A rough OFI shroud fitcheck was done on a earlier version of the structure. We found out a lot of issues with various types custom hardware. Most common problem - tapped not all the way through where it needs to be (D1700233, D1700244), bad threads on D1800111. The hardware has been re-tapped with clean taps.

Some photos of the assembled shroud are attached (did not use viton and coated hardware for the fit check because some parts were still at the C&B etc.)

Luis,

I got 4 different quotes for the build and stuffing of pcb's from two different designs, the next tables show the differences.

I am not 100% sure if the quote that I received from Manufacturing Services include the SMD stuffing material, this price is quite different if we compare this with Screaming Circuits and Advanced Circuits.

I will ordering the boards using Sunstone and Digikey, for us to built these here...

Luis

During the last visit to LLO and HLO we noticed the even though we could set the CMRR utilizing the Digital potentiometer, the communication with digital potentiometer was loose from time to time.

Reason why I have been working on a pair of pcb board desings to improved SPI communications over longer distances, these design includes the device LTC6820.

The noisy enviroment at the End chamber from X or Y arm affect the digital signals that comunicate with the pot (MOSI, CS, CLK).

The communication ocurres between the microcontroller located outside chamber (and connected to D1002283) to the digital potentiometer inside the EMF(that is located inside end chamber) . 

The enhancement  of SPI communications is intented by using a Master pcb board that will be located outside the chamber. This board will be connected into the D1002283 chassis EFM programming input. 
The signals then will travel thru the 25 pin cable until they reach the Slave pcb board inside chamber.  This board will be conected before the EFM DB-15 input connector.  
After the CMRR set up operation is finilazed this boards can be removed and leave the EFM as was originaly set.

Right now I am requesting price quotes for the pcb design and the assembly. 

Below can be seen the both schematic designs.

Luis

This week I went back to the EFM programming interface cable and I created a new interface to be used with the EFM. I am using a case from C4 Labs and cable that Rich had it in his office, see D1800175.
Also I started on making a graphic user interface with python to control digital potentiometer (set and fuse resistance). Kind of tricky since I am just starting to work with python, but I am learning a lot. I am still working in this but soon it will be ready for implementation.

Programing_cable.JPG

python_gui.PNG

Rich, Calum, Luis

EFM SN 003 passed final electrical test(Full assembly).

After finally understand how to conduct the electrical test and set the efm cube on a very simetryc layout, We collected some data  from EFM SN 003.

The transfer function at 100Hz results are listed below, (these results match our simulation on LTSpice).

The ringdown test was done by charging the sense plate while the calibration plate was grounded, we found that the RC time constant is around  20seconds .

X+ 20.2sec

X- 19.3sec

Y+ 20.7sec

Y- 20 sec

Todd, Rich, Calum, Luis

DB25 to DB15 & DB25  cables were tested and passed Continuity and HiPot Test (These cables are for EFM2 and EFM3 in chamber connections).

SN S1800632 D1800090,

SN S1800628 D1800089,

SN S1800633 D1800090,

SN S1800629 D1800089.

Bob, Calum, Carl, Luis, Rich

After passing the RGA scan the EFM3 was removed from oven F to be assemble and then electrical test.

Bob, Calum, Luis, Stephen.

This week (05-29-18) EFM3 passed accumulation test, see description on T1800242.

The criteria to pass the accumulation test is 3e-10.

EFM3  Leak rate =9e-11 Torr L/s.

EFM3 was set into oven F for a bake at 50C for 24hrs and RGA scan. Prior to placed the electrometer in oven F,  EFM3 was cleaned with freon 113 using a cloth.

EFM2 blank flange was torque to 190 Lb-inch.

EFM2 ready for accumulation test, the electrometer was set into oven D.

This time we torqued to 185 in lbs Cf. before 180 in lbs 

starting level pump down was (after 1 re-start and vent) = 1 e-10 torr l/s with pressure 6.2e-3 Torr

Then he into bag 

stayed flat for ~ 1min after 1st helium

Highest num got was  5e-10   torr l/s after 2 min

Pass 

Calum and Luis

After the success of the Helium Leak Test, we continue on fill EFM 003 with Neon 1cfm for 2min 15sec, video is also attached.

Luis, Calum, Rich

After a successful helium leak check which took unbelievable care to pull off (mainly by Calum's persistence and good judgement), EFM 2 was charged with neon gas today.  A flow for 125 seconds at 1 SCFM was performed and other steps per the procedure.  The resulting bag measured approximately 60cm long with a radius of around 20cm. 

The volume of neon introduced into the bag was 6e4 cm^3 based on flow and time, and 7.5e4 cm^3 based on final bag volume, which is within reason for a crude measurement.  The EFM was placed in its dedicated Pelican case and is ready for transport to the 40m lab for neon accumulation test preparation in the morning.  The test data for EFM 2 is being loaded into the revised E1800151 procedure for upload to DCC

Initial pump down was 3.5e-10 torr l/s with a pressure 7.1 e-3 torr.

after 1 minute it reachs 5.6 e-10 torr l/s  then continues to change until the pressure hits (at 2 minutes)  6.2 e-10 torr l/s ;

the pressure stay steady at the 2:40minutes mark then gradually start to change, 6.0e-10torr l/s was seen at the 3 minutes mark.

after the 3 minutes, we read 6.0e-10 torr l/s with a pressure of  6.9e-3 torr.

A) Start pump down 4.1e-10 torr l/s

after 1 minute 1.5e-9 torr l/s

B) vent and pump down while still in bag with he

this time levels out at 5.5e-10 torr l/s within 1 minute p1=7.5e-3 torr 

EFM 003 2nd try with he leak test

2a) pump down and then fill bag

start 1.4e-10 mbar l/s

after 1 minute or so went up to 1.4 e-9 mbar l/s

2b) stopped, vented and then pumped down with he still in bag - also changed units to torr l/s

after 30 seconds  or so flat out at around 1.2e-9 torr l/s, was still at this value after 5 minutes

Now have really good system for leak testing with new bags, tapes, hoses out window etc 

Luis, Calum, Rich

Changed out the resistors in EFM 2 from 249k to 49.9k (R15), 240k to 43k (R12), 100pF to 470pF (C5 and C8) to allow wider CMRR range and reduced gain.

Bagged EFM 3 for helium leak test

20 sec helium. Pump rate started at 2e-10 mbar liter/sec helium and rose to 2e-9 over a few minutes.  Pressure then started slowly dropping to 1.9e-9.  We consulted with Dennis on this and he feels the result is marginal.  We elected to remove the unit from the bag and start again.  We checked the torque on each bolt using the same torque wrench as we used on EFM 1 in case there was a calibration error on the torque wrench.  Some bolts did move by 5 to 10 degrees or so.  The feeling we got was that the observed pressure rise may have been due to one of the couplings used to attach the leak checker to the EFM, so we bagged all the couplings as a precaution.  We will wait until the pressure is back down in the minus 10 range and try again.

We tried again, but are still having problems.  The leak rate is now ~7e-9 so we abandoned the helium leak check of EFM 3 and are moving to do the electrical tests of EFM 2.

Below can be seen the video from EFM3 test process.

Nichole, Calum, Rich

The electrical testing for EFM 3 is complete.  EFM 3 has now been torqued down on 5 out of 6 flanges to 180 inch-pounds and was left attached to the helium leak checker.  The leak checker was running well around 8e-10 mBar-l/sec as it was left.  A new procedure (E1800151) has been written to describe the following measurements.  The data below was used to seed the procedure with reasonable numbers.  A SPICE simulation was performed to validate the anticipated magnitude and phase plus noise performance.

Testing EFM3 Continued

Calculated RC pole frequency based on R=49.9k, C=470pF to be 6.79kHz. Using this as a basis for TF acceptance testing. X and Y pots set to 0561 HEX at start of measurement to get +/- gains to be nominally the same. SR-785 on 400 points, 100Hz to 10kHz, 1V source.

In on X+ -> Out Differential -> 100Hz, 6.024dB, -0.9 deg

In on X+ -> Out Differential -> 6.754kHz, 2.979dB, -45.3 deg

In on X- -> Out Differential -> 100Hz, 6.020dB, 179.1 deg

In on X- -> Out Differential -> 6.754kHz, 2.921dB, 134.3 deg

In on Y+ -> Out Differential -> 100Hz, 6.021dB, -0.8 deg

In on Y+ -> Out Differential -> 6.754kHz, 3.278dB, -43.4 deg

In on Y- -> Out Differential -> 100Hz, 6.021dB, 179.2 deg

In on Y- -> Out Differential -> 6.754kHz, 3.229dB, 136.5 deg

Next, the noise measured differentially at the output of each axis is measured with both positive and negative inputs shorted

X-axis noise at 100Hz -> 215 nVrms/rtHz

Y-axis noise at 100Hz -> 215 nVrms/rtHz

Measured the time constant of each input:

X+ -> 5.8 sec

X- -> 6.2 sec

Y+ -> 6.0 sec

Y- -> 5.8 sec

​CMRR (added by Luis on 052418)

X = -65dB @ 10Hz

Y = -75dB @ 10Hz

Luis, Rich

The attached notes are a first look at EFM3 after Luis soldered the new parts (C5 and C8 were 100pF and now 470pF for both X and Y differential amplifiers.  The resistors (R12 and R15) were 240k and 249k respectively, and are now 43k and 49.9k respectively. The resistor change was done to broaden the range of gain mismatch that we could compensate with the CMRR trim potentiometer. We measured the DC offsets with the inputs shorted and the DC response to a 1VDC input on all axes

Stephen, Luis,Liz.

We bag the EFM this morning and Liz Help us to shipped the Electrometer to Hanford.

Luis, Rich

Today we explored setting the common mode rejection ratio (CMRR) of the Y-axis on EFM1.  We quickly realized we have a shortcoming in the range of CMRR we can adjust.  Based on the schematic used in EFM1 for the differential amplifier, we are only able to vary the gain (which in turn specifies the window aperture of gain that we can accommodate) over a 0.11dB range.  The intrinsic imbalance of a pair of inputs is usually around 0.3dB or so, so this window is not large enough to reach a minimum in the CMRR.  We are going to ship this unit as it is, but are immediately going to fix the problem and ready EFM2 as a replacement at LHO.

Luis, Stephen, Calum, Liz, Rich

Today we had a first look at EFM 1, 2, and 3 in terms of the DC transfer function (observing 1VDC injected into each axis input results in 10V differential at the output), and the exponential decay associated with a step voltage change applied to a single calibration plate.  We spent the first part of the day in a typical state of confusion as we struggled to interpret the behavior of EFM 2 and 3.  EFM 1 was just clearing the final clean and bake process in the 40m Lab.  EFM 2 and 3 appear to be functional, however the background electrical noise in the Downs 2nd floor clean room seemed to be getting rectified and contributed a current to the input that lead to saturation.  At some point, we got the word that the mission critical EFM 1 was on it's way over from the 40m Lab, so we abandoned EFM 2 and 3 for now.

On EFM 1, we first verified all axes are functional with a simple DC injection.  1V injected into each output produced the requisite 10V differentially at the EFM outputs.  Then we moved on to the exponential decay test.  We first estimated the capacitance of each of the components on a single axis input to the EFM.  Corroborated by direct measurement where possible, we calculated based on area the following (1 inch PEEK standoffs on the sense plate to the body of the EFM, 1/2 inch PEEK standoffs from the sense plate to the calibration plate:

Sense Plate to EFM Body - 3.6pF

Sense Plate to Grounded Calibration Plate - 5.7pF

Ceramic Vacuum Feedthrough (measured) - 4pF

This yields a total expected capacitance of around 14pF including a bit for the input to the opamp internal to the EFM.

Well, so much for that notion.  Nature had a different picture of the capacitance.  As can be seen in the attached decay plot taken with an oscilloscope in response to a 5V step applied to the calibration plate, the RC time constant is around 23 seconds.  Assuming we got what we paid for in terms of the 10% tolerance, 10^12 ohm resistors, that would imply a capacitance of around 23pF.  Quite a bit more than we expected.  We then embarked on a process to try to figure out where we went wrong.  We had been shielding the sense and calibration plates during the decay measurement by creating a grounded aluminum foil tube that acted as an electric field shield.  We wondered if this was contributing and parasitic capacitance, so we repeated the decay with no shield.  Now we had a 19.8 second RC time constant.  Not much different, but better.  Next, we hung the sense and calibration plates off the side of the work bench in a effort to see if the proximity to the grounded bench was a factor.  Now we had a 18.8 second RC time constant.  Next we removed the voltage source from the calibration plate to effectively unground it so it would be an unreferenced metal object that theoretically should add no capacitance. This also required that we make a big step change in the local EFM electric field environment (Rich wiggling around) such that it would essentially saturate as we no longer had a calibration plate to drive.  This brought the time constant down to 13.6 seconds.  Lastly we did the absurd and completely removed the calibration plate even though Rich was sure it would make absolutely no difference.  Now the time constant decreased to 8.8 seconds to his utter amazement.  Here's a summary:

Perhaps, and most likely by coincidence, the time constant of 8.8 seconds actually is beginning to agree reasonably with the anticpated capacitance of the sense plate plus feedthough (3.6pF plus 4pF)

Calum, Rich, Bob, Luis

EFM Cube No.1 was loaded into oven F for a 24hr 50C bake. The bake started around 11:00AM today 051418.

WIP (photos will be posted, more information will be gathered)

1. 14 May 2018: Stephen and Luis gathered serial numbers from Cube 2 / Cube 3 / spares (all of which were Class A baked in Bake-9049) (ICS records will be updated)
   1. Cube 3
      1. D1800043-v1 SN003
      2. D1800032-v2 SN013
      3. D1800032-v2 SN004
      4. D1800032-v2 SN001
      5. D1800032-v2 SN012
      6. D1800054-v1 SN002
      7. Cube 408003 unserialized
   2. Cube 2
      1. D1800043-v1 SN004
      2. D1800032-v2 SN009
      3. D1800032-v2 SN010
      4. D1800032-v2 SN002
      5. D1800032-v2 SN014
      6. D1800054-v1 SN003
      7. Cube 408003 unserialized
   3. Spares
      1. D1800032-v2 SN011
      2. D1800032-v2 SN007

Accumulation test data 

https://dcc.ligo.org/LIGO-T1800212

Luis, Rich,

Yesterday we started a new test. This test was setup trying to confirm the resistance from the 1 tera ohm resistor. This time we have the idea of adding a known capacitor in parallel with the setup. Now we have 2 equations with 2 unknowns. 
1st equation is  R x C =6sec; the charging time of 6sec was found in the experiment developed yesterday. 2nd equation is R x (Csense + C) = charging time. Csense is the capacitance from the flange to the sense plate, 
we measured the separation of these plates and found it to be 0.656inch apart. With this data we found the capacitance of 5.4pF.  We measured the new  RC time and we got 24sec. 
With this new data we have the 2nd equation as  R(5.4pF + C)=24sec if we solve for C considering that R=6/C we found that C= 1.8pF making R =3.3teraohms. Having the time of 24sec makes the resistor larger than 1 tera ohm and also 
make us think that this new setup is adding more capacitance to our system. The way we applied the input signal to the sensor created a new capacitor? we need to measure the new capacitance!!!
It is important to mention that adding a sense plate to our original set up(without isolating the pcb boards)made a very good detector but the signal is too noisy, with this signal it was dificult to identify the asymptotic line when trying to fit the curve. 
We tried to isolate the sense plate for any external signals but the shield applied to setup was not enough, the signal behaved almost similar as before.     

Today we started with a good idea of adding the calibration plate to the set up (the separation between this plate and the sense plate is about 0.5inch). We shielded the set-up completely and after taking this new ring down measurement we got around 18pF which give us a time constant of 18.2seconds at the 63% (1.79v) charge time from the top down to asymptote. These measurement was considering the max peak voltage at 2.62v and the asymptote line at -0.22v. With this new piece of information we conclude that our Resistor is indeed a 1Teraohm resistor. 18.2seconds/18pF= 1.011x10E12ohms.See scope image. 

Luis, Rich

Yesterday, we spent quite sometime messing around trying to measure the input capacitance of the EFM.  Each EFM input feeds into an AD549 opamp.  These opamps have a very high input impedance made using JFETs.  Essentially, one is looking into the gates of a bunch of JFETs in parallel.  The gate of a JFET can be envisioned as a very high impedance resistor (>10^15 ohms) in parallel with a picofarad or so.  We thought we would be clever and measure this capacitor.  It turns out that the opamp had different ideas...

While the input of the opamp is reasonably modeled by a parallel resistor and capacitor, the capacitor value depends on the amount of voltage present on the opamp input.  When we would try to measure the input impedance using a simple capacitance meter, the meter would apply a sinusoidal voltage that is used to calculate the capacitance of the circuit hooked to the meter.  As the voltage of the sinusoid varies, so does the capacitance.  This dynamic situation confused the hell out of the capacitance meter, which staunchly refused to place any limit on the measured capacitance other than approximately a value from minus 1uF to plus 1uF.  The actual expected value should be around 1pF.

Eventually, we gave up trying to measure the input capacitance and instead applied a step change of input voltage to the electrometer input (using a piece of aluminum foil near the copper rod that forms the input to the EFM).  We would watch and plot the voltage decay as a function of time.  Because we don't know the resistance (assumed to be 1 x 10^12 ohms), nor do we know the input capacitance, we can only derive the product, RC, of the two.  Here's what we saw:

This grainy image shows some measured data with a fit overlayed in orange.  If the resistance is indeed 10^12 ohms, then the calculated capacitance would be 6pF.  We were able to measure the capacitance of the ceramic feedthrough to be about 4pF, so this means the total input capacitance of the EFM circuit board and opamp would be around 2pF.  Quite reasonable really.  Next, we are going to bolster our confidence that these resistors are actually producing 10^12 ohms by using a second resistor in series with the EFM input to form a 2:1 divider.  If this doesn't end tragically, we will post the results soon.

Bob, Calum, Rich, Luis.

As Rich mentioned, the EFM helium leak test has passed.

I am attaching a video that was taken during this test, in this video we can see how the vacuum pressure was kept during the experiment eventhough the tracer gas was in the vecinity of the EFM.

DCC document E1800144 details some aspects from the test process and describe the final results.

Finally, Today (051118) I was able to compress the video file and this can be viewed here. 

I started to assemble the interface board power regulators and sense board in clean room located in dows 2n floor. EFM Cube2 and Cube3 are partially assemble, sense boards and interface board are connected thru the interface wires. One cable for each sense board.

Luis, Rich

In an effort to understand where the ~20pF capacitance comes from as measured at the site of the sense plate to body, we measured the capacitance of the ceramic feedthrough all by itself.  After checking two examples, we concluded that the feedthrough capacitance is about 3pF.  The datasheet for the AD549 opamp says the amplifier input capacitance is around 1pF.  A calculation of the approximate capacitance of the sense plate to the body is around 7pF ignoring fringe fields, and the capacitance of the L-bracket inside the cube to the inner flange is around 0.5pF.  This total (~11.5pF) leaves 8.5pF unaccounted for.

Luis, Bob, Stephen, Calum,

The EFM No1 was loaded into oven D and the bake process started with set point of 50 Celcius, Nitrogen was used during the process. It is possible that the neon acumulation test will be happen tomorrow (050818).

The sense feedthrough pins were grounded using aluminum foil (which connects to all 4 pins).

In the mean time we will start to assemble the Power regulators on Interface board(D1800040) and set the brackets to connect the D15 feedthrough flange.

Also several items (related to EFM) were unload from oven F and move to float bench for overnight cooling down.

Rich, Bob, Calum, Todd, Luis

EFM#1 has passed the helium leak checking phase and is being moved over to the 40m Lab for the neon accumulation test.  The electrical inputs will be banded with clean aluminum foil to protect against inadvertent static electric charge damage.

Todd, Luis, Calum, Rich

Filled EFM #1 with neon today.  This is the first time we have done this so a procedure (E1800138) was written.  Basically we wrapped the EFM with a band of foil across all electrometer (X and Y) inputs to ground them, then put the EFM into a bag.  We used a vacuum cleaner to remove the air from the bag, then sealed off the port used for the vacuum cleaner.  A hose to inject neon from the gas bottle was prestaged such that it went into the top flange port of the EFM (the blank off flange).  We filled the bag with neon while recording time and flow rate.  The neon was being delivered at ~4psi which corresponded to ~1 +/-0.1 standard cubic feet per minute as measured by a sightglass flow meter.  It took 1 minute and 50 seconds to fill our bag.  Using flow and time this corresponded to 1.83 cubic feet of neon.  We also measured the dimensions of the bag which had formed a rough cylinder (2.5ft long, 0.5ft radius) which equates to 2 cubic feet which is in good agreement with the flow based number.

After filling the bag with neon, we found that it was better to snug up the bolts using a closed end wrench thus keeping the neon inside the EFM, then to do the final torquing immediately after the EFM is removed from the bag.  The torque spec used was 180 inch-pounds to bolt the CF flange to the cube.  This spec was derived from the MDC vacuum specification for the cube and flange arrangement we used.

Made series of laser irradiation on a 3” fused silica uncoated optic https://dcc.ligo.org/LIGO-S1700118-v4

The optic is planned to be used as a viewport at LDF. In air laser damage test is required before using the optic as a viewport on the LDF vacuum chamber. Beam on the target is 127 by 91 microns ellipse. See estimated equivalent laser power input table and layout picture in the related doc https://dcc.ligo.org/LIGO-T1700184-v1

No damage was observed using DF microscope. The inspection was done using manual control – no scanning. After each irradiation run a picture was taken with a FLIR camera.  Post irradiation pictures taken with DF microscope are posted herte https://dcc.ligo.org/LIGO-S1700118. 

Warming up of the beam dumps up to 57 deg C was detected. No other elements of the set up were changing temperature (including the target – 3” fused silica optic and it`s mount)

Added a magnetic base for the 3" diameter target mount. The mount is angled in order to dump the reflection from uncoated fused silica target. Beam on the target is 127 by 91micron ellipse. Irradiation is planned at 25, 50, 75 and 100% power (100%=50W). Equivalent power densities and irradiation runs see in the dcc doc https://dcc.ligo.org/S1700118-v1

Worked on a laser layout for in-air fused silica viewport optic laser damage test. Made a temporary enclosure to prevent high power laser scatter and to stop any possible fragments of potentially damaged viewport. A “labyrinth” shape was built instead of making holes for the laser tube. The top of the enclosure will be covered with one large black panel as well. The beam was focused to 2w=100 microns at the position of the target (red line)

Inspected batch 2 of thinner elliptical mirror received from Brand Laser Optics & Mfg coated on one side. Uncoated surface is matte vs glassy coated side. Shiny green reflections can be observed on the coated side. The mirrors have no chipping but a bit dusty from the package. I have marked the front and back side of the mirrors on the box.

Elliptical mirror received from Brand Laser Optics & Mfg is coated on one side. Uncoated surface is matte vs glassy coated side. Shiny green reflections can be observed on the coated side. I have also confirmed it using a 1064 laser and a laser card. The mirror has chipping at least on one edge. Surface looks dusty from the packaging but clean. I have marked the front and back side on the box.

New mount has been reworked  https://dcc.ligo.org/D1700002 in order to use 2 v-lamps. See modefied mount https://dcc.ligo.org/LIGO-D1700002-v5

See T1600060 the transfer function measured with four different configurations: - elliptical mirror, no viton (black curve)
- elliptical mirror, viton under tip of the PZT (blue curve)
- elliptical mirror, viton (full length) (green curve)
- 2" mirror, viton (full length) (red curve)

First resonance appears at about 2 kHz which is very close to the internal resonance frequency of the PZT (3 kHz with no load according to the specs). Adding viton in the grove did not dump the resonance (black curve vs green and blue) however it can be an option. Changing the mirror from 2" to a lighter elliptical is still a significant improvement even with the new mount (red curve vs others).

PZT`s transfer function measurement with a test v-block was done in tree differen configurations:

- no viton (blue curve)

- viton under PZT (orange curve)

- viton only under the front part of the PZT (green curve)

Conclusion: 3rd configuration is the winner. It provides stiff clamping of the back of the PZT plus dumping. Need to design new v-block type mount with two clamps. Promissing first resonance at 3kHz with the new mount

PZT body was taped to a v-block. An extra mirror used in the layout for stirring the beam. Pictures show a comparison between this tape mount and a sleeve mount in a U200 mount. Taping the PZT damped the two resonances (300 Hz and 1 kHz). The internal PZT resonance was stiffen.

This log entry reflects on recent efforts to image the 1064nm spot on samples undergoing BRDF measurement within the CIT CASI scatterometer. The cameras used are (reiterating from eLOG ENG_Labs/33)

• Basler ace acA640-100gm, same as used at sites, borrowed from Johannes at 40 m. Equivalent to acA640-120gm, which is the current Basler designation of this camera.
• Basler scout scA1400-17gc, originally purchased by OpLev design team, leftover from aLIGO efforts, borrowed from Eric/Gabrielle in West Bridge.

The set up and measurement is described by the following procedure:

1. A sample of oxidized stainless steel was found in the lab and taped to an existing sample mount using kapton tape.
2. Images were taken with the room lights on and then without the room lights.
3. Imaging settings were maintained for each photo taken with an individual camera, but gain values and packet size varied between cameras.

Hardware for the cameras included:

• a lens with adjustable focus (Rainbow, 50mm focal length, C mount, 2/3" to 1" aperture, 1:1.8 image; 40m loan)
• where indicated, a ~70% transmissive bandpass filter (Newport 10LF25-1064).
• mounting post with fixed clamping platform (will add photo), approximately 1 meter from the sample.

The acquired images may be summarized by the following points:

• The 1064nm light was represented by the color scout camera as white-ish pixels with a bluish tinge.
• The scout camera appeared to have a number of dead pixels (someone with more camera expertise may have better terminology; I am referring to the scattered red and blue pixels that do not appear to be real light).
• The focal length of the two cameras was different, hence the difference in perceived spot size.
• The bandpass filter reduced intensity of the image and, with the light off, did not seem to provide any useful noise cancelling.

Image 07: Lights on, scout OpLev GigE Camera Link to casi_test_07_laptop_lights_high_exposure_10_mw_high_gain.bmp

Image 08: Lights off, scout OpLev GigE Camera Link to casi_test_08_no_lights_high_exposure_10_mw_high_gain.bmp

Image 09 Lights off, bandpass filter, scout OpLev GigE Camera Link to casi_test_09_no_lights_high_exposure_10_mw_high_gain_bandpass_filter.bmp

Image 13: Lights on, ace Site GigE Camera Link to casi_test_13_site_gige_foyer_light_auto_exposure_10_mw_auto_gain.bmp

Image 15: Lights off, ace Site GigE Camera Link to casi_test_15_site_gige_foyer_light_auto_exposure_10_mw_100_gain.bmp

Image 19: Lights off, bandpass filter, ace Site GigE Camera Link to casi_test_19_site_gige_no_light_auto_exposure_10_mw_100_gain_bandpass_filter.bmp

Measured lasaer power at the end of the layout with defould pyrex viewport, fused silica AR coated viewport and no viewport. See https://dcc.ligo.org/T1700003-v1 for more info.

Laser power measured using a “PM100USB” power meter and an S314C sensor (±3% measurement uncertainty at 1064 nm). Gray dots represent the power indicated at the laser display which can be inaccurate. There are additional mirror losses at the beam transport via RTS setup. That is why power at LDF setup location will differ from the values shown on the laser display.

Black curve on figure1 is the power measured at the end of the LDF laser layout (including a window from the viewport cover) with no viewport (figure 2). Blue and red curves represent the power measured with a fused silica and pyrex glass correspondently. In both cases the glass was added at the end of the laser layout (see figure 3).

Figure 1: 1064 laser power measured as a function of the input value

Figure 2: Power-meter installed at the end of the LDF laser layout (correspond to the black curve on figure 1)

Figure 3: Power-meter installed at the end of the LDF laser layout plus the fused silica glass (correspond to the blue curve on figure 1)

Almost 100% transmission trough the fused silica viewport has been observed. Losses in fused silica were less that the resolution of the power meter. In contrast, losses up to about 15% were observed in the pyrex glass. An integrating sphere may be used for more accurate measurement.

Calibration function between the display input and the power in vacuum chamber after passing the viewport (±3% measurement uncertainty in not considered):

P_LDF[W]=-4.796812+ 0.549683 P_{display input} [%]

Summary

I was successful in connecting two Basler GigE cameras to my personal laptop and viewing/saving images using pylon, the UI downloaded from the Basler website here. The details of the cameras are documented below:

• Basler ace acA640-100gm, same as used at sites, borrowed from Johannes at 40 m. Equivalent to acA640-120gm, which is the current Basler designation of this camera. ~$500 for qty of 1, 3 week lead time.
• Basler scout scA1400-17gc, originally purchased by OpLev design team, leftover from aLIGO efforts, borrowed from Eric/Gabrielle in West Bridge. ~$3300 for qty of 1, 3-4 week lead time.

There will be more to follow in subsequent log entries regarding integration into the CASI BRDF measurement system.

Some Helpful Resources

aLIGO GigE camera layout and usage. Joe Betzwieser from LLO has a guide to the camera usage and system set-up at the sites located at T1300202.

Basler set-up guide. Basler has a very good set up guide which covers details and some troubleshooting. Accessed at this link, by navigating to [ Support > Downloads > Current pylon Windows ]

Description of Set-Up

Once all of the materials were in hand, a PoE injector (TP-Link TL-POE150S; 40m loan) was connected in series between my laptop and the camera in question. I also had success using the power supply supplied with the camera and a direct ethernet connection to the laptop. A lens (Rainbow, 50mm focal length, C mount, 2/3" to 1" aperture, 1:1.8 image; 40m loan)  was installed on the camera, and the focus ring was adjusted as needed (no iris adjustment required).

The pylon IP Configurator was used to force IP addresses at the onset (Static IP used for the ace and DHCP used for the scout, based on defaults upon opening). The camera status was "OK", then the pylon Viewer was opened, the camera was opened (may need to refresh the list) and placed in Continuous Mode for image collection (ctrl-S to save an image) See the Basler set-up guide (link at the top of this post) for full details of camera set-up.

Difficulty with Set-Up

The only real difficulty I had turned out to be due to exceeding the bandwidth of the set-up. The common error was:

Error: 0xe1000014 "The buffer was incompletely grabbed. This can be caused by performance problems of the network hardware used, i.e. network adapter, switch, or ethernet cable. To fix this, try increasing the camera's Inter-Packet Delay in the Transport Layer category to reduce the required bandwidth, and adjust the camera's Packet Size setting to the highest supported frame size."

Accessing Transport Layer in the camera drop-down and following the recommendations eliminated the bandwidth issues. I was confused at first because there is a second Transport Layer category that lives outside of the camera drop-down. This one is not useful.

Better yet, following Joe's recommendations in "Notes on camera settings" on page 11 of T1300202 (link at the top of this post) made for a seamless camera run.

Results

The color image from the scout camera was taken from about 3m distance with my unsteady hand holding the camera.

Link to scA1400-17gc_image_from_test_office_emergency_map_01.bmp

The black and white images from the ace camera were taken from the same 3m distance, also mounted in an unsteady hand. The image tagged 02 has an increased Gain (250 --> 700) which increased the contrast while causing the image to be less sharp and detailed.

Link to acA640-100gm_image_from_test_office_light_switch_01.bmp

Link to acA640-100gm_image_from_test_office_light_switch_02.bmp

Admittedly, these images don't really mean anything. There will be more to follow in subsequent log entries regarding integration into the CASI BRDF measurement system.

Two front screws on the PZT sleeve mount out and back in to checj the repeatability. No big difference has been observed.

I measured the transfer function of the PI`s PZT with two configurations of the setscrews that hold the PZT`s body in the sleeve mount: 4 screws and only 2 back screws. The natural frequency resonance slightly moved to the lower frequency range with 2 screws setup. Also there is a bit more mess in the high frequency range with the two screws configuration. But it is hard to see clear in this range using the current setup.

The temperature was measured by two thermocouples: integrated thermocouple at the ring heater and external thermocouple at the back side of the radiative plate (gold plated). Four data sets were taken (with no gasket - gray,  with indium gasket - green, tin -indium alloy gasket - blue and tin-indium "heat spring" - magenta on the graph). The higher temperature is the ring heater and the lower temperature is the external thermocouple at the plate. The two temperatures are different due to the heat losses caused by low thermal contact between the ring heater and the radiative plate. For example, when the ring heater at 120 deg, the radiative plate is only about at 75 deg C. Using a gasket prevents overheating of the ring heater. Tin-indium gasket reduces the heater temperature to 90 deg C, indium - to 80. Tin-indium has lower thermal resistance than indium by is nor as soft as indium therefore indium performs better. Making a tin-indium "heat-spring" improves the contact such that it performs as good as pure indium, at the same time it is easier to handle (not too soft). But tin-indium melting temperature is 117 deg C and pure indium melting temperature is 156 deg C. Therefore indium is the winner.

Fig1. Temperature measured at the ring heater and radiative plate with elevated current (200 mA, 250 mA, 300 mA)

Fig2. In heat-spring gasket installed between the heater and the plate

Fig3. InSn gasket installed between the heater and the plate

Fig4. InSn heat-spring gasket

Swapped 2" mirror with an elliptical mirror (25mm minor axis)

setup i  - 1" mirror

setup ii - 2" mirror

I have swapped 1" mirror (5 mm thick) with a 2" mirror (12 mm thick). I just remover the old mirror and glued the new one in place without moving od disassembling the PZT sleeve mount and u200 mount. the resonance at about 300 Hz came back particularly in Y and a new resonance appeared at about 800 Hz in both: Pitch and Yaw, X and Y. This is probably the PZT`s natural frequency resonance. The specs mentioned 1.6 Hz  resonance under the load of a 25 mm (8 mm thick) mirror.

Setup 6 - 1" mirror, magnetic base , glue and so on (final setup used for presentation )

Setup i change - back to the orange base

Setup ii change - Ø2" mirror 12.5 mm thick

Isolating the table did not change the results at all

More epoxy!.. Nor much improvement byt the remaning peak moves. So that 300 Hz is related only to the connection of the two mounts.

After halfway success with a drop of epoxy between the two mounts  I took it apart again, cleaned the dry epoxy and put really a lot of fresh epoxy. This removed "pitch" and "yaw" resonances on X, however small peaks are still observed on Y.

The PZT mount holds in the Newport mount with only one set screw. I released the screw and put a drop of 5 min epoxy between the two mounts. After this the "pitch" resonance was almost gone and the "yaw" resonance reduced.