The mode profile of the green beam going through 40m cavity was measured.

According to the fitting the coupling efficiency to the cavity is 98.46%, but still the beam looks loosely focused.

This measurement has been done by using the oplev legs (entry #2957) to allow the beam to go through the 40m walkway.

With a beam scan set on a movable cabinet, I measured it along the 40m chamber.

Since the plot looks not so nice,  I am going to work on this measurement a little bit more after I improve the mode matching.

Here is the parameters from the fitting

I believe the error for the travel length was within 0.5 meter. The length was always measured by a tape measure.

A thing I found was that: spatial jittering of the beam gets bigger as the beam goes further. This is the main source of the error bar for the spot size. 

We should completely turn off the air conditioner when working on green locking.

Even if green beams propagates inside of chambers, the air conditioner does affect the spatial jitter of the beam. 

The attached picture was taken when Steve and I were seeing how the green beam jittered. 

At that time the beam was injected from the end table and going through inside of the ETM, the ITM and the BS camber.

Eventually it came out from the camber and hit the wall outside of the chamber. It was obvious, we could see the jittering when the air cond. was ON.

I guess I succeeded in locking of the cavity with the green beam 

 Strictly speaking, the laser frequency of the end NPRO is locked to the 40 meter arm cavity.

Pictures, some more quantitative numbers and some plots are going to be posted later.

After the alignment of the cavity I could see DC fringes in its reflection. Also I could see the cavity flashing on the monitor of  ETMY_CCD.

I drove the pzt of the NPRO with f=200kHz, and then the spectrum analyzer showed 200kHz beat note in the reflection signal. This means it's ready to PDH technique.

And then I made a servo loop with two SR560s, one for a filter and the other for a sum amp.

After playing with the value of the gain and the sign of the feedback signal, the laser successfully got lock. 

To make sure it is really locked, I measured the open loop transfer function of the PDH servo while it stayed locked. The result is shown in the attached figure.

The measured data almost agrees with the expected curve below 1kHz, so I conclude it is really locked.

However the plot looks very noisy because I could not inject a big excitation signal into the loop. If I put a big excitation, the servo was unlocked.

The current servo is obviously too naive and it only has f^-1 shape, so the filter should be replaced by a dedicated PDH box as we planed.

Congratulation! Probably you are right, but I could not get this is a real lock or something else.

1) How much was the fringe amplitude (DC) of the reflected beam? (Vref_max=XXX [V] and Vref_min=YYY [V])
    Does this agree with the expectation?

2) Do you have the time series? (V_ref and V_error)

Andri and I mounted the Raicol Crystal #724 in one of the new Covesion Ovens. The procedure was the same as before - see elog entry here.

There was one issue - the glass plate that goes on top of the crystal is coated on one side with ITO (Indium-Tin Oxide) and it's not 100% certain that this was mounted in the correct orientation. It is virtually impossible to tell which side of the glass is coated.

The base plate of the oven was tapped for an M3 hole. We retapped it for an 8-32 and bolted it to a post and that one of the New Focus 4-axis translation stage. The assembly is currently bolted to the PSL table, awaiting use.

 Here are some more plots and pictures about the end PDH locking with the green beam. 

-- DC reflection

 I expected that the fluctuation of the DC reflection had 1% from the resonant state to the anti-resonant state due to its very low finesse.

This values are calculated from the reflectivity of ETM measured by Mott before (see the wiki).

In my measurement I obtained  DC reflection of V_max=1.42 , V_min=1.30  at just after the PD.

These numbers correspond to 7.1% fluctuation. It's bigger than the expectation.

I am not sure about the reason, but it might happen by the angular motion of test masses (?)

--- time series

Here is a time series plot. It starts from openloop state (i.e. feedback disconnected).

At t=0 sec I connected a cable which goes to the laser pzt, so now the loop is closed.

You can see the DC reflection slightly decreased and stayed lower after the connection.

The bottom plot represents the feedback signal measured before a sum amp. which directly drives the pzt.

-- length fluctuation  

One of the important quantities in the green locking scheme is the length fluctuation of the cavity.

It gives us how much the frequency of the green beam can be stabilized by the cavity. And finally it will determine the difficulty of PLL with the PSL.

I measured a spectrum of the pzt driving voltage [V/Hz^1/2] and then converted it to a frequency spectrum [Hz/Hz^1/2].

I used the actuation efficiency of 1MHz/V for the calibration, this number is based on the past measurement.

RMS which is integrated down to 1Hz  is 1.6MHz.

This number is almost what I expected assuming the cavity swings with displacement of x ~< 1um.

-- flashing

A picture below is a ETMx CCD monitor.

One of the spot red circled in the picture blinks when it's unlocked. And once we get the lock the spot stays bright.

 Its OK, but the real number comes from measuring the time series of this in the daytime (not the spectrum). What we care about is the peak-peak value of the PZT feedback signal measured on a scope for ~30 seconds. You can save the scope trace as a PNG.

 I leave notes about a plan for the green locking especially on the PSL table.

 (1) open the door  of the MC13 tank to make the PSL beam go into the MC.  Lock it and then optimize the alignment of the MC mirror so that we can later align the incident beam from the PSL by using the MC as a reference.   

 (2) Remove a steering mirror located just after the PMC on the PSL table. Don't take its mount, just take only the optic in order not to change the alignment .

 (3) Put an 80% partial reflector on that mount to pick off ~200mW for the doubling . One can find the reflector on my desk.

 (4) Put some steering mirrors to guide the transmitted beam through the reflector to the doubling crystal. Any beam path is fine if it does not disturb any other setups. The position of the oven+crystal should not be changed so much, I mean the current position looks good.

 (5) Match the mode to the crystal by putting some lenses. The optimum conversion efficiency can be achieved with beam waist of w0~50um (as explained on #2735). 

 (6) Align the oven by using the kinematic mount. It takes a while. The position of the waist should be 6.7 mm away from the center of the crystal (as explained on #2850). The temperature controller for the oven can be found in one of the plastic box for the green stuff. After the alignment, a green beam will show up.

(8) Find the optimum temperature which gives the best conversion efficiency and measure the efficiency.

(7)  Align the axis of the PSL beam to the MC by steering the two mirrors attached on the periscope.

A poor lonely SR785 was found this morning roaming around in the lab in evident violation of the fundamental rule which requires all the equipment on carts to be brought back inside the lab right after use.

The people and the professors related to the case should take immediate action to repair for their misdeed.

A progress on the end PDH locking :

by using a modified PDH box the green laser on the X-end station is locked to the arm cavity.

So far the end PDH locking had been achieved by using SR560s, but it was not sophisticated filter.

To have a sophisticated filter and make the control loop more stable, the PDH box labeled "#G1" was installed instead of the SR560s.

After the installation the loop looks more stable than the before.

Some details about the modification of the PDH box will be posted later.

Although, sometimes the loop was unlocked because the sum-amp (still SR560) which mixes the modulation and the feedback signal going to the NPRO PZT was saturated sometimes.

Thus as we expected a temperature control for the laser crystal is definitely needed in order to reduce such big low frequency drive signal to the PZT.

A thermal feedback was installed to the end PDH locking and it works well. There are no saturations 

As I said the feedback signal was sometimes saturated at the sum-amp because the drive signal going to the laser PZT was large at low frequency (below 1Hz).

So I made a passive low pass filter which filters the signal controlling the temperature of the laser crystal, and put it before the temperature drive input.

Now the amount of the feedback signal got reduced when it is locked, and there are no saturations. It's very good.

(thermal property of the crystal) 

According to the specification sheet for the 1W Innolight, the thermal properties of the crystal are:

  Response coefficient : 3GHz/K 

  Temperature control coefficient : 1K/V

  Thermal response bandwidth: 1Hz 

(filter circuit and actuator response)

In order to feedback the signal blow 1Hz, a low pass fiter is needed. 

The attachment:1 shows the filter circuit I made.

Since I found that the drive input had an input impedance of 100kOhm, so I put relatively big resistors to have a moderate gain.

The expected actuator responses are also attached.

The blue curve represents the response of the PZT, the green is the thermal response including the low pass filter and the red curve is the total response composed of both the responses.

I assume that the PZT response is 1MHz/V according to Mott's measurement.

Also I assume that the thermal response intrinsically has two poles at 1Hz according to the specification listed above.

In the total response, there is a little gain reduction around 2Hz due to the cancelation of each other, but it still looks okay.

GJ!

On the wiki I summarized about the modification of the PDH box which is currently running on the end PDH locking. 

http://lhocds.ligo-wa.caltech.edu:8000/40m/Electronics/PDH_Universal_Box

The box was newly labeled "G1" standing for "Green locking #1".

We added a channel on c1psl in order to monitor the temperature of the PPKTP sitting on the PSL table.

To take continuous data of the temperature we added the channel by editing the file: target/c1psl/c1psl.db

We named the channel "C1:PSL-PPKTP_TEMP".

To reflect this change we physically rebooted c1psl by keying the crate.

Is this a setpoint temperature that we can change by writing to the channel or is it a readout of the actual temperature of the oven?

kiwamu:

This is a readout channel just to monitor the actual temperature.

The power of the green beam generated on the PSL table should be about 650uW in terms of the shot noise.

       One of the important parameters we should know is the power of the green beam on the PSL table because it determines the SNR.

The green beam finally goes to a photo detector together with another green beam coming from the arm cavity, and they make a beat signal and also shot noise.

So in order to obtain a good SNR toward the shot noise at the photo detector, we have to optimize the powers.

If we assume the green power from the arm is about 650uW,  a reasonable SNR can be achieved when these powers are at the same level. 

To get such power on the PSL table, a 90% partial reflector is needed for picking it off from the PSL as we expected.

  power dependency of SNR

      Suppose two lasers are going to a photo detector while they are beating (interfering).

The beat signal is roughly expressed by

      [signal]  ~ E_1^* E₂ + E₁ E_2^*,

                     ~ 2 ( P₁ P₂)^½ cos (phi), 

 where  E_{1 ^and} E₂ represent the complex fileds,  P₁ and P₂ represent their powers and phi is a phase difference.

This equation tells us that the strength of the signal is proportional to  ( P₁ P₂)^½  .

At the same time we will also have the shot noise whose noise level depends on the inverse square route of the total power;

          [noise] ~ ( P₁ + P₂)^½.

 According to the equations above, SNR is expressed by

        SNR = [signal] / [noise] ~ ( P₁ P₂)^½  / ( P₁ + P₂)^½.

If we assume P₁ is fixed,  the maximum SNR can be achieved when  P₂ goes to the infinity. But this is practically impossible.

Now let's see how the SNR grows up as the power P₂  increases. There are two kinds of the growing phase.

    (1) When P₂ <  P₁ , SNR is efficiently improved with the speed of  P₂^½.

    (2) But  when P₂ >   P₁ , the speed of growing up becomes very slow. In this regime increasing of  P₂ is highly inefficient for improvement of the SNR.

Thus practically P₁ ~ P₂  is a good condition for the SNR.

At this point the SNR already reaches about 0.7 times of the maximum, it's reasonably good.

 power estimation

         According to the fact above, we just adjust the green powers to have the same power levels on the PSL table.

 The table below shows some parameters I assume when calculating the powers.

         Attached figure shows a simplified schematic of the optical layout with some numbers. 

By using those parameters we can find that the green beam from the arm cavity is reduced to 650uW when it reaches the PSL table.

To create the green beam with the same power level on the table, the power of 1064 nm going to the doubling crystal should be about 150mW.

This amount of the power will be provided by putting a 90% partial reflector after the PMC.

In order to increase the green power on the PSL table, I moved the position of the Second Harmonic Generation (SHG) crystal by ~5cm.

After this modification, the green power increased from 200 uW to 640 uW. This is sufficiently good.

     As I said in the past elog entry (# 3122), the power of the green beam generated at the PSL table should be about 650 uW.

I measured the green power by the Ophir power meter and found it was ~200 uW, which made me a little bit sad.

Then I performed the beam scan measurement to confirm if the crystal  was  located on the right place. And I found the postion was off from the optimum position by ~5cm.

So I slided the postion of the SHG oven to the right place and eventually the power got increased to 640 uW.

some notes: 

(power measurement)

        The outgoing beam from the SHG crystal is filtered by Y1-45S to eliminate 1064nm.

According to Mott's measurement Y1 mirrors are almost transparent for green beams (T~90%), but highly reflective for 1064nm (T~0.5%).

All the green power were measured after the Y1 mirror by the Ophir configured to 532nm, though, the measured power might be offseted by a leakage of 1064nm from the Y1 mirror.

I didn't take this effect into account.

(beam scanning and positioning of crystal)

          Here is the properties of the incident beam. These numbers are derived from the beam scan measurement.

w0h             = 52.6657      +/- 0.3445 um

w0v             = 52.4798      +/- 0.1289  um

z0h             = 0.574683         +/- 0.001937 m

z0v             = 0.574325         +/- 0.0007267 m

Where the suffixes "h" and "v" stand for "horizontal" and "vertical" respectively.

The distances are calibrated such that it starts from the lens postion.

Both waist size are already sufficiently good because the optimum conversion can be achieved when the waist size is about 50um ( see this entry)

The measured data and their fitting results are shown in attachement 1.

         According to my past calculation the center of the crystal should be apart from the beam waist by 6.8mm (see this entry). 

So at first I put the oven exactly on the waist postion, and then I slided it by 6.8mm.

(to be done)

        I need to find an optimum temperature for the crystal in order to maximize the green power.

Previously the optimum temperature for the crystal was 38.4 deg. But after moving the position I found the optimum temperature is shifted down to around 37deg.

The feedback signal going to the laser PZT at the X end station was measured in the daytime and the nighttime.

It's been measured while the laser frequency was locked to the arm cavity with the green light.

The red curve was measured at 3pm of 8/July Friday. And the blue curve was measured at 12am of 9/July Saturday. 

As we can see on the plot, the peak-peak values are followers

              daytime:  ~ 4Vpp

       nighttime:  ~1.8Vpp

It is obvious that the arm cavity is louder in the daytime by a factor of about 2.

Note: the feedback signal is sent to the PZT only above 1Hz because the low frequency part is stabilized mostly by the crystal temperature (see this entry)

The optimum temperature for the doubling crystal on the PSL table was found to be 36.8 deg.

I scanned the temperature of the crystal from 44 deg to 29 deg, in order to find the optimum temperature where the frequency doubled power is maximized. 

(method) 

The method I performed is essentially the same as that Koji did before (see this entry).

(1) First of all, I enabled the PID control on the temperature controller TC200 and set the temperature to 44 deg.

(2) After it got 44 deg, I disabled the PID control.

(3) Due to the passive cooling of the oven, the temperature gradually and slowly decreased. So it automatically scans the temperature down to the room temperature.

(4) I recorded the power readout of the power meter: New Port 840 together with the temperature readout of TC200. The power meter was surely configured for 532 nm.

(result)

The measured data are shown in the attachment. 

The peak was found at T=36.8 deg where the power readout of  532 nm was 605 uW.

Compared with Koji's past data (see this entry), there are no big side lobes in this data. I am not sure about the reason, but the side lobes are not critical for our operation of the green locking.

 (to be done)

 Adjustment of the PID parameters

 It seems like you might inherit an offset by using step (3) b/c of the temperature gradient between the crystal and the sensing point. Depending on how large this gradient is you could increase the linear coupling from temperature to intensity noise from zero to a significant number. Phase noise should not be effected.

SInce these things (ovens) are so low time constant, shouldn't we

1. Lock to a temperature
2. Let the oven equilibrate for however long - a few tau maybe - my oven has a time constant of 60 sec, don't know if this is fast or slow compared to that
3. Measure P_532/P_1064
4. Change the setpoint
5. Go back to step 1

What did you use to filter the 2f components from your error signal? A homemade low pass or what?

Kiwamu:

I am using a homemade low pass filter.

It's just a RC passive LPF with the input impedance of 50 Ohm.

[Koji, Jenne, Kiwamu]

This is to describe the work that went on in the Cleanroom today.  Kiwamu's entry will detail the tidbits that happened in the chamber.

We engraved the periscope mounts with the mirror info for the mirrors which were placed in the periscope.  We also engraved the barrels of the optics with their info, for posterity.  Koji carefully put the mirrors into the periscopes.  Since we have wedged optics, the goal was to have the front HR surface of the mirror parallel to the plane of the mount, and leave a bit of space behind one side of the optic (if we just pushed the optic fully in, the HR surface wouldn't be flat, and would send the beam off to the left or right somewhere).  Once the mirrors were mounted in the periscopes, we checked the vertical levelness of the outcoming beam.  For the first periscope (the one which has been installed on the BS table), the beam was deflected upward (2.5)/32 inches over 55inches.  This corresponds to a 1.4mRad vertical deflection.  The second periscope (which will eventually be installed on the OMC table) had a deflection of 1/32 over 55inches, or 0.6mRad.  We did not check the side-to-side deflection for either of the periscopes.

We also engraved one more DLC mount with mirror info, and put a mirror into the mount.  This is one of the optics that was placed onto the BS table today, which Kiwamu will describe.

We removed TT#3 from the BS chamber so that it could have rubber vertical dampers installed, and be characterized.  For future reference, the #'s of the Tip Tilts refers to the serial number of the suspension block piece, which forms the top horizontal bar of the frame. 

If I understand your elog, you are just calculating the the offset in position space that you get by having a refractive index.

Did you end up changing the mode matching so that the rayleigh range (which changes with refractive index) was confocally focused inside the crystal (e.g. Zr = 15 mm?

- As you said, I just calculated the waist position in the crystal because the speed of light changes in a medium and eventually the waist position also changes.

- Yes, I did. Once you get a beam with the right waist size, you just put your crystal at the waist position with the offset.

  In fact you don't have to think about the rayleigh range inside of the crystal because what we care is the waist size and it doesn't change.

 I thought we cared about satisfying the confocal focusing parameter, that is to say we want to set Zr = 2L_crystal. If Zr changes inside the crystal, this is the number we care about..isn't it NOT the waist size, but the rayleigh range we care about? I am not entirely sure what youre response is saying you did...

1. Calculate Zr = pi * wo^2/(lamba/n)
2. Do mode matching to get this wo in free space
3. Calculate the offset you need to move the oven by using n
4. Move hte ovens

OR

1. Calculate Zr = pi*wo^2/(lamba)
2. Do mode matching to get this in free space
3. Calculate the offset you need to move your ovens using n
4. Move your ovens

I guess the waist size would also let me know - are you using 69 um or 53 um waist size?

Okay, I guess I understand what you want to know. I did the following steps.

1. calculated the conversion efficiency as a function of the waist size. I found w~50um was the optimum waist.

  Note: the confocal relation Zr = pi * wo^2/(lamba/n) = L/2  gives us almost the same optimum waist. 

elog #2735

  2.  Did mode matching to get w=50um

elog #2959

elog #3188

  3. calculated the offset

elog #2850

 4. Moved the ovens

 I turned ON the laser at the X end station, which had been OFF for several weeks because of the crane business.

Now the green beam hits the ITMX and I got a reflection back to the end table.   

This green beam will be a nice reference when we install the green periscope in the chamber.

If it's necessary, feel free to correct the alignment of the green beam during my absence.

 With a help from Joe, I made a diagram of the simulated plant for green locking in order to get better understanding and consensus.

Eventually these simulated plants will help us developing (sometimes debugging) the digital control systems.

Here is the diagram which tells us how we will setup and link the control/plant models and on which machine they will be running.

Basically upper side represents the realtime control, and the lower side is the simulated plant.

The models are talking to each other via either a local shared memory (orange line) or the reflective memory network (purple line).

 Each model is stil not systematically named, at some point we have to have an absolute standard for naming the models.

- current model names -

  GCV = Green Control model at Vertex

  GCX(Y) = Green Control model at X (Y) end

  GPV = Green Plant model at Vertex

 - things to be done -

1. let the RFM work

2. revise the old plant models : SUP, SPX(Y) and LSP

 I put some optics to get the green SHG on the PSL table.

Now a green light successfully comes out from the doubling crystal.

Since the optical layout of the PSL table was dramatically changed, I had to re-setup the green SHG stuff. 

 - what I did

I put a steering mirror after the 90/10% pick off mirror, and then a half wave plate and a convex lens.

The lens is approximately on the right place.

 To get the green beam easily, temporarily I raised the current of the NPRO up to 2 A.

I connected the oven to the heat controller, set the temperature to 36.8 deg which is the set point previously used.

After putting and aligning the oven, I finally obtained the green beam.

At the end of the work I set the NPRO current back to 0.9 A and relocked the PMC.

- things to be done

 1. more precise mode matching

 2. optimization of the temperature 

I improved the mode matching of the incident beam to the doubling crystal on the PSL table.

As a result it apparently got better (i.e. brighter green beam), but it's not the best because now the beam is a little too tightly focused on the crystal. 

I am going to work on it again someday after seeing the beat note signal.

- The measured waist sizes are

    41.94 [um] for vertical mode

   42.20 [um] for horizontal mode

while the optimum waist size is 50 um (see entry #3330).

The plot below shows the beam scan data which I took after improving the mode match.

[ Kevin and Kiwamu ]

 We made the setup for the green PLL stuff on the PSL table.

 Now the two green beams are happily going to the RFPD.

Tomorrow we try to catch the beat note signal 

  - - - what we did

 * took the two light doors out from the OMC and the MC chamber in order to let the green light go through there.

 * using aluminum foils we covered the space between the OMC and the MC chamber in order to protect from dust

 * aligned the steering mirrors inside of the chamber because the height of the green light coming out from the chamber had been a little bit low at the PSL table.

 * at the PSL table we put several steering mirrors and a beam splitter which combines the two green lights

 * installed Hartmut's RFPD and applied -150V bias on it.

 * put a lens on each path of the green beam in order to make the beam size approximately the same at the RFPD

 * closed the light doors

- - - Notes

 * At the beginning, an output signal from the RFPD was pretty small ( less than 1mV at DC ), so I replaced a feedback resistor that was 241 Ohm by 24 kOhm.

   As a result the signal became about 10mV when the green lights go into the PD.

* Actually the power of the green beams are so weak.

  I measured them by using a Newport power meter, it said something like 4 uW for both of the green lights. 

  One of the reasons is that the transmitted light from the PMC which generates one of the green lights is already weak. It's about 480 mW ( while more than 600 mW was reflected by the PMC ! ).

  I am going to make sure if these numbers are reasonable or not.

 I made some KAIZENs (what does kaizen mean ? ) for the PSL green setup.

I replaced the lenses for the modematching of the two green lights at the PSL table, and the beams now look pretty identical.

Also I tuned the temperature setpoint of the doubling crystal and eventually the green light increased to 14 uW at the PSL table.

Once I finish the modification of the RFPD tomorrow, I am going to search for the beat note signal.

( details )

 - In-vac green mirrors

   I found one of the green steering mirror, which stands at the corner of the MC table, was clipping the green light.

  So I steered another mirror, which sends the beam to the clipping mirror after the downward periscope.

 I touched also the last steering mirror in the OMC chamber to correct the alignment.

 - temperature of the doubling crystal

   I took a quick temperature scan in order to find an optimum point for the crystal temperature.

  The scan was performed by just turning the heater off after I heated up the crystal up to 40 deg. 

  Using the NewPort power meter I found the optimum point around 37.3 deg. So I set the temperature to that point.

 - mode matching lenses

 As written in this entry , Kevin and I had put some lenses to make the two green beam almost the same size at the RFPD.

 But today while I was checking these mode-profiles by using a sensor card, I found they were not so matched.

 Therefore I replaced these lenses to match them more.

The idea of this replacement is t to let them have a long Rayleigh range, such that they can efficiently and easily interfere because of the flatness of their wave fronts along a distant.

 For the green light from the chamber, I put one more lens to form a Keplerian beam shrinker (see here about the Keplerian lens configuration).

 They look pretty identical now.

 I did a health check for a 80MHz VCO box. 

I started taking care with the black VCO box, which has been sitting on the SP table and will be used for converting the green beat signal from frequency to voltage.

The circuit in the box basically consists of three parts: low pass filters (LPFs), a VCO and RF amplifiers.

Today I checked the LPF stage. It looks pretty healthy.

Tomorrow I will check the VCO part, especially I am curious about the VCO range.

 (soldering)

 Since somebody ( surf students ?) removed some resistors, the VCO was just freely running without being applied any voltage.

I put some resistors back on the circuit board by soldering them.

Now the resistors are placed in the same configuration as the original schematic (link to LIGO DCC) except for the wideband signal path, which has a differential input.

I left the wideband path disconnected from the VCO.

(transfer function measurement)

The LPF part in 'external mod' path contains two stages in series:

one is for cutting off demodulated signals above fc=80MHz and the other one is for PLL servo with pole=1Hz, zero=40Hz.

In order to activate this path I shorted 10th pin of the analog switch: MAX333A.

During the transfer function measurement I injected signals to 'external mod' input and took the output signal from a test point pin TP7.

The plot below shows a fitting result of the measured transfer function of the whole LPF stage. I used liso for the fitting.

The measured filter's shape agreed with the design. (though I haven't checked 80MHz cut off)

(Kiwamu, Yuta)

Background:
  For green locking, we are planning to feedback frequency differential signal to ETM suspension for the final configuration.
  We don't have ETM suspension control system right now, so we are going to feedback the signal to X-end laser frequency for a test.
  We have two loops for the servo;
    1. coarse locking using frequency counter, feeding back to the laser temperature
    2. using VCO, feeding back to the laser PZT
  Today, we checked frequency counter SR620 and see how to get the small beat note signal(-48dBm; see elog #3771).

What we did:
  1. Using Marconi(RF signal generator), put RF signals to SR620 and see how small signal SR620 can see.
    It depends on the frequency. For 80MHz signal, you need more than about -9dBm.
       60MHz  >-12dBm
       70MHz  >-10dBm
       80MHz  >-9dBm
       90MHz  >-8dBm
      100MHz  >-7dBm

Since we are going to lock the frequency difference between X-end and PSL to 80MHz, we need at least +40dBm amp before putting the signal into SR620.

RF amplifier ZHL-32A has around +28dBm +28dB gain at 80MHz, so we need 2 of them.

  2. Marconi -> ZHL-32A -> ZHL-32A -> SR620 and see how small 80MHz signal SR620 can see.
    Around -68dBm. This should be enough.

  3. SR620 has "STRIP CHART" output on the rear panel. The output voltage is proportional to the mean frequency of the input.
    The output range is 0-8V. So in order to get 4V for 80MHz, set SCALE to 20MHz.

Plan:
 - find green beat again and see if SR620 can see it with double ZHL-32A configuration

ZHL-32A is a high power (well..., actually middle power) amplifier with the max output power of +29dBm (~1W!).
It seems to be overkill.
Your signal is so small so you don't need ZHL-32A, but can use small amp which we may have somewhere in the lab.

And the description:
"RF amplifier ZHL-32A has around +28dBm gain at 80MHz"
The unit is wrong.

Background:
  Yesterday, I said I will use ZHL-32A for amplifying beat note signal, but as Koji pointed out, ZHL-32A is for medium high power.
  So I changed my mind to use ZFL-1000LN instead. ZFL-1000LN is a low noise RF amp whose maximum power is 3dBm.
  Also, we included a splitter in our consideration.

What I did:
1. Set up a new setup. ZFL-1000LN has a gain of +24dB at 80MHz and splitter ZFRSC-42 has a loss of -6dB. So;

beat note signal -> ZFL-1000LN -> ZFL-1000LN -> ZFRSC-42 => SR620 and VCO

2. Measured frequency-output relation. When the input signal was 80MHz -48dBm, the output was -8.7dBm. For other frequencies;
       60MHz  -3.3dBm
       70MHz  -5.7dBm
       80MHz  -8.7dBm
       90MHz  -5.5dBm
      100MHz  -3.5dBm
  So, we can see frequency up to >100MHz by SR620 using this setup.

3. Checked harmonics peak levels of the output using an RF spectrum analyzer. When the input signal was 80MHz -48dBm, the height of the peaks were;
     80MHz -8.8dBm
    160MHz -30dBm
    240MHz -42dBm
  Other peaks were smaller than the 3rd harmonics. Also, RF PD that detects beat note signal has a cut-off frequency at around 100MHz. So, we don't need to worry about wave transformation for this setup.

Yes. I corrected my previous elog.

 I calibrated the VCO frequency as a function of the applied input voltage.

The range is approximately +/- 5 MHz, which is large enough to cover the arm's FSR of 3.75MHz.

======== measured parameters ======

center frequency: 79.5 MHz

VCO range: 74MHz - 84MHz

coefficient : 1.22MHz/ V (+/- 2V range)

nominal RF power: -0.66 dBm

(Note: The measurement was done by using Giga-tronics hand-hold power meter.)

P.S. There is a document about the 80MHz VCO box. This may be helpful. 

link to LIGO DCC

(Kiwamu, Yuta)

We succeeded in coarse locking the green beat frequency, using a frequency counter and feeding back the signal to the X-end laser temperature.

Setup:
  beat note -> RF PD -> SHP-25 -> SLP-100 -> ZFL-1000LN -> ZFL-1000LN -> ZFL-500LN -> ZFRSC-42 -> SBP-70 -> ZFRSC-42 -> SR620 -> c1psl(C1:PSL-126MOPA_126MON)
  c1auxey(C1:LSC-EX_GREENLASER_TEMP) -> X-end laser temp

  The frequency counter SR620 converts the beat frequency to voltage.
  We added some filters (SHP-25, SLP-100, SBP-70). Otherwise, SR620 doesn't count the frequency correctly.

What we did:
  1. Getting green beat note again
     Set PSL laser temp to 31.81 °C and X-end laser temp to 37.89 °C.
     Set PPKTP crystal temp to 37.6 °C, which maximizes output green beam power.

  2. ADC channel and DAC channel
     Disconnected one channel going into VME-3123 (at 1X1) and used c1psl's C1:PSL-126MOPA_126MON as ADC channel for the output from SR-620
     Made a new DAC channel on c1auxey named C1:LSC-EX_GREENLASER_TEMP, and disconnected one channel from VME-4116 (at 1X9) to use it as DAC channel for X-end laser temperature control.

  3. Coarse lock by ezcaservo
     Ran;
        ezcaservo C1:PSL-126MOPA_126MON -s 150 -g -0.0001 C1:LSC-EX_GREENLASER_TEMP
     "-s" option is a set value. The command locks C1:PSL-126MOPA_126MON to 150 (in counts), using 0Hz pole integrator.

Result:
  The beat frequency locked on to ~77MHz. The frequency fluctuation of the beat note during the servo is ~3MHz with ~10sec timescale.
  VCO has  ~+/-5MHz range, so this coarse locking meets the requirement.

  Here's a plot of the error signal and feed back signal;
 

Wow! Great guys!!

Can I expect to see the spectra of the frequency counter output with and without the servo?

RA: I think the SBP-70 is a bad idea. It limits the capture range. So does the SHP-25. You should instead just use a DC-block; the SR620 should work from 1-200 MHz with no problems.

Also, we have to figure out a better solution for the DAC at the ends: we cannot steal the QPD gain slider in the long run and the 4116 DAC at the ends has all 8 channels used up. Should we get the purple box for testing or should we try to use the fast DAC in the EX IO chassis as the actuator?

I measured the open loop transfer function of the 80MHz VCO's PLL while locking it to Marconi.

 This measurement is for a health check and a characterization of the PLL

The transfer function looks good, it agrees with the designed filter shape.

(measurement setup) 

 The frequency of Marconi is set to 79.5MHz which is the center frequency of the VCO.

The signal from Marconi is mixed down with the VCO signal at a mixer ZLW-3SH.

Then the demodulated signal goes to a 80MHz LPF to cut off high frequency components.

And it goes through a control filter which has 1Hz pole and 40Hz zero (see this entry).

The 80MHz LPF, the controls filter, the VCO and the RF amplifier are all built in the box.

 In order to measure the open loop transfer function I inserted SR560 before the 80MHz LPF.

Using T-splitters the input and the output of SR560 are connected to a spectrum analyzer SR785.

(results)

 Exciting the system using a source channel of SR785, I measured the open loop transfer function.

The unity gain frequency was measured to about 20 kHz.

It agrees with the designed filter shape (though the gain factor is a little bit underestimated).

Apparently there is a phase delay at high frequency above 10kHz, but it is okay because the phase margin is quite acceptable up to 100kHz.

However I found that the control range was quite narrow.

The PLL was able to be kept in only +/- 1MHz range, this fact was confirmed by shifting the frequency of Marconi during it's locked.

I will post another elog entry about this issue.

 (notes)

 Marconi power = 6dBm

 VCO power after RF amp. = -0.6 dBm

 Marconi frequency = 79.5 MHz

 Phase detection coefficient = 0.4 V/rad (measured by using an oscillo scope)

Bad; there should be a passive ~1 MHz LP filter between the mixer and anything that comes after. The SR560 + mixer does not equal a demodulator.

The hold-in range of the PLL must be greater than +/- 4MHz in order to bring the arm cavity to its resonance. 

(Hold-in range is the range of frequencies over which the PLL can track the input signal.)

However as I mentioned in the past elog (see this entry), the PLL showed a small hold-in range of about +/- 1MHz which is insufficient.

In this entry I explain what is the limitation factor for the hold-in range and how to enlarge the range.

(Requirement for hold-in range )

 We have to track the frequency of the green beat signal and finally bring it to a certain frequency by controlling the cavity length of the arm.

For this purpose we must be able to track the beat signal at least over the frequency range of 2*FSR ~ +/- 4MHz.

Then we will be able to have more than two resonances, in which both the end green and the PSL green are able to resonate  to the arm at the same time. 

And if we have just two resonances in the range, either one of two resonances gives a resonance for both IR and green. At this phase we just bring it to that frequency while tracking it.

  Theoretically this requirement can be cleared by using our VCO because the VCO can drive the frequency up to approximately +/- 5MHz (see this entry)

 The figure below is an example of resonant condition of green and IR. The VCO range should contain at least one resonance for IR.

(In the plot L=38.4m is assumed)

(an issue) 

However the measured hold-in range was about +/- 1MHz or less. This is obviously not large enough.

According to a textbook[1], this fact is easily understandable.

The hold-in range is actually limited by gains of all the components such as a phase detector's, a control filter's and a VCO's gain.

Finally it is going to be expressed by,

                         [hold-in range] = G_pd * G_filter * G_vco

 At the PD (Phase Detector which is a mixer in our case) the signal does not exceed G_pd [V] because it appears as G_pd * sin(phi).

When the input signal is at the edge of the hold-in range, the PD gives its maximum voltage of G_pd to maintain the lock.

Consequently the voltage G_pd [V] goes through to G_filter [V/V] and G_vco [Hz/V].

This chain results the maximum pushable frequency, that is, hold-in range given above equation.

In our case, the estimated hold-in range was 

                      [hold-in range] ~ 0.4 [V] * 3 [V/V] * 1 [MHz/V]

                          = 1.2 [MHz]

This number reasonably explains what I saw.

In order to enlarge the hold-in range, increase the gain by more than factor of 5. That's it.

* reference [1]  "Phase-Locked Loops 6th edition" Rolan E. Best

In order to enlarge the hold-in range I modified the control filter and increased the gain by factor of 25 in the PLL.

It successfully enlarged the range, however the lock was easily broken by a small frequency change.

So I put a low frequency boost (LFB) and it successfully engaged the PLL stiffer.

Now it can maintain the lock even when the frequency disturbance of about 1MHz/s is applied.

(enlargement of the hold-in range)

I modified the control filter by replacing some resistors in the circuit to increase the gain by factor of 25.

        - R18 390 [Ohm]  => 200 [Ohm]

    - R20 1000 [Ohm] => 5000 [Ohm]

    - R41 39 [Ohm] => 10 [Ohm]

 This replacement also changes the location of the pole and the zero

    - pole 1.5 [Hz] => 0.3 [Hz]

    - zero 40 [Hz] => 159 [Hz]

 Note that this replacement doesn't so much change the UGF which was about 20 kHz before.

It becomes able to track the input frequency range of +/- 5MHz if I slowly changes the frequency of the input signal. 

However the PLL is not so strong enough to track ~ 1 kHz / 0.1s frequency step.  

(make the PLL stiffer : a low frequency boost)

One of the solution to make the PLL stiffer is to put a boost filter in the loop.

I used another channel to more drive the VCO at low frequency. See the figure below.

The 80MHz VCO box originally has two input channels, one of these inputs was usually disabled by MAX333A.

This time I activated both two input channels and put the input signal to each of them.

Before signals go to the box, one of the signal path is filtered by SR560. The filter has G=20000, pole=0.3Hz. So it gives a big low frequency boost.

Once the PLL was achieved without the boost, I increased the filter gain of SR560 to 20000 because locking with the boost is difficult as usual.

 I've set up a PID script that senses the EX-PSL Green Beat note (from the frequency counter) and actuates on the temperature of the end laser to drive the beat note to a given setpoint.

• I've added the necessary EPICS channels to c1iscaux and rebooted it so that the channels are live. They are listed in a new database file slow_grnx_pid.db
• This database was added to the list of those loaded by startup.cmd.  
• The PID script, GRNXSlowServo, is just a modified version of FSSSlowServo.
• The version I've been running is currently in /cvs/cds/caltech/users/abrooks/.
• There's also an MEDM screen in this directoy, C1LSC_EX_GRN_SLOW.adl, there that shows the PID settings.

Right now the script only passes the initial sanities checks, that is:

1. It runs.
2. You can enable/disable it without any errors and it starts actuating.

The settings all need to be tuned up - e.g. maximum_increment, hard_stops, time_step, PID constants.

Additionally, the units in the whole thing are pretty useless - some of the channels are in VOLTS, others in WATTS. I'd like to change all these to be in Hz. 

EPICS channels added:

• grecord(ao,"C1:LSC-EX_GRN_SLOWKD")
• grecord(ao,"C1:LSC-EX_GRN_SLOWKP")
• grecord(ao,"C1:LSC-EX_GRN_SLOWKI")
• grecord(ao,"C1:LSC-EX_GRN_PID_SETPT")
• grecord(ao,"C1:LSC-EX_GRN_TIMEOUT")
• grecord(stringin,"C1:LSC-EX_GRN_SLOWVERSION")
• grecord(bi,"C1:LSC-EX_GRN_SLOWLOOP")
• grecord(bi,"C1:LSC-EX_GRN_DEBUG")
• grecord(bi,"C1:LSC-EX_GRN_SLOWBEAT")

Stabilizing the beat note frequency using Yuta's temperature servo (see this entry)

I was able to acquire the PLL of 80MHz VCO to the real green signal.

Some more details will be posted later.

 I checked the slow servo and the PLL of 80MHz VCO using the real green beat note signal.

 The end laser is not locked to the cavity, so basically the beat signal represents just the frequency fluctuation of the two freely running lasers.

 The PLL was happily locked to the green beat note although I haven't fedback the VCO signal to ETMX (or the temperature of the end laser).

 It looks like we still need some more efforts for the frequency counter's slow servo because it increases the frequency fluctuation around 20-30mHz.

 (slow servo using frequency counter)

   As Yuta did before (see his entry), I plugged the output of the frequency counter to an ADC and fedback the signal to the end laser temperature via ezcaservo.

The peak height of the beat note is bigger than before due to the improvement of the PMC mode matching.

The peak height shown on the spectrum analyzer 8591E is now about -39dBm which is 9dB improvement. 

     The figure below is a spectra of the frequency counter's readout taken by the spectrum analyzer SR785.

When the slow temperature servo is locked, the noise around 20-30 mHz increased.

I think this is true, because I was able to see the peak slowly wobbling for a timescale of ~ 1min. when it's locked.

But this servo is still useful because it drifts by ~5MHz in ~10-20min without the servo.

Next time we will work on this slow servo using Aidan's PID control (see this entry) in order to optimize the performance.

In addition to that, I will take the same spectra by using the phase locked VCO, which provides cleaner signal.

(acquisition of the PLL)

 In order to extract a frequency information more precisely than the frequency counter, we are going to employ 80MHz VCO box.

 While the beat note was locked at ~ 79MHz by the slow servo, I successfully acquired the PLL to the beat signal.

 However at the beginning, the PLL was easily broken by a sudden frequency step of about 5MHz/s (!!).

I turned off the low noise amplifier which currently drives the NPRO via a high-voltage amplifier, then the sudden frequency steps disappeared.

After this modification the PLL was able to keep tracking the beat signal for more than 5min.

(I was not patient enough, so I couldn't stand watching the signal more than 5min... I will hook this to an ADC)

 [Aidan, Kiwamu]

Kiwamu and I roughly calibrated the analogue output from the SR620 frequency counter yesterday. The input channel, intuitively named C1:PSL-126MOPA_126MON, now reads the measured frequency in MHz with an error of about 0.1MHz - this is, I think, due to the bit noise on the D/A conversion that Kiwamu discovered earlier. That is, the output range of the SR620 corresponds to around 100MHz and is digitized at 10-bit resolution, and ...

100MHz/(10^2) ~= 0.098MHz. [Sad Face]

Calibration:

We set the EPICS range to [-100, 100] (corresponding to [-5V, 5V]), connected a Marconi to the Freq Counter, input a variety of different frequencies and measured the counts on the EPICS channel.

The linear fit to the calibration data was F = 2.006*EPICScount - 0.2942. From this we worked out the maximum and the minimum for the range settings that give the channel in MHz: EGUF = -200.8942 and EGUL = 200.3058. The previous range was [-410, 410]

I restored C1:PSL-126MOPA_126MON to its original settings (EGUF = -410, EGUL = 410) and added a new calc channel called C1:LSC-EX_GRNBEAT_FREQ that is derived from C1:PSL-126MOPA_126MON. The calibration in the new channel converts the input to MHz.

grecord(calc, "C1:LSC-EX_GRNBEAT_FREQ")
{
        field(DESC,"EX-PSL Green Beat Note Frequency")
        field(SCAN, ".1 second")
        field(INPA,"C1:PSL-126MOPA_126MON")
        field(PREC,"4")
        field(CALC,"0.4878*A")
        }

I rebooted c1psl and burtrestored.

The PID loop is ready to be run on the green beat note but, since the tanks are open, there is no green transmission from the end getting to the PSL table. Nevertheless, here's the screen for the PID loop. The loop script is still in my directory /cvs/cds/caltech/users/abrooks/GRNXSlowServo

The medm screen is attached. It shows the current beat note frequency in MHz ()

In c1auxey/ETMYaux.db I added a couple of channels. These are all displayed on the MEDM screen. I added them to autoBurt.req as well.

• C1:LSC-EX_GRNLSR_TEMP_NOM: the zero-volt setpoint temperature of the end laser (as set on the front panel of the Mephisto controller). This must be entered manually in EPICS as there is no way to read it remotely. [Sad Face]
• C1:LSC-EX_GRNLSR_TEMP_CALC: the sum of the zero-volt set point temperature and the offset temperature set by the input voltage from C1:LSC-EX_GREENLASER_TEMP

I rebooted c1auxey to get these to work.

Once we get the green beat back again, the PID loop should servo on the end laser temperature to drive the Beat Frequency to the Frequency Setpoint, C1:LSC-EX_GRN_PID_SETPT, which can be set by the pink slider.

RA: All MEDM screens must be in the proper MEDM directory!! Also, all perl scripts must have a .pl extension!!! Also, all scripts must be in the scripts directory even if they are in development!!! And all scripts should use 'env' rather than have absolute pathnames for the location of perl, csh, tcsh, python, etc.