Printable View
Does anybody have any experience building a frequency doubling system that would accomplish this? Or is it beyond the scope of the hobbyist..
10W 976nm pump diodes aren't hard to find, and .5-1W of 488 cyan would be very nice in a projector imho :)
http://www.alphalas.com/products/las...-crystals.html
Attachment 44933
It will be nearly impossible to frequency double a multimode diode like that.
The crux of nonlinear optics is that they require extremely high intensities to get reasonable conversion efficiencies. For example, the intracavity power of a normal 532nm laser pointer can easily be 100w, with a beam waist of only a few tens of micrometers in the doubling crystal.
In general, diode lasers are not capable of generating high enough intensities, so directly doubles lasers use some sort of cavity enhancement to make it work. The Coherent D3 (http://www.repairfaq.org/sam/laserdio.htm#diocod3) uses an external fabery-perot cavity locked to the laser diode, creating a sort of 'fake laser' which has an intracavity power high enough for the nonlinear process work. The novalux diodes use an extended cavity diode, so that the crystal can be put inside the diode laser cavity, which they were able to get working by using a fancy large mode area VSCEL which can handle very high intracavity powers (http://www.repairfaq.org/sam/laserdio.htm#diopr488). The advancement of PP-KTP waveguides (similar to the crystal you linked) have drastically reduced the necessary power to efficiently generate second harmonic, such that it is possible to start with a single mode diode of with powers as low as 50mw (or as high as you can find a suitable diode for) and couple into the waveguide and get reasonable conversion efficiency, such as is achieved with the melles griot BDD series (https://marketplace.idexop.com/store...ils?pvId=35354 and http://laserpointerforums.com/f40/48...tem-85658.html for a picture of the insides). Of course with all of these techniques there are extremely stringent demands on the frequency purity of the diode (the D3 requires a truly stable SLM diode with many meter coherent length, the BDD can theoretically work with any diode which you can couple into a single mode waveguide, although practical limitations mean that you need a frequency stabilized SLM diode as well), the novalux is the least sensitive but still requires temperature stabilization down to a few hundredths of a degree kelvin for stable operation.
Wow thanks for the very detailed and concise explanation Krazer. I Guess we'll be waiting for the 488nm DI diode then. Thanks :)
and Cheers !
Pardon my ignorance, but if it is so hard, why are the 532 DPSS modules a dime a dozen from china now? Seriously, you can get 532 laser pointers for less than $10 on eBay. I thought that was also a frequency doubling system?
Diode beams are typically broad band and even single mode and focused produce relatively low intensities. The 532 is a Nd crystal laser that is PUMPED by a diode. The fundamental bandwidth (wavelength spread) is very narrow and when the nonlinear crystal is then inserted into the cavity, between the cavity mirrors, the high intensities and narrow bandwidth make SHG efficient. I agree with krazer that the periodically pooled non linear crystals are pretty amazing. I switched over from bulk KTP to PP lithium niobiate and a crystal 1/3 as long produces the same conversion efficiency and avoids the low damage threshold of the KTP.
I'm waiting for the gap between 532 and 638 to be filled. This is a huge spread and if addressed the whole 532 vs 520 debate becomes moot.
Didn't you build a high power yellow laser Eric? I remember a video. Was that crystal or dye? I don't remember. I do remember it was pretty amazing ..
Yes they are freq doubled. The Chinese have 532 dpss down to a mass production science. It's a lot harder to build a 532 from scratch at home (I've tried) .Quote:
tribble wrote:
Pardon my ignorance, but if it is so hard, why are the 532 DPSS modules a dime a dozen from china now? Seriously, you can get 532 laser pointers for less than $10 on eBay. I thought that was also a frequency doubling system?
And with the advent of the d.i. 520nm diodes, the complexity of building the 532 dpss will be non profitable and out-dated so my assumption is that they are now having to take a loss instead of a profit in order to clear excess 532 stock from their shelves.
I built two "yellow" lasers. One is simply a LS pumped dye laser and the other is a Raman shifted vanadate, similar to the DPSS I demonstrated in the video. Yellow-orange is the problem. The lasers I built are simply too complex or expensive for practical projectors. A Pr+ laser pumped with a blue diode would be as simple as a DPSS for up to about 200mw. Losses in the gain crystal produce problematic heating above this level. But a yellow diode (not submerged in LN2) would be the answer. There was a recent thread that discussed the fact that common AlGaAs lasers generate lines and emission all the way down to the green. Could a junction, especially a long one without cavity optics, be induced to lase in the yellow-orange with the appropriate external optics? Someone here on the forum works for DILAS and maybe other experts could address the potential of this set up. Has it been tried? I know that external cavity optics have been used with IR diodes to facilitate wavelength stepping/shifting for extreme brightness fiber launching.
I am curious why we haven't seen more of the 594nm frequency summation DPSS lasers as well as the raman shifted lasers. Are these designs, the raman shifted laser and the frequency sumation laser difficult to produce?
Fundamentally, these are only one step beyond a 532nm DPSS laser. In fact, Nd has a number of weak lines that can be doubled just like 532, but these lines are weak enough that special coatings are needed to allow these lines to lase and it is necessary to inhibit the stronger 1064 line from lasing by reducing the reflectivity of the cavity coating at this line. Ramen shifting can occur in the gain crystal itself and so an additional element is not necessary, but the Ramen process red shifts by converting the energy difference between the fundamental and the shifted wavelength into heat. This increases the thermal problems that already are a significant problem with Nd lasers. An alternative to self Ramen is a separate CVD diamond in the beam path. This material has become so cheap that it is competitive with the gain crystal as far as cost and produces a better beam especially at higher powers. So, this can be doe (I did it), but it is just too complicated to be competitive with a mixed 532 and 638 beam for these wavelengths. Summation or wavelength mixing can be done and is becoming more feasible with periodically pooled non-linear crystals, but the wavelength and temperature stability requirements are difficult to meet and so this is restricted to a laboratory technique.
Thank you for the information, Eric. I would imagine that the thermal requirements would make modulation of the pump diode difficult, if possible at all. I think we'd be left with AOM as a modulation option. It sounds like a more realistic expectation, outside of beam mixing, would be a yellow diode.
I would love to see some video of your Yellow DPSS, if you ever get around to it. Your fiber fed vanadate was a very interesting video to watch not to mention all of your mitsu red videos!
If anyone is interested, I have 4 of the DFB stabilized 976nm diode laser modules used in the Melles 488 lasers in the stash here. They've got the relay protection board attached so I can't see the exact model number, but if someone is interested, I'll open a package and find out.
Are these intra-cavity doubled or extra-cavity doubled? Is the only additional component required a nonlinear crystal such as BBO or is a cavity required as well?
PS. I am now deep into a pulsed dye project. The first video will be in a couple of days and so no time now for a review of the DPSS.:(
Hey Sir P -- I wouldn't mind seeing that vid of the LS pumped yellow dye again .. do you have that link handy? What's the new dye project if I dare to ask ??
Here's the link:
http://www.youtube.com/watch?v=xAYxfsHyVSA
The world record for a dye laser pulse is 400J. I can't find any reference to a higher pulse energy. I recently converted a small medical dye laser (the next video) and am now upgrading it. If successful then I'll upgrade it again and just maybe...
10W of lemon-yellow aaaahhghhhhh .......
Attachment 44941
That looks so much better than red+green mixed imho ..
Attachment 44942
Thanks for the link Sir Planters. Awe inspiring.
Now on to breaking the world record. Set the bar high for yourself much ?? ;)
OPSL. This is 10W @ 577nm.Quote:
Originally Posted by planters
Attachment 44943
the gold line from copper vapor lasers is very beautiful (578.2), but it's useless for anything but beams.
Attachment 44944
Quote:
Originally Posted by Junktronix
Comparing your dye laser to the LS600's I've worked with, I noticed that you removed the BRF from the dye cavity. My understanding is that the BRF narrows the gain region and allows the wavelength to be tuned to some extent. I always wanted to try pulling it to see what happens. Are you getting a broad output spectrum? Is this with the LS standard dye pack or did you mix your own?
Quote:
Originally Posted by planters
Still working on a test harness for these, but the one I removed from the relay board seems to be working. They are Bookham (Oclaro) LC95A76ULR-20R which are rated to make 300mW. I've also contacted Alphalas and am awaiting a quote for a kit of parts including the PP-KTP to build a 488nm laser using these diodes.
Quote:
Originally Posted by Junktronix
Junktronix,
I pulled the BRF because I wanted the gain bandwidth to OPTIMALLY overlap the custom optics I got from CVI. The output coupler has a partial reflectance that is unavoidably broad ie it is about the same at 560nm and 590nm. I let the dye and cavity find the highest gain wavelength. This can be adjusted to some extent by altering the concentration of dye, but when maximum power is the target the optimal concentration is then again limited. You keep adding concentrate (I mix my own) until the power stops increasing. The result ended up right at the 575nm I was looking for (I was a little lucky here), but I would have been happy at 585 and 565. The maximum output for the system at 575nm is 20W.
Juntronix- Does Alphalas make PPKTP waveguides? As far as I know they only make 'bulk' PPKTP crystals, which is not really what you want for doubling extracavity CW pulses.
Planters - that's impressive. Most I ever saw was about 10W @ 635nm. Sadly the dye lasers are gone and the last of the laserscope equipment is going away soon.
Krazer - They apparently offer a kit of parts to make 488nm from the DFB diodes. Still don't have details though.
10 W @ 635 is about where my stock dye laser came in. This was fiber fed from the LS and this decreases the output. Then, the COT has to be added SLOWLY because the Q switched pulse duration is about 300nsec (to my measurements). This is around the cross over where the advantageous triplet quenching of COT is balanced against its fluorescence inhibition. It definitely helps though and will increase the output by about 40%. Also, even if the dye packs include COT (I do not know what is in them), COT slowly deteriorates in air and so after a few weeks you might want to top it off. My lamps were nearly new and this sure doesn't hurt.
Back on topic - I've finally got the right connectors and tested one of the 976nm DFB stabilized diodes. It's working and making rated 300mW at 450mA, well under the 700mA max spec. The spectrum is weird though. Most of the power is concentrated at 976.2nm, but there is some leakage at ~970nm. If I adjust the temperature (there's a TEC in the package), there are a number of points where the leakage goes away. I don't think that's normal and is likely why the diodes ended up here. The peak wavelength does not move at all over temperature which is good. I'll check the other 3 ASAP.
Also, I heard back from Alphalas and the PP-KTP crystal costs 1000 Euro. I was hoping they'd have a kit including the fiber collimator and lenses, but they suggested getting that stuff from Thorlabs. Does anyone have a pointer to a patent or tech paper that would describe the doubling scheme? I'm wondering if this is a simple as fiber -> fiber collimator -> focusing lens -> PP-KTP -> collimating lens -> anamorphic correction. As opposed to some sort of aligned cavity or specially coated mirrors.
Junktronix,
You'll have one in three or four days to inspect, I'm sending you the 488 module.. In most cases Extracavity PPLN needs a actively steered resonant cavity to build up the energy for threshold of doubling. In this case they appear to be steering the diode to match the cavity, which is backwards but should work. I'll toss some papers in the box.
Most European Crystal exotic crystals come out of plants in a single town in the former USSR. They have a network of resellers and its very profitable for the Russians and the Resellers, hence the markup. We'll talk via phone.
There was a company in Canada offering Blue PPLN modules cheap. I'll look them up.
Steve
Well, this is getting interesting!
I don't know if this is helpful but I googled 976/488 SHG and found these pics of Lazeerer's Melles Griot 488:
Attachment 45063
That is pretty awesome. I am guessing the fiber is for beam shaping?
No clue. Maybe between Mixedgas and Junktronix we'll get a clue ..
The pump diode is fiber coupled. SO the laser can be more compact, it eases alignment, and helps with mode scrambling.
Steve
Some interesting things in this thread, directly doubling diode lasers has been around for quite some time (see the second post in this thread for a brief history) but historically it has been extremely expensive due to the tolerances required to make it work. Check out the Coherent D3 laser linked in the aforementioned post for an idea of how tricky it is to make an extracavity doubled diode laser work reliably.
Luckily, technology has improved a lot since the D3 was made. DFB lasers, (http://www.rp-photonics.com/distribu...ck_lasers.html ), have improved in quality and performance, so instead of the 100mw diode that was available back with the D3 was made, now we have 300mw ones which are commercially available, like the ones that Junktronix has found. Just be careful, a lot of 'frequency stabilized' lasers are not actually singlemode lasers (SLM), but rather a multimode laser which has its center frequency stabilized (like http://www.thorlabs.com/newgrouppage...tgroup_id=8042 ) intended for pumping erbium/ytterbium doped lasers, which would be worthless for this application. PS - the fiber coupling eliminates eliminates the need for careful mode matching between the diode and the enhancement cavity, greatly reduces the alignment sensitivity between the coupling into the enhancement cavity (since you just need to couple the fiber output into the cavity now), it removes the thermal issues of having the diode in the same substrate as the cavity, etc.
Another improvement made since the D3 was created was improvements in optical crystals.
First - a brief summary of the limiting factors of second harmonic generation: First, (if you have not already) read about how SHG works at http://en.wikipedia.org/wiki/Second-harmonic_generation we are using ), you can ignore the sections about surface second harmonic generation and the section which consideres depletion for now. Note that the generated second harmonic I(2w,l), goes as a function proportional to the effective second order nonlinearity Deff, the length of the crystal l, and the input intensity I(w), each squared. This means that to improve the conversion efficiency, you need to have the highest possible Deff, the longest possible crystal length, and the highest possible input intensity. Further note, there is a fundamental limitation that even a perfect (singlemode, gaussian profile, see http://www.rp-photonics.com/gaussian_beams.html) laser can only be focused so tightly, limited by diffraction, and that the length over which the beam stays focused (rayleigh range) is inversly prorptional to the focused spot size. Because of this, simply making the crystal longer does not help you with the conversion efficiency, because you will need to make the beam larger to keep it focused through the crystal and everything ends up canceling out (at least to first order, for a well optimized initial design). In fact the issue is actually even worse than that, because of other parasitic effects associated with second harmonic process (will touch on it later). So, the main factors in determining the conversion efficiency are the input beam intensity, and the effective second order nonlinearity, and doubling either of them gives you a 4x (squared) increase in the second harmonic conversion efficiency.
The D3 uses a LiNbO3 crystal (or at least that is what we think it uses) which has a pretty high nonlinear optical constant of about 1pm/v, but PPKTP has a value about 10 times higher. Furthermore, the LiNbO3 uses birefringent phase matching (http://www.rp-photonics.com/birefrin...hing.html?s=ak), which is limited by spatial walkoof (http://www.rp-photonics.com/spatial_walk_off.html?s=ak), which further limits how long of a crystal you can use, where as the PPKTP uses quasi-phase-matching (QPM, http://www.rp-photonics.com/quasi_ph...hing.html?s=ak) so that the beams propagate directly along the crystal axis, and thus has no spatial walkoff.
So, right off the bat, by switching to PPKTP over LiNbO3 you get about a 100x better performance than a D3.
But, you still need an enhancement cavity. Why? Well, lets go back to to the second harmonic conversion efficiency calculations:
Assume, we have:
300mW input power
1cm long crystal (this is an arbitrary length, but will give us a starting point), made of PPKTP (Deff = 10, n1~=n2~=n3~=1.8)
If we want a rayleigh range of 1cm, Zr=pi*r0^2/lambda = 1cm, lambda = 860nm -> r0 = 52um, so the radius of the beam is 52um. This is a pretty rough approximation, but should be fairly close to the optimum beam size.
So, the intensity is simply the power divided by the area, P=300mw/(pi*r0^2) = 3500w/m^2
Now, going back to the second harmonic generation formulai, we know the second harmonic conversion efficiency, I(2w,l)/I(w,0) = (2*w^2*Deff^2*l^2 )/(n^3*c^3*e0)*I(w) for perfect phase matching (deltaN = 0), which works out to about 0.0000003 (here is a link to google calculator for the calculation ( https://www.google.com/webhp?sourcei...watts%2Fm%5E2) )
So - no matter how hard you try to extracavity double a diode with a free space beam you are not going to get anywhere. This leaves you 3 options:
1. Put the second harmonic generator crystal inside the laser cavity. This helps you 2 ways, first the intracavity power is much higher than the output power (for a set of cavity mirrors with a modest 99.9% reflectivity, you get a Q=~1000, and get a 1000x boost in your intracavity power), boosting your conversion efficiency up to 0.0002, and second the required conversion efficiency is reduced by the intracavity power, since you are starting with a 1000x stronger beam, so that .0002% conversion efficiency is corresponds to a 20% 'effective' conversion efficiency compared to your extracavity power. Looks a lot better now, huh? This is basically how the Novalux directly doubled diode lasers work.
2. Put your harmonic generator in an external cavity. This has more or less the same effect as putting it in the laser cavity, but allows you to get away from the inherent issues associated with intracavity doubling (the fact that half of your power goes out the wrong end of the cavity, tricky cavity design needed to keep the generated second harmonic out of the laser medium, getting a good cavity Q with the gain medium in the cavity, etc). This is what the D3 did, the only downside is that you need to lock the length of the cavity to the frequency of the laser, and this requires sub-nanometer locking precision on the cavity length (ex, for a cavity with an enhancement factor Q=1000, you need roughly 860nm/1000 = 0.86nm precision on the length of the cavity to keep it running!), not to mention the engineering nightmare of keeping the whole thing stable under operation.
3. Use a waveguide based second harmonic generator (ex, https://physik.uni-paderborn.de/ag/a...cation-in-ktp/ http://www.advr-inc.com/waveguide.html etc). You can get around the limitation on beam size vs crystal length by adding a waveguide in the nonlinear medium, which serves to guide the light down the length of the waveguide. Using this technique, it is possible to keep the spot size at roughly 1um radius over an arbitrarily long crystal (up to a few cm long, determined by practical issues), giving you that last factor of 1000 boost in intensity needed to get the conversion up to reasonable levels. This is how the melles griot BDD series (linked in the second post, and a few posts up) works, they use a roughly 2cm long crystal with a few um wide waveguide in the crystal, which works out to just about 100% theoritical conversion efficiency--no cavity needed. Also, the optics are dead simple for this design, you simply take your fiber coupled pump, and use a microlens (or lensed fiber) to couple the output of the laser into the waveguide, and then take the single mode output of the waveguide and collimate it like any other fiber coupled device.
Not quite as simple as a normal dpss laser :rolleyes:
.. damn .. my head just exploded ..
--updated--/ edit -
So, it looks like option #3 is the way to go .. How about not building the whole thing from the ground up/ scratch. If a person was so inclined, could he not find bits and pieces to build one of these without having to fab the whole thing? Also, we are talking about a 300mw pump diode? What would the output be? Not much power for show use I would imagine ..Quote:
Originally Posted by krazer
Thanks for the detailed explanation - saved me a lot of time trying to understand all this. Alphalas says their PP-KTP is bulk crystal rather than waveguide type. They cautioned me that coupling the pump laser into the waveguide is non-trivial. I suppose it's the same as coupling a beam into a single-mode fiber which requires a precision 5-axis mechanism. Does anyone know a source for waveguide PP-KTP?
I got the laser head from Steve and a matching controller from eBay and they are on the bench making 6-7mW at the diode max current of 650mA. The head has <900h so I'm not sure why it's not producing it's rated 20mW. Might have to try powering the head with my own drivers and see if tuning the temperature makes it any happier.
So are you actually seeing some 488nm being produced from your setup right now Kevin?
Yes coupling into the waveguide takes some doing, it is pretty similar to coupling into a single mode fiber. If the waveguide is designed well the mode should be more or less round, and with a simple lensed fiber or aspherical collimator/focusing in a precision 5 axis mount will do the trick. The mode matching between the fiber and the mode of the waveguide takes some careful thought (picking the correct lenses, lest you have to try coupling into the waveguide with 10 different lenses to find the one that works). If you look at the pictures of the MDD from melles griot it looks like they are using a lensed fiber which was aligned using a precision alignment jig then glued in place. They also used a crystal cut at a slight angle, most likely to reduce back reflections which could destabilize the pump laser, and cause etalon effects inside the waveguide. They also added a prism pair on the output to correct for the elliptical/astigmatic beam, this could just be from the angle cut crystal, but more likely it is because the waveguide is not so great at 488nm (since it has to also guide 976nm) and has an elliptical mode as a result, which luckily can be completely removed with a simple prism pair. The good news is that since it is a single mode waveguide you can get a perfect 1.0 m^2 out of the crystal with simple corrective optics (prism pair, etc).
As far as getting ppktp waveguides, http://www.advr-inc.com/waveguide.html and http://www.hcphotonics.com/products2...at=1751&sn=938 have them, and I think there are a few other companies in europe/china that will do it although I have never purchased components from them. Prepare yourself for some sticker shock, a custom waveguide can easily run into the $10k range, more if you want one that comes packaged and fiber coupled.
As far as the power output is concerned, I am not sure what is considered state of that art for these devices in CW operation but 10% conversion seems to be pretty par for the course for these devices, and that is not counting coupling losses (another 5-50% at each end depending on how good your are at it). In theory there is no reason you shouldn't be able to get 100% conversion to the second harmonic (ie, 100mw IR -> 100mw blue) but there are a lot of practical fabrication issues that you have to fight against and I think that if you were able to demonstrate anything more than about 25% your paper would get published in a decent optical journal.