Flood photo album: https://photos.app.goo.gl/BZAG8DyQzFVTfMNz6 (This link is read-only who has no access to the account)
Facilities placed a blower and dehumidifier in B265B. I checked the airflow and the air around the tables is comparitively still. The North table is covered and the South table is over pressurized by HEPA filters, so there should be little risk of dust being stirred up.
This morning, facilities removed all the porous ceiling panels that had been soaked/damaged by water (in B265B: above WS1 and WS2, see Attachments 1+2; In B265A, see Attachment 3). Specifically in B265A, an enclosure was created (Attachment 4) and a dehumidifier was placed inside. All monitors/equipment underneath the panels were thoroughly covered, and the floors were swept up afterward.
No work was done above the North table in the QIL. I asked about it and facilities said they would look into it, but it wasn't on the schedule for today. A member of facilities also pointed out that the sink in the QIL was running black liquid (Attachment 5). It looks like soil/dirt entered the water pipes? This seemed to also be outside of their scope for today.
Muddy Waters is not new, but if the facility can fix it we'd take it.
Facilities will be returning on Monday 4/4 between 8-9 AM to remove all ceiling panels above the workstations in B265B (QIL). Replacement of the panels is not yet scheduled, but in the meantime the open ceiling will be covered and the workstations will still be accessible.
Pictures attached. WS1 and WS2 have been turned back on, since the replacement for the ceiling panels will not arrive for another few weeks according to Facilities.
As noticed by Kate a few times last year, the north side of the lab has hot air comiing out of the HEPA vents and the south side has cold air. This seems to be a problem with the setpoints for the sensors or the hot water actuators.
Let's remember to call physical plant after the current roof leaking situation settles down.
1. I have achieved 50% fibre coupling efficiency for the 532nm beam. I have attached the simulation of the estimated solution in ABCD and the profiles after the first two lenses.
Solution - 50cm lens at 73.4cm, 35cm lens at 105.4cm 17.5cm at 150.2 cm.
The beam profiles approximately agree wth the given solution.
2. I measured the attenuation of 1064nm light in the 532 nm single mode patch cable. These are the reaadings I took of input 1064nm power vs output power.
10mW - 0.10uW
30mW - 0.14uW
40mW - 0.15uW
50mW - 0.26uW
I will later calculate how this would lead to a loss in measured squeezing.
3. While trying to characterise the fibre beam splitter, I found that for the for an input power of 50mW in the 1064nm fibre, there was a total of 22mW coming out of the output ports. On checking the 1064m fibre, we found that the mode shape coming out of the fibre wasn't clean. Inspection through a fibre microscope showed no damage to the fibre ends. We have changed the fibre, but lost the 75% mode-matching in the process. I'm getting a maximum of 50% with the new fibre.
Noise analysis of photodetectors in the ATF Lab (I have collected data from the FFT network analyser. I will plot it and put it up on the elog)
Set up the PZT phase control when I return to Caltech.
Make the Homodyne Setup.
Calculate Squeezing Losses.
Convince people that these losses can be reduced.
Successfully finish SURF project
Its now possible to install DTT (aka diaggui) ont your personal laptop using macports. This would allow you to access data from the LIGO sites, the 40m NDS, as well as the W. Bridge sub-basement labs (once Jon and I set up a way to move the data to the CIT cluster and NDS it).
as usual with MacPorts, YMMV, but this way works for me on more than one machine:
sudo port install gds
Error: Failed to archivefetch gds: root5 must be installed with +python27.
sudo port uninstall root5
sudo port install root5 +cocoa+gcc6+graphviz+gsl+http+minuit2+opengl+roofit+soversion+ssl+tmva+xml+python27
This works on my old laptop (El Capitan), but so far not on Sierra.
After that you can run diaggui to open up your old .xml files, look back at old data, etc.
These are some some summary notes on bandwidth and noise budget of transformer coupled photodiode detectors and and example made for the WOPO experiment.
A lesser known strategy in photodetector design is to transformer couple the photocurrent generated by a photodiode before applying amplification. It offers a potential benefit in terms of improved input current signal-to-noise at the amplifier input. Monolithic style RF amplifiers only come in a couple of fixed input impedances, determined mostly by mass market commercial imperatives. Transformer coupling their inputs allow us to adapt common 50 Ω, low-noise figure, wide bandwidth RF amplifiers for use in TransImpedance Amplifiers (TIAs), boosting their current signal to noise at a tradeoff of bandwidth. For the addition of this simple passive element we get an AC coupled output with added overall gain that adds (almost) no additional noise.
We don't, however, get a free lunch. There are tradeoffs in bandwidth and dirt effects of transformers that need to be carefully considered against the requirements. We also miss out on a DC coupled output that is useful for diagnostics and initial alignment.
A schematic of the basic circuit is shown below:
The photodiode is reverse biased and connected to the primary coil of a low resistance RF transformer of coil ratio* N:1, where N is the integer ratio of extra turns on the primary coil. With less turns on the output secondary coil the voltage is stepped down but the current is scaled by a factor of N. This means that for a fixed current noise at the input of a given amplifier we are able to increase the relative size of the input current: this means better signal to noise.
Shot nosie is given by
where e is the charge per electron (1.602•10-19 C) and Ip is the photocurrent. Ip is given by
where eta is the quantum efficiency of the diode (from 0 to 1.0), P is power incident, h is plank constant (6.626•10-34 J.s) and c is the speed of light. At some choice of incident laser power shot noise will become the dominate noise source.
Vendors tend to quote responsivity of diodes which is current per unit laser power
for 1064 nm light the maximum is 0.858 A/W and responsivity is scaled by the quantum efficiency of the diode.
Op amp first stage amplifiers have quoted voltage and current noise. Typically we choose FET input op amps for their lower current noise: these can be order 0.8 fA/rtHz and 5 nV/rtHz. For monolythic RF amplifiers there is a terminating impedance at the input. Thermal noise of this termination sets the limiting signal to noise.
Thermal noise (Johnson–Nyquist) associated with a 50 Ω termination resistance at the input of the RF amplifier given by
where kB is Boltzmann's constant (1.38•10-23 J/K), Tn is the amplifier noise temperature (Ideal ~ 293K) and R (=50Ω) is the input resistance of the amplifier. For 50 Ω termination at room temperature, current noise is 18 pA/rtHz. Referenced to the input at the photodiode, the transformer reduces this apparent noise by a factor of N (the number of coils on the primary side).
Also, there is a voltage induced by thermal noise at the input termination. This goes as and these fluctations in voltage accross the diode capactitace induce a convertion to current noise, where C is the inherent capacitance of the PD. At the diode the voltage is stepped up going from the amplifier to the diode. In the frequency domain this means the input refered current noise induced from amplifier voltage noise is given by
Thus, there is some tradeoff between current and voltage noise. Ignoring dark current of the PD, the total noise is
The tradeoff is reduced bandwidth.
There is another bonus here: reduction in the equivalent capacitance of the PD as seen by the TIA. Voltage noise at the input pins of the first stage op amp can be converted to equivalent current noise by the intrinsic capacitance of the photodiode. Applying a voltage across a capacitive junction induces a current
that, in the Fourier domain, is
where is the amplifier voltage noise density [V/sqrtHz]. The effective step up of the amplifier voltage noise going back to the PD input g is N for a N:1 transformer the source impedance apprent at the amplifier input stage for a N:1 is N
The migration of charge carriers to the two sides of a photodiode N-P region means that natural l capacitor is formed across the active region of the photodiode
Optimization given desired noise profile and bandwidth
Dynamic range estimate
[xxx] Chaoyong Chen, Shaoping Shi, and Yaohui Zheng, "Low-noise, transformer-coupled resonant photodetector for squeezed state generation", Review of Scientific Instruments, Volume 88, Issue 10 (2017). https://doi.org/10.1063/1.5004418
[xxx] Shreyas Potnis and Amar C. Vutha, "Broadband low-noise photodetector for Pound-Drever-Hall laser stabilization", Review of Scientific Instruments 87, 076104 (2016) https://doi.org/10.1063/1.4960088 https://arxiv.org/pdf/1607.01816.pdf
[xxx] Malcolm B. Gray, Daniel A. Shaddock, Charles C. Harb and Hans-A. Bachor, "Photodetector designs for low-noise, broadband, and high-power applications", Review of Scientific Instruments 69, 3755 (1998); https://doi.org/10.1063/1.1149175
Audio band transformer wisdom: http://jensen-transformers.com/wp-content/uploads/2014/09/Audio-Transformers-Chapter.pdf
Some typical 'chip' sized RF transformers: https://www.coilcraft.com/pdfs/pwb.pdf Range is limited, you might want to wind your own but this is challenging
* Note: in this post I refer to the primary coil as the one with more turns connected to the input side. Some of the literature makes a habit of defining the secondary coil as the larger number of terms (the assumption is that you always want to step up voltage) and refer to hooking up the transformer with secondary on the input.
What is going on with this fossilized squeezing experiment? I reckon that there is a loss diagram out there somewhere. Le'ts see some ellipses and arches:
We used the setup specified in the previous eLog post. The design inputs a digital signal which is stored as a control signal to compare against. Next, the digital signal is passed through the DAC and comes out as an Analog signal through terminal DAC0. Since only the DAC or the ADC can run at one time, the DAC is then paused until the ADC converts the signal back to digital, at which point the ADC is paused and the DAC resumes functioning. Theoretically, this conversion should be happening at 100 Hz, and in practice, this number will be very close to 100 Hz. See Figure 1 for the setup diagram.
We passed two signals through the LabJack – 1 Hz sine wave and 10 Hz sine wave. These signals were converted to an analog signal through the DAC and then converted back to a digital signal using the ADC and the return signal was saved. We then ran Rayleigh statistic tests on this obtained data, which measured the continuous probability of a circular distribution for random variables. In GWpy, the Rayleigh statistic is a calculation of the coefficient of variation of the power spectral density (PSD) of a given set of data. It is used to measure the ‘Gaussianity’ of those data, where a value of 1 indicates Gaussian behaviour, less than 1 indicates coherent variations, and greater than 1 indicates incoherent variation. It is essentially a p-value for a frequency range where a Rayleigh value closer to zero means that we can reject the null hypothesis that the signal is not of the corresponding frequency. These methods were obtained from GWpy (https://gwpy.github.io/docs/stable/examples/spectrogram/rayleigh.html and https://gwpy.github.io/docs/stable/examples/frequencyseries/rayleigh.html#gwpy-example-frequencyseries-rayleigh). This produced both Rayleigh spectrograms and Rayleigh spectrums. We also produced Rayleigh spectrograms and spectrums for the digital signal inputted into the LabJack to compare against the data read back from the LabJack. The attached 8 plots are these spectrograms and spectrums – 4 for each frequency containing 2 expected and 2 observed spectrograms and spectrums. We can see that the LabJack is mostly consistent with converting the data. We see clear signals in the observed plots that correspond to 1 Hz and 10 Hz. We can see in the spectrograms that the LabJack occasionally drops data, but very rarely – only around 3 times for sampling sessions of around an hour. The data drops no more than a few milliseconds each time (usually only one packet).
I found some relevant work done on the discussion of acoustic and thermal sensitivity within a optic fiber. For reference, it's important to have an idea of the geometrics of the fiber. The core, cladding, and coating are 11 +/- 1 um , 125 +/- 1 um, and 245 +/- 10um in diameter, respectively. Though the cladding is pure silica, the coating is Ge-doped. ( http://www.thorlabs.com/drawings/65f0b20de1051938-7503B425-91B4-7335-B52672A2FD1F6447/SM2000-SpecSheet.pdf ). Additionally, the fiber's sensitivity to some pressure depends on its characteristic elastic coefficients ( Young's Modulus, , and Poisson's Ratio, ) and Pockels coefficients, and . The elastic coefficients for fused silica can be found online but are also referenced below.
To have an idea of the approximate sensitivity we could expect, we can consider some preform silica cylinder that is undergoing a uniform pressure. Many of the references I found measured that pressure from variations of the relative phase change between two interferometer arms- one static and the other undergoing a pressure difference relative to it. The optical phase retardation per unit of pressure (in dyne/cm^2) can be expressed as:
This comes out to be approximately .
Obviously there's more to the picture than just that. We need to consider the differences in refractives indices and and elastic coefficients between the different materials present in our optic fiber, as well as account for the radial and axial displacements in the "cylinder" caused by some pressure. The approach in reference  consideres a two layer cylinder, and assumes that the center is of homogenous material much like the example above. We can place the optic fiber in a cylindrical coordinate system as shown below:
(Insert crudely made cylinder coordinate system here)
Following reference , we'll assume that the axial stress at the very ends of the fiber is zero. ( at , where is the full length of the fiber). This is referred to as the radial model, a form of boundary condition. Since axial symmetry is also assumed, the stresses and strains will be functions of and . For a single material solid cylinder, the solutions for the differential equations for the axial/radial displacements are expressed as products the trigonometric and modified Bessel functions. When we consider a multilayered cylinder, the general solution will be the the series expansion of those products, and their coefficients are determined by the boundary conditions.
It was shown both experimentally and theoretically that a 2D model of the above scenario (i.e. a plane strain ) gives nearly identical results as the 3D model, and involves much less calculation power. The reference goes in-depth at how the following equation was is derived, but to keep this concise, the average induced fractional phase change can be expressed as:
where are the axial and radial strain, respectively. When the strains are a function of position along the axis of the fiber, we would need to average of its length:
[Work in progress. -Vinny (7/19/18)]
I just came across some notes I made about optimal fiber length for the TMTF fiber stabilization scheme. I'm putting them here for future reference. The idea is to construct a fiber Mach-Zehnder interferometer with one arm much long than the other. The arm length mismatch gives us a frequency discriminator to which we can lock our laser to. The question is what length is best, given expected losses, for maximum signal-to-noise?
Larger path length mismatch always increases the slope of the ideal MZ interferometer in frequency space. However, at some point the propagation losses in real fiber reduce the fringe visibility, degrading the signal slope and, therefore, signal-to-noise.
Vinny made some initial calculations that indicated that optimal path length mismatch for SM2000 fiber MZ interferometer was 117 m and in other estimates on order of kilometers. I redid these calculations myself to see if there was an nice analytic expression for the long arm length given a known loss. It turns out there is:
Where alpha is the loss per unit length that goes as . For SM2000 fiber, which is the standard type that Thorlabs ship, the losses are estimated at 37.5 dB/km @ 2000 nm. This makes an alpha of 8.6e-3/m and the estimated optimal fiber mismatch length of 116 m. This agrees with the last number Vinny publish on the elog (ATF:2207).
I've attached a Mathematica notebook (as a pdf and .nb file) so that we have a common point of reference for future questions of design. This way we don't have to redo it every time. In the book there is an estimate of the expected frequency equivalent shot noise if we operate the 2 µm MZ with 100 µW of power. In there I find that, at the shot noise limit, we are looking at order 50 mHz/rtHz noise floor. This is an order of magnitude worse that the next best standard 1064 nm fiber.
These calculations were only considering shot noise and not fiber acoustic noise or PD dark noise. If we up the input power to 1 mW then we should expect a frequency equivalent shot noise of 15 mHz/rtHz.
During construction of this experiment for the SURF project Vinny built a single TransImpedance Amplifier (TIA) for each MZ: in loop and out of loop. We then subtracted a DC offset with a stable voltage reference to place ourselves roughly at the half fringe. This was ok as a quick dirty first pass. However, the ANU people pointed me to a better way. We make a subtraction measurement between the two outputs of the MZ. This doubles the signal slope and to first order is immune to intensity fluctuations of the laser.
We had trouble locking because when we feedback to the laser current to control frequency the power is also affected. By making a ballanced measurement this way would make us immune to power changes in the laser and always keep us at the half fringe.
At this stage the experiment is turned off and unmanned. When we get another undergraduate we should get them onto make a pair of matched low noise TIA to make a ballanced readout scheme from the MZ. We need to also look into getting some more PD to be able to also do the out of loop detection at the same time.
I would like to give a brief update on the ongoing effort for the assembly of the cry-vacuum chamber at the QIL. Gabriele has been kind enough to spare some time to help me with safe crane operation, considering the chamber and plates are 70 kg each.
At first, I prepared an aluminum spacer (made out of bosch extrusion) which is 90 mm in height. The idea is to lift the bottom plate (which has 16 through holes and counterbore for bolting purpose) such that the bolts can be easily inserted from the bottom. The spacer is strong enough to take the load of the entire assembly (i.e. around 200 kg, bottom and top plate + collar). The assembly will be resting on this spacer, even during the in-situ baking procedure. After baking and pump down testing is complete, the spacer will be removed and the vacuum chamber will be resting flat on the optical bench.
Next, with the help of a crane we tried lifting the collar using eye bolts and nylon slings, however the suspension point was way too high giving us no room for the crane to lift. This required shortening the length of the nylon sling. Firstly, I got shorter length eye bolts (2.5 inch against 4 inch) and secondly I used carabiners to tie up the sling. Using this set up we were able to lift the collar successfully.
The collar has groove for single dovetail O-ring (pictured attached) at the top and bottom surface.
The specifications for the viton O-ring was provided by Nor-Cal. I bought the viton O-ring (2-474V, 24.94’’ ID) from Kurt Lesker, the dimensions of which is given below,
O-RING,FKM,24.940"ID X .275" W(ACT),25"ID X 1/4"W (NOM),ISO 630
The challenge is to keep the O-ring held in the groove at the bottom of the collar (against gravity) during it’s mating with bottom plate. However, the O-ring always falls off. At this point I am not inserting the O-ring completely inside the groove (and I will come to this point next) and just slightly pushing it in. I want the O-ring to move in naturally from the pressure created by the plate and collar. I contacted Zach since he has faced similar issues with his tank (Chris gave me this information) and he used vacuum grease to keep the O-ring sticking on to the groove. However, outgassing from the grease could be a point of concern.
Next, I tried to insert the O-ring completely inside the grove, strangely in this case I find it to be longer by at least an inch. The inner diameter of the groove is 25 inches.
I contacted Nor-Cal to cross-check the size they have recommended. They realized that they have made a mistake with the size and have agreed to ship one size smaller O-ring (i.e. 2-473 having an ID of 23.94) which should give a stretch of 4%. I will use this once we receive it.
I found a Youtube video of the O-ring installation in a half or full dovetail groove. Given below is the link, since this is different from the usual method, hence important. The advantage of using dovetail groove is that the O-ring will stay inside during the assembly without falling off.
I asked Nor-Cal about in-situ baking to reduce outgassing and they recommended that with viton O-ring seals 200C is max and 150C or lower is safe. I will go with 100 C. Steve doesn’t agree and doesn’t like the idea of baking it during pump down. As per him, once a turbo pump was destroyed (I guess at 40m) after some parts of the O-ring got damaged and got sucked inside. However, I think keeping things at 100 C should be safe as other folks at the cryolab have done the same, successfully.
On the Friday cleaning, we vacated the east optical table. The Si scatterometer was disassembled and the Si block was moved and stored to the cryo lab.
This is just a note about damage tolerances for fibers so we have a reference of the amount of power that can be used.
Thorlabs gives the maximum theoretical CW power as 1 MW/cm² and a 'practical' safe power level of 250 kW/cm² (or a quarter of the max). They don't seem to provide information about wavelength dependence and assume its the same for all wavelengths. We don't expect to be affected by any of the exotic ultra high power effects like bend loss induced damage and photodarkening. The air/glass interface is where the damage will occur. This can either be because of heating of the ferrule/connector (causing epoxies etc to break down and damage the interface by depositing on the optical surface) or regular mechanism that are the same as bulk optics (dielectric break down and thermal effects).
The intensity profile of light confined in the fiber is defined by the Mode Field Diameter (MFD) -- the cross-sectional diameter of the light that includes the core of the fiber and a region just beyond the cladding the mode occupies. MFD of 1064 PM fiber (PM980-XP) is 6.6 ± 0.5 μm @ 980 nm and for the 532 nm PM fiber (PM460-HP) is 3.3 ± 0.5 µm @ 515 nm.
Fiber effective area is
which is 8.5x10^-8 cm^2 for PM460-HP and 3.4x10^-7 cm^2 for PM980-XP. Taking the conservative 'practical' damage threshold this indicates a maximum power of 21.4 mW into the 532 nm fiber and 85.5 mW into the 1064 nm PM fiber. The absolute maximum is just a factor of four more than this: 85.6 mW into 532 nm PM fiber and 342 mW into 1064 nm fiber. If the fiber ends are kept clean then we should be fine if the power level is kept below 85 mW.
Of particular concern is the power handling capability of the fiber beam splitter (PN1064R5A2). There will be some waste 532 nm light coming out of the WOPO and I don't want these potentially multi-mode components to exit the cladding at the point of the coupler and damage the surrounding material. The 1064 nm maximum power rating of this 50:50 PM beam splitter is listed as 1 W (for the connectorized fiber), so we should be well clear of that threshold for the LO light. For 532 nm its less clear. The equivalent 532 nm PM 50:50 beam splitter (PN530R5A2) has a rated power of 100 mW @ 530 nm for bare or connectorized fibers. As the MFD of the 1064 nm version of this PM beam splitter has a much larger MFD and the exiting 532 nm light will already be expanded in the cable patching the WOPO to the BS, we should be well clear of this damage threshold point.
So bottom line is that we need to keep power below 85 mW going into the WOPO device and keep all the end connectors super clean and it will be fine.
Andrew measured the output voltage noise of photodiodes through an AC coupled pre-amplifier (G=200, 0.7 nV/rtHz) while varying the intensity of white light falling on it. The setup uses a regular 2xAA flashlight (torch) with an incandescent bulb to illuminate the photodiode. The flashlight focus is adjusted to give a non-expanding beam and a 25.4 mm (1") lens is then used to focus the collimated beam down. The total power of light was controlled by trimming it with an iris.
Below is the fitting of data to estimate the shotnoise intercept current. Here the fitted function is the quadrature sum of input referred shot noise and dark noise modeled as an equivalent shot noise quanitiy i_snint (called intercept current):
I have included zip file with data and notebook in it as well. With measured shotnoise intercept current of 48 ± 7 uA and 31 ± 7 uA, we are good to go to use these for measuring shot noise with light power in the order of 5.0 mW falling on the photodiodes. With 5.0 mW (~3.9 mA) on each PD, this gives us a clearance of 10dB from the dark noise floor on average for each detector. Plots with fits are attached below.
The transimpedance amps for the 2um (unamped) InGaAs detectors were made and evaluated.
Attachment 1: The circuit diagram
The usual transimpedance configuration. The detector (Thorlabs DET-10D) is an extended InGaAs which is sensitive up to 2.2um. I believe the detector is biased to 1.8V although it is not obvious and the 12V battery is used. The feedback resistor was chosen to be 5kOhm so that the circuit can handle up to ~2mA (~1.7mW). The feedback capacitance pf 100pF for compensation was chosen kind of arbitrary to keep the circuit stable and also the RC cut off to be more than 100kHz. The output resistance is 100Ohm. The selection of the opamp is described below.
Attachment 2: The amplifier noise Part I
The amplifier noise (the first unit called Amp #2) was evaluated with the opamp swapped with OP27 (BJT), LT1128 (BJT), OPA604 (FET), and LT1792 (FET), chosen from the 40m stock. For the given environment, the FET amps exhibited better performance while the BJT amps suffered from more line noise coupling and the larger 1/f noise. Particularly, LT1792 reached at the level of ~2pA/rtHz, with lower line noises. This looks the best among them. Note that the 5kOhm feedback resistor gives 1.8pA/rtHz current noise.
Attachment 3: The amplifier noise Part II
Then the second unit (called Amp #1) was made. This unit has more high-frequency noise. It turned out that the noise was coming from the power supply which was the +/-12V from the rear panel of an SR560 which was connected to the AC power. The noise dramatically went away with the battery mode operation of SR560 (by disconnecting the AC power). The floor level was 2.2pA/rtHz and it was slightly higher than the quadratic sum of Johnson noise of 5kOhm and the voltage noise of the amp (4nV/rtHz). This noise level was just sufficient for the purpose of the 2um detector.
Attachment 4: The detector noise levels
Now the detector #1 and #2 were paired with the amp #1 and #2, respectively. In fact the detector 1/f noise was way too large compared to the amplifier noise. There is no hope to detect shot noise level of the mA photocurrent.
Attachment 5: The detector response
The detector response of each PD+AMP pair was measured using Jenne's laser and Thorlabs PD10A (~150MHz). There was some systematic error of the absolute level calibration, therefore the transfer functions were adjusted so that they have 5kOhm transimpedance at ~1kHz. The phase delay is ~30deg at 100kHz. This partially comes from the combination of 100pF//5kOhm and the ~4MHz bandwidth gain of the opamp. If we want faster response we need to modify these.
The amplifier sets for the thorlabs 2um PDs were delivered to the lab.
- PD1 and Amp1, PD2 and Amp2 are the proper combination. If a high quality power supply is used, it is not an issue.
- The cables for the external bench supply or the 9V batteries have been made.
The 80 MHz brimrose AOM driver, which came with the AOM, can actually drive between 75 MHz and 85 MHz. It has an input port, which accepts between 0V and 10V, for altering the frequency.
Previously, before Friday 2019-08-09, I had set the offset to + 5V, using a signal generator because that was what was available in the QIL lab. The signal generator, for some unclear reason, had then moved this offset to +4.4 V, when plugged into the AOM driver. Attachment 1 shows a power spectrum, from the BB PD, One notices a large peak negatively detuned from the 80 MHz signal frequency. Attachement 2 shows a zoom in, around 70 MHz, revealing this peak to be two peaks, one near 78 MHz, and another smaller one near 79 MHz.
I decided to adjust the voltage, input to the frequency port, to minimise the RF sidebands I observed. The results of this adjustment can be seen in attachments 3, and 4. Notice that there no observable sidebands. This was achieved with an offset of 4.52V.
Attachement 5 is the manual for the AOM, and its driver. (Removed by KA)
InAsSb PDs were housed in the PD cages. The cages were engraved to indicate the batch (Sb3512 or 3513) and the serials (A1, A2, ...).
The PD legs does not have an indicator for the pin1. So, the tab of the PD case is directed "UP". Also the direction of the tab is marked on the cage. The tab of the short plug was also aligned to Pin1. However, the PD case is too thin and the PDs can rotate in the cases.
So the face photo was also taken so that it indicates how Pin 1 looks like from the PD face. (Attachment 4)
Also made the cable for the LaserComponents PD and the InAsSb PD. Pin n shows up as Pin n of DB9 Male connector.
Once we have the PD test is the bias circuit (with a monitor) and some patch panel kind of preparation, we can start working on the PD test.
DB9 switchable breakout box is ready. We are ready to do some PD test now.
Koji set up an experiment measuring the dark current of the photodiodes. A bias voltage is given and the current is converted to voltage via a TIA, where it is measured. Also note that in order to provide a high quality bias voltage, we LP the output of the device with a second order sallen key filter cutoff at 1Hz.
The QE and dark current of all the InAsSb sensors were measured. All the measurements were done in room temperature.
- The incident beam power of the 2004nm beam was 0.95mW.
- The beam was focused down to 50um gaussian radius, which was confirmed by DataRay BeamR.
- The angle of incidence was ~0deg.
- The element side (nominally Pin 2, 3, or 6) were connected to the vias boltage (negative) and the common ground was connected to the transimpedance amplifier (Shalika OP140 R=5100Ohm)
- The dark current was highly dependent on the reverse bias voltage. The QE was also bias dependent.
- Sb3512 A2 have different behavior compared to others. Alex mentioned that Sb3512 is the test batch. We can exclude this sensor from the test.
- The best QE was ~0.7 for Sb3513 A3 P2 (Pink) and Sb3513 A2 P6 (Purple). Both have the area of 500um^2. These two particular elements have low dark current of <1mA. The dark noise of this specific sensor should be measured.
Some issues of the measurements
- The transimpedance amp (TIA) has suspicious behavior. The saturation voltage was ~17V rather than <-15V. This indicates that the voltage regulators possibly have leakage of the input voltage (+/-18V) to the output line. This needs to be checked, particularly before the dark noise test.
- TIA saturation: The bias voltages could not be raised to ~1V for some PDs because of the dark noise and the saturation of the TIA. The transimpedance should be lowered by a factor of ~5.
- Because of the low bias voltages of these saturated cases, the max QEs were not reached. This also prevented from checking if there was any clipping loss. This should be checked again with the lower transimpedance.
- TBD: The angular dependence and the reflectivity of the sensor should be checked. It is difficult to carry out these tests without a sensor card.
The lenses were arranged so that the spot on the PD can become smaller. A quick measurement on a (500um)^2 element showed the QE of ~80%
With the strong focusing lens of f=40mm, the beam was once expanded to a few mm. Then f=75mm lens focuses the beam to ~30um (radius). (See Attachments 1&2)
With this new beam, the QE was quickly checked. The new measurement is indicated as "Sb3513 A2P6new" in the plot. It showed the QE of ~80%.
The AOI was scanned to find any maximum, but the AOI of 0deg was the best at least with the given beam. I'm not sure yet why 500umx500um requires such small beam radius like 30um. Awesome
Here are the commands.
sudo /sbin/rmmod c4tst c4iop && cd /opt/rtcds/caltech/c4/target/c4iop/scripts/
./startupC4rt && cd ../../c4tst/scripts/ && ./startupC4rt && systemctl start firstname.lastname@example.org && systemctl start email@example.com && systemctl restart firstname.lastname@example.org
System diagram of the PD QE test with the IRLabs cryostat.
PT-SE (MS/PT-SE) connector data sheets
Amphenol catalog http://www.amphenol-industrial.com/images/catalogs/PT.pdf
Detoronics Hermeic Sealed Connectors (DT02H-18-*PN) http://www.hselectronics.com/pdf/Detoronics-Hermetic-Connectors.pdf
AF8 crimping tool (expensive!) https://www.mouser.com/ProductDetail/DMC-Tools/AF8?qs=gvhpkjpQEVSjrLbsepewjg%3D%3D
AF8 alternative https://www.jrdtools.com/?gclid=Cj0KCQiA2vjuBRCqARIsAJL5a-IQ9ztCEYKdo645v_RhUBJS3eMIars1LubjlKZoorS-lnx6ClDDiMUaAlZiEALw_wcB
Thermistor link: https://www.tec-microsystems.com//Download/Docs/Thermistors/TB04-222%205%25%20Thermistor_Specification_upd2018.pdf
TEC spec: Mounted TEC type: 2MD04-022-08/1 https://www.tec-microsystems.com/products/thermoelectric-coolers/2md04-series-thermoelectric-coolers.html
2MD04-022-08/1 dTmax = 96, Qmax = 0.4W, Imax = 0.7A, Umax = 2.0, ACR = 2.29 Ohm
The external Dsub cable is ready except for the 32pin connector to be plugged-in to the chamber. See QIL ELOG 2460 for the pin assignment.
While I'm still waiting for the proper connector for the vacuum feedthru of the IRLabs cryostat, I have connected to the Dsub9/15 split cable to another Dsub9 connector so that I can test the cooling of the InAsSb sensor in air. Also, the 2004nm laser, a fiber-coupled faraday isolator, and 90:10 beam splitter was moved to the cryostat table and fixed on a black al breadboard. [Attachment 1]
The InAsSb TEC was controlled by the TEC controller of ITC-50. I didn't change the PID parameters of the controller but the temperature nicely setteled to the setpoint. The sensor has a 2.2kOhm thermister. And the max current for the TEC was unknown. The TEC driver had the current limiter of 0.3A and it was not changed for now. With this current limit, the thermistor resistance of 10Kohm was realized. This corresponds to the temperature of about -20degC. According to the data sheet given by Alex, the resistance/temperature conversion is given by the formula
1/T = 7.755e-4 + 3.425e-4*log(R)+1.611e-13*log(R)^3
To satisfy the curiosity, the dark current of a (500um)^2 element was measured between -250K and -300K. At -254K, the dark current went down to the level of 40uA (1/15 of the one at the room temp). For the measurement, the bias voltage was set to be 0.5 and 0.6V. However, it was dependent on the diode current. (Probably the bias circuit has the output impedance). This should be replaced by something else.
The quantities we want to measure as a function of the temperature:
- Temperature: 2.2k thermister resistance / 100ohm platinum RTD
- QE (Illuminating output / Dark output / Reference voltage / Reference dark output)
- Dark current (vs V_bias) -> Manual measurement or use a source meter
- Dark noise (PSD) 100kHz, 12.8k, 1.6kHz, 100Hz
I borrowed KEITHLEY 2450 source meter from Rich. The unit comes with special coaxial cables and banana clips. Most of the peripherals are evacuated in the OMC lab.
The dark current of A2P2, A2P3, A2P6 were measure with different temperatures (300K, 270K, 254K). The plot combined with the previous measurement ELOG QIL 2425.
== How to use the source meter ==
- Two-wire mode: Connect the wires to the diode
- Over voltage protection: [MENU] button -> SOURCE / SETTINGS->Over Voltage Protectiuon 2V
- Sweep setting: [MENU] button -> SOURCE / SWEEP -> e.g. Start -750mV, Stop +500mV, Step 10mV, Source Limit 1mA -> Select Generate
- Graph View: [MENU] button -> VIEWS / GRAPH
- Start measurement: Note: The response of [TRIGGER] button is not good. You need to push hard
This starts the sweep, or a menu shows up if your push is too long -> Select "Initiate ..."
- Data Saving: [MENU] button -> MEASURE / READING BUFFERS -> Save to a USB stick
[Raymond, Aidan, Chris, Koji]
P6 element (500um)^2
- We looked at the current amp (FEMTO) output. The amplifier saturated at the gain of 10^3 V/A. Looking at the output with a scope, we found that there is a huge 1.2MHz oscillation. Initially, we thought it is the amplifier oscillation. However, this oscillation is independent of the amplifier bandwidth when we tried the our-own made transimpedance amp.
- Shorting the cryostat chamber to the optical table made the 1.2MHz significantly reduced. Also, connecting the shield of the TEC/Laser controller made the oscillation almost invisible. This improvement allowed us to increase the amp-gain up to 10^7.
- Then the dominant RMS was 60Hz line. This was reduced by more grounding of the cable shields. The output was still dominated by the 60Hz line, but the gain could be increased to 10^8. This was sufficient for us to proceed to the careful measurements.
- The dark current was measured by the source meter, while the photocurrent (together with the dark current) was measured under the illumination of the ~1mW light on the PD.
- Attachment 1 shows the dependence of the dark current against the swept bias voltage. We had ~mA dark current at the room temp. So, this is ~10^5 improvement.
- Attachment 2 shows the dependence of the apparent QE against the swept bias voltage. The dark current was subtracted from the total current, to estimate the contribution of the photocurrent in the measurement.
- Attachment 3 shows the dark noise measurement at the reverse bias of ~0.6V. Up to 1kHz, the noise level was below the equivalent shotnoise level of 1mA photocurrent.
All the data and python notebook in the attached zip file.
The QE measurements from the first couple of photodiodes are attached below.
QE = [I_photocurrent]/[P_PD] * h *nu/e
P_PD = Power incident on photodetector = 0.966*power_incident on cryo window
Power incident on cryo window = F(voltage on reference PD)
% load JPL data
f0 = dir('*dark*.txt');
f1 = dir('*photo*.txt');
f2 = dir('*cond*.txt');
% get temperature vs time
tempList = ;
pList = ;
for ii = 1:numel(f2)-1
% load JPL data
f0 = dir('*dark');
f1 = dir('*bright*');
% get temperature vs time
tempList = ;
refPDList = ;
for ii = 1:numel(f1)
Item lending as per Ian's request: Particle Counter from OMC Lab to QIL
The current particle class of the room was measured to be 800.
The particle counter went back to the OMC lab on Aug 10, 2020.
Still trying to figure out how to set up the particle counter remotely. The current particle count is 576.
Note: the particle count is the number of particles detected over 0.3um size.
West Bridge flooding Apr 6th due to rain in the night
Looks like the first responder was Calum. The attached photos were sent from him.
We can use Thorlabs SAF1450S2 gain chip to generate 1418 nm light using an ECDL design similar to the one described in Kapasi et al. Optics Express Vol. 28, Issue 3, pp. 3280-3288 (2020) (ANU 2um ECDL design).
I have contacted Disha and Johannes to get the actual measured data for the PZT transfer function of this ECDL design. Fig.5b in their paper plots the transfer function of the PZT. Since, in ECDL PZT directly changes the cavity length, it has a more powerful actuation strength (2 orders of magnitude more) with actuation of 560 MHz.V upto 1 kHz. It however had a very low pole at 1 kHz and two mechanical resonance-antiresonance pairs near 1 kHz and 2 kHz. I modeled a transfer function by eye using Fig.5b of the paper. Page 1 in the attached pdf shows this modelled transfer function.
Next, we need to change the PDH loop for the auxiliary laser lock with the 40m cavity since the PZT has changed. I modelled one from scratch. This simple analog loop's performance is shown in orange in pages 2-5. This loop seemed stable from all the metrics I know, viz: phase margin of about 55 degrees (Page 2), no strong peak in close loop transfer function (page 3), and no remanant oscillations in time domain response (page 4).
I also modeled a similar loop but with digital compensation of the resonance-antiresonance features. This loop is plotted in green on pages 2-5. Both these loops have 300 kHz of bandwidth just by using PZT. I beleive this could be increased but I have not taken into account any saturation of PZT.
From Fig.4. of the paper gives a frequency noise estimate for free running ECDL. They mentioned that a roll-off below 10 Hz was due to their thermal feedback to remain in linear range of their frequency noise emasruement method. I modeled the noise of ECDL hence by
where the flicker noise contribution is similar to NPRO noise but ECDL has a white noise of 15 Hz/rtHz due to natural linewidth of spontaneous emission or Schawlow-Townes linewidth (with several broadening factors). I think this is an inherent limitation of ECDLs.
Page 5 shows both unsuppressed and suppressed frequency noise estimate for ECDL with the loops mentioned above and current values of NPRO noise are also plotted for comparison.
Following up on the last post, here I presented a near back of the envelope calculation of how different choices of AUX cavity finesse and laser source for mariner would affect the prospects of calibration scheme.
As mentioned in the last elog post, here I considered using an NPRO seeded auxiliary laser source (converted to 1418nm by whatever method), ECDL based on ANU design with a modified PDH loop and same ECDl with a digital compensation of PZT resonances. I have taken the residual frequnecy noise of these lasers as the dominant noise source in the calibration scheme. Craig and Gautam in their proposal for SoCal wanted the AUX laser to be locked to the arm cavity in a PDH shot noise limited way. That would be necessary for 4km interferometers and would be easier to achieve there with higher laser powers and higher cavity finesse.
Here I considered three cases. First assumes about 3% transmittance of 1418nm in ITM and ETM HR coatings for mariner. This gives a finesse of about 100 and a cavity pole of 18.9 kHz. I believe this is the existing case at 40m. Next we consider transmittance of 0.5% and 0.05% (500 ppm) of 1418nm in ITM and ETM HR coatings for mairner. These cases give finesse of 625 and 6.28k respectively with cavity poles at 3 kHz and 299 Hz respectively.
Page 1: Consideres the case of finesse of 100. The green dashed line shows the amount of drive strength (in m) required at different frequencies if we use ECDL with PZT resonance compensation, to get an SNR of 1000 in 100s of integration time.
Page 2: Same as above but for Finesse of 625.
Page 3: Same as bove but for Finesse of 6280.
Page 4: Comparison of different finesse cases for the ECDL with PZT compensation option. Dashed curves represent requried drive strength (in m) for different cases.
Page 5: Same as above but for NPRO seeded auxiliary laser.
Note: For the NPRO seeded auxiliary laser, we have assumed that the noise of conversion to 1418 nm is similar to noise due to SHG process which is not dominant. There would be an effect of multiplying with a factor ranging form 1-1.5 due to frequency conversion but I have ignored it here for simiplicity. Also, NPRO case is limited in bandwidth due to PZT resonances. We might be able to get away with them using digital compensation like the case study for ECDL. But I haven't attempted that here as we do not know our NPRO PZT's resonance features yet.
would be easier to achieve there with higher laser powers and higher cavity finesse.
But I haven't attempted that here as we do not know our NPRO PZT's resonance features yet.
I don't know why it would be easier to have higher finesse with longer arms. Something about beam size???
The NPRO PZT TF's are all in the 40m elog - there are many measurements of TF made over the past 10 years. Its like Raiders of the Lost Ark - you have to believe its there while searching.
If we use ECDL for auxiliary frequency in 40m and hope to stabilize it up to 1 MHz with digital compensation of PZT, it is important to take into account any phase effect of the nearby FSR at 3.97 MHz. This should ideally be included in the Input Mode Cleaner loop considerations as well. These effects would be more prominent in longer cavities like aLIGO and LISA where FSR is very low and should we attempt to stabilize a laser lock beyond cavity's FSR.
I did a no assumptions calculation for getting a general transfer function fo PDH error signal in units of [W/Hz] assuming 1 W of incident power. This calculation would soon be uploaded here. I'll put down here primary results.
For incident field on a Fabry-Perot cavity (with fsr of ), reflected electric field transfer function (unitless) is given by:
Then, PDH error signal for a modulation frequency of at a modulation index of , in units of [W/Hz] (i.e. error signal power per Hz of error in laser frequency from cavity resonance) is given by:
after demodulation and low pass filtering. Note this transfer function is a complex quantity as it carries phase information of the transfer function too. The real signal is obtained by multiplying this signal at with and taking the real value of the product.
Having done this, we can see how in the real PDH error signal, there is a low pass at cavity pole, given by and a notch every fsr. The notch creates a zig-zag in the phase of the tranfer function and has a HWHM same as cavity pole. After this point, I just fitted a ZPK model to the transfer function to obtain a empirically derived model for PDH error signal transfer function. Apart from the cavity pole, this model needs to have resonance and antiresonance features present at each FSR with resonance having a linewidth of cavity pole while anti-resonance having a linewdth of . Here's how the ZPK model would look like:
I've attached my notebook where I did the fitting analysis and the overlap plot of real PDH error signal TF and the modelled approximation.
I have a preliminary calculation to post here. This does not include noise sources from cavity fluctuations and main frequency noise. But it gives some idea about shot noise and frequency noise of AUX laser conttribution to the noise in calibration.
An issue was raised with last calculation about the fact that our sensing of PDH signal isn't ideal and in the real world there is scattering, clipping extra adding excess noise in the PDH loop. This noise primarily comes by the intensity noise imparted on promptly reflected light from the cavity via various shaking optics etc on the table before it goes to the PDH reflection RF photodiode.
This noise's coupling to the PDH loop is identical to how shot noise of light couples into the PDH loop i.e.:
I moved the brand new TED200C on the workbench to Crackle for 2um ECDL (permanently)
The TED200C temp controller used in the 2um PD test setup will stay there (permanently)
FEMTO DLPCA200 low noise preamp (brand new)
Keithley Source Meter 2450 (brand new) => Returned 11/23/2020
were brought to the OMC lab for temporary use.
I was thinking about getting this new current pre-amp from NF:
It seems to have a good noise performance and has a built in low pass filter and also a remote interface.
The FEMTO seems less fancy, but their noise performance is actually 2-3x better.
Is the reverse bias programmable? FEMTO has a bias trimmer on it. It's useful in the usual application, but for automation, the configuration of the input becomes cumbersome.
doesn't seem so, but they sell this one:
which has a USB interface and pretty good voltage noise spectrum
I'm attaching my rough first draft of the QIL photodiode testing schematic. Please provide comments for fixes/improvement!