Homebrew CO2 Laser Design and Construction Notes.




I built my laser in summer of 1980, while I was still a student at Georgia Tech.  First I assembled the tube itself, and got it working using a rough-and-ready neon-sign transformer as a power supply, some months before I managed to design and build a DC power supply.


One of the requirements for graduation was a course called Project Laboratory, or EE 4430.  It was expected that all EE candidates would design and build something unique for this course.


I approached Dr. W. R. Callen, an EE professor who had taught a course on lasers at Tech (and who was co-author, with another EE professor and a professor from the Physics department, of a book on the subject). I asked him if he would sponsor my project.


            “What are you planning to build,” he asked.


            “A carbon dioxide laser,” I replied…


Dr. Callen just shook his head.  I knew why; no undergraduate at Tech had ever succeeded in scratch-building a laser before.    Well, I quickly admitted that the oscillator was already up and running!  Most of my remaining work only involved the DC power supply, anyway.


One of the proudest and most exciting events of my life was demonstrating the finished product in Dr. Callen’s and Dr. O’Shea’s Laser Physics class!


*          *          *


The Laser Tube


My laser uses a blown-glass tube, which was made for me by Tech graduate student Wayne Penn (who also suggested the clever “bellows-less” mirror-mount design).  He used a glass lathe to fabricate the tube in about an hour.


Almost any large city will have a scientific glass-blower who can (in between dope-bongs, which he will be making when scientific business is slow!) make a similar part for relatively low cost.


Another option is to adapt something from off the shelf, such as a chemist’s Liebig condenser.  These are not expensive—I have seen suitable units, half a meter long, on the Web at prices ranging from $50 to $60.


The Liebig condenser is especially well-suited for a laser intended to be run on direct current.  Cut off the angled tip on the condenser’s “output” end and mount one mirror cell there, just as I have done in my custom-made laser tube.  This mirror cell then serves as the anode.  Mount the other mirror cell at the “input” end of the condenser, which flares out.  The larger size of this end of the tube will allow you to fit a cathode made of brass tubing, as I have done in my laser, having a diameter a bit larger than the diameter of the condenser’s bore.  


Install gas fittings into each of the mirror cells.  The resulting half-meter tube should put out about 10-15 watts!


A final option is not to use blown glass in the tube at all.  You can “build up” a tube using individual pieces of straight glass tubing, making them concentric (as in the water jacket) using turned doughnut-shaped insulating spacers. The gas fittings may be built into the metal mirror cells. Joints may be sealed with epoxy resin, RTV silicone, or O-rings.


My personal preference is for the blown-glass tube. It’s likely to have significantly fewer problems with leaks and power-supply arcing.


Output power in the slow-flow CO2 laser is essentially proportional to the length of the electrical discharge, for around 0.30 to 0.50 watts per centimeter.  It is pretty much independent of the diameter of the discharge bore.


The size of the bore is often chosen so that the laser will oscillate on only the fundamental transverse mode (TEM00).  This may be accomplished by dimensioning the bore such that the tube’s Fresnel number (the square of the bore radius divided by the product of the distance between the mirrors and the laser wavelength) is around 0.80. 


The Mirror Cells


In my laser mirror mount design (see the EE 4430 Project Report), I used separate Buna-N O-rings to seal the mirror and to allow angular adjustment; this approach allowed me to remove a mirror without disturbing the optical alignment.  This capability is not essential, and the construction of the mirror cell may be considerably simplified by using the same O-ring both to seal the mirror and to adjust it.


I made my mirror mounts on my own lathe.  To make the tube mounts, I borrowed time on a small milling machine at school.  If you do not own any machine tools, you can certainly have any metal-work done at a local shop, but you should really consider the possibility of assembling a small shop of your own—miniature lathes and milling machines, perfectly suited for making small parts such as these, may be purchased for just a few hundred dollars.


Some experimenters prefer bellows-sealed mirror cells. A perennial problem is finding that metal bellows…


There is an air control thermostat in every air-cooled Volkswagen engine.  It employs some volatile fluid, sealed inside a length of copper-plated steel bellows, to open and close a vent as the cooling air temperature rises and falls.  The bellows is easily de-soldered from the rest of the mechanism, and it’s perfect for use in small laser mirror cells.  It is long enough that it may be cut in two, providing a bellows for each end of the tube.   “Replacement” VW thermostats may be bought for a few dollars.


The Optics


One hobbyist (see Levatter) made his own optics!  He ground and polished them from commercially-available glass blanks, and then coated them with copper using the sputtering technique in a home-made vacuum chamber.  His was an “aperture-coupled” design in which both mirrors were totally-reflecting; the output was extracted through a hole in one of the mirrors.  To seal the hole, he made an IR-transparent window from a slab of crystalline salt.


At present, it is probably easier to use commercially-prepared optics in the laser, and now they are manufactured in sufficiently large quantities that costs have become relatively low--it should be possible to buy a complete set for a couple of hundred dollars.


In my laser, I used a gold-coated glass total reflector and a broken fragment of a multi-layer dielectric-coated (MLDC) germanium partial reflector.  Current industrial practice calls for a protected gold- or silver-coated silicon total reflector and an MLDC zinc selenide partial reflector.  Germanium partials are probably OK, but that material exhibits significantly higher absorption and does not transmit visible light.


The partial reflector or “output coupler” is typically flat.  It has an MLDC partial-reflector coating on the side facing into the laser cavity, and an MLDC anti-reflection coating on the other side.  Zinc selenide has a high refractive index, so without the AR coating there would be unacceptably high reflection losses at the output interface.


Output coupler reflectivities range from about 80 to 95 percent in CO2 lasers less than a few meters long.  Mine has a reflectivity of 92 percent, which works well for my 1.15 meter unit.  In general, it is preferable to under-couple than to over-couple the laser output, because the general function of output power versus output-coupler reflectivity is decidedly “lop-sided” in shape, with a much steeper slope on the over-coupled side of the curve (see Fahlen or Garrett).


It is actually possible to calculate the optimum output-coupler reflectivity (see Silfvast for the equation), but you need to know the small-signal gain of the laser medium and the absorption and scattering losses in the cavity, factors that may be impossible to quantify without measurements.  Should you have two output couplers (each having a different value of reflectivity) available for your laser, you may measure the output power using each of them and then set up two equations in two unknowns, which will yield reasonably accurate values for the gain and loss characteristics.  These may then be used to perform a final calculation to obtain the optimum reflectivity.


The total reflector should be concave (spherical), with a radius of curvature at least somewhat longer than the cavity length (that is, the distance between the mirrors).  My mirror has a radius of 2 meters, not quite twice the length of the cavity.


Aligning the optics is easy.  Aim a visible laser beam (a HeNe beam is excellent), from a few feet away, down the bore of your tube from the output-coupler end.  Rig up a paper screen at the visible laser, with a little hole in it, such that the alignment beam passes through the center of the hole on its way to your tube.


Next, adjust the position of the laser being aligned so that the visible beam passes right down its axis. I use stacks of index-cards under each end of the laser mounting extrusion. It is much easier to do this than to try to position the alignment beam itself!


After aligning the bore, install the total reflector and adjust its mounting so that its reflection of the alignment beam falls right back onto the alignment beam itself; that is, center the reflection on (or in) the hole in the paper screen.  The total reflector will then be aligned with the bore.


Finally, install the output coupler, ensuring that the reflecting side faces into the cavity, and adjust its mounting so that the visible back-reflection from this mirror falls on the hole in the screen, too.


A “short” (< 2m) CO2 laser aligned in this way will almost certainly be in good enough alignment to “lase” as soon as the discharge is struck.  Then, once the laser is oscillating, small “tweaks” of the mirrors will allow adjustment for optimum power and mode quality.


The Gas-Handling System


Bad news:  the flowing-gas CO2 laser requires a vacuum pump to achieve the low pressures (10-30 torr or so) required for its operation.  Do not waste your time with refrigerator compressors or similar kluges; they’re nothing but trouble-- when I tried a rotary refrigerator compressor, all I got for my efforts was a lab full of acrid smoke!


Good news:  suitable vacuum pumps are showing up with increasing frequency at on-line auctions, and they aren’t even very expensive.  The condition of any prospective unit is probably not critical, either; if a belt-drive high-vacuum pump will run at all, it will almost certainly be able to pump down to a few torr even if it’s badly contaminated or has been abused. 


When preparing a surplus pump, clean the outside of the unit with mineral spirits and check it for leaks.  Drain the oil and replace it with new vacuum pump oil (available from suppliers such as Duniway Stockroom for about $10 a gallon); if the spent oil looks or smells really contaminated, it’s probably a good idea to run for a while with the first refill, drain the pump,  and then refill it again.


The easiest way to obtain the laser gas itself is to buy or lease a cylinder of pre-blended laser mixture from a local supplier of industrial gases.  In my opinion, building your own mixing manifold and then preparing your own mix from gases of whatever source is another waste of time and effort, unless (of course) you are specifically interested in experimenting with the composition of the gas mixture.


I had no trouble obtaining my gas mixture from any of several local vendors in Atlanta and in Denver.  A common standard mixture is 4.5 percent CO2, 13.5 percent N2, and 82 percent He, which works well in my laser even though it may not be optimum (Penn suggests using a mixture of 14:14:72 in narrow-bore CO2 lasers).  Never attempt to use a mixture containing carbon monoxide!


You will need a pressure regulator compatible with the fitting on your gas cylinder.  A single-stage regulator is adequate and will be less expensive than a two-stage unit.


A means of measuring the pressure in the tube is convenient but not essential.  Resist the temptation to use a mercury manometer, because the toxic metal vapor will contaminate everything in your system, and any sudden pressure change in the manometer tube could break it and scatter its contents. A much safer manometer may be made with vacuum oil of the same type used in the mechanical pump (see Levatter), but all-electronic gauges are preferable.  When selecting an electronic gauge, be aware that many such devices (thermocouple and Pirani gauges, for example) are designed and calibrated for use with standard “air” and that the presence of helium may cause erroneous readings.


The gas connections are simple.  You connect the vacuum pump to one end of the laser tube, and you connect the gas regulator to the other end through a suitable leak valve—you may use a hardware-store needle valve, or select a real gas-metering valve from the Swagelok catalogue.  Adjust the regulator for an output pressure of about two psig, evacuate the laser to the limit of the vacuum pump, and then admit gas through the leak valve until the sound of the pump changes.  Turn on the power supply, and then adjust the gas flow for the highest pressure at which you can maintain a stable discharge—after a little practice, you will acquire a feel for this sequence of operations.


If your vacuum pump is large, it might be a good idea to place a “throttle valve” on the pump side in order to limit the gas flow.  While large gas flow rates can be advantageous for laser power output, they consume the gas mixture more rapidly.


Unless you’ve rigged up a special output vent, the vacuum pump simply expels the spent gas mixture into the ambient air.  This is generally no problem in normal use because the volumes of gas are small, but you should still ensure that the area is well-ventilated, especially if the laser is to be run for a long period of time.


After shutting down the laser, open your gas valve and pressurize the tube to atmospheric pressure or above (as indicated by a bubbling sound in the non-running vacuum pump).  If you leave a partial vacuum in the tube, oil or oil vapor can be drawn into it and cause damage.


The pieces of tubing that connect the laser to the pump and to the gas valve should be heavy-walled enough that they do not collapse when pumped down.  Also, they should be significantly longer than the distance between the laser’s electrodes, or you risk having the electrical discharge go down one of them rather than down the bore of the laser!  The mirror mounts should be regarded as electrically “hot” even if they are not directly connected to the electrodes, because a discharge can form between them and the electrodes should there develop a suitable external current path (such as your body!).


The Cooling System


The CO2 laser requires active cooling to keep the temperature of the discharge tube below around 30 degrees C or so.  It is possible to build an air-cooled tube using fans and large heat sinks, but since the heat sink can have a thermal resistance of no greater than a few hundredths of a degree C per watt, it is usually more practical to use liquid cooling.  In my laser I use ordinary tap water, with a flow rate of about 0.75 liters per minute.


Generally you want the laser tube as cool as possible.  Indeed, laser cooling may be a more fertile field for amateur experimentation than the composition of the gas mixture; I have read that cooling the laser to dry-ice temperatures will double the output power.


Remember that when cooling to temperatures below the dew point, there will be condensation of moisture on the cooling lines and on the outside of the cooling jacket. This could cause corrosion and electrical insulation problems.


The Power Supply


Gas lasers require a high-voltage power supply.  For a CO2 laser, the voltage will be around 10 kV to 15 kV per meter of discharge, at currents ranging from about 10 to 40 mA.


When first attempting to operate the laser, try using a neon-sign transformer as a power supply, assuming that your tube design permits the use of alternating current (symmetrical electrodes).  The NST has many advantages—it is inexpensive, easy to obtain, and automatically limits its output current when operating into “negative resistance” loads such as a gas laser discharge.  A variable transformer (Variac® or equivalent) should be used on the primary side of the NST to permit more precise control of the tube current.


A 15 kV, 30 mA NST works well for lasers less than a meter in length. 


Operating a CO2 laser at maximum efficiency requires a direct-current discharge.  You may rectify the output of any high-voltage transformer to obtain this direct current.  Solid-state rectifiers may be used to accomplish this, and they should be rated well in excess of the highest expected voltages in the circuit; for example, when bridge-rectifying the output of a 15 kV NST, the individual rectifiers should have a specified PIV of at least 22 kV and preferably a bit higher!  The minimum average current rating should be 100 mA or so.


Smaller individual rectifier diodes, such as the popular 1N4007 (PIV 1 kV, current rating 1 A), may be connected in series to obtain “stacks” having a higher PIV rating.  When doing this, it is necessary to equalize the voltages across the individual devices using parallel resistors and capacitors.  In my view, the complexity of this arrangement (for the 15 kV NST, we would need 88 1N4007s, 88 high-voltage resistors, and 88 capacitors!) renders it totally impractical—I suggest that you obtain suitable pre-assembled rectifier stacks instead.  Their higher cost will pay back at once in simplicity, reliability, and reduced effort.


Filtering the DC output is not absolutely necessary, but it’s a good idea when applying active current regulation, or if you want a steady power output with minimal modulation.


If you use a conventional high-voltage transformer in your power supply rather than an NST (as I did in my first power supply), and you do not have an electronic current regulator, you will require a ballast resistor in series with the laser in order to limit current.  Its resistance should be such that it drops from 20% to 50% of the total power supply voltage when the laser is operating (I used 100 K ohms).  The larger the resistor, the more stable the discharge will be, but much of the power supply’s output power is wasted.


You should rate the ballast resistor so that it can accept the full output of your power supply without damage; for a 15 kV, 35 mA supply this means that the resistor must be capable of dissipating over 500 W (!), at least for short periods.


Active current regulation, which I have mentioned previously, works far better than simple ballast resistance.  Any experimenter going to the trouble of building his DC power supply from scratch should consider using this powerful technique (see Posakony).


High voltage power supplies require special care in design and construction in order to make them safe and reliable.  Be sure to leave adequate clearance between components to prevent arcing (I used 2 inches per10 kV), and to insulate internal nodes wherever possible.  Enclose the entire unit in a grounded metal cabinet to prevent accidental contact.  It’s always a good idea to use interlock switches to disconnect power when covers are removed or doors are opened, and to provide bleeder resistors across any filter capacitors to dissipate their charge when the unit is turned off.  As a minimum, you should have an insulated “shorting stick” you can use for discharging the capacitors before you work on the unit.


The leads from the power supply to the laser should be insulated for at least twice the highest expected voltage.  Belden CRT cable and automotive ignition cable (be careful to obtain the non-resistive type) are satisfactory.




No discussion of CO2 lasers is complete without addressing the many hazards…


First, the power supply is lethal.  Insulate everything as well as you can!  Make protective covers for the mirror mounts and any other components that may be at high potential.  If the design of your power supply is such that one terminal can be grounded, connect that terminal to the electrode or mirror mount you’ll spent most of your time working around (on my laser, it’s the output mirror).


Again when analyzing potential high-voltage hazards, be sure to consider unintentional leakage paths, such as down the gas tubing or between electrodes and the mirror mounts.


The laser’s output is more destructive than you might imagine.  While the wavelength is such that it is not specifically considered a retinal hazard, the sheer power of the CO2 laser beam is more than sufficient to broil your eyeball right in its socket!  When working with exposed beams, you should always use protective eyewear. Fortunately, this is easy to arrange; inexpensive wrap-around safety glasses are sufficient.


Special “laser” goggles are not required when working with CO2 lasers, because all of the commonly-available plastic and glass materials are completely opaque to 10.6 µm radiation.  That said, Lexan® and similar polycarbonate materials are preferable for use in eyewear and safety screens, because they are mechanically tough and resist CO2 laser burning well (unlike acrylics, which burn fiercely).


If the beam strikes your skin, you will instantly receive a painful burn.  As a minimum you may expect blistering, and you will likely be charred!  Should the beam strike any flammable material, there will be immediate ignition.


The beam of a CO2 laser is absolutely invisible, so there is a premium on knowing exactly where it’s going.  There must also be a suitable backstop for it.  Perhaps the best such “beam dump” is a hefty metal box whose insides are blackened with soot and which has a small hole for admission of the beam, but a fire-brick is more practical.  When using a fire-brick there may be significant scattering from the point where the beam strikes it, and proper eye protection is essential.


Post a laser warning sign such as the one I have included on this Web site!


Finally, be sure to secure the gas cylinder so that it cannot fall or be knocked over.  Should the valve be broken off, the pressurized cylinder could become a lethal projectile.




Here are some laser references I have found particularly worthwhile.


Arnold L. Bloom, Gas Lasers.  New York:  Wiley (1968).


C. G. B. Garrett, Gas Lasers.  New York:  McGraw-Hill (1967).


These early books, identically titled, cover pretty much the same ground through most of their text.  Then, Bloom becomes more applications-oriented while Garrett becomes more design-oriented.


Theodore S. Fahlen, “CO2 Laser Design Procedure,” Applied Optics Vol. 12, No. 10, pp. 2381-2390 (October 1973).


Because lasers are feedback systems, many of their design parameters strongly interact with one another, and arriving at an optimum design requires a really thorough understanding of just how they interact… Fahlen’s paper describes an approach to CO2 laser design in which a series of nomographs show us these relationships graphically!  It may be used, like a spreadsheet, to see how changing any one parameter affects each of the others.


Levatter, Jeffrey, and C. L. Stong, “The Amateur Scientist:  A Carbon Dioxide Laser is Constructed by a High School Student in California,” Scientific American  Vol. 225, No. 3, pp. 218-224 (September 1971).


Jeffrey Levatter’s homebrew CO2 laser project is the one that first inspired me to build a laser of my own.  Levatter did everything the hard way (he made his own mirrors, mixed his own gases, and used a refrigerator compressor as a vacuum pump), but he got exciting results.


This column was later reprinted in a Scientific American book [Jearl Walker, Light and Its Uses.  San Francisco:  W. H. Freeman (1980)].  The book is long out of print, and it will probably be easier to locate the original magazine.


Donald C. O’Shea, W. Russell Callen, and William T. Rhodes, An Introduction to Lasers and Their Applications.  Reading:  Addison-Wesley (1977).


This is my professors’ textbook.  It has an excellent, easily understood discussion of laser theory; presented with an enjoyable style and with considerable wit (for instance, they note that since most laser devices are actually oscillators rather than simple amplifiers, the acronym should more properly be derived from “Light Oscillation by Stimulated Emission of Radiation”—except that nobody wants to be associated with a LOSER!).  Unfortunately, the book has never been revised, and its chapters on specific laser types and on various laser applications are dated.


Michael J. Posakony, “A High Voltage Current Regulator for Laser Gas Discharge Tubes,” The Review of Scientific Instruments Vol. 43, No. 2, pp. 270-273 (February 1972).


Posakony’s paper presents a practical vacuum-tube based electronic current regulator for use in the laser power supply.  It permits more precise control of the discharge current, with much better stability and efficiency, than a ballast resistor.


I based the design of my own dual-discharge current regulator (see the Current Regulator Page) upon the ideas discussed in this paper.


William T. Silfvast, Laser Fundamentals.  Cambridge:  Cambridge University Press (1996).


In my opinion, Dr. Silfvast’s book is the finest text on basic laser theory available at this time.  It’s up-to-date, complete, and loaded with examples and practical information.  Its mathematics will probably be daunting for anyone unfamiliar with calculus.


I have just learned that the book is now available in a revised second edition (2004)!