NAOJ GW Elog Logbook 3.2
A small problem was encounter : While screwing the cover part of the hole of cavity to the mode cleaner, a screw stayed stuck.
This was due to the fact that too long screw were proposed on the design. Instead of using M4*12, we used M4*8 screw for this part. We also used ethanol to avoid to stuck another screw.
Also, we couldn't find M3*16 screws so we used M3*15 screws.
To protect the wire used for the piezo power supply, we used a small piece of aluminum folded. One part is screwed to the optical bench while the other part hold a adaptor between BNC and the 2 wires for the piezo.
The table is to be updated with the values of yesterday (in blue).
low probe power | high probe power | |||
signal | AC | 23mV | AC | 230mV→295mV |
DC | 440mV | DC | 5400mV→5200mV | |
AC/DC | 0.052 | AC/DC | 0.043→0.057 | |
noise With sample |
AC_rms | 0.1mV | AC_rms | 2.8mV→1mV |
DC | 440mV | DC | 5400mV→5200mV | |
AC_rms/DC | 2.30E-04 | AC_rms/DC | 4.5e-4→1.9e-4 | |
ppm | 884ppm | ppm | 2000ppm→667ppm | |
noise Without sample |
AC_rms | 3μV | AC_rms | 70μV→ 200μV |
DC | 0.65V | DC | 6.5V | |
AC_rms/DC | 4.60E-06 | AC_rms/DC | 1e-5→3e-5 | |
ppm | 18ppm | ppm | 46ppm→108ppm |
This week, in order to check the AOM characteristics, we install the AOM after a beam splitter on the green path. By using a beam splitter before the AOM and 2 powermeters ( 1 one reflexion, the other on the transmission at the output of the AOM ) and checking their ratio, we were able to characterize the AOM despite still having power fluctuations on the green beam. The optical setup used is described in an attached figure.
By changing the RF power send to the AOM, we were able to characterize the AOM 1st order with the use of a gaussian fit ( even if this wasn’t really a gaussian, it helped to locate the maximum) as following :
- Maximum efficiency : 73 % @ RF Power 28.4 dBm ( 692 mW)
The AOM test sheet said that we could expect a 1st order efficiency superior than 85% at 633 nm. In this case, our alignment was approximative as we wanted to check only the response of the AOM to different RF power.
Then we tried to put the AOM on the right position on the optical bench. As the AOM need a small input beam size, we put it in the middle of 2 lenses ( f = 100 mm ) .
At that position, we couldn't see anymore any diffraction order.
First, we checked the green Power Density sent to the AOM. We measure 10W/mm² when the AOM test sheet limit this power density to 2.5 W/mm². Hopefully, we reduced quickly (after few minutes) the laser power down to 2 W/mm². In regard to this, we contact AA Opto-Electronic, manufacturer of this AOM. Following their advice, we check that the crystal was still transparent without any visible damages.
Then, we tried to put the AOM back on the characterization position. We were able to see again diffraction orders. We realize again the characterization of the 1st order efficiency and obtain :
- Maximum efficiency : 69 % @ RF Power 28.3 dBm ( 676 mW) We expect that the difference might be due to misalignment.
After that, we checked the polarization of the green beam using a PBS because this AOM needs a vertical polarization to work. We found that in both positions the green beam has a vertical polarization as we expect.
The last difference is the divergence of the beam. Indeed the beam is really more diverging in the right position (5.6 mrad) than on the characterization position (1.6 mrad) compared to the diffraction angle (16.6 mrad).
To correct this problem we will try to change the lenses configuration in order to obtain a smaller divergence of the beam on the right AOM position.
case | std(AC) | DC | std(AC/DC) | AC | std(AC) | AC/DC | std(AC/DC) | std(ppm) | std(ppm) |
pump | OFF | OFF/ON | OFF | ON | ON | ON | ON | OFF | ON |
without IR filter | 3.4uV | 0.77V | 4.4uV | 0.042V | 128uV | 0.0545 | 1.67E-04 | 16 | 621 |
with IR filter | 2.7uV | 0.45V | 6.1uV | 0.024V | 99.5uV | 0.053 | 2.26E-04 | 23 | 873 |
We found out that the DC changes from 0.46V to 0.44V when switching off the pump. This happens only when there is the sample, this means that some pump is scattered from the sample.
low probe power | high probe power | |||
signal | AC | 23mV | AC | 230mV |
DC | 440mV | DC | 5400mV | |
AC/DC | 0.052 | AC/DC | 0.043 | |
noise With sample |
AC_rms | 0.1mV | AC_rms | 2.8mV |
DC | 440mV | DC | 5400mV | |
AC_rms/DC | 2.30E-04 | AC_rms/DC | 4.50E-04 | |
ppm | 884ppm | ppm | 2000ppm | |
noise Without sample |
AC_rms | 3μV | AC_rms | 70μV |
DC | 0.65V | DC | 6.5V | |
AC_rms/DC | 4.60E-06 | AC_rms/DC | 1.00E-05 | |
ppm | 18ppm | ppm | 46ppm |
Today we checked again the signal level at the above conditions and we found almost the same values of the table above but the AC noise with low probe power and without sample was higher: around 200 μV instead of 70μV.
Then Kuroki suggested to cover the Imaging Unit to protect from wind (as it was in the original setup last year) and the noise became between 50-100μV , then we removed the cover and the noise remained on the same level 50-100μV. We think we should cover better the optical parts, in order to avoid temperature fluctuations which might affect the noise.
The table is to be updated with the values of yesterday (in blue).
low probe power | high probe power | |||
signal | AC | 23mV | AC | 230mV→295mV |
DC | 440mV | DC | 5400mV→5200mV | |
AC/DC | 0.052 | AC/DC | 0.043→0.057 | |
noise With sample |
AC_rms | 0.1mV | AC_rms | 2.8mV→1mV |
DC | 440mV | DC | 5400mV→5200mV | |
AC_rms/DC | 2.30E-04 | AC_rms/DC | 4.5e-4→1.9e-4 | |
ppm | 884ppm | ppm | 2000ppm→667ppm | |
noise Without sample |
AC_rms | 3μV | AC_rms | 70μV→ 200μV |
DC | 0.65V | DC | 6.5V | |
AC_rms/DC | 4.60E-06 | AC_rms/DC | 1e-5→3e-5 | |
ppm | 18ppm | ppm | 46ppm→108ppm |
We tried to calibrate the AOM again. Since the PD has too much effect on the power fluctuation, we decided to put two power meter in 0 and 1st order, so when we change the RF power, we can see the difference between these two orders. But we cannot find a good position to put the power meter that two order is separated enough and also the beam size is smaller than the aperture of the power meter. So we put another beam splitter after the MZ BS, and two power meter on two path of this beam splitter. Then with spectrum analyzer, we did the ratio between these two power in time domain to see the real change effected by the RF power.
In the past days we tried to characterize the locking loop of the filter.
The loop transfer function for the filter cavity (sketched in figure1) is compose by different blocks
- G1 [Hz/V] = piezo actuator
- G2 [Hz/Hz] = SHG
- G3 [Hz/W] = cavity
- G4 [W/V] = photodiode + demodulation
- H [V/V]= servo
- inject noise on perturb
- measure EPS1/EPS2
- inject noise on perturb
- measure piezo mon/EPS2
- inject noise on RAMP
- measure EPS2/piezo mon
The amplitude of the loop transfer functions plotted so far are actually the square of the real amplitude. The problem comes from the way I treated data saved by the spectrum analyzer. Each file is composed of 3 columns: frequency, real part (a) and imaginary part (b) of the TF. Of course amplitude and phase are recovered by doing:
Amplitude = sqrt (a^2 +b^2)
Phase = angle (a+i*b)
Due to an oversight, I had replaced the sqare root with the absolute value in the amplitude computation. This explain the unexpected behaviour (1/f^2 instead of 1/f) of the openloop TF around the UGF.
We will upload soon new TFs measurements (taken by Yuefan and Marc on monday night) properly plotted.
The activity of locking characterization of the past days pointed out some issues that are worth to be reported.
open loop (urad) | closed loop (urad) | |
PR yaw | 1-2 | 0.5-1 |
PR pitch | 4-8 | 2-3 |
BS yaw | 3-4 | 2-3 |
BS pitch | 5-6 | 4-5 |
INPUT yaw | 2-3 | 1-2 |
INPUT pitch | 4-5 | 3-4 |
END yaw | 2-3 | 1-2 |
END pitch | 4-5 | 3-4 |
After a long alignment work we were able to make the IR and the green beam flashing in the cavity at the same time. In this video are shown superposed flashes in trasmission both of the green and the IR (the green beam as been cut at the second 5).
In this configuration we were able to lock the cavity on the green TEM00 but we coudn't check the IR condition since there was too much green light transmitted by the dichroic mirror before the camera installed on the end bench.
After solving this issue and installing the AOM we will start to look for the frequency shift needed to have both the IR and the green resonant on the fundamental mode.
Then we powered up the AOM, there was the other orders appearing, the one close to the 0 order is the 1st order which we need. We put a PD at the 1st order, connected it to the oscilloscope and started to turn the AOM to have maximum power in the 1st order. After finding the best position of the AOM, we tried to change the RF power we were sending to increase the power more. But when we tried to do this, we found out the power fluctuation is about 40-50% percent, it was much more than the tolerance.
We put PD both in the infrared path and the green path to check the fluctuation of the green is coming from the infrared or not. Then we found out since the PD aperture is quite small, so it will increase the degree of uncertainty we saw. So we changed the PD to the powermeter which has larger aperture. This time the infrared is quite stable and the green is fluctuating but much less than before. We guessed that the fluctuation of the green is coming from the changing of the SHG cavity alignment when people moving around or touching the bench. We are going the install the MZ and stabilize the green.
Then other thing we checked is that the SHG loop gain we are using now is 50. But with this gain,if we changed the power sent to the PZT, the cavity cannot pull the error signal back to zero when the cavity is locked. If we increase the gain, until it reaches 10000, there will be no oscillation appear, the cavity will be locked more stable, but one problem is that it will be quite easy also lock on the other mode. We will try to find a the optimal gain considering also the filter cavity.
- In order to increase the pump power safely, we cover the laser path with aluminum black walls and black paper.
- We also cover the probe part in order to stop air flow from hepa filters which is a possible noise source. Kuroki-san helped me on this tasks.
- I put a small translation stage below the half ball in the imaging unit in order to adjust better the alignment and I tried many times the alignment until I found a good signal for the surface reference sample, comparable with the one of last year. Also the bulk reference sample gives almost the same signal as last year.
- The parts for the 1310nm probe laser were delivered, I glued the golden prism mirror on a half inch post, I'm waiting for the glue to cure under the neon lamps
The PDH filter cavity signal has been calibrated injecting a line at 28 kHz (above the ugf ~ 10 kHz of the loop) on the “ramp” input of the electronic servo. The ramp input is summed to the PZT correction signal.
The amplitude of the 28 kHz line in Hz is obtained using the formula:
S_Hz = V_RMS (V) * sqrt(2)*100*2e6 Hz/V = 1.25e-6*sqrt(2)*100*2e6 = 353 Hz
Where V_RMS is the line amplitude measured by the Agilent spectrum analyser. The factor sqrt(2) is obtained to pass from the V_RMS to the line amplitude (the factor has been also experimentally verified looking the same line with the spectrum analyser and the oscilloscope).
The factor 100 is the reduction of the PZT_moni output . 2e6 Hz/V is the calibration of the PZT after the SHG.
Measuring the line at 28 kHz in the error signal and compensating for the cavity frequency pole is it possible to find the calibration factor K in V/Hz. The formula used is :
S_V = K(V/Hz)* S_Hz /sqrt(1+ (f/f_0)^2)
where f_0 = 1.5 kHz and S_V = sqrt(2)*38.9e-3 V
--> K = 2.9e-3 V/Hz
which seems to be in agreement with the calibration obtained looking the PDH signal when the cavity is freely swinging. In that case we see a peak-to-peak of the PDH of ~ 4 V for 1.5 kHz of the cavity line which correspond to a K = 2.7e-3 V/Hz. Note that when the cavity is freely swining we have also rining effects which can perturb this measurement.
We have also checked that reducing the frequency of the line sent to the PZT (with the same amplitude) to 14 kHz, the amplitude of the line of the error signal is multiplied by 2, as expected given the cavity pole. A more quantitative analysis (fully taking in account the effect of the loop) is necessary to check the position of the cavity pole.
Another test was to increase the amplitude of the line by a factor 10, thus having a 29 kHz line with amplitude of 3 kHz (two times the cavity width of the cavity). The cavity stays locked and the calibration factor measured is the same with the one measured with the line with an amplitude of 300 Hz. Increasing further the amplitude of the 28 kHz line to ~ 7 kHz (4 times the cavity linewidth) makes the lock more fragile, and sometimes the cavity unlocks. Moreover, an oscillation with a frequency of ~ 1 Hz appears in the error signal (but it is not accompanied with a similar oscillation in the transmitted power).
We observe that the PDH filter cavity signal has an offset of ~ 170 mV. See picture.
The offset is present even when the 78 MHz signal sent to the EO modulator is swithed off (and the 78 MHz sent to the local oscillator is ON). When both signals are OFF, we see a slowly varying offset between 200 mV and -200 mV, which also have an higher frequency oscillation. To be investigated.
We have measured the spectrum of the PZT correction signal sent to the laser when the cavity is locked, using the output PZT_mon (1/100 of the PZT correction signal). The spectrum is in the attached plot.Since in this region the gain of the loop is very high, the signal is proportional to the cavity length/frequency noise.
The calibration is 1 MHz/V (given by the manufacturer).
at 100 Hz we have ~ 700 nV/sqrt(Hz) corresponding to 70 Hz/sqrt(Hz), at 1 kHZ we have 100 nV/sqrt(Hz) corresponding to 10 Hz/sqrt(Hz)
The shape of the spectrum is compatible with the free running laser noise ~ 7-10 kHz /f Hz/sqrt(Hz) up to a few kHz. According to aother measurement, after ~4 kHz the spectrum is limited by a flat noise, which is compatible with the noise of the 100 kOhm resistor at the output of the PZT_moni signal. For f<10 Hz probably the mirror control noise and the seismic noise are limiting the spectrum.
We also see several 50 Hz harmonics. It is not clear if this harmonics can be reduced rearranging the grounds and if they have an impact on the RMS of the error signal of the filter cavity locked. To be investigated.
Summary of yesterday night work (thu 29-->fri 30). The goal was to make a characterization campaign for the cavity lock, in order to make it more stable.
1) Beam stability
In the past we observed an evident jitter of the beam. From a comparison of the spectra we were convinced that this was caused by the residual motion of BS and PR. In the past days we where able to improve the stability by improving the local control filters (a dedicated entry will follow).
We observed that the beam direction (observed by misaligning the input mirror) was drifting and we decided to test a new strategy to keep the mirror position. We change the local control filters in order to avoid to gain at low frequency (we changed a pole at 0.1 Hz with a double zero at 0.1 Hz and we controlled the mirror position not by adding an offset of the loop but simply sending a DC signal to the coils.
We coudn't see a major improvement in the performances.
We also observed the intermittence presence of spikes in the error signals from BS and PR which makes difficult to keep the cavity alignment.
Eventually the old controls (with integrators at low frequency) were restored.
2) Laser servo gain transfer function
We have set the gain of the servo in order to have ~10 kHz bandwidth. See the transfer function in fig.1. (in 1/f^4 mode)
At a first look, the TF behaves as expected. The data have been stored in the floppy disk and they will be compared with the model. The phase margin at ~10 kHz is about 40 degrees.
The transfer function has been measured with the Agilent 35670A spectrum analyser, with a swept sine with 50 mV ptp.
3) Servo parameters
- modulation depth = 1 V pp at 78 MHz (reduced with respect to before). This should correspond to a modulation depth of m= 0.185 rad.
- LO = 8.5 Vpp at 78 MHz (increased with respect to before)
- Demodulation phase = 111 deg
--> With this data the error signal is 3-4 V ptp, for a transmited signal of ~ 3-4 V depending on the alignment of the cavity (note that we did not checked the green laser power yesterday night)
- attenuation of the input signal =9.1
- PZT gain = 0.7
- thermal control gain = 3
- Threshold on the transmitted signal ~ 2 V
4) Auto-relock
With this configuration the cavity automatically locks when the transmitted power crosses the resonance. When the cavity unlocks, it relocks automatically. Note that the servo is always in the 1/f^4 configuration. The video shows the cavity locked, then the input mirror is on purpose misaligned, then it is re-aligned and the cavity re-locks.
5) Stability
During yesterday night lock the cavity was very stable. The plots 2 and 3 show the transmitted power (in cyan) and the error signal (in yellow) for 500 s. No actions were performed to realign the cavity on the second plot. Max transmitted power was ~ 4 V.
On Tuesday 27th june we managed to lock the laser on the filter cavity length.
In the first attachment there is a plot of the transmitted power during the lock acquisition, in the second there is a picture of the transmitted beam when the cavity is locked. A short movie of the the lock acquisition can be seen here.
Friday, June 23, 2017
Following the following procedure, I aligned the Imaging Unit for the HeNe probe beam.
1) move IU micrometer closer to the end of translation which would give you enough translation range to move the whole IU farther away in case you test thick objects;
That will complete rough alignment of the IU. The fine tuning is done by maximizing AC signal coming from the surface calibration piece. For that, try different micrometer positions around one you started with. For every position you have to center the probe (maximize the DC), maximize AC if needed. The maximum R should be close to the original R for the surface calibration. Then make scan with the bulk calibration piece.
According to the theory the signal is maximum when the detector at the Rayleigh length of the perturbation, experimentally we can check this changing the position of the blade and aligning again the imaging unit and measuring the signal. So, in order to maximize the signal, I repeated the procedure changing the position of the blade from 18mm to 12mm and 6mm but I got a lower signal, so I aligned it back to 18mm.
The absorption signal of the reference is similar to original value (the one we had since we bought the system) even if I changed by the 20% the waists of pump and probe,
Parameters: LD current = 0.8A, power without sample = 33mW
We changed the layout of the end bench, added one more mirror after the beam splitter to divide the beam into two, one is received by a screen and a camera was set to look at this screen, the other goes directly into the CCD. The mirror we used has the maximum reflection when it is putting in 45 degree, so we turned it a bit in order to have large enough power also for the transmission. The one received by the screen we took it as a reference for the alignment of the cavity and the CCD is to find a good mode matching. In this configuration we will need three images sending back from the end room at the same time(two on the end bench and one for the second target), the electrical board we are using now for the video only has two channels. So we took the same board from the west end, but it seems one channel of that board is broken, but it is enough for our current need. The fiber used to send the video signal now is '1-13,1-14,1-15,1-16', each board need two fibers. Also each monitor only can receive two channels, so we also took the monitor at the second target of west end to the central room.
2.Analog signal board
On the reflection path of the beam splitter on the end bench we put a PD before and use the receiver box to send this signal to central room for locking the cavity. But since the the aperture of the PD is limited, the spectrum we saw will also be effected by the alignment of the beam. So yesterday we changed the PD to PSD in order to have the information of the beam position. But in this way, we need to send back three analog signals together. We found two board for this and one of them have four channels. We tested the board with a sine wave sent from the end room board, and received it in the central room. The fiber they used for this board is too short to connect into the rack('1-11,1-12'), but the fiber system of TAMA is too complicated, so we just simply changed with longer spare fibers, now we are using '3-13,3-14' for this board.
From the board it seems we can change the gain and offset of each channel, but when we sent a 4V peak to peak signal, it pretty hard to see if it is changed or not, so maybe we can only changed in very small scale. Also it seems four channels all have different setting offset and gains, but we need further check about this.
So we decided to increase laser current to have higher power in infrared and also in green, the current we used before is 1.040A which gave a 8mW-green power at the end of the bench, now we increased the current to 1.2A which gives out a three times green power around 27mW at the end of the bench.
With this power we easily found the beam at the second target and also received it at the end bench.By looking through the window of the end chamber, we better adjust the beam on the end mirror and let it pass more or less the center of it, luckily we got the flash of the cavity with the first try.
The other thing we did before closing the chamber is that we sent the beam out of the chamber to the corridor again after the BS, tried to superpose the green and infrared both in the near field and the far field with the last two infrared steering mirror on the bench. Although considering the air fluctuation in the corridor, two beams moved a lot, but we did our best to make them overlap at 300m.