF for all of the above mentioned suspensions, and if need be, use the "secret" functionality to request State -2, and switch off all COILOUTF compensation filters).

- Using AWGGUI, excite an individual OSEM coil on the M2 stage, using a band-limited uniform noise excitation, where the frequency limits of the bands are ~1-2 Hz wide, centered around the expected frequency. We used a 4th order elliptic band-pass, with 1 dB ripple. One can typically drive amplitudes of 5e6 counts without saturating the DAC. Channels are of the form
    H1:SUS-${OPTIC}_M2_COILOUTF_${OSEMDOF}_EXC.

- Use the cursor in the DTT ASD template to find the resonant feature that results from your excitation. 

The above data was gathered at a 0.01 or 0.005 frequency resolution, so one should tried the numbers as though they have an uncertainty of f_res +/- 0.01Hz.

An example template is shown in the attachment (i.e. during the measurement of PR3's V3 mode), but all templates have been saved to:
    /ligo/svncommon/SusSVN/sus/trunk/${SUSTYPE}/H1/${OPTIC}/SAGM2/Data/YYYY-MM-DD_H1SUS${OPTIC}_M2*Search.xml


Note -- the only suspension mode for which we don't have an answer is H1 SUS MC3's Highest Roll Mode (R3). We tried several different excitations in order to find it, but turned up unsuccessful everytime:
    - Driving an M2 coil as described above (we tried UR), with band-limited uniform noise,
    - Driving all M2 coils in PITCH (via the M2_DRIVEALIGN_P2P_EXC channel), with band-limited uniform noise,
    - Driving all M2 coils in PITCH, with a swept sine excitation, and dense frequency vector between 39 and 42 Hz.
We tried the above both in the standard State 2 acquire filtering and also in the secret State -2 and the compensation filters off.