My goal was to investigate the effect of placing a 1k ohm resistor as the input of the DC port of the Bias Tee. The expectation was it would decrease the bandwidth around DC, and this is supported by LTspice simulation. According to simulation, at around 300 Hz there becomes a few dB difference between having the resistor and not having it. From the paper 'W. Zhang, M. J. Martin, C. Benko, J. L. Hall, J. Ye, C. Hagemann, T. Legero, U. Sterr, F. Riehle, G. D. Cole, and M. Aspelmeyer, "Reduction of residual amplitude modulation to 1 × 10-6 for frequency modulation and laser stabilization," Opt. Lett. 39, 1980-1983 (2014)'(https://www.osapublishing.org/ol/abstract.cfm?URI=ol-39-7-1980) in Figure 2B we see the frequency noise without servos active. The noise falls off after around 50 Hz, which indicates the change of bandwidth to 300 Hz should not be an issue. I also showed the effect for 500 ohms to see how the transfer function changes intermediately between 0 and 1k ohm resistances.
Here is the result of taking the transfer function of the EOM Driver (DCC: D1200794-v3) with the bias tee and a dummy EOM from the RF input to the RF monitor output. A Black 143-10J12L tunable inductor (438 nH - 788 nH) was used to bring the peak over 37 MHz. The data was taken using an Agilent 4395A Analyzer which was controlled using the .py .yml and .ini files provided in the zip.
Edit Thu Aug 1 16:11:12 2019 ScottA :
4 Vpp is actually 16 dBm, so we need to feed 16 dBm power from OCXO to the EOM. So we'll need ~10dB attenuation but everythign will still work.
The plan is to put a half-wave plate before the mirrors which steer the beam into the faraday isolator before the South cavity. The mirror which will act as a pickoff is only reflective to S polarization at 45 degrees, so with the HW plate, we can get enough light in transmission to measure RFAM with a photodetector.
Our previous plan turned out to be ineffective due to the mirror not transmitting all of the P polarization, but only a fraction. We now used the PBS within the faraday isolator (FI) as a pickoff, and rotate the HW plate at the input to control the power reflected. When the HW plate is maximized for transmission through the FI, I measured 13.4 mW into the FI, 12.00 mW at the output and 1.18 mW picked off. I then used a mirror to steer the picked-off beam onto the 1811 new focus photodetector. Using the DC coupled output I maximized the output on an o-scope, which was at the level of 400 mV. The AC output was then connected to the HP4395A analyzer from which the data was taken.
Edit: I have now put a focusing mirror with a focal length of 5cm before the PD, the o-scope now reads 3.41 V. A pdf of the layout with these updates can be found at https://nodus.ligo.caltech.edu:30889/ATFWiki/lib/exe/fetch.php?media=main:experiments:psl:ctn_optical_layout.pdf
This is the time series from the AC port of the 1811 new focus photodiode which is set up for RAM measurement. A strange waveform is present, it has a peak to peak voltage of around 2.9 V and a frequency of around 60 kHz.
I replaced the EOM driver that was attached to the south EOM with the Bias Tee and the driver which I tuned with the Bias Tee and Dummy EOM. After some minor tuning, I was able to move the transfer functions peak back over 37 MHz. I took the spectrum of the RF monitoring port while a 16.5 dBm signal was present on the RF in port from OCXO after placing an 8 dBm attenuator. We noticed that the max peak at 37 MHz was around -3.49 dBm which corresponds to a signal with an amplitude of 6.4122 V by converting to voltage, multiplying by sqrt(2) and dividing by the value of the simulated TF at 37MHz which is about 0.0328. This is less than the expected 20 V amplitude from a 16.5 dBm signal with the transfer function from the previous post (https://nodus.ligo.caltech.edu:8081/CTN/2376). This motivated us to take transfer functions at different input powers to see if it does change. We can see that it does a fair amount for large input signals. I can offer no explanation at this time.
I set up the transfer function for three scenarios using a -5 dBm, 5 dBm, and 15 dBm signal for each:
In every case, there is a decrease in the magnitude of the transfer function with an increase in signal.
EOM Driver Link: https://dcc.ligo.org/cgi-bin/private/DocDB/ShowDocument?.submit=Identifier&docid=D1200794&version=
List of EOM Drivers in the lab: https://nodus.ligo.caltech.edu:30889/ATFWiki/doku.php?id=main:experiments:psl:electronics
To determine if the loss in gain seen by the new eom driver (SN:06) was due to the driver itself or just because of the addition of the bias tee, I took the transfer function of the old eom driver with and without the bias tee. This plot shows the LTspice simulated transfer function with and without the bias tee, as well as the effect of adding the bias tee to either driver. Here are some take-aways:
It seems we cannot introduce a bias tee without facing a reduction in the gain of the driver.
I investigated the possibility of increasing the EOM Drivers gain by replacing the R5 resistor with a different value, as this should be what effects the gain of a common base amplifier.
It seems we cannot use this method to increase the gain of the driver.
We were suspicious that the reason the transfer function was decreasing with an increase in power was actually due to the monitoring circuit, specifically the MAX2470 RF Buffer. To test this hypothesis I took a power sweep with an EOM driver (SN07) which only contained the 5V regulator and the Buffer, along with all the components associated with these devices. There was no transistor, as well as no diodes at the input power supply. The overall change from -5 dBm to 15 dBm was only around 0.1dB. This indicates the monitoring circuit does not provide an explanation for the reduction in gain seen at higher power levels.
i checked the resonant frequency of both cavities in order to see if we can lock both using the existing frequency actuator (AOM) for the first one. Used the slow frequency actuator of the NPRO (temp) to scan the frequency.
refcav1 is resonant @ 0.7578V
refcav2 is resonant @ 0.5068V
assuming about 1GHz/V the resonances are about 250MHz different. So we have to use the thermal actuator on one of the cavities in order to tune it. I started a calibration scan for the heater on friday in order to set the correct heating power as the time constant is more than an hour and trial and error method would take too long...
I don't know if its worth the trouble, but we do have a ~200 MHz AOM. Sam Waldman had us buy one of these for doing the OMC g-factor measurements.
i think the problem is that we don't have the VCO for the FSS for 200MHz. so i think it"s easier to heat one of the cavities. the temperature required to match them should be only a couple of Kelvin difference. and by heating one of those cavities the frequency noise due to ambient temperature fluctuations might be uncorrelated as well.
i had a problem taking the data from one of the photodiodes over the weekend. We had a loose connection for the cavity1 transmitted light PD. i only checked on a scope and assumed that the (already) connected cable to the DAQ is OK as i saw some fluctuations while scanning. so i have no data from that long scan over the weekend. we repaired this today and also made the FSS for the first cavity work. now everything is working and i started a new measurement...
sweep frequency is 50uHz which is almost 3h for the full ramp from min to max
C:PSL-FSS_RMTEMP is used for the heating voltage set point (not the heater voltage!). calibration is [-273-(voltage*100)] (is still calibrated for temperature sensors)
we will add the temp sensors on the cavity tank tomorrow...
currently the setting for the NPRO temp slider is:
changes for refcav2 are large due to the missing insulation and heater on that chamber. The LIGO refcav heaters don"t fit as the position of the connections to the main chamber tube are different (and in quantity too). changes over the last couple of hours are .7350 to .7570, which is abou 20MHz and so more than the tuning range of the VCO
in order to improve the stability of the chamber temperature the current plan is to add a heater and insulation for the second chamber. room temperature changes were about 2Kpp over the last couple of days. I already ordered flexible insulating foam for the chamber (the round parts). What we need is one or more heaters. We could somehow add half of the original heater to that chamber but i would like to go for a more final solution as we need one for the other chamber in the ATF as well. The plan is to buy standard heaters with adhesive backside and stick them to the chamber. Price is about $70 for a 10"x10" heater (MINCO). The chamber surface is about 22"x25" in total, cut into smaller areas by the 4 vaccum tubes and 2 legs. I think we can cover most of it with a total amount of 4 to 6 heaters of different sizes.
I think its important to think of a good solution, since we may want to retrofit some other chambers.
For the electronics to drive the heater, let's make sure not to use the power supply solution that is in use now in LIGO and at the 40m. We should make sure to make the design such that the residual temperature noise from the heater is below what we expect for coating thermal noise, assuming we use Fused Silica spacer of 1m length.
This should be quite easy in principle, especially with the radiative shields on the inside.
What is a "good" solution for you? A custom-made heater fitting the whole thing like the ones at the sites? They are expensive and it will take several weeks/month to get them. The only advantage i see is that you can remove them from the chamber but do we need this feature? The other heaters are standard parts and you just stick them to the chamber, so you have a good thermal contact to it. As we only have two of the "other" chambers with a different layout for the small CF flanges i think this is good enough for our tests. For a new super-cavity chamber we can choose a different design. I think it does not make a big difference in stability. The design of the insulation is more important..
Yeah, i'm currently thinking off designing a first prototype for such a low-noise driver. As i think we will use the DAQ for temp controll thats the only part we need (and the power supply for the supply, but we have one of those). The current heaters i want to use have enough heating power when used with up to 24V, which is a good value as easy to buy. The noise can be easy as low as a couple of 10nV/sqrt(Hz) with standard parts..
current settings for the slow actuator to bring the cavities on resonance are:
For the heater, my main concern is on the residual thermal gradients after stabilization. The Alnis, Hansch, et al papers described how they had trouble with using multiple heaters and multiple sensors.
Our experience from the 40m is also that the single out of loop sensor (AD590) shows a much larger signal than the residual in-loop noise. This is only an issue below 10 mHz, so its not worth worrying about if it makes things take longer in the near term.
my hope is that if we choose the right heaters with the same heating power density that we will not have large thermal gradients. e.g. if we choose 4 heaters in total, 2 small ones and 2 large ones and we choose the right values for the resistance (you have the choice of 5 different values) then e.g. we first could connect a small and a large one in series and then both pairs in parallel or so. I don't wanna have individual drivers for the individual heaters. That should give us a fairly even temperature distribution of the entire chamber.
So far our plan is also that we add 5 or more sensors on each layer of insulation/heating in order to measure the thermal distribution and gradient throughout the entire insulation system.
after getting the o-ring from the drever lab for the adapter to the turbo pump to pump the second chamber i figured out that the installed CF2.75 valve is leaking.
So while pumping with an external pump everything is fine but after disconnecting the external pump the ion pump can"t handle the leak rate and the pressure increases to a level where the ion pump switches off.
The leak is so large that you can"t disconnect the external one and quickly close the open port with a blank one until the pressure reaches the limit for the ion pump.
Because we don"t have a spare valve i decided to close this leaking port and don"t use this valve for further pumping. Instead i build an adapter to the small valve for a tiny hose (<1/4") already installed.
By using this adapter the pump rate is tiny but i still have the chance to lower the pressure below the limit of the ion pump.
By now the pressure is low enough to switch on the ion pump. The current is less than 10mA by now and decreasing, so i think the pressure should be ok tomorrow.
In parallel i started building the thermal insulation for that chamber.
found four old Minco heaters (model HR5494-106) (from 1995) . This type with 106 Ohms is not in their system anymore.
But corresponding to their data the maximum current for this type of heater is about 7.5A. So driving this heater with 24V would give us 5.4W of heating power beeing well below the limit. Using the standard power supply for heating refcavs we can get even more power. Due to the age (14years!) the adhesive back is not sticky anymore so i will use aluminum tape for first tests.
added the four Minco heaters and a first layer of 1" thick foam insulation to the second refcav chamber. Pressure is down to 4e-8 torr. started heating with ~20W to bring it above room temperature.
slow actuator settings for being resonant:
heater refcav1: programming voltage 1.95V
heater refcav2: power supply voltage 16V
added a single temp sensor on refcav2 (AD590) last week in order to monitor the temp of refcav2. We tried to change the heater power on both cavities to match the frequencies. So far we have no tempctrl. Changes of the room temperature have too large (and different) influence on the individual cavities. So it"s almost impossible to predict the room temp changes and change the heater power to match both cavities. We need a real temp stabilization of the chambers...
the name of this sensor is C:PSL-FSS_RCTEMP
room temp next to both chambers is measured with C:PSL-FSS_RMTEMP
there is a full LIGO RefCav available here at LHO. They don"t use it anymore and its just sitting here on the table with the pump running. Rick offered me to ship the whole thing to Caltech in order to replace the second cavity in the PSL lab. The advantage would be that we have two identical setups, a good heater and a fitting insulation. In addition we could get analog control.electronics for temperature stabilization.
pictures taken from the existing power supply.
we repaired the sensors on the first refcav and renamed and added channels to the DAQ. The channels are now:
C:PSL-FSS_MINCOMEAS -> temperature of first refcav (average of 4 AD590 sensors), cal. in degC
C:PSL-FSS_HEATER -> heater power supply control voltage (heater voltage is x10)
C3:PSL-GEN_DAQ1 -> temperature of second refcav (one AD590), cal in degC
C3:PSL-GEN_D2A1 -> heater power supply control voltage (heater voltage is ~ x2.5)
settings for the FSS:
common gain : 1dB
fast gain : 15dB
phase : 1.5V
VCO power : 4.5V
RF amplifier gain : 7V
we added two endcaps for the first refcav and additional insulation for the ion pump and valve.
here a plot of the temperatute stability in peter's lab (upper graph) if nobody is working in the lab (christmas) and if someone opens the door to the lab (the large spikes in room temp). lower graph shows the temp of the refcav (in-loop, no out-of-loop sensor so far. Except the large spikes stability is about 4mK/pp. I will add a single ool-sensor today (will be C:PSL-FSS_RCTEMP). We used this channel so far for the second cavity but have a second temp box since christmas...
the last couple of days we fixed a couple of thinks:
We found a wrong calibration of the temp readout of the first refcav, which gave us a wrong residual noise level compared to the other one. The absolute value was almost correct so we didn't realize this before. Now we are limited by the DAQ noise so the next step will be improving the gain settings in the temp sensor readout-box.
A big problem is the changing room temperature as soon someone opens the door even for only a couple of seconds. The temp control of the lab goes crazy and changes a couple of degrees and is oscillating half a day after that. The delay to the cavities is about 30min and a couple of deg changes of the room temp also change the cavity temperature, only a couple of tens of mK but enough to shift both cavities away from each other as the room temp couples different into both due to the different thermal insulation (insulation and time constants are different). We opened the room temp sensor and adjusted the temperature to i higher level, removed surrounding parts to get a better air flow to the sensor. We don't understand why opening the door has such a large effect on the room temp as 10 seconds of open door don't change the room temp by 2 degC or so. So if anybody has an idea plz let us know even if it seems to be stupid.
In order to improve the temp stability of the analyzer cavity we added a second layer of insulation and wrapped the whole thing in aluminum foil, see picture below. The insulation is 2" thick, except a small part at the large flanges at the end where its only 1" thick. We also changed the gain of the temp sensor readout in order to reduce the influence of DAQ noise. If nobody is working in the lab we have a stability of about 6mKpp within hours. As soon someone is working in one of the labs this changes to tens of mK for working in other labs to about a hundred mK if working in the PSL lab. We are currently working on improved servo settings...
in order to build a custom fit insulation for the cavities i've built some hot wire foam cutting machines to cut the 2" thick polyurethane thermal insulation. The wire i use is a .010 piano wire typically used for suspensions (thats what we had i the lab). Nominal current is about 1.5A at a couple of volts, so any simple power supply does the job.
Here are some pictures of the large foam cutter to cut the 48" x 48" boards into smaller peaces and the circle cutting device:
original panel size
cutting into smaller segments:
surface comparison before and after hot wire cutting:
some cut parts:
ready to cut the rest...
refcav: 0.9512 @ 70.02 degC
acav: 0.9252 @ 39.996 degC
with 0.858V/GHz -> ~30MHz difference
1pm: changed refcav temp to 69.9 degC -> SLOWDC=0.9488
4pm: changed refcav temp to 69.5 degC -> SLOWDC=0.928 (but not final)
7pm: acav now at 0.923, refcav at 0.928 -> difference now ~6MHz
--> good temp values are ~69.5degC for the refcav and ~40degC for the acav
heater value refcav: ~2.983
heater value acav: ~6.245
both @21.6 degC room temp
after some trouble with fluctuating temperatures and a brocken cable to the vco i could finally lock both cavities. So now we can take first data.
temperatures are still fluctuating and the software loops have to be optimized for the new thermal insulation. Currently only a p-servo is running and the channel names have to be finalized...
slow actuator signal: C:PSL-FSS_SLOWDC
room temp: C3:PEM-SENS1_TEMP
actuator signal: C:PSL-FSS_HEATER
temp signal: C:PSL-FSS_MINCOMEAS
actuator signal: C3:PSL-GEN_D2A1
temp signal: C3:PSL-GEN_DAQ1
- more insulation to the refcav chamber added
actual slow actuator values for both cavities:
refcav : 0.4385
acav : 0.2539
-> ~215MHz difference
with ~6mK/Mhz -> refcav temp reduced by 1.3K
refcav temp now 67.7, slow actuator 0.304 -> reduced temp further to 67.5
refcav : 0.243
acav : 0.251
refcav: 0.2456 temp:67.4
acav: 0.2582 temp: 40.0
01/27/10 9 pm
refcav: 0.2423 temp:67.55
acav: 0.2578 temp: 40.0
refcav: 0.2525 temp:67.65
acav: 0.2578 temp: 40.0
01/29/10 4 pm
refcav: 0.2485 temp:67.65
acav: 0.2335 temp: 40.0
I guess if you have a frequency counter with a GPIB interface or a simple flip-flop XOR based phase/frequency discriminator, you can feed the output to a 3113 and use at as an input to a slow EPICS PID to bring the beat frequency to within range.
Actually, we need a frequency discriminator for the Green Locking so it might be good to brainstorm about this with Aidan and Koji.
good idea - we now know where we have to be +- acouple of 100mK as the impact of changes in room temp is not equal for both systems so we have to slightly adjust the temperature of one cavity.
Right now we have a much larger problem with the AOM. if we lock both cavities and the AOM frequency changes slowly over time (both cavities drift a bit) we get a huge amplitude modulation for the refcav, meaning a drift in power from almost 0 to 2.5V (the full range) topped by a sine-like modulation, looks like an etalon with about 10% modulation depth, every 1.4Mhz. It's not the power of the VCO, that changes only by about .2%, but you see this modulation there as well, but almost covered by the noise of the DAQ already. It's a power modulation of the diffracted beam as you can see it in the reflected and transmitted light with the same sign, so it's not pointing or so. I think it could be an impedance matching problem causing some standing wave resonator. We try to investigate it today. I will post a graph later today...
measured the beam pointing caused by driving the AOM frequency modulation input. data is uncalibrated so far, just a screenshot of the dataviewer. PDHOUT is the VCO input signal...
The power at the AOM and the reflected beam from the AOM vs frequency are measured, but they seems to be unrelated to the problems.
When the VCO is modulated by a triangular function +/- 6V, a small fluctuation is observed in transmitted and reflected beam. There might be some small constant phase shift between the two.
at 78 MHz the modulation width is about 0.3 MHz, with +/- 2.4% amplitude
at 79.5 MHz, the width is 0.238 MHz, +/- 1.1 % in amplitude,
at 81 MHz, the width is 0.193 MHz, +/- 1.5% in amplitude.
The first plot shows the comparison between the transmitted beam and the reflected beam between 78 - 82 MHz. The second and the third plots are the whole data from the transmitted beam and the reflected beam respectilvely. The data is taken aroud 2010-02-03 20:00:00 UTC.
measured tf of the analyzer cavity including VCO (wideband input, around 80MHz center frequency), double-passed AOM, cavity, pd (Thorlabs PDA10CS) and mixer for different source levels (to see if there are nonlinearities).
acav is now re-aligned. As Tara stopped aligning the AOM i will use the NPRO pzt to lock it (instead of the VCO). Then i will also see where the shit of the other tf comes from (hopefully).
we spent the whole night to re-align everything. By now everything is back online, both cavities were locked.for a short time. We still have the pointing problem. It's different now, seems to be smaller but is now almost the same for x and y. Will do this later today ....
The 10kHz peak in the old spectrum seems to be related to the crossover frequency of the fast actuator and the pc of the fss loop.
The PDH box is modified, all filter stages now have socket adapter boards to easily change the filter frequencies by changing the capacitor value within seconds to optimize everything. A zero for compensating the cavity pole is also installed. A modulation input too.
After adjusting the mirror for reflecting beam back to the AOM, the QPD signal shows the better alignment. Before the voltage different readout is about 1.2 V, now it's reduced to ~500 mV.
We'll try to add a translational stage for the mirror for better alignment.
values for VCO pwr 4.7V
RCTRANSPD signal: 5.17V
DC-level of RF-PD:
refcav unlocked: 590mV
refcav locked : 202mV
current cavity temp: 67.7degC
slow actuator value : 0.8966V
added a detector behind the curved mirror to measure the changes in deflected power (1.order) while changing the frequency of the VCO. PD is a large area Si diode to reduce pointing effects. Signal is connected to QPD-SUM instead of the QPD signal (QPD channels for x and y still connected and valid, only sum disconnected !!). We didn't trust the sum anymore as we can see slightly pointing related changes there as well.
in order to get an idea about the part/subsystem causing the RF power modulation while sweeping the VCO frequency i changed the length of the cable to the AOM. I decided to just extend the existing cable using a cheap SMA cable, not one of those semi-rigid ones. Now i thought due to more losses, low quality cable, longer, more connection etc. the shifted power should be less, but it seems to be the opposite, see screenshot. The purple curve is the transmitted power through the refcav, red the VCO input signal, the rest unimportant. On the left third the old cable, then a short break where i extended the cable, the center part with the longer cable, another short break and then with the original length. First i thought it's some aligment related thing, touching the cable, changing the alignment of the AOM etc. But it's not. The PD behind the curved mirror (not shown) shows exactly the same. Any ideas?