Purpose & Design goals

The purpose of this system is to enable efficient selective sintering of transparent and reflective materials such as glass and steel. To reach this goal I'm designing a compact, high-efficiency, powerful and affordable gas laser which can operate in the mid-IR region.

State-of-the-Art and Theory

Usually in gas lasers most of the input energy is lost from the system in the form of heat conduction through the walls of the optical resonator. The existing designs of especially HeNe lasers seem to use an approach of dumping ever more energy in the system and removing the heat in a controlled manner to achieve temperature stability and beam power. These designs expose the hot lasing gas to very large surface areas compared to effective beam volume, making them inefficient.

According to the art of building gas lasers, you get very little power increase when you make the laser beam wider. This is because the atoms spend a long continuous time in the active lasing region so they get little opportunity to reset their energy levels, which is needed to ready them for emitting more photons. Using a thin beam and a very long waveguide is the efficient approach, but folding the laser beam in a zig-zag pattern inside the optical resonator through various topological means has not given as much space saving as one might expect. These existing solutions are brilliant in their own right, with very robust geometry often involving convex mirrors, so that when the device undergoes thermal expansion the beam does not attenuate. 

Home-built CO2 lasers often have a beam which fades in and out because of this. To get a stable beam you need to de-couple the mirrors. Desktop HeNe devices used for education tend to use an elaborate and very costly rig of machined parts to do this.

This project will hack high-voltage amplifiers to drive common piezoelectric discs as actuators, and mount the mirrors to the actuators. There are two pairs of parallel mirrors at a diagonal to the other pair, with one pair kept at nλ and another at nλ ± ½λ (The latter is also referred to as 'half-wavelength mirrors' in this document.). This creates a tightly-folded beam path which yields high efficiency and high beam power since the ratio lasing region volume vs. resonator surface area is large. This configuration also reduces device size vs. beam power. The solid medium version of this laser is called total-internal-reflection face-pumped laser (TIR-FPL), where the laser is confined to the medium by angle and index of refraction, similar to how an optical fiber works. In TIR-FPL the parallel surfaces are still a problem because even though the index of refraction lets light perpendicular to the surfaces pass right through and no mirrors are used along that path, some of the light is still reflected. These reflections are called 'parasitic oscillations' in slab lasers. The patent US7505499 assigned to Panasonic Corporation postulates no parallel polished surfaces to address this issue. We of course solve the same problem by keeping the parallel mirrors and detuning them. 

If you put two mechanically coupled end mirrors in parallel against each other, making an 'infinity mirror', then thermal effects will make them sometimes lase and sometimes not as the optical resonator heats up and expands. This happens because the light as they constructively interferes and destructively interfere since this effect is controlled by the distance between the mirrors. Mirrors separated by nλ ± ½λ, meaning a length of any number of full laser light wavelengths with a half-wavelength either added to subtracted to its length, will always destructively interfere and cancel light amplification. Mirrors separated by a number of full wavelengths, eg. for HeNe 632.8 nm x some large number, will always increase light amplification, adding more and more intensity to the laser beam as the process of Light Amplification through Stimulated Emission of Radiation does its thing.

Stimulated Emission is the quantum trickery which is the most central to the laser's operation. Stimulated Emission means that when a photon of a given wavelength passes an atom which has an excited valence electron energy at a level corresponding to that incoming photon's wavelength, an entropic effect occurs so that the incoming photon's direction and phase affect the photon that gets released from the electron's excited energy level. The emitted photon then has, with only minute variation, the same direction and phase as the photon which stimulated this emission. - There is a wide-held belief that electrons always emit photons in a random direction, and Bose-Einstein statistics proves this belief false.

In lasers, photons therefore replicate themselves by passing back and forth between parallel mirrors. By tuning the mirrors out of phase by 180 degrees we can make the photons stop each other from replicating between the mirrors, so that we have hot gas ready to lase but no lasing actually takes place. If we add another pair of mirrors at 0 degrees from each other relative to the wavelength of the laser gas, with the pair at 180 deg  between them at a small angle, they multiply into a very long path along which light amplification takes place.

If thermal effects are uncontrolled then the mirrors shift around from their positions at 180 and 0 degrees, often by several wavelengths, also getting misaligned from their parallel configuration. When the mirrors are actuated with nanometer precision the time they spend misaligned or in the wrong phase is minimal.

The beam can further be folded in layers so that the beam's length is increased by a factor of two. In this advanced configuration the volume of the lasing region can be made almost as large as the non-lasing region, where the gas goes to pick up new energy for another round of emitting photons. Much less heat needs then to be removed from the system, simplifying the structural design, the power supply, the cooling system, and the logistics of the device. The electro-mechanically stabilized waveguide makes for a high-end laser, while the reduced complexity of the supporting systems makes it low-cost.


Currently the prototype only exists as CAD rendering, but most of the major technical issues have a solution already.The central components around which the design revolves are aluminum profiles and piezo benders. 

Among aluminum profiles there are many standard geometries to choose from and many manufacturers supply additional hardware which installs on the profiles to enable e.g. vacuum supply. Usually HeNe lasers are pre-ionized by wrapping aluminum foil around the dielectric of the discharge tube, but this design instead supplies the dielectric as an aluminum oxide slurry which coats the inside of the profile. Since this makes the device more robust to e.g. impacts handling such as international logistics can be cheaper.

The piezo benders are what makes for the actuation of the mirror assemblies. The mirror assemblies are easily attached and made rigid to the aluminum profile because of the standard connection elements that slide into the rails of these profiles. Low-cost piezo benders come in dimensions up up 50mm, which eases the electrical supply requirements by providing more leverage for each volt of electrical potential.

The windows of the waveguide are not installed at the traditional Brewster angle, but instead they are at a right angle. In other lasers this would result in serious losses due to parasitic oscillations, but in this design those oscillations are cancelled as soon as they arise by the half-wavelength mirrors. This greatly simplifies manufacturing and allows for wider tolerances than the rather exact ones of Brewster windows, again driving down cost.

The design and placement of the electrodes which energize the gas only have a placeholder at the present. - T5 tubes with wire wrapped around them. The glass of the T5 tubes should hold up to the thermal requirements since they are designed to contain a much more corrosive mercury plasma. One possible configuration to use for energizing the plasma is to make the two T5 electrodes the same voltage while the aluminum profile is grounded instead of left floating. 

For the power supply a commercially available unit will simply be bought new. The lasmix gas will also be sourced from a commercial supplier.

Electronics & Software

This design can fortunately benefit from a great amount of previous work done in both analog electronics and software. There exists a reasonably mature open source Direct Digital Synthesis software package for the Parallax Propeller microcontroller, which also recently became open hardware. 

Most of the work that remains is the design of a filter and amplifier which can drive the piezo benders at near-DC frequencies, which means that direct-coupling has to be used. Fortunately there exists an operational amplifier which can supply a 140V voltage swing, and it costs about as much as a cup of coffee! Since we need six of these for three plus three channels of actuation, and for the longer CO2 wavelength twice as many actuators may be needed to accommodate for the longer stroke, the price really matters. Suitable IC filters cost about as much, as does the negative voltage supply needed for push-pull operation. The use of a 50mm piezo and push-pull operation allows for maximum utilization of the op amp. See the component list for details.

The Propeller MCU can supply many more channels than the max of 12 anticipated. The ASC+ dev board which the prototype will use also comes with an 8-channel 12 bit DAC, but with proper filter design higher accuracy might be obtainable from the MCU's remaining IO pins. This return path for data is very important since the MCU might need to inform an application layer about the beam power which essentially drives the feed-rate of a CNC machine. In a fault condition the MCU should know about the fault, be able to e.g. shut down power and get a message to the operator about the fact.

The design goal of the electronics package is a stand-alone Arduino Shield compatible PCB with six channels which turn Pulse Width Modulation or Pulse Density Modulation into a large-amplitude voltage, capable of driving e.g. a same-supply, diode voltage-drop compensated Class B Power FET amplifier at ±60VAC. The shield should also be ready to provide six channel of feedback to the MCU it is connected to so that e.g. the power factor may be observed on the high-voltage side.

This package is intended to be useful for e.g. variable-frequency supply for AC motors. Two 110V three-phase AC motors could be driven with the addition of a Propeller ASC+, this shield, and appropriate FETs. This slight generalization should allow for an economy of scale, capable of again, bringing costs down.

Finally and most critically a method to calibrate the mirrors is needed. One approach is to use input beams from inexpensive 405nm diode lasers into the optical resonator and detect deviations with an interferometer. The error can be further amplified with the help of the parallel mirrors, and this would result in even more stable beam power. Detectors in the visual range are common and inexpensive too.

An alternative approach is to only read beam power and scan the mirrors until optimal power is achieved. For certain applications where short interruptions in beam power can be tolerated, this could afford additional cost-saving.

A hybrid approach would cut down on the time of the interruptions to allow for many more applications. The duration of the interrupts is related to the resonant frequency of the actuator and mirror pair and complicated by the scan pattern. A ball-park estimate for a full calibration aided by one laser diode is that it would take less than a second provided that the weight of the mirrors is carefully considered.