149
edits
Zjs log: Difference between revisions
Jump to navigation
Jump to search
no edit summary
No edit summary |
No edit summary |
||
| (43 intermediate revisions by the same user not shown) | |||
|
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 19:52, 19 March 2015 (CDT)
*Reading a lot: papers, theses, etc. trying to understand the line broadening for our system. I really want to know the decay rate to be able to plug it into our simulations. More on that subject later.
*Got another isolator so i'm able to start working with the wide-tune laser. So far i see a mode-hop free tuning range of about 5GHz, that should be sufficient for our purposes, but i really want to get it working better. Initially didn't seem to work very well, but i'm hopeful that increasing the load is helping give more repeatable/reliable piezo scan performance.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 11:50, 11 March 2015 (CDT)
Quick Update:
*Wide-tune laser is designed, drawn and built. Seeing initial laser feedback. Now to see if I can get it performing well and to see if i can optimize the tuning range.
*Re: Laser linewidths: The interferometer has been set up for measuring linewidths at 1064, i think other Zach aligned it for us. I looked at two lasers we have on hand: the spare laser set up in our default way and the laser set up for rotating the grating about its face. Both had linewidths less than a MHz. Also looked at the experiment laser, initially the linewidth was quite large (few MHz), which may explain some difficulties with the crystal and why we haven't seen any narrow burned features. Attenuating the fast feedback gain to the diode and changing the laser driver seemed to bring the linewidth down to less than a MHz. When locked there was some evidence of further narrowing but the shape appears suspect and difficult to fit. We should probably look at this further.
*Re: EIT and experimental simulations: we looked at the delay between green and fluorescence but it was difficult to conclude when the fluorescence 'turned on'. Figuring out the effective gamma to use in simulations is still an unresolved question. We were able to measure AOM switching and observed times of 100ns or so, as expected. Another question is what effective density to use, have an estimate but this needs to be looked at more carefully as it contains a lot of assumptions. I don't think we're going to have much hope of seeing things until we have parameters really nailed down and a decent handle on a simulation.
*Re: In the crystal: trouble seeing anything when focused, but why?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 15:44, 2 March 2015 (CST)
Recent work, three main irons in the fire:
*wide-tune laser design: finally have Zemax working, need to figure out more exact cavity optical path length measurement to inform CAD laser design.
*green beam on-off: installed switches, have the NI-6002 putting out TTL pulses, doesn't seem very stable so far, perhaps pursue a clock-linked design, maybe we could get better stability. Built a TTL or-gate circuit to allow switches to be controlled by delay box or 6002, so far it doesn't work properly.
*EIT parameters: investigating with my simulation to try and decide on good parameters for seeing EIT in the experiment. So far I see evidence of EIT, but there are still a lot of parameters to play with.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 14:22, 16 February 2015 (CST)
Last few days: Little progress, having trouble digesting the new information. We have been making progress but there is still so much we want to do. Did order some parts to hopefully build a laser with a larger mode-hop-free freq sweep and as a very simple NI DAQ to use to turn RF switches on and off.
Ideally we will have evidence of EIT before the end of the month. Also would love to have a wider absorption scan of the overall (inhomogeneously broadened) absorption feature showing things like a spectral hole and side/anti- holes.
'''Things to better understand:'''
*How much power should we use? I think we need to know this if we want to know how long of EIT pulses to use as well.
*Spectral hole width: this is determined by hole-burning power, i.e. more power=larger width, but what exactly is the relationship? Does this even matter that much when talking about pumping?
*Fraction pumped back: When looking at transmission, it appears that a lot of population can be pumped back into the hole. However, this is somewhat misleading because we may have enough optical depth at that laser power to absorb our laser many times over in the crystal adjacent to that spectral hole in freq. Presumably, if we were to turn up the laser power of the probe scan to the point that it would be not totally absorbed, we would see that absolute contribution of the population pumped back from the other hyperfine ground states. Thus far, when turning up the probe power somewhat (perhaps a x10 factor) we don't see much of an intelligible change. I would expect the overall amount pumped back to appear smaller. Perhaps we just don't have enough probe sweep power, to see the difference.
Alternatively, we should look at the fluorescence. This could give us an observation that doesn't have the relative background. Fluorescence should relate directly to how much is there. We will i think want to understand this better for collecting data for our upcoming EIT experiments anyway.
'''Things to build:'''
*ULE cavity.. this could really help with our laser drift..
*Wide sweep laser, this would be great but will be some work.
*Implement switches on RF sources so we can turn them on and off through a combination of the delay box and computer.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 17:47, 9 February 2015 (CST)
OK, spent a little time today with the He. Some observations:
*spectral hole width depends on burning power. Green power of a few uW is sufficient to create noticeable features in the absorption. But what about burning all the way through, etc. hard to see as you don't really get much fluorescence when you try and scan with low power..
*We don't have our pump AOM in the right spot, we'll have to put it in after the control AOM in the control beam path if we want to get the correct spacing.
*Overall population MAY be decreasing, this would make sense if population was accumulating in the other hyperfine state, we should get the other AOM set up to investigate this.
Lots of things to get sorted out. Also, thinking about building a laser better suited to wide scanning...
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 18:46, 6 February 2015 (CST)
Week in review. It's been an exciting week, we're finally doing some science! after spending a lot of time building things.
*Mon-put the ref cavity in a can; i should probably get some windows for that to further isolate it from the environment. We also did a little work with the crystal.
*Tue-more work with the crystal, hole-burning etc, temp in the lab spiked, destabilizing things.
*Wed-temp was wacked up, did some reading on spectral hole burning.
*Thur-Got back to the crystal. Spent a ton of time trying to get the laser int he exact right spot. The mode-hop-free tuning range is about the same size as the feature we want to see so it doesn't work very well for scanning, as you have to get it int he exact right spot and what you're looking for also has to be in the middle of the scan range. Remarkably, it cooperated for a little while. We were able to slowly ramp the laser piezo and lock the SHG cavity to the moving laser. This lets us scan the whole inhomogeneous linewidth and observe a burned feature 40MHz wide in the overall inhomogeneous linewidth. Needs to be improved if we are going get decent data, but we were able to see something!
*Fri- step back to document, try and digest what we've been looking at, some notes with scope shots:
[[File:2_2_data_summary.pdf|thumb|400px|]]
Note: the laser mode-hopping doesn't appear to cause a problem for the fiber amp, this is good news!
A few questions/issues:
*Need a laser with a larger mode-hop-free tuning range if we are going to be able to scan over the whole thing.
*Need to resolve why it appears we can pump so much back into our hole; need to reconcile abs and fluorescence observations and straighten that out.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 17:33, 30 January 2015 (CST)
Experimenting with spectral hole burning. Burning with a 10mW laser, looking with a sweep that is ~1-2uW.
Observations:
*Spectral hole is perhaps wider than we expect: We need to look at the laser linewidth, but i was thinking we would we see a hole ~1MHz wide and instead its more like ~10-15MHz wide.
*Temperature change-> ref cavity length-> laser wavelength->hole position: We heated the cavity with the heat gun for a fraction of a minute, see the transmission peak (spectral hole) fall out of the scan range. Also as the cavity length was changing and laser changing color, when we burn a hole the hole is large and shallow, this makes sens as the laser was effectively scanning.
*Hole persists for a long time: We see the hole minutes after its creation, as it drifts out of and into view.
*Big question: does burning a hole wipe out an adjacent hole?, no we can burn holes next to each other: two spectral holes
[[File:Two_spectral_holes.pdf|thumb|400px|]]
Also, i changed out the mirror for the invar ref cavity and it worked basically immediately, probably dumb of me to order that other cavity mirror. I wonder if that would have better stability performance? We should probably put it in a can and see.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 18:23, 29 January 2015 (CST)
We are finally back to doing science again! In the last week or so we have the laser locking well to a cavity and seeding the fiber amplifier. Fiber amplifier provides way more power to the SHG cavity. There is still ringing on the SHG output, but it seems to work OK; we could potentially mess with the lockbox parameters to get a cleaner lock. Nick has the AOM beams set up.
Today we finally went looking for hole burning again. We see evidence of it, but it doesn't look as narrow as we would hope, appears to be more like ~5-10MHz wide, when it should be more like the laser ~1MHz (locked actually more like 50kHz?- need to measure this). Perhaps this cavity is fluctuating significantly, (few MHz) or there is some other issue. Presumably, we could up the burning power but burn for less time, giving us a narrower feature? So far we have not observed such behavior yet. So far we have looked with about 1uW of power and burned with much more, up to ~1mW or so for a second.
We're doing science again! Next:
*look at inhomogeneous broadening again, might as well with the more stable laser.
*mess with lock parameters to get more stable/narrower laser?
*ref cavity change?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 17:30, 21 January 2015 (CST)
Added a pick-off to monitor laser on spectrum analyzer/wavemeter. Seems well behaved compared to the previous laser. This is quite encouraging. Have the fiber amp seed launch moved toward to the inside of the table, should protect it a little/keep it out of the way. Next is to mode-match the fiber laser output to the SHG cavity.
Spent some time dialing in the seed, current parameters:
{| class="wikitable"
|-
! 1055.073nm Laser Details !! !! !!
|-
| Settings: || 64.79mA || 22deg C || 44.8V DC on laser piezo
|-
| Power Budget:|| 2.05mW seed|| .95mW to lock || ~125uW to monitor launch
|}
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 13:20, 20 January 2015 (CST)
Swapped out the diode laser for the new one, it seems to be working well. Is much more efficient, 50mA of current provides more output power than 180mA to the previous laser. Lock seems to be working well as well, but it's hard to tell if that's the laser or just that there's a little more power going to the cavity.
Hooked up and seeded the fiber amplifier, with it quite in the lab, the laser and amp stayed locked for an hour no problem. Output power seemed basically constant, to within a percent or or ability to measure.
Question: so if the laser drops lock will that mess up the amplifier? -checked this, if the lock drops, the seed is still there, just not locked so we're ok.
Do need to add a pick-off in the system for the wavemeter/spectrum analyzer.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 13:33, 19 January 2015 (CST)
Came in on Saturday and built a new lock-box circuit with the new board. Just re-hooked it up and it seemed to work! First time i've seen locking via PDH.
Couple of circuit issues to fix:
*need to move the switch that grounds the ramp BEFORE the input resistor to that op-amp
*jumpers that i though would be clever are really unnecessary/wrong, a bnc short at the input does the job.
*put piezo fb switch after gain section so we can observe the gain change the signal magnitude
*note: error signal DC offset pot direction changes with invert switch: this is not ideal..
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 11:59, 15 January 2015 (CST)
So it appears that the mirror reflectivity for the invar cavities is so high it is making them hard to work with. Ordered a broadband (lower reflectivity) mirror from thorlabs, we'll see if that helps.
In the meantime, i've been working with the other "new" laser i have been working on, i would like to get it in the bullpen. There was really no reason it wasn't working, i just needed to keep playing with it until i found some feedback. Interestingly, i think the tilt of the grating had insufficient travel to get feedback at the correct wavelength, apparently this was why it wasn't working/showing any feedback.
Initially i noticed feedback with the polarization of the laser diode oriented 90 deg to what I expected, this lead me to compare to the other diodes but they are in fact the same. I then changed the grating angle to get the correct output wavelength. Maybe i was just unable to get the grating angle close enough before to observe any feedback.
Initially the feedback was not very good, it was not a single peak (in output power vs tilt space), but a hill with many sharp peaks that have different output wavelengths, each local peak about a 20%increase in output power. Eventually i was able to get better performance by messing with tilt and collimation as well as grating angle. I adjusted the asphere position to try and get better feedback but that suffered from the fact that just pushing on the lens and changing its tilt in the mount changed how well the feedback worked; this is a pretty significant problem with our lasers, relying on adjusting collimation by turning the lens in its mount. I followed up by carefully measured the grating angle and tried to get it closer to what it's supposed to be. At some point, while moving the grating to a closer angle to what i expect, the feedback performance greatly improved. Now see about a factor of 6 due to optical feedback, .25mW to >1.5mW output power at about 22mA input current.
For a 1200 rulings/mm grating and 1055nm output, i'd expect about 39.3deg off the normal, but ended up at about 37.7. This is not unreasonable as i'm not able to measure the angle super accurately. Looking at the other laser design, that turns about the grating face, that one appears to have an angle of about 41 deg. Interestingly, these seem a lot more angled than our other 1055nm laser.
Lesson learned re: tuning up lasers, if the grating angle isn't right to get feedback, you're not going to see anything. Useful to make sure you're at the right angle (or close) if you want to have hope of finding feedback.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 19:53, 12 January 2015 (CST)
Crunch time. So i machined 2 invar reference cavities, completed before the break. David is using one and i'm working with the other. So Far I have not been able to get good mode-matching, not good enough coupling to lock to it. Potential issues:
-higher finesse than the other cavity so the ramp has to be better?
-perhaps we need input wedge to avoid disturbing the seed laser?
Also, ordered new circuit boards for the lock box, tried to incorporate most everything i could think of. Omitted the op-amp based ffb and generally streamlined it. Hopefully it works better.
Getting locked to a ref cavity is our number one priority.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 15:59, 17 December 2014 (CST)
So feedback electronics issues are driving me crazy and i'm giving that a rest. Serendipitously, the material for the invar fabry-perot cavities just arrived and i've started working on that. Hopefully after doing this, looking at the circuit with fresh eyes will yield some ideas/progress.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 18:56, 12 December 2014 (CST)
Network analyzer is working better. So far haven't been able to get everything working, but at least we have an additional tool at our disposal. Lots of things to try.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 20:46, 11 December 2014 (CST)
Re: Network Analyzer, ok, this thing is driving me crazy. If you go to too low of a stop freq, the structure disappears, although it doesn't give you an error or anything. This appears to be because if you scrutinize low frequency you need to decrease the resolution bandwidth to compensate. i.e. i think you need to look more accurately, hold the frequency more steady, longer. This means that the scan will take longer; perhaps my impatience got in the way. In any case, if you set it to auto, it should take care of getting an adequately small bandwidth given your other scan parameters. Note: you can also reduce the number of points to make the scan take less time.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 01:26, 11 December 2014 (CST)
for lack of other ideas i've been putting the lockbox circuit in multisim to verify that the gain and phase-margin are sufficient for our application. No startling revelations as I looked at it when building the first box. A few ideas:
-50ohm coupling resistor?
-insufficient gain? Check error signal magnitude on the SHG lock and see if they're vastly different..
-something just not connected? since the ramp and the FFB work, there's a pretty narrow amount of stuff that may be messed up, look in the circuit?
-evaluate the circuit with a network analyzer?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 12:46, 9 December 2014 (CST)
Another day, another crack at circuit debugging..
-tried putting inverter before the lockbox, that doesn't seem to fix it.
-invert fast feedback? putting inverter right before the fast feedback messes up the laser, linewidth looks like crap.
-upped the gain in the slow feedback branch by x100, this makes it the same as the hexagon group lockbox schematic, still not seeing any slow locking at higher gain.
-boo, perhaps i should simulate the circuit to look at bode plots/examine the circuit with a network analyzer?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 16:01, 8 December 2014 (CST)
Re: circuits One issue is coming into focus: the source of the 10s of kHz noise on the circuit rails, that seems to bleed through into the error signal. It appears that its origin is the heater in the lm399s. I'm going to try running them without the heaters and see what happens. Running without the heaters squelches the kHz level noise and doesn't seem to prevent any ffb locking. Maybe this is the way we should run them.
I tried an inline inverter. Tough to use as the input to the driver cannot go negative and the lock box doesn't really put out a negative DC. As a result, when you flip (the positive) signal you get something negative or close, not much room to operate. Also, one of the inverters seems to rail which is problematic.
Also, interestingly, looking at the previous lockbox, there doesn't seem to be any discernible square-wave bleed-through to the error signal monitor. Seems to be i had the input set up wrong when i converted the differential amp to one input. Instead of grounding the other opamp input, i was actually looking at the differential of the power supply rails, that where the square-wave noise bleed through was coming from. That's taken care of, but the slow feedback still doesn't work. hmm. gains? try inversion again?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 19:07, 4 December 2014 (CST)
Have not been able to get the slow feedback working. Hmm.. Tried a variety of things:
-Scrutinized the switching
-Tried locking with slow feedback by itself inverted, didn't see any evidence of locking.
-I know the LM399s seem to be mysterious noise generators. Tried putting some caps across their heaters, as i've seen this in some circuit drawings, didn't seem to help with error signal issue, not sure if it helped with rail noise.
Perhpas:
-circuit power supply could be better?
-try inverting slow branch
-too much/little gain - change cap and resistor values in fb branch
-bad component? but the ramp works fine, most all of that is working
-some strange coupling? it is weird we can see discontinuities on the error signal that correspond to the square pulse, do we see those in the other lock box?
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 19:36, 3 December 2014 (CST)
Made a little bit of headway today with the PDH.
* Tuned up the FP cavity modes; ended up moving the cavity and mode-matching lens a fair amount, probably about an inch, putting the distance from the front mirror very close to where ABCD predicts it should be. Lesson learned: turns out it is much easier to move the lens than the cavity. Was able to get it pretty much single-mode at times.
* Strung together the PDH electronics and it worked! well we could see an error signal anyway. Error signal looks pretty much like the Black paper. Did not have to struggle with the phase-shift in order to get a proper-looking error signal.
* Stuck the lock-box in the set-up. We can turn on the fast feedback and see the central resonance broaden, as we expect, as we zoom in on the peak (i.e. adjust the piezo DC and shrink the ramp), the fast feedback intermittently locks on its own. This is pretty encouraging, so at least the FFB seems to be working. However, when we turn from ramp to lock, it doesn't lock. There is a problem with the slow (piezo) feedback branch. Probably the first thing to look at is the switch.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 22:13, 2 December 2014 (CST)
Spent much of the day messing with the Fabry-Perot cavity. I removed the cavity and checked to see what the beam looks like in the position of the cavity; it looks very similar to what my ABCD matrix script predicts if i remove the input coupler, so i am confident that my mode-matching is decent. I mentioned before that i did not see a Finesse above about 20, it turns out these were not the cavity resonances we were looking for. Nick suggested looking at more extreme orientations. I turned the cavity 90deg in its mount, i.e. rotated the whole can, and then searched for more peaks. Then i saw much sharper peaks, finesse>500. My leading theory is that when i glued the piezo back on, i must have glued it off-center. This led to the cavity resonance mode being at a weird angle and thus hard to find. Perhaps the crappy peaks we were seeing were etaloning between a cavity mirror and the windows? or some other cavity-like effect? Those peaks were very clean, we didn't see neighboring modes, they also benefited slightly from turning the polarization, two differences from what we see now.
Now the finesse looks like it's more like 500, 1.5GHz FSR cavity so about ~few MHz wide resonances, comparable linewidth to the laser. Also the cavity features are narrow enough that we can resolve the EOM sidebands. I stuck a cube in the beam path and now i can resolve corresponding dips in the reflection off the front of the cavity. Spent time tuning up the cavity mode-matching but was unable to kill a feature sitting on the side of the main peak.
Hopefully we can get the electronics all set up now that this cavity is working better.
--[[User:Zjsimmons|Zjsimmons]] ([[User talk:Zjsimmons|talk]]) 17:46, 1 December 2014 (CST)
Yikes, it's Dec, where has the semester gone. I have the "1030" Fabry-Perot that we have from QC set up. It doesn't look that great, but hopefully works well enough for us to set up the PDH electronics. Planning to work on that tomorrow.
| |||