After the feedback from last meeting and Liu's help narrowing down what I should do to improve the model. I made some changes: First with Liu's help I made the proportions of the test mass and ring heater much more reasonable and parametrized by constants. Second from Cao's paper "FROSTI Nonimaging Reflector Design" Liu showed me what I should do to define the elliptical mirrors. Third during Friday's modeling/programming meeting walked me through getting the different irradiance plots to work. 

Heater 1= 72.8; 80.6

Heater 2= 69.5; 80.8

Heater 3= 70; 83.2

Heater 4= 70.6; 78.7

Heater 5= 69.9; 80.6

Heater 6= 71.1; 78.2

Heater 7= 68.5; 76.8

Heater 8= 70.1; 82.8

Heater 1= 73.6; 81.8

Heater 2= 70.4; 82.1

Heater 3= 71; 84.5

Heater 4= 71.5; 80

Heater 5= 70.5; 81.7

Heater 6= 72; 79.4

Heater 7= 69.2; 78.2

Heater 8= 71.1; 84.2

The Red Pitaya ecosystem has been upgraded to OS 2.00-35, with a key feature being greater freedom in adjusting the sampling frequency for signal analysis. Before, decimation factors could only be applied if they were a power of 2 (i.e 2,4,8,16,...) up to 65536. Now, the factors can be any power of two up to 16, and any whole number greater than 16 up to 65536. Further information can be found here.

Some changes have been made to the FLIR readout code to help improve its functionality:

• More accurate temperature readings than before due to updates in the calculation procedure. A bug was causing one of the parameters to not update correctly; this is now fixed.
• Saved data now stored in HDF5 files rather than CSV.
• User can now enable automatic data storage by specifying a collection interval (in minutes). The choice of manually saving data is still present if desired.

Summary

Today I finished implementing an RTS model to read out the integrated FROSTI RTDs (temperature sensors) via the CyMAC. The model is named "MSC" and is located at cymac:/opt/rtcds/usercode/models/c1msc.mdl. We successfully tested it with the heater elements operating in vacuum at low power (12 VDC), finding them to reach an average steady-state temperature of 160 C.

From the cymac host, the MEDM control screen can be accessed with the terminal command "sitemap" (from any directory).

Measurement Technique

Each FROSTI heater element [299] contains an internal two-wire RTD placed near the front emitting surface, which enables the temperature of the blackbody emitter to be directly monitored. From the measured temperature and the emissivity of the uncoated aluminum nitride surface (known to be ~1 in the IR), the radiated source-plane power can also be estimated.

The resistance of each RTD is measured via a ratiometric technique. The RTDs are powered in series with a 1 kΩ reference resistor located inside the readout chassis [305], whose temperature is not changing. The signal is obtained by taking the ratio of the voltage difference across each individual RTD to the voltage difference across the reference resisitor. The advantage of this technique is that the ratio of  the voltage differences is insensitive to changes in the current through the resistors (since they are all in series; see [271] for wiring diagram).

Implementation Detail

The signal flow is shown in Attachment 1. The eight RTD signals enter through ADC channels 0-7, along with the reference resistor signal on channel 8. The first set of filter modules apply a calibration gain to convert the signals from raw ADU counts to units of input-referred voltage. The ratio of each RTD signal to the reference resistor signal is then taken. The second set of filter modules multiply the voltage-difference ratios by the resistance of the reference resistor, 1 kΩ ± 0.01%, to obtain the RTD resistances in physical units of ohms.

Finally, a freeform math module is used to invert the quadratic relation between each RTD's resistance and temperature. The final signals passed to the third set of filter modules are the RTD temperatures in physical units of degrees C. The temperatures of the tungsten RTDs are estimated assuming TCR coefficients of A=0.0030 C^-1 (±10%) and B=1.003E-6 C^-2, which were provided by the manufacturer.

One DAC channel is used to provide the excitation voltage for the RTD measurement, which is visible on the far right of the control screen. At its maximum output voltage of +10 V, the DAC can drive a maximum current of 10 mA.

I performed another continuity test on the RTD chassis wiring, and everything seems to be set up correctly. The chassis should be ready for installation.

Below is the current state of the RTD readout chassis wiring. Initial continuity tests seem good, will run through one more time to confirm.

Today the FROSTI RTD readout chassis underwent a redesign:

Instead of the original ratiometric method, which involved wiring the FROSTI RTDs in series, each element is individually powered by separate excitations. Each element additionally possesses its own reference resistor of 100 Ohm. Now, if an RTD experiences an electrical short, it should not affect the measurements of the others in sequence, as it had with the original design.

The custom front and rear panels for the RTD readout chassis arrived last Friday. I installed them in the chassis frame to check their fit. They fit very well, so all that now remains is to complete the internal wiring and test the connections.

The chassis panel designs are archived to LIGO-D2300452 and LIGO-D2300453.

We noticed that the RTD temperature readings given on the Cymac were off, and traced the issue to miscalibration in the relationship between the resistance and temperature of each RTD (Callendar-Van Dusen eqn). Below is the table of values inferred from independent measurements of temperature and resistance to rectify this problem. This data was then fitted to better determine the coefficients present in the temperature-resistance relation:

       R_0 (ohm)   Alpha    Beta

RTD 0   80.8674   0.001315   4.273e-6

RTD 1   79.5704   0.001887   3.7873e-6

RTD 2   81.7334   0.002014   2.1724e-6

RTD 3   74.3060   0.003677   3.6022e-8

RTD 4   81.1350   0.001761   2.3598e-6

RTD 5   77.9610   0.002423   -7.5192e-7

RTD 6   78.7980   0.001373   6.2909e-6

RTD 7   83.8616   0.001890   3.3529e-6

       R_0 (ohm)   Alpha (1/C)

RTD 0   79.3962   0.002031

RTD 1   78.2874   0.002530

RTD 2   80.9775   0.002381

RTD 3   74.2947   0.003684

RTD 4   80.3199   0.002157

RTD 5   78.2106   0.002297

RTD 6   76.6825   0.002438

RTD 7   82.6645   0.002458

Measurements taken can be found here. An uncertainty of 1 C was assumed for temperature.

Another re-fit, but this time the quadratic coefficient (beta) is set to 1.003e-6:

       R_0 (ohm)   Alpha (1/C)

RTD 0   79.7386   0.001863

RTD 1   78.6248   0.002359

RTD 2   81.3254   0.002211

RTD 3   74.6127   0.003509

RTD 4   80.6652   0.001988

RTD 5   78.5450   0.002127

RTD 6   77.0144   0.002268

RTD 7   83.0204   0.002288

Using the data taken during the FROSTI testing at Caltech, I attempted to find a better calibration of the RTD sensors, given our past issues with inaccurate readings. The fit parameters, alpha and beta, are still all different from the initial values given to us by Fralock (alpha = .003, beta = 1.003e-6, R_0 was not given), but the true values will differ based on factors such as part geometry.

Below are a basic diagram of what the RTD measurement circuit logically looks like and an example schematic of the actual wiring. The schematic wiring will be placed internally into a chassis, connected to the RTDs via DB25 cable.

After updating the wiring in the RTD Chassis, a signal is now seen at each ADC input. However, there seems to be a discrepancy between the voltages I measured out with the multimeter (see below). Next steps include:

• Finish final debugging
• Calibrate ADC inputs with known voltage source (likely to use DAC).

Voltage Readings:

RTD 1: 0.576 V

RTD 2: 0.578 V

RTD 3: 0.598 V

RTD 4: 0.563 V

RTD 5: 0.477 V

RTD 6: 0.463 V

RTD 7: 0.456 V

RTD 8: 0.491 V

Reference Resistor: 5.463 V

Total Voltage: 9.665 V

After further modification of the RTD readout chassis (i.e. adding resistors, placing reference resistor in front of RTDs), here are the following direct measurements:

RTD 1: 0.484 V

RTD 2: 0.486 V

RTD 3: 0.503 V

RTD 4: 0.474 V

RTD 5: 0.495 V

RTD 6: 0.483 V

RTD 7: 0.476 V

RTD 8: 0.510 V

Reference: 5.847 V

Here are the Cymac signal readings:

RTD 1: 74

RTD 2: 67

RTD 3: 73

RTD 4: 45

RTD 5: 82

RTD 6: 75

RTD 7: 70

RTD 8: 71

Reference: 884

The one (possible) discrepancy here is the readout for RTD 4 via Cymac, since it's signal reading is ~30 counts lower than the others. I do not believe this is a wiring issue due to the direct measurements taken.

After initial analysis from last week on a single RTD, I then extended to looking at all 8 in series with R_ref (set to 1 kOhm). Shown below are the edge cases for the setup:

• RTDs are all at ambient lab temperature. This would correspond to a minimum resistance value.
• RTDs all read out 400 C. This gives the maximum resistance value.

The results show that indeed only a few mA of current is drawn even at room temperature (a little above 5.5 mA), and this will continue to decrease with increasing temperature. The voltage across a single terminal, at a maximum, is only about 5.4 V.

Attached below are updated plots for the FROSTI RIN measurements for Jan 14 group meeting.

Attached below are updated ASD plots for the FROSTI RIN measurements. The parameters set for this are the following:

• DFT Size (N): 16384
• Sampling Frequency (F_s): 7.629 kHz
• Resolution (F_s/N): 0.47
• CH0 DC Voltage w/FROSTI ON: 113.6 mV
• CH0 DC Voltage w/FROSTI OFF: -5.7 mV
• CH1 DC Voltage w/FROSTI ON: 113.0 mV
• CH1 DC Voltage w/FROSTI OFF: -5.7 mV

Each measurement was recorded over a roughly two-day period. Before each spectrum was computed, the time-series signals were AC-coupled (i.e. the DC offset was subtracted from the data). The low-pass filters are still attached for dark and light noise measurements. ADC noise is measured with two 50-ohm terminators attached to the Red Pitaya inputs rather than the IR photodetectors.

I tried adjusting the gain settings on the photodetectors to check if this would help improve the RIN spectra measurements. Overall, it doesn't look like it does, and if anything, looks worse. I assume this is so because as the gain is lowered, the amount of detectable signal from the FROSTI becomes smaller and smaller.

We've added two low-pass filters in hopes of reducing any potential aliasing that may be introducing additional noise into the power spectra for the RIN measurements. It still looks like the noise levels are too high. Attached below are some recent measurements taken with the FROSTI powered on and off.

I downgraded the Red Pitaya back to OS 2.00-18 due to runtime errors during measurement. Once I did this, the device appeared to work much better than it has the last few weeks. First, it appears we can actually see the cutoff of the added low-pass filters that were added in to the RIN setup. Second, there does appear to be a difference again between the FROSTI ON state versus the dark state (i.e. FROSTI OFF). A long measurement of the ADC noise floor in the current configuration still needs to be recorded, but it does appear that the recent highlighted issues with the Red Pitaya have been solved.

For some preliminary tests, I moved the IR photodetectors outside of the cleanroom and onto the other optical table. The basic goal was to obtain a signal from both photodetectors. To achieve this, one of the heater cartridges used for early FLIR measurements months ago was hooked up to a power supply (PS). The PS was set to supply 0.20 A with a voltage of 2.8 V; the corresponding power is thus 0.56 W. With this, I was able to measure a signal using the Red Pitaya, the device that will be used for following RIN measurements.

I have begun moving parts into the cleanroom for the upcoming FROSTI RIN tests that will be conducted within the next few weeks. While waiting for the rest of the equipment to arrive to perform the full-scale tests, I have additionally moved the FROSTI under the shelf above the optical table, where it will stay for the meantime. As always, please use caution when in the cleanroom. Aside from the FROSTI, the IR photodetectors that will be used for the test are delicate and costly to replace.

[ Luke, Jon ]

Started work at 11:00

As mentioned in my previous post the RGA was not clearing the table because of a tilt in the cross. So I removed the 90 deg elbow, loosened the bolt securing the Tee to the ZLR, and removed the cross from the gate valve. I then spun around the Tee on the bottom so that the cal leak would be pointing in the right direction and connected the rotating 2.75" flange of the cross to the gate valve. I then added the small turbo pump to the top of the cross. 

Then with the help of Dr. Richardson, we made some adjustments to how the Tee was oriented with respect to the cross and how the cross was with respect to the table. So that the cal leak and RGA would fit on the table. After that we made some changes to the bolting of the ZLR to reducing cross replacing the 2.00" bolts and nuts with 1.75" bolts and nut plates. We did face some difficulty with half of the 2.00" bolts which required a bit of torque to get them out. 

Finished work at 2:20 

Things to do:

Before we start pumping down the system again Dr. Richardson wanted two other changes to be made. To replace the 2.75" to 1.33" conical reducer with a zero length reducer. He also wants the bolts that hold the main turbo pump to be replaced with nut plates. 

I plan on starting this work on Friday (7/12) before pumping down over the weekend

[Luke, Luis, Jon, Tyler]

On the 22nd Dr. Richardson showed Luis, Tyler and I how to preform a leak test with the RGA. We did the initiall test and found a few leaks two were particuarly bad highlighted in orenge below. To try and remidy this we planned on replacing the copper gaskests on the leaking flanges. We then began taking the two problem flanges off, but seven of the eight bolts holding on the turbo pump were over tightened and had seized. So after we got them off we postponed the rest of the work to the next day.

On the 23nd Luis and I reattached the badly leaking flanges with new copper gaskets. We then preformed the Helium leak test with the RGA. As seen in the table below we weren't able to majorly changed the leaks in the two flanges.

[Luke, Tyler]

On the 27th Tyler and I ran the RGA leak test again with the electron multiplier on. These were the leaks we measured. The chamber's overall pressure was at ~6e-8 torr. 

[ Luke, Cynthia, Michael, Xuesi, Anthony ]

Started work around 12:45

We vented the chamber first and once it reached atmospheric pressure we started taking apart the RGA line.

We removed the parts in this order: Calibrated leak, RGA, Pressure gauge, 2.75" Blank on Tee, 2.75" Tee, all 1.33" Blanks on the cube, 4.5" feed through port, and finally the cube. Everything went smoothly after every part was taken off we covered the ends in Aluminum foil to maintain cleanliness. 

We then started the assembly of the new line. The parts were added in this order: Reducing cross, ZLR, Tee, Cal leak, RGA, 90 deg Elbow, and Pressure sensor.

We left off the Turbo pump as of now because we weren't able to find a 1" post not in use and we didn't want to put too much weight on the cross. We also noticed that the RGA might not fit on its probes because of a slight tilt to the cross and by extension the Tee. If this ends up being a problem we might need to remove all the parts connected to the cross so that we can reposition it so that the 2.75" rotating flange is connected to the gate valve.

Ended work around 4:00

1. Run a second RGA degas cycle, but next time with the main volume valved off with only the RGA volume being pumped (through the bypass line). This will prevent "boiled off" particulate from entering the main chamber and will also increase the pumping rate for the RGA volume, reducing the amount of particulate that resettles on the RGA filament.
   
   
2. I also noticed that the SRS manual states that the filament is designed to be long-lived and it is recommended to leave it on any time the RGA is on. By leaving the filament on all the time (i.e., hot), we could reduce the amount of particulate that is evidently settling on it between scans. I am checking whether LIGO Lab does this in their own chambers.

As a follow up ELOG 261, I have received advice from one of the vacuum experts at LIGO Hanford on best practices for our RGA:

1. For future RGA degassing, definitely keep the main volume isolated, since it could contaminate the main volume with everything that just got cooked off of the filament. So the procedure should be to (i) close both gate valves, (ii) ensure the angle valve on the bypass line is open, (iii) initiate the degas cycle on the RGA, (iv) pump the RGA volume through the bypass line, until its pressure returns to its pre-degas level.

2. Repeated degassing of the filament will definitely wear it down much faster, so do this operation sparingly.

3. As long as the pressure of the RGA volume is in the UHV range (~1e-9 torr), best practice is to leave the filament on. This keeps it hot which helps prevent particulate from settling on it. However the electron multiplier should stay off when not actively taking scans, as it will wear down if left on all the time.

[Luis]

I rebooted Spica and I was able to connect the RGA again. After Connecting the RGA, I left the filament on.

[Luis, Luke}

Three RGA scans were taken. The improvement in the amount of HC in the vacuum is visible across the different measurements. Images are attached.

The following images compare the RGA scans from 9/23/24, after the first bake with the new vacuum system, with those from 3/14/24, after bake 12 with the old system.

The first image shows a graph of the raw data and includes the calibrated leak for both curves. As we can see, our new system meets LIGO standards of cleanliness.

The second graph contains the plot of the normalized data.

[Luis, Luke]

RGA Volume: T = 21°C, P = 1.74e-9 Torr
Main Volume: T = 22°C, P = 5.60e-9 Torr

We performed an RGA scan of the system. As we can see, we met cleanliness standards. The S/N ratio was roughly calculated and it is about 100. I will update the code to make this calculation and to have it displayed on the plot in the next scan we perform.

After this, we closed the RGA to the main volume valve, degassed the filament, and after the pressure came down to what it was initially, we reopened the valve. After the degassing, we took another scan. The overlaid plots are attached below.

The noise floor seemed to be about the same. I will take another scan tomorrow to verify whether it was the same or if the system just needed a little more time to be completely clean.

[Luis,Shane]

An RGA scan was taken on Dec 4th. The results are attached at the end of the e-log. As we can see, the scan results are under the allowed limits. After the scan, Shane and I worked on the vacuum connections. The filament was turned off, and the RGA was disconnected from Spica.

After we finished the tests, we reconnected the RGA and attempted to turn the filament back on. However, the program was unable to connect to the instrument. We checked for any loose connections but did not find anything. The RGA connection needs to be troubleshot and fixed.

The following figure displays data acquired on 10/09, and it shows that we are no longer below the cleanliness standard. Also, the system's pressure went up to 1.9x10^-8 Torr (approximately 7x10^-9 Torr previously). This might be due to a leak somewhere in the system. More tests will be performed later.

[Luke, Luis, Mary, Ma]

On 10/22/24 Luke, Ma, Mary and myself ran a RGA scan, the results are displayed below. The overall pressure of the vacuum was 2.0e-8 and the temperature readings were 26C for the RGA and 25C for the main volume.

As we can see, the vacuum is passing cleanliness standards again.

[Ma, Luis]

February 20, 2025 - 3:30 PM

Ma performed an RGA scan with the vacuum system containing the heater elements. The results are attached below.

As we can see, we are not yet under cleanliness standards. The passing threshold is 4 × 10⁻¹⁰, we are very close to it.

[Ma, Luis]

February 24, 2025 - 1:00PM

Ma performed an RGA scan with the vacuum system containing the heater elements. The results are attached below.