June marked this year’s camping trip with “the boys”. That camping trip is the reason I started the whole Laser Project to begin with, and I decided this was the year to unveil what I've been working on for two years.
Since the last update, I’ve made some good strides on the design and keeping the system stable at ever higher pump powers. However, even with these improvements, output power has been stuck at under five watts for two years. I’ve decided to move the best design to the actual laser and take a break. There are three main reasons for this:
I’ve amassed a huge number of parts and test gear with nowhere to store them. I’ve got parts in cardboard boxes and to use anything I need to sprawl it all out on the rug. It’s messy and I’ve already bought parts more than once because I couldn’t find something.
My simulations are showing much higher powers than the real designs, so something is wrong there. It’s stupid to keep running different simulations when the results don’t match reality.
I realized incremental updates are OK. It’s fine to incrementally improve the performance and continue iterating.
With that, I transferred a promising design into the laser chassis, took it camping, and yes, it worked. Power output is 3W CW, which is 3x what the former design had. The rest of this post covers the design and my plans going forward.
The Design
The design I decided on has three areas each with different beam waists. KTP gets about 100µm, Gain gets about 400µm and there is a section for a saturable absorber that is 50µm. I don’t currently have the saturable absorber installed so this laser runs continuous wave. Why? Because with the absorber installed I saw less power, worse beam quality and greater risk of damaging optics. But the cavity design was very tolerant of thermal lensing and the waists stayed somewhat constant even with higher pump powers, so I kept the design even though I’m not using the QS waist.
Here is a comparison of the old and new designs:
Old and new laser resonators. The old is the top, the new one is the bottom.
The new design has much shallower angles and is actually a lot longer cavity. Most mirrors are 1/2” due to tight angles. I’ve moved to optical rails which gives me a lot more precision over mirror placement. Keeping the gain in the middle of the two z-fold arms helps with lensing: the thermal lens is in the middle of the resonator instead of the end, so the distances are shorter. A nice thing about the way I’ve built this is I now have two base plates for mounting optics and I can swap in new designs easily.
Here is a video of the finished product with the cover off so you can see it in action:
Video of the laser in action with the cover removed.
MATLAB Simulation
I’ve invested heavily in simulating designs. The simulation I have now allows me to define the optics for a cavity and it can predict lensing, power outputs, pulse widths, etc. Here is the definition for the laser shown above:
section1 = Optics.Section;
section1.Left = Optics.Mirror(Inf, 0, .998, false, "M1");
section1.Crystal = Optics.Materials.KTP(.010, "KTP");
section1.Right = Optics.Mirror(.2, 8.5, .998, true, "M2");
section1.SectionLength = .109;
section1.CrystalOffset = .003;
section2 = Optics.Section;
section2.Left = Optics.Mirror(Inf, 8.5, .995, false, "M3");
section2.Crystal = Optics.Materials.Nd_YVO4(.012, .27, "Gain");
section2.Right = Optics.Mirror(Inf, 8.5, .995, false, "M4");
section2.SectionLength = .05;
section2.CrystalOffset = .019;
section3 = Optics.Section;
section3.Left = Optics.Mirror(.1, 8.5, .998, false, "M5");
%section3.Crystal = Optics.Materials.Cr_YAG(90, .0018, "QS");
section3.Right = Optics.Mirror(Inf, 0, .998, false, "M6");
section3.SectionLength = .049;
section3.CrystalOffset = .003;
section3.CrystalOffsetFromRight = true;
assembly = Optics.ParallelSectionAssembly;
assembly.Sections = [ section1, section2, section3 ];
laser = Definitions.Laser;
laser.optics.Add(assembly);
laser.pumps = Definitions.Optics.Experiments.Pumps([50, 50]);This design gives the following analysis summary:
Summmary
--------
Average Output Power : 3.96W
Average Circulating Power : 103.39W
Threshold Power : 392.61mW
SHG Efficiency : 3.98%
Pump 1 Waist : 400.00µm, 400.00µm
Pump 1 Optics : 50.00mm, 50.00mm
Pump 1 Overlap : 94.96%
Pump 2 Waist : 400.00µm, 400.00µm
Pump 2 Optics : 50.00mm, 50.00mm
Pump 2 Overlap : 94.96%
Thermals
--------
Thermal Lensing : 324.77mm, 354.20mm
Thermal Destabilization Current : 46.86A
Pump 1 Power : 11.83W
Pump 2 Power : 10.85W
Gain Waste Heat : 5.44W
Cavity
------
Cavity Length : 555.50mm
Minimum Gain Diameter : Inf
Gain Waist : 398.05µm, 363.35µm
SHG Waist : 81.54µm, 93.97µm
SHG Waist Ratio : 18.88
Q : 6.45e+07
Slope Efficiency : 17.77%
Divergence : 6.83mrad, 7.48mradAnd some pretty graphs to go along with it:
Note, however, that if I enable the Q-switch by uncommenting it, I get really quite good performance:
Summmary
--------
Average Output Power : 12.84W
Average Circulating Power : 52.22W
Threshold Power : 2.47W
Peak Output Power : 665.60W
Pulse Repetition Rate : 129.88kHz
Pulse Width : 91.80ns
Pulse Energy : 98.87µJ
Fleuence : 660.16mJ/cm^2
SHG Efficiency : 32.61%
Pump 1 Waist : 500.00µm, 500.00µm
Pump 1 Optics : 40.00mm, 50.00mm
Pump 1 Overlap : 96.20%
Pump 2 Waist : 500.00µm, 500.00µm
Pump 2 Optics : 40.00mm, 50.00mm
Pump 2 Overlap : 96.20%
Thermals
--------
Thermal Lensing : 315.93mm, 344.19mm
Thermal Destabilization Current : 62.29A
Pump 1 Power : 19.00W
Pump 2 Power : 17.44W
Gain Waste Heat : 8.74W
Cavity
------
Cavity Length : 555.50mm
Minimum Gain Diameter : Inf
Gain Waist : 564.34µm, 410.53µm
Q-Switch Waist : 183.34µm, 130.94µm
Q-Switch Waist Ratio : 9.65
SHG Waist : 57.77µm, 82.52µm
SHG Waist Ratio : 48.60
Q : 8.48e+07
Slope Efficiency : 37.91%
Divergence : 7.31mrad, 10.05mrad
Damage Thresholds
-----------------
M1 : 1.62J/cm^2, 7.50J/cm^2
M2 : 27.16mJ/cm^2, 7.50J/cm^2
M3 : 32.46mJ/cm^2, 7.50J/cm^2
M4 : 44.48mJ/cm^2, 7.50J/cm^2
M5 : 221.86mJ/cm^2, 7.50J/cm^2
M6 : 316.08mJ/cm^2, 7.50J/cm^2
KTP : 1.54J/cm^2, 5.00J/cm^2
Gain : 32.87mJ/cm^2, 10.00J/cm^2
QS : 315.15mJ/cm^2, 5.00J/cm^2I got nowhere near this output power when I tried it, though. Also, while all the intensities are well under the damage thresholds for the optics, I’ve pitted M2, the KTP and the QS crystal. I think this is because I’m not getting nearly the KTP conversion I think I should. One more reason to revisit the simulation.
What’s Next?
There are three major efforts I need to complete next before I can work on the next iteration.
Organize and clean my work area.
Build a bench laser controller.
Revisit the MATLAB simulation.
Cleaning
When I started this project all optics fit into two 14x11x3” plastic bins. Then I bought more optics, more test equipment and more custom parts. My “work area” has spilled out into an array of cardboard boxes and piles of parts. I’ve already had problems where I re-purchased parts I couldn’t find, only to find unopened packages later. Here is the mess that was my work area, and this doesn’t include the five stacked boxes of parts in the garage:
What a mess!
I’ve since replaced the folding conference table and plastic bins with an official workbench:
Laser Controller
My test environment for testing new laser designs is a bit of a hot mess. A test environment requires careful temperature regulation of three components and a high current power supply. My test environment consisted of a prototype circuit board, a controller cobbled together on a solderless breadboard and three different lab power supplies. I could only temperature control one element; the other two were just “chilled” with no temperature regulation. If the laser was started cold it needed about five minutes of run time to get to a working temperature, and if I left it alone for too long the gain crystal would get so cold it would ice up. While this was enough to get things working, it is not not a stable enough environment for controlled measurements. I also attribute this poor setup to a cracked Nd:GdVO4 crystal, which really sucked because I had that crystal custom made.
My plan is to build a lab laser controller using the extra circuit boards from my laser. It will have built-in pump diodes so I can directly connect fiber optics to it to power a laser, and it will have two “ports” for TECs and sensors to regulate temperature of two additional components. I can save a lot of money and engineering cost by reusing existing board designs, although the TEC controller board is just the PC board and no parts, so I need to assemble that (if I recall this is about $1000 in parts though - that board is pretty busy).
Another area I may invest in here is to add integrated power measurement. When I measure power for a new design it’s very manual: start the laser, read the power output and write it to a spreadsheet, then increase the drive current and repeat thirty or so times. I could completely automate this.
MATLAB SIMULATION
Laser physics is governed by a series of rate equations: power goes in, there is some amplification and there are losses. In most of the laser texts, these rate equations are solved analytically for a fixed scenario. My current MATLAB simulation cobbles together various solutions and builds out a full simulation. This leads to lots of approximations though. I want to redesign the simulation so individual elements can provide portions of the rate equations and to simulate results from running the bare rate differential equations. That’s what a tool like MATLAB excels at and it will also teach me a ton about laser physics.
So that’s it! Once all that is done I will resume work on improving power.