Displaying report 1-1 of 1.
Reports until 15:24, Tuesday 21 September 2021
jeffrey.kissel@LIGO.ORG - posted 15:24, Tuesday 21 September 2021 - last comment - 11:45, Tuesday 28 September 2021(59993)
Beckhoff Infrastructure to support reading back (ZM2, ZM4, ZM5) HPDS P-SAMS Drive Voltage Monitors and Strain Gauges
J. Kissel for D. Sigg

Daniel is slowly but surely building up a basic control (well, drive and readback) system for the lower (M2) stage PZT actuator of ZM2, ZM4, and ZM5, which are all the HAM Piezo Double Suspension (HPDS).

In this entry, I cover the changes he's made to the TCS Corner Beckhoff PLC. 

The PZTs on the M2 stage are just a single, longitudinal actuator inside the barrel of a metal optic holder that pushes on the back side of a modular, curved, pre-loaded, mirror/optic clamped to the barrel (you can find the designed radii-of-curvature after pre-load in T2000101). Thus the drive changes the radius of curvature of the optic and can be used to change the mode-matching of either the beam going from the OPO to the new Filter Cavity (FC) via ZM2 OR to mode-match the filter-cavitied SQZ beam to the IFO via ZM4 and ZM5. 

As is fundamental to PZTs, the position of the PZT is some hystoretic, complicated function of the drive voltage. However, they're built with strain gauge systems that track the strain on the PZT which is proportional to the optical power (in diopters, D) caused by the distorted mirror's radius of curvature. There's commentary about how one can calibrate the readout into physical strain from the voltage readback in the Rich Abbott design note for our PZT and strain gauge system in T2000228, but folks don't seem to interested in performing that math yet. One can see discussion of the typical proportionality between strain gauge voltage readback and optical power in the Fabrice Matichard paper P2100315, and one can find the specific curves for H1's ZM2, ZM4, and ZM5 in Camille Makarem data in https://dcc.ligo.org/LIGO-E2100288.

All this being said, the readback channel for the strain voltage readback may now be found in the EPICs reproduction of the Beckhoff channel 
and as far as I know, there's no digital calibration installed yet, so the voltage is as it comes out of the front of the Aplus Dual Strain Gauge and HV Piezo Driver D2000555, readout by a Beckhoff EL3104 ADC (el3104.html), as shown in the PSL / TCS / IO chassis layout D1101114.

In addition, there're also EL3104 ADC readbacks of the PZT drive voltage monitor paths, in various forms, which appear in EPICS as 
in units of "(Differential) Voltage (between the two +/- legs) at the input of the EL3104 ADC."

From deep inside the Aplus Dual Strain Gauge and HV Piezo Driver D2000555, we see that the chassis contains our old faithful PZT driver PCB board, D060283 (though, importantly with components for the monitor path changed -- to be reflected in a incoming later version). In *that* board, these readbacks come from the AC and DC pickoffs of the high-voltage PZT drive that heads out to chamber:
    - The AC path takes the high voltage, applies a passive highpass filter with (a zero at 0 Hz and) a pole at 1/(2*pi*R51*C26), and increases the amplitude with a gain of 200. As confirmed by designer Luis Sanchez, in the built-but-not-yet-drawn version, R51 = 2kOhm, and C26 = 0.47 uF, so the pole frequency is ~170 Hz.
    - The DC path takes the high voltage, and reduces the amplitude with a gain of - 0.02 (yes, an inverting op-amp is used, for whatever 2006 reason, so the signal's sign is flipped).

Now, recall from LHO aLOG 59992 that the intended dither frequency is ~40 Hz... so this newly configured high-pass filter will actually filter out our intended dither. Also note that while the Beckhoff ADC may be run a nice, high, sampling rate (in this case they're run at fs = 1 kHz), any readback channels from beckhoff must pass through the slow, 16 Hz EPICs gateway.

As such, Daniel's modified the AC voltage monitor path after arriving in the ADC to be an RMS channel instead. Pulling from the Beckhoff code within the AWC library, svn/slowcontrols/trunk/TwinCAT3/Source/Library/AWC/AWC/POUs/PSamsFastFB.TcPOU, we see 
    7      HVACDiv:        LREAL := 0.0025;
    21     cur := HVACDiv * EL3104(PSamsFastIn.Vmon_Ac);
    22     PSams.Voltage_Ac := SQRT(0.999 * PSams.Voltage_Ac * PSams.Voltage_Ac + 0.001 * cur * cur);

which, for a code running at fs = 1 kHz, and A = 0.001, means that the AC_voltage signal is *not* band-passed or anything prior to the RMS, and the RMS is effectively low-passed with a corner frequency of fs/(2*pi) * ln[1/(1-A)] = 0.160 Hz (as per math methods from LHO aLOG 1265).

The 0.0025 factor serves as a conversion factor such that the resulting RMS voltage calibration matches the DC path voltage. Still working with Daniel as to why it's that particular value (I think it should be 200 / 0.02 or 0.02 / 200 -- the ratio of gains between the AC and DC paths).

Currently, one can only find these channels via diving down many many layers auto-generated screens from the "ETHERCAT Overview" from the SYS menu of the sitemap, but I'll also make sure to add these readbacks and explanation to the future HPDS_OVERVIEW medm screens.
Comments related to this report
aidan.brooks@LIGO.ORG - 11:45, Tuesday 28 September 2021 (60048)AWC, ISC

Added a cartoon to the DCC that illustrates the above implementation. There is a second page in the DCC document that shows a slightly more sophisticated version which includes calibrated channels and the ability to request specific ROC.


Displaying report 1-1 of 1.