Problem after Problem

Oh, holy cow, can we please stop being plagued by problems? Alright, I can’t remember where my last post left off and I’m too lazy to go look it up but I need to get down the basics of things that have been going on recently.

First off, lets start at the end. I found today that one of my windows on the vapor cell is coated with Strontium. I’m not sure how or why. The controllers are keeping the window flanges at around 430 C, and the windows do not have any extra insulation on them. So even though they are recessed, they are quite a bit colder. I touched a thermocouple to the metal/glass junction and the coated one read ~ 272 C while the other was harder to get a reading on but it was only slightly hotter (maybe 10-15 C hotter). Because the temperature difference is not incredibly significant I wonder if there is something else that might make it more of a preference to stick to the one viewport. I turned off the heat on the back of the cell completely and capped the coated window with some foil to boost its temp a little. We’ll see how long it takes to blow all the Sr off of it.

I had been noticing previously a decrease in the size of the error signal. I had believed this might be due to decreased amounts of Sr in the cell so I started playing with hotter temperatures on the back side. Now, it seems in retrospect that what was really happening is that temperatures in the cell were just so that the window was slowly coating. As the window coated the error signal size decreased until I boosted temperatures and finished the job. I love seeing the glow of the atoms in the cell, but maybe we’ll have to run with caps on the viewports to keep the center of the sapphire hotter and less prone to condensation. We’ll see how things look when the windows are clear and the only heat source.

There have been other window scares as well. Yesterday, the temperature controller on the MOT’s slower viewport ran away and the temperature hit 694 C – almost 250 C above the temperature rating of the flange and almost 300 C above the rating of the glass to metal seal. This window is now nick-named “lucky”, because despite this the vacuum did not break (how? I’ll never know). In order to control the cooling of the window, we cut power to the controller and quickly hooked the heater’s leads up to a variac. Despite managing to accomplish this in about 15 seconds, the thermocouple temperature dropped about 50 C. Miraculously, the vacuum was still good. Using the variac I slowly brought the temperature down to just below 400 C and left it there. I don’t want to cool it all the way down to room temperature and temp fate with opening up an overstressed glass to metal seal.

In other news, I spent all last week implementing a homodyne lock to rid the error signal of unwanted noise and junk . It appears as though the circuit I built adds a small offset to the signal but I can zero that out with alignment. The circuit is saved as something like “9khzNoiseReductionCircuit” or something to that effect. It actually runs at ~7 kHz and basically consists of an astable multivibrator  which generates a square wave to control 4 digital switches. One of the switches chops the signal to the voltage controlled oscillate (VCO) which drives the acousto-optic modulator (AOM). Two other switches alternately pass the the error signal to generate one signal for when the AOM is on and a second signal for when the AOM is off. One of these signals then gets inverted and they are summed back together. A low pass filter then takes care of high frequency noise.

The circuit seemed to be doing what it was designed to do but didn’t actually work until we figured out that low frequency noise was still passing through the circuit. It turns out that the square wave was not putting out a 50/50 duty cycle and so was giving preference one of the signals generated by the chop. By adding a potentiometer to adjust the oscillator we were able to zero out all the noise. Unfortunately this achievement came at the point where other problems prevented us from moving on.

Aside from the viewport coating on the vapor cell, I noticed that the laser was only putting out about 1/2 it’s intensity in the 461 nm. The output from the tapered amplifier (TA) was still good however. I spent yesterday with the top and side off of the doubling cavity box, getting the laser power back up to snuff. Most of it turned out to be alignment both of the input beam and the cavity. As it turns out we run at about 75% of the peak 461 nm power normally. The initial hunch was that this might be due to local heating in the doubling crystal when the laser is locked, but I was able to verify that it really has to do with stability of the master laser and that with a lot of pain and patience we can regain that power for a short time until the master becomes less stable again.

NO SIGNAL

The title makes me feel like I work for SETI (the search for aliens). That is, except that the people at SETI expect to see no signal and I do. I have not been able to see an error signal associated with the lamb dip. The lamb dip is on the order of ~2 the natural linewidth of the transition and is shifted by ~90 MHz from the center of the absorption signal (AOM driven at 181 MHz). So we can see that the width is roughly on the same order as the shift which is a good check all around. I was looking for the signal using our homemade homodyne detector. Detector has two high pass filters in it: the first filter has a 3db point of ~40kHz and the second has a 3db point of 340kHz. Neither of which should pose a problem for a 1MHz signal. The mixer in the detector is a SYPD-1 from minicircuits whose response starts at 1 MHz, but should still be just fine.

To make sure that we are sweeping far enough around the 181 MHz AOM frequency I did some fourier transform plots in octave so that we have an idea from the spectrum analyzer how far we are modulating the 181 MHz signal at 1 MHz. We think that initially we were sweeping about .5 MHz. Now I’ve changed the signal so that we think we are sweeping around 2 MHz or so.

As an aside, I changed the board generating the the AOM signal. It consists of a 1 MHz signal that is added to DC voltage which goes to a voltage controlled oscillator (VCO). The output of the vco is then amplified and sent to the AOM. I used an extra op-amp on the board to buffer the DC voltage and then I changed the input and feedback resistors on the summing amplifier to values that were more reasonable to work with.

Watching the lamb dip on the raw signal it’s not hard to be believe that the dip is about 10 mV tall (maybe slightly bigger) coming out of the photodiode. So we want to know, what kind of signal should get generated in the mixer. Well, if we take the second derivative of the lorentzian and set it equal to zero we get the point at which the largest amplitude signal gets sent to the mixer. We can multiply this by the sweep we are imposing on the AOM signal and then by the gains in the homodyne detector. The output of the homodyne detector is cleaned up with a low pass filter that has an RC time constant of ~100 Hz I think but I will double check. In short, we are expecting to see something that is tens of millivolts.

Before I forget, now that I’m thinking about the signal that gets sent to the mixer… we actually get features out of the mixer that show up at places where the laser isn’t happy. So the laser intensity sharply decreases and then comes back in kind of an inverse “lamb dip” shape. For those features to show anything in the mixer they are much larger than our lamb dip and maybe give us some idea of how large the dip has to be in order to see something.

Since we didn’t see anything we decided to see what we could with some modular components. We took the output of the large area battery backed photodiode (this was maybe a mistake even though it should have been plenty fine for 1 MHz) and sent it through 2 amplifier each with a gain of 24 dB. This was then mixed with the original MHz signal in a ZAD-1 mixer from minicircuits. Did not observe anything.

We knew from the position of the lamb dip that we had the right offset from the AOM, but we decided to try and see if we could determine that we were actually sweeping the frequency around the main carrier frequency. So we aligned the pump and the probe beams on a small area battery backed photo diode (good to 500 MHz into 50 ohms), and put the output on the spectrum analyzer. We could not see anything that did not indicate we were simply amplifying RF pickup. The signal did not go away when blocking the beams and could not be seen unless both amplifiers were used. When it did show up it was about –90 dBm. This was very confusing because the 181 MHz difference between the 2 beams should have been KILLING the frequency spectrum and we would have to carefully align in order to see the sidebands well. Not sure what is going on, but this is getting ridiculous.

Research Update 9/6/2011

I pulled apart the cell to clean it and create a better vacuum system. When I pulled the strontium out of the cell several pieces were dark red, most pieces were black, and a small surface area was clean Sr. Also, it was easy to notice that the along with the pad of strontium that had been sprayed on the cold tee there was a ring of black and red as well. I immediately submerged the Sr from the cell in our remaining dodecane. I cleaned the cell out with nitric acid. For an interesting story regarding that check this post. We have been concerned that our “glove sink” doesn’t do a good enough job evacuating oxygen from environment so this time I used a bag over the top of the sink. I fit the bag so that it could line the sink then I filled it with argon and deflated it again twice to get a good Ar environment. Then with positive Ar pressure I cut holes in the bag for my hands and vision. I took only the cleanest pieces of strontium and tried to scrape them off even more before depositing them in the chamber.

I ended up only using a fraction of the total Sr I initially had in the cell. This time I put the sapphire viewports on the cell. The idea is to run in a true vapor cell configuration but be able to evaporate the black colored (still don’t know what it is) and oxide layers from the Sr without coating the windows. I put the cell back on the pump with the chamber evacuating opposite the valve as opposed to through it. I put a blank on the valve end and heated the the valve assembly as well as part of the bakeout.

In the first phase I heated the viewports to 430 C and the rest of the CF flanges to ~330 C. Once the pressure had reached ~2x10^-7 T I cooled the cell so that I could unwrap the tee area between the viewports. I disconnected the heater on the flange joining the tees as well and brought the viewports back up to 430 C.

I then turned on the cartridge heater to ~715-720 C while using freeze spray and ice cubes to try to cool the chamber wall facing the cup. Eventually I realized that this was probably not necessary and the chamber is now sitting there in this state.

The pressure jumped to ~4.6x10^-6 T when I turned up the cartridge heater. Since there is not a lot of strontium to fill the cup most of the vapor will be directed down the tee and deposited in the vapor cell while only a small amount will be lost to areas outside the chamber. This is also true of the oxide layers, however, the cell will never again reach temperatures high enough to make them active. Right now the pressure in the cell is only down to 3.8x10^-6 T so there is a way to go before I’m  going to turn it off.

A couple of outcomes are possible: 1) despite there being no direct line of sight to the viewports they still end up coated. 2) the pressure never comes all the way back down in a reasonable amount of time because the oxide layers are so thick. 3) all of the Sr and oxide layers are transferred from the cup to other parts of the cell and vacuum system. 4) the oxide layer is evaporated and a significant amount of Sr still remains in the cup.

Research Update August 29th

This is more of a note than an update of any kind, but I haven’t been able to sit down and write about what’s been going on research wise lately. The 461 nm laser system works wonderful now as long as you know you to finesse it. We went through the whole system and after passively stabilizing the diode as well as possible and rebuilding the doubling cavity we charted out the stable cavity modes that the diode has over a series of currents and temperatures. While the diode can scan mode hop free for ~30 GHz when things are JUST right, it still doesn’t follow the doubling cavity very far when locked. However, it follows it far enough.

The problem has been that while the laser scans over the resonance in the vapor cell the doppler broadening is about 300 MHz FWHM and the lamb dip is about 190 MHz FWHM. In short I took the cell over a broad range of intensities and alignments, we looked at the linewidth of the laser on a cavity, and we actively pumped (6 L/s) on the cell while running it all with no effect on the linewidth of the lamb dip. However, I noticed today that while we were pumping out on the cell when I first turned on the Sr, as the laser scanned across the resonance, not only could I see fluorescence all the way from window to window, but the brightest part of the fluorescence was concentrated on a very small segment of atoms. However, by the time we got to looking at thing on the scope the fluorescence was limited to the spray of atoms exiting the bottom part of the cell and the brightest part of the fluorescence was very spread out. This indicates that as the cell heats a Sr vapor that is mostly only Sr flows through the cell, however, when we get to temperatures where the Sr absorption is significant something else flows with the Sr.

Basically, watching the length of the fluorescence become limited across the cell is like watching the mean free path diminish. This is not good, and it happens even while pumping out on the cell which indicates that it is something that is generated with the Sr vapor. I’m going to convert the cell into hybrid jet/vapor cell design in an effort to overcome this problem and I’m going to refurbish it in the meantime. We have some lead left over in the cell, and I’m going to cut down the Sr in the cell big time to make sure that only the cleanest Sr is present.

And…. We’re BACK!

Oh, man. What a long past couple of months it has been and for a variety of reasons. Somehow we magically had set the laser JUST right all the time we had been running it before. It is actually an extremely fickle and ornery laser system despite the diode being AR coated. I’ll just list the basic changes we made to the laser system.

1. Replaced the doubling crystal. The AR coating on one end of the original crystal had been been damaged. We don’t know why but we suspect it was related to changing the crystal’s temperature. The new crystal was cushioned in several wraps of .1 mm thick purified lead foil to act as a buffer between the crystal and the aluminum mount.

2. Rebuilt the doubling cavity. Since we were aligning the cavity to a new crystal we took a moment to improve the cavity too. We made the cavity just the right length to be at the most stable beam waist for a 125 mm focal length lens. This required the cavity to be square instead of bowtie which wasn’t an issue because the mirrors are all flat. This also made the piezo scan twice as far as the cavity length with a FSR of 1.2 GHz.

3. Modified the doubling crystal temperature controller. Since we had been worried about the crystal damage having occurred due to temperature rates we installed an adjustable voltage regulator set at ~-18 V down from -48 V to ensure that the crystal temperature would never change faster than 10 C/min.

4. Floated both the doubling cavity box and the master laser system box on Sorbothane. Although the boxes float independently the alignment to the doubling cavity is surprisingly stable. The difference in mechanical noise isolation was immediately noticeable.

5. Mounted the diffraction grating on a piece cut at the right feedback angle on a cage mounted to the laser head face. I made the cavity as short as possible.

6. Placed a wavemeter in the master beam to watch stability and wavelength along with cavity coupling and blue output power. Verified the .1 nm/1 wavenumber FWHM of the cavity temperature.

7. Mapped out wavelength resonances of the diode as a function of current and temperature.

All of the work paid off in the end. The pictures and video show fluorescence in the vapor cell today. The infrared wavelength scanned 10849.13-.18 cm^-1 to cross the atomic transition in the cell. With the cartridge heater at ~500 degrees C (cold end at ~200 C) fluorescence could be seen extending all the way across the cell, not just the central source even with chilled water running to cool the target area.

 

Here you can see fluorescence in the reflections off of both windows.

Sr fluorescence on the 461 nm resonant transition.

Research Update June 9 & 10th

Frustrated while looking for the saturated absorption signal in the cell. Despite differing levels of absorption and laser intensity we could not find a lamb dip with the probe laser. We did however work out that to get the right laser frequency, whether we shift the probe or the pump, the aom needs to be upshifting the frequency either way.

On top of that the etalon from the cell is horrible. Blah. The etaloning was getting worse throughout the time we were working with the cell. When we looked at transmittance through the cell we were getting 50%. We left the windows baking overnight with the rest of the cell cold and transmittance was slightly worse than 50% in the morning. After cooling down the cell, we unwrapped a window and could visibly notice some darkening of the windows and also that the copper had been affected by the strontium.

We ordered sapphire windows and silver coated gaskets to refurbish the cell with after the conference. I also ordered some lead foil so that we don’t have to pound out our own. The sapphire should be fine as long as Sr doesn’t condense on it when it is above 600 C. Otherwise Sr that does find it we can blow off with temperature. I left the cell for the week with the valve at 400 C to blow Sr off of it before we vent and put on new windows. Unfortunately, the sapphire has a lower transmittance than glass (~80% at our wavelength).

There is also the idea that we could lock using vapor in the MOT. But since we don’t know that anybody does that we wonder what problems lie therein.

Research Update June 8

Kind of mundane work today, or maybe it’s just that I’m only writing to keep up the habit and I just really want to go to bed. Let’s get through the highlights.

I closed the angle valve and the pressure in the ion pump went up a decade and half before coming down. It finally settled at ~1 x 10^-8 T! What the heck. Why the huge pressure bump? Why is one pump just as good as two? Either way I wrapped up and stored the turbo pump. It’s nice to have that off the table and out of the way.

I spent way too long this afternoon trying to get the breadboard back on under the MOT chamber. The trick that finally worked? Wedging the faztek nuts in approximately the right place with little balls of aluminum foil to hold them while I installed the board. Even then I still only got 3 out of 4 screws in. I’ll have to check once more to make sure the board doesn’t need that last one.

Scare with the 461 nm laser system today. It was on… and then it wasn’t. and there was no IR power registering on the ammeter. I quickly turned the system off. Then I noticed that the temperatures on both the TA and the master were off the set point. The TA was at ~11 C and the master at ~22 C. A tweak and both came back. However, this put the master about 2 wavenumbers lower than the transition. I finally got it back close by setting the temperature around 25 C. I’ll have to remember to write that down tomorrow so that I don’t forget. When I was running the system later there was a sudden decrease in power on the TA and the current was suddenly running about .6 A lower than it should have been. Very scary and very frustrating. I don’t know if it’s electrical noise or soft power cycling that seems to be rewriting new values to the DAC, but if it turns off the TA while it’s coupled I’m screwed.

Tried to see absorption in the cell one last time and finally had to cool it and reinstall the cartridge heater. I was just getting an optically dense strontium vapor when I had to shut down for the day. Here’s crossing my fingers for tomorrow.