Note on Pytest

Often our code uses pytest and Lee's version wield.pytest. If given a file with these tests and they don't run, add the attached files to the folder. They tell pytest what to do and how it handle it correctly.

to make a test make a function with a name that states with "T_". Then run the command:

or replace the period with the file name and it will run all the tests in that file. Us the code in post about the GQuEST Spectrum models to see an example.

Correcting the Flat FOM gain to always be 1

[Ian, Lee]

NOTE: this should have been posted on June 21 2023 as this is the date that it was writen, but it was in my drafts for some reason. I have added latex since then

In the H2 system the gain for the flat FOM (F2) is 1/F1_gain, where F1_gain is the gain of the BNS FOM (F1). This works fine for the H2 norm because that is made up of the composite RMS from both FOMs so as one FOM's gain increases and the others decreases they cancel out. However, this is not the case in the H_infinity case. The H_infinity norm is the maximum value for the frequency response from the Zinf input to the Flat FOM output this value must be less than or equal to the gamma that we are using. Currently we are searching for a gamma around 1. The equation for the frequency response along the  Zinf input to the Flat FOM output is:

\[ \left | \frac{K(\omega)P(\omega)}{1-K(\omega)P(\omega)} F_{\text{flat}}(\omega)\right |\leq \gamma \]

where \( K(\omega)\) is the controller response, \(P(\omega)\) is the plant response and \(F_{\text{flat}}(\omega)\) is the Flat FOM's response. If the environmental noise is large like at low frequencies then \( K(\omega)P(\omega) \) must be large to control it. Thus at low frequencies the above equation becomes,

\[ \lim_{KP \to \infty }\left | \frac{K(\omega)P(\omega)}{1-K(\omega)P(\omega)} F_{\text{flat}}(\omega)\right | \approx\left |F_{\text{flat}}\right | \leq \gamma \]

Thus if we use the F2_gain=1/F1_gain then as F1_gain gets very small F2_gain gets very large. When F2_gain is very large, the gamma we are searching for (around 1) is much less than the actual H_infinty norm which is on the order of the F2_gain and the solver fails. This is what was causing the problems with the RMS plot. 

We switched the F2 gain to be all ones and acquired attached plot.

Note: when this PDF is printed the second color bar has the same colors as the first color bar.

---

Improvements for code and plots for paper:

1. Fix the color bar problem (noted above) for printing.
2. Make the color bar discreet
3. decrease the number of F1_gain points so that the plot is not so crowded 
4. Maybe add an H2 boundary line. See how it looks
5. change the F1_gain color bar. maybe connect like F1_gain points with a subtle line.
6. Add an example of the model maker (getSPOFF) to Buzz
7. make the code transpose the model right before solving in the H_infinity case

Geontropic Spectrum models

Here is python code as a wield.pytest that generates the Geontropic noise model from arXiv:2209.07543 [gr-qc, physics:hep-ph, physics:hep-th].

You'll need to install wield.utilities and wield.pytest to run. You can just drop  T_GQuEST_concept.py and GQuEST_models.py  in a folder and call pytest. The output is

Total power displacement 2.3306928419630585e-18
Effective_BW: 57.28 MHz  by c/L 0.954696717877494
Effective_BWsq: 14.68 MHz  by c/L 0.24468351704291535
Frequency of Max: 11.29MHz
3db_bandwidth: 12.70MHz
Total power displacement2 2.3306928419630585e-18
3db points:  [[ 5.49516136]
 [18.19782777]]
Effective_BW: 57.28 MHz  by c/L 0.954696717877494
Effective_BWsq: 14.68 MHz  by c/L 0.2446835170429154

GQuEST_model.pdf Shows the model with and without the IR cutoff. It also includes the integrated RMS. Note that the power spectrum falls off only as 1/F, so has a logarithmic divergence. We presume this logarithmic divergence gets cut of a c/w, for c the speed of light and w the beam radius. That's at about 100GHz for GQuEST's beam. That divergence is physically important, but not practically measurable.

GQuEST_model.pdf shows the same plot, but with overlayed axes to allow easy comparisons with the units of the Geontropic noise model paper. The code shows how to set up the plots for such fancy axes.

there are also plots showing the model in the Holometer.

 

Both of the plots also show that there is a reasonably close analytic formula for the noise. In python it is

Omega = 2*np.pi * F
Omega_SPEC = c_m_s/(2**0.5*L)
Canalytic = 2 * Cmax * Omega**2 * Omega_SPEC / (4 * Omega_SPEC**4 + Omega**4 - Omega**2 * Omega_SPEC**2)**0.5 / (Omega**2 + Omega_SPEC**2)**0.5

Which can be updated to be

\(\Omega = 2 \pi f\)

\(\newcommand{\obar}{\overline{\Omega}}\)\(\obar = \frac{c}{L}\)

\(S_L(f) \approx \frac{2 \Omega^2 \obar}{\sqrt{\Omega^4 - \frac{1}{2}\Omega^2 \obar^2 + \obar^4} \sqrt{2\Omega^2 + \obar^2}} S_L(f_{\text{pk}})\)

 

and Cmax is the maximum of the power spectrum. The above are in power, not amplitude, units.

I'll edit or add a comment with the Latex form once I have it.

 

Support manual for Boostik HP Amplifier

I requested the operating manual for our old-style Boostik HP amplifier. When we update the SOP's to include the amplifier, this should probably go with them.

 

Standard Operating Procedures for B102

https://dcc.ligo.org/M2300159

Rough draft for our swing space's SOP. We moved the fire extinguisher in the hall into the room. Just need the actual laser hazard sign out front now I believe and we can go laser hazard (no amplifier) in the swing space. Although we probably want to do some sort of cleaning before we expose optics to the dirty room.

Moving the Plant to Environmental Shaping Filter for Riccati Solver

In the Buzz code we move the plant to shape the seismic noise without shifting the controls noise is:

\[ S_c = \left |\frac{EP}{1-KP}\right |^2  S_\mathrm{env} + \left |\frac{MKP}{1-KP}\right |^2  S_\mathrm{meas}  \]

where \(K\) is the controller, \(P\) is the plant, \(E\) is the environmental noise shaping filter, \(M\) is the measurement noise shaping filter, \(S_c\) is the controls noise PSD, and  \(S_\mathrm{meas} \) and \(S_\mathrm{env} \) are the unshaped PSDs of the respective noise. If we move the plant to the env noise shaping filter such that \(E'=PE\),  \(P=1\), and  \(K'\) is a new controller calculated for this system then the controlls noise becomes

\[ S_c' = \left |\frac{EP}{1-K'}\right |^2  S_\mathrm{env} + \left |\frac{MK'}{1-K'}\right |^2  S_\mathrm{meas}  \]

In order for these two ststments to be equlivilant

\[ K'=KP \]

I think this means there needs to be some sort of elimination of the \(P\) from the \(K'\) since what we are looking for is \(K\) and not \(K'\).

More cryolab cavity locking

I moved the beam profiler to in reflection of the cavity to see if there was a discernable difference between the REFL beam when the cavity is locked and unlocked. There was not (see attached, should also more into this new beam profiler software and better ways to record data). I have not recovered alignment of the TRANS beam on the beam profiler. Aaron lent me a camera to use in transmission but we suspect the power incident on them is most likely too low for the poor QE camera. It seems like the TRANS beam should be much easier to find than this as the beam profile and camera diode are quite large. The cavity still appears to be locking on the mode we want.

To combat this I increased the power from the laser to 2.2mW coming from the source (allthough this puts you close to the maximum range on the TEC controller). This corresponds to ~1.3mW incident on the cavity. The 1811 PD saturates at far lower than this so I have a .5 ND filter in front of the PD.

Aaron thinks (from the elog I believe) that they were able to get up to 7mW out of the laser in the past. May be worth looking into this, but currently the range on the TEC controller doesn't allow this.

PDH Lock on the CMTF Table with a Red Pitaya

We used a red pitaya to lock a Mephisto laser to a reference cavity. Once the TEM00 mode is within the Mephisto's piezo's scanning range, a python script will automatically lock the laser to the reference cavity. The system held lock for over 30 minutes and was only unlocked when we disconnected the computer from the red pitaya to run other tests. We successfully locked and unlocked the laser from the cavity several times and the locking script worked every time. The lock acquisition is nearly instantaneous upon initiating the python script. Chris also successfully ran the script and locked the cavity, and doing so doesn't require much know-how about the specific setup. The script automatically calculates the PID set point and the initial voltage to feed the piezo when initiating a lock. The feedback loop is implemented as just an integrator, and could certainly be tuned up further. We have not yet implemented slow feedback control to the laser's temperature, so the laser will eventually loose lock if its frequency drifts relative to the cavity's by more than about 80 MHz. I uploaded the locking script to a github repo for the lab. It should be generally useful for PDH locking optical cavities with FPGAs. Attached files show the curve fit used to calculate the PID set point and initial piezo voltage, the locked TEM00 mode on a camera, and a schematic of the controls scheme.

Steady lock achieved on west cavity

[Lee, Sander, Torrey]

We are able to lock on the cavity mode from my previous post. This is a dump of the moku/loop gain settings for future reference. 

Cryo Cavity Alignments

[Ian, Torrey]

We found the beam in transmission of the cavity when it flashes. Alignment of the beam into the cavity has so far been unsuccessful. This screenshot is an example of the beam we see. So far the alignment process has been:

1) Scan the cavity at large amplitude with the current to locate the REFL dip.

2) Lower the scan amplitude and adjust the temperature according such that the REFL dip stays in the scan.

3) Repeat until a scan is no longer needed and manual park the cavity on resonance using the temperature.

4) Attempt to adjust the input mirrors while parked on resonance.

There a few problems with this. First, the smallest step size we can take on the temperature is too large. Makes it difficult to park on resonance. Second the stability of the laser's frequency isn't the best for this process. When attempted to manually park on resonance it will often drift out of range in a few seconds. The next logical step would be to lock the cavity on this mode, but we have been unsuccessful in doing so. Either the loop shape isn't quite right or the mode resonant in the cavity is just too garbage to lock on. 

 

Few more examples of things we saw: Aliasing effects, some garbage but was fixed due to aliasing effects of our scan frequency and the scanning slit frequency. Unsure what this is, we've seen it a couple of times but I'm not sure it is significant. Would be good to find out what kind of lenses are in the cavity chamber.

GQuEST CMTF Laser Alignment to Cavity

Edoardo and I are aligning the laser in the CMTF to an optical cavity. We can get a lot of optical power into TEM01 and TEM10 modes, but can't manage to get power into a TEM00 mode, or find any spike corresponding to this mode on an oscilloscope when sweeping the laser's frequency by tuning its fast PZT input. We're driving the PZT from 0V to +100V and the laser's frequency shift per volt was previously measured to be 1.66 MHz/V (the spec in the manual is 1-2V/MHz). Calculating the frequency difference between the TEM00 and TEM01 modes, we find that it should be 150 MHz, so in fact we should not expect to see a TEM00 spike on the oscilloscope if the TEM01 mode is roughly centered on the scope, since our scanning range is only 165 MHz. To calculate this, see eqs. (10) and (11) in https://opg.optica.org/ao/abstract.cfm?uri=ao-23-17-2944 Given this information, we plan to adjust the laser temperature so the TEM01 mode is slightly out of the scanning range and then look for a spike corresponding to the TEM00 mode. We will update with our success or lack thereof.

Pump Quotes

Key: Pump (Pumping speed, minimum pressure, cost, size)

 

Agilent

Rougher:

IDP-3 (3.6 m^3/h, 0.25 Torr, $3610, 5.5 x 7 x 14 in) ($3,286 on Pro Vac)

IDP-7 (9.1 m^3/h, 0.02 Torr, $6,074 10 x 10 x 17 in)

IDP-10 (10.2 m^3/h, 0.02 Torr, $6,883, 10 x 12 x 17 in)

IDP-15 (15.4 m^3/h, 0.01 Torr, $8,705 , 13 x 14 x 19 in)

TriScroll 400 (15 m^3/h, 0.013 Torr, $8064, 11 in x ? x 14 in)

 

Turbo:

TwisTorr 74 (70 L/s, 4*10^-10 Torr, $5832, 2 kg)

TwisTorr 84 (80 L/s, 4*10^-10 Torr, $6252, 2 kg)

TwisTorr 305 (300 L/s, 8*10^-11 Torr, $11301, 5.8 kg)

TwisTorr 404 (400 L/s, 8*10^-11 Torr, $12727, 22.6 kg)

 

Integrated (seems like a bad deal unless they include a lot more than the individual pumps):

TPS-mini: TwisTorr 74 + 0.6 m^3/h diaphragm)

TPS-conpact with TwisTor 74 (3 m^3/h, $13944)

TPS-conpact with TwisTor 305 (3 m^3/h, $17,636)

TPS-flexy IDP-3 + TwisTorr 74: ($11937)

TPS-flexy IDP-3 + TwisTorr 305: ($17486)

TPS-flexy IDP-7 + TwisTorr 74: ($14423)

TPS-flexy IDP-7 + TwisTorr 84: ($15441)

TPS-flexy IDP-7 + TwisTorr 305: ($19594)

TPS-flexy IDP-10 + TwisTorr 305: ($19490)

 

Pfieffier (asked for quotes)

Rougher:

HiScroll 6: (6.1 m^3/h, 0.01 Torr, ?, 19 kg) ($5,888 from high vac depot)

HiScroll 12: (12.1 m^3/h, 0.004 Torr, ? 24 kg) ($7,709 from high vac depot)

HiScroll 18: (18.1 m^3/h, 0.004 Torr, ? 23 kg) ($9,144 from high vac depot)

MVP 030-3 (1.8 m^3/h, 1.5 Torr, ? 4.3 kg) (diaphragm)

MVP 070-3 (4.3 m^3/h, 0.75 Torr, ? 16.4 kg) (diaphragm) (Website doesn’t list under compatible with Turbos)

 

Turbo:

HiPace 10 (final pressure 4*10^-5 Torr)

HiPace 30 (32 L/s N2, 4*10^-10 Torr, ?, 3.2 kg)

HiPace 60 P (64 L/s N2, 7.5*10^-7 Torr, ?, 2.2 kg) *note bad pressure*

HiPace 80 (67 L/s N2, 4*10^-10 Torr, ?, 3.8 kg) ($6,555 from high vac depot)

HiPace 80 Neo (67 L/s N2, 8*10^-8 Torr, ?, 1.7 kg)

HiPace 300 (260 L/s N2, 4*10^-10 Torr, ?, 19.2 kg) ($13,110 from high vac depot)

ATH 500 (350 L/s N2, 8*10^-9 Torr, ?, 37.5 kg)

 

Integrated:

HiCube Eco backing pumps MVP 30 or smaller

HiCube 80 Classic

HiCube 80 Pro

 

Edwards

Rougher:

mXDS (3 m^3/h, 0.075 Torr, $4945, 8 kg)

nXDS (6-20 m^3/h, 0.005-0.022 Torr, $7,800-$14,000, 25kg) ($6000-$10,300 from high vac depot)

4 sizes

XDS (40 m^3/h, 0.02-0.06, $16,500-$18000, 48) ($11,692 from high vac depot)

 

Turbo:

nEXT55 (55 L/s N2, 1*10^-7 Torr, $8,174, ?)

nEXT85 (47-85 L/s N2, 4*10^-10 - 4*10^-9 Torr, $8,582, 2.9-4.4 kg) ($6,763 from high vac depot)

nEXT 240 (240 L/s, 4.5*10^-8 Torr, $13,195, 5.7 kg) ($11,000 from high vac depot)

nEXT 300 (300 L/s, 4.5*10^-8 Torr, $13,858, 5.7 kg)

nEXT730 (680 L/s N2, 1-6*10^-10 Torr, $20,742, 19.6 kg)

 

Integrated (don’t come with scroll)

T-Station 85

T-Station 300

 

 

Leybold

Rougher:

Screw Pump:

Ecodry 25 plus (25 m^3/h, 0.01 Torr, ?, ?) (I reached out for pricing)

 

Scroll Pumps:

ScrollVac 3 plus (3.5 m^3/h, 0.08 Torr, $4744, ?)

ScrollVac 3S plus (3.5 m^3/h, 0.08 Torr, $5219, ?)

ScrollVac 7 plus (36.1 m^3/h, 0.02 Torr, $7586, 26 kg)

Also have larger pumps

 

 

Iwata

Rougher

Scroll Meister

ISP-50 (3.6 m^3/h, 0.11 Torr, ?, 12 kg) ($4,835 from high vac depot)

ISP-90 (6.5 m^3/h, 0.037Torr, ?, 13 kg)

 

TMP-B70 (70 L/s N2, 10^-8 Torr, ?, 5.2 kg)

I reached out for pricing

 

I also checkedtequipment.net but nothing was cheaper than high vac depot

Cryolab Setup Diagrams and Lock Settings in multi-instrument mode

Attached are

1. The laser lock box settings via screenshot in multi-instrument mode using the PLL as the modulation setting.

2. Diagram of multi-instrument mode for reference.

3. Ariel photo of the set up.

4. Inkscape drawing of the set up.

Also should note when scanning the cavity using the current it should be safe to use the full range (2V). The TEC controller should be set to ~10.5kOhms, but double check the range on the integrator is roughly centered.

SNSPD Calibrations

Andrew Muller and I have completed our calibrations for the new SNSPD system which utilizes a Yokogawa frame controller. The results can be seen in the first attached image. On the left side of the table we list the numbers achieved on the older system and on the right the Yokogawa. These values include Dark Count (DC), Theoretical or computed number of expected photons/second, maximum and minimum achieved photon rate (as a percentage of theoretical rate), as well as the applied bias voltage on the SNSPD. The results for the dark count are varied between the two systems as the older unit had shielded fiber optic cables and the new system had not received this upgrade yet thus we saw an increase in dark counts by an order of magnitude. Both systems were given the same target rate of 100k photons/second and through calibrations made in python that are seen in the code "photonrate.py" and the newer "photonrate_yokogawa.py" a theoretical maximum photon rate (109k) was calculated based on the readouts on the power meter while cycling through each attenuator module to in order to account for differences or fiber loss. The code then automatically sets and attempts to achieve the desired rate. The efficiencies are then calculated by the resultant photon rate, with respect the the theorized rate, readout by the counter (seen as item 1 in the second image). Maximum and minimum values represent that of the highest and lowest achieved rates when adjusting the polarization of the optical fiber exiting from the laser which affects the rate readout on the counter (but not the power meter). Our results showed that the newer Yokogawa system was capable of achieving very similar and slightly higher efficiencies than the older unit which came in about roughly 72% and 71% respectively. The Thorlabs power meter is shown as item 2 in the second image, followed by the older attenuator and switch modules as item 3, and the new Yokogawa system, item 4. The Yokogawa frame controller has a number of compatible modules which include an in unit laser, power meter, switch and attenuators. The Yokogawa power meter seems to have a broad range of power input (larger than the Thorlabs meter) and more tests will be done to test its accuracy and limits. Code will be uploaded to git this Tuesday.

Notes from Meeting with Maty Lesovsky

I had an impromptu meeting/tour with Maty Lesovsky. The following are the notes I took

 

Good needle valves for bringing things up to pressure

All metal angle valves

Avoid rubber oring

 

10^-10 l mbar/s cm^2 is LIGO’s desorption

 

Have metal bake out site

 

All metal bake out valves

KJL has rubber o ring which sucks in contaminants

Varian is a good brand

VAT makes large ones

 

Use research grade N₂ if you want to backfil

Can hook up to needle valve

Clean air cheaper

 

They have an SRS RGA 200

Realistically only gets to 100 amu

Hanford uses 1000, but this SRS should be plenty

 

Have custom controllers for bake out

8 W/ft

Takes 110-220 V

PID controller from computer, can tell ramp time and other parameters

They have prints of the circuits

 

They have thermistors on the outside

Get surface temperature, not in air

Tac weld for surface temperature

Omega or McMaster

 

They also have insulation we can use

 

200°C is standard for small stainless

This is 150°C because it’s so large

They think they’ll keep the turbo on during the bake out

Biggest concern is machining oils

 

Fiberglass tape (from McMaster)

Goes up to 300°C

Great for holding thermocouples

Has gone through hundreds of bake outs over years, been fine

Avoid fiberglass insulation

 

Iwata Scroll Meister pumps, equivalent to the IDP-7

Tri scroll

Have mufflers to trap chemicals out of them

 

Smart to have a safety shutoff in case you lose power

Cut power to turbo

Choke turbo if it comes back on with pressure

 

Do a rate of rise test on partial volumes and the whole system

Good to test if something is degrading

 

Have a battery backup for a couple hours to run the turbo so the turbo breaks don’t wear out

 

For controllers, have Terranova Model 934 wide range which are good

MKS PDR 2000 also nice and easy

Invest in a case to display everything

 

They have a lab view program to monitor the pressure and the RGA

Pfeiffer Vacuum Calculator Initial Impressions of Pump Requirements

Using https://vacuum-calculator.pfeiffer-vacuum.com and the previous post's volume and surface area data, I have some inital impressions of our pump requirements for the 8 m long interferometer.

Other assumptions:

Desoprtion: 1e-9 mbar*l/(s*cm²)

Leak rate: 3e-9 mbar*l/s

1. Our final pressure limitation will be a function of desorption, not leak rate, because our surface area is so huge

2. The time to reach 6 Torr, where Pfeiffer starts the turbos, is given approximately by 1 hr * (4.3 m^3/h rougher pumping speed / pump roughing speed)

3.

A. 3 HiPace 80 DN 63 get to a pressure of 10^-7 Torr in 12 hours, 10^-8 Torr in a week

B. 1 HiPace 80 DN 63 gets to a pressure of 10^-7 Torr in 36 hours, 5*10^-8 Torr in a week

4. A 1 m, 6.45 mm tube between the roughers and the turbos doesn't greatly affect the pump down time or the final pressure

CMTF Mode Match to Optical Cavity

On Friday, Chris and I worked to mode match our laser to an optical cavity. At the end of the day, we placed our lenses in roughly the location suggested by a mode matching calculation, but the mode reflected off of the cavity appears much larger than the incident mode, so we believe we need to move the lenses further to improve mode matching. The attached file shows the calculated mode matching solution given a beam profile of the incident beam and the waist size and location of the reference cavity.