Professor Mark Csele's Homebuilt Lasers Page

Nd:YAG Solid-State Lasers

The Nd:YAG laser is an optically pumped laser resembling, in many ways, the ruby laser. The rod itself is a special type of glass (Yttrium Aluminum Garnet) doped with Neodymium. It is optically clear although a tint may appear depending on doping concentration - it invariably appears violet when photographed with a flash. The rod is pumped optically by a flashlamp or CW high-pressure krypton lamp. In newer lasers laser diodes are used as the optical pump. Mirrors are high-relfection dielectric type and usually transparent to visible light. Although impossible to grow a rod yourself, rods and cavity optics are available surplus. This makes the YAG laser a good choice for those wanting to enter the field of higher powered lasers.

My laser, an old Control Laser CL-511QT originally rated for 70W of CW output, was acquired surplus. The laser was complete including the rod, head, mirrors, and an optical rail. In its original form the laser was pumped by a 3000W high-pressure CW krypton lamp. The lamp and the rod were parallel to each other in a gold-plated elliptical cavity which coupled the pump light onto the rod. Deionized water flowed around the rod and lamp to cool the laser (yes, it surrounded the electrical terminals of the lamp: pure water is an insulator). A pump and heat exchanger arrangement was used to transfer heat from the DI water to city water. The lamp was powered by a supply which was fed from 230 volt 60 amp single-phase AC. The AC was converted into about 300 volts DC which supplies the lamp. When the lamp is operating normally, about 15 to 20 amps of current flow through it.

Kr Lamp for a YAG
Being a high-pressure lamp, it required a very high voltage (about 20 kV) to start it. High voltage pulses are generated by the circuit in which a capacitor charged to 600V was discharged into the primary of a pulse transformer by using a krytron tube (a trigger device similar to a thyratron). A voltage generated on the secondary of the transformer, which was in series with the main lamp supply, ionizes krypton gas inside the tube which allows the main discharge to begin.

I was given the laser (free :) because it was overheated and damaged. Apparently the cooling water supply was cut off during operation. The intense heat of the lamp melted a portion of the laser head damaging it as well as the lamp and other internal parts. When the melted plastic was cut away using a Dremel tool it was discovered that the rod was still intact! It was decided that rather than rebuild the laser into a bulky unit with a large cooling system (including heat exchanger) and massive power supply for the CW lamp that the unit would be converted into a flashlamp-pumped type.

Nd:YAG Laser
Elements of the laser are shown here. The laser head itself consists of the 50mm long Nd:YAG rod, an elliptical gold-plated cavity to reflect pump light into it, and a high-pressure krypton lamp. Krypton is used as it's output spectra matches the absorption spectra of Nd:YAG quite well. The optical rail is for mounting the head as well as mirrors and optional components such as a Q-switch. The resonator cavity mirrors are dielectric mirrors (one output coupler and one rear mirror). They reflect infrared light (this laser has an output at 1064nm) but transmit visible light quite well allowing a 'targeting' laser such as a HeNe beam to be sent through the laser itself (i.e. through the rod) and be emitted at the same point as the YAG beam will when fired.

Nd:YAG laser in use to cut ceremonial ribbon
A newspaper article detailing the opening of our new Photonics lab at the college in which college president Dan Patterson used this rebuilt YAG laser to cut the ribbon.
Credit: The Tribune, Welland ON, Oct. 3, 2001, Photo by JT Lewis

WARNING: Construction and operation of any laser device is hazardous. Do not attempt to construct or operate a laser without adequate safeguards and safety practices. YAG Lasers involve high voltages, laser radiation and other hazards. The author specifically disclaims any and all liabilities associated with the construction and use of such devices. The design is presented here in the interests of providing information on operational principles only.

TEST #1: Using a Single Xenon Flashlamp

In order to ensure the laser rod and optics were in working order (since the optics for such a laser are extremely expensive to replace) the laser was retrofitted to use a conventional xenon flashlamp. An old Honeywell Strobonar photoflash unit provided both a long flashlamp (50mm long, the same length as the YAG rod), a capacitor which stores about 75J of energy, and a suitable charging and trigger supply.
The xenon lamp was inserted into a quartz tube of 6mm inside diameter. That tube was fixed parallel to the YAG rod (in place of the old Krypton lamp) via two fuse clips which were mounted in the cavity to hold the tube at one foci of the elliptical reflector (i.e. where the original lamp was mounted).

The photo below shows the rod and a xenon lamp inside an elliptical gold-plated cavity. The xenon lamp shown is blackened from experiments in which too much energy (over 250J) was pumped into lamp.

The laser itself was able to operate with the configuration although only at low output power levels. With 75J of energy input to the lamp, a focussed output beam was able to burn tiny holes in paper (accompanied by a wisp of smoke). Pulse energy was measured (using commercial Melles-Griot Energy Meter) at 150mJ with 75J of input energy. Attempts to pump more energy (Greater than 100J) into the lamp have caused lamp failure.

Such a low output power can be explained in two ways. First, by the absorption characteristics of Nd:YAG itself which absorbs light primarily in two bands at 730 to 760 nm and 790 to 820 nm. Kyrpton lamps have high output in these near-IR areas and are almost perfect for the job. Kr flashlamps are, however, quite rare on the surplus market (and quite expensive new) and so Xenon was used. Xenon lamps produce a continuum of white light so that only a small fraction of the light emitted by the xenon lamp is absorbed and used to produce laser action. This tradeoff will result in lower efficiency of the laser (and hence lower output). The second problem, which was not evident until later (read on as this is described later on this page), was that the Nd:YAG rod in this laser was designed for CW use and hence has a _low_ dopant concentration. Rods for use in flashlamp-pumped lasers are usually quite heavily doped while those for use in flashlamp-pumped lasers aren't. This can be seen in the photo as a distinct lack of colour in the rod. The laser does work though with the arrangement described although it appears to have a high lasing threshold requiring about 50J just to lase. (although I now question that ... read on ...)

TEST #2: Using High-Pressure Krypton Lamps

My plans were to use the original krypton lamp as a flashlamp for the laser. After all, Kr is the best choice. Despite the fact that this is a high-pressure lamp it was _hoped_ that it could be used as a flashlamp. The main electrical charge would be supplied by several capacitors totalling 1000J and the trigger pulse will be generated in a similar manner to the original laser. Hence the high-voltage pulse will ionize the lamp which will then fire on energy stored in the capacitors. This quasi-CW operation (a long pulse) would allow high peak powers but low enough average powers that air cooling should be sufficient.

The krypton lamps were found by experiment to be difficult to trigger. External triggering could be accomplished but an incredibly high voltage (enough to generate a 5cm spark !) was required. It was produced by a G.M. High energy ingition coil driven from a 2.0uF capacitor charged to 400V. To alleviate the need for such high voltages, the EG&G Optoelectronics Linear Flashlamps design guide was consulted andpseudo-series triggering was attempted.

The main capacitor was a bank of four 450V, 220uF capacitors in series and the trigger supply was provided by a TV flyback coil driving a high-voltage tripler. The output from the tripler produces a spark over 3 cm long (more than the required 20kV to trigger the the lamp) and was used in place of the trigger transformer shown (it was connected directly to the lamp). The main capacitor was connected to the lamp via six Varo VG-6 6kV high-voltage diodes in series. The tripler has a self-contained diode to prevent the main charge from discharging into it.

Using this arrangement it was found that the lamp has a minimum voltage of 600 volts (at the main capacitor) to fire. Voltages below this would not allow the lamp to fire. When the voltage was increased beyond 1500 volts (approaching 100J of energy) the Varo diodes failed shorting the circuit.

Although the pseudo-series arrangement works, finding a suitable diode has been rather tricky. Current flowing from the main capacitor through the lamp during discharge can be over 150 Amps! Using a standard downrating formula (I2t) this translates into a required diode with a rating of 10A at 30kV! If such a diode does exist, it would likely be prohibitively expensive. Suggestions from one colleague were to use selenium rectifiers but these are difficult to obtain at high-voltage ratings as well.

A promising possibility for firing this lamp was to use series triggering in which current from the main capacitor flows through the secondary of a pulse transformer. To initiate lamp discharge, the pulse transformer is used to induce a large potential across the lamp. Unfortunately this proved unsatisfactory as enough induced voltage was not developed to relaibly trigger the lamp. I settled on standard external triggering with the ignition coil and proceeded to experiment.

The Krypton lamp itself proved unable to take the high currents produced during flash pumping. Normally, such a lamp is designed for a current of 15 to 20 A but during a capacitor discharge the lamp experiences currents at least ten times that level. The thin-glass of the lamp coupled with the already high-pressures inside the lamp lead to rapid lamp failure (The thinness of the walls of the lamp were evident in the shards left over after catastrophic failure).

At 100J of input energy the lamp fired once then failed catastrophically (ie exploded violently!) on the second pulse. Another amateur laserist I talked to had similar results with Krypton lamps. Even with a suitably large inductor in the discharge path (to slow the discharge to several milliseconds in length and limit the maximum current to about 100A) the lamps still fail. In short, these lamps (designed for CW use) are not suitable for high currents and hence cannot be used as flash lamps. They make nice explosions on the workbench (safety glasses are a must around these lamps) but not good flashlamps.

Another "Wile E." moment: Oh well, back to the old drawing board!

TEST #3: Using Multiple Xenon Flashlamps

Having abandoned attempts to use the high-pressure Krypton lamps as a pump source an attempt was made to use multiple xenon flashlamps to pump the YAG rod. A single (inexpensive photographic strobe) lamp can only handle about 75J so multiple lamps would be required. I _could_ have used a single, large,xenon lamps which will handle 300J but these cost about $200 U.S. new so in the interests of 'keeping it cheap' multiple lamps might be the ticket.

The configuration used is to arrange six small xenon flashtubes, each pumped with about 50J of energy. The flashtubes are arranged around (and in close proximity with) the rod and a reflector of polished aluminum tube placed around that to reflect as much light as possible onto the rod.

Xenon lamp pumped Nd:YAG laser
Here is the arrangement of the lamps around the YAG rod. The reflector is removed for clarity. The plexiglas mount on the sides hold the six lamps parallel to the rod in the center. Only two of the six lamps are visible here ... the front lamp was also removed to reveal the rod in the center.

Nd:YAG laser rod
The laser rod itself, in the case of this laser, is mounted in two stainless steel tubes to hold it in the main housing.

Lamp mount and reflector
The reflector is seen here with the lamp mounts. The reflector was fashioned from the casing of an old electrolytic capacitor. The insides were highly polished using a rotary tool with polishing compound on a cloth buff.

Nd:YAG laser system
The entire system showing the lamps around the rod to the right and the bank of photoflash capacitors used to pump the lamp inside the channel which forms the base of the laser. The output coupler mount is seen to the left.

The laser is working although power levels aren't as high as expected. Indeed the focussed beam of the laser burns holes in paper however more was expected - with six times the input energy of the single lamp arrangement considerably more output was expected . I haven't completed experimenting with this setup and will post more results of this latest experiment later.

The Effects of Dopant concentration (and why hindsight is always 20/20) ...

A contact on the internet ( suggested that the low power outputs in multiple-lamp configuration are due to the nature of the laser rod itself. His rod, designed for flashlamp pumping, was distinctly coloured (meaning it was heavily doped with Nd) whereas the rod in my laser was quite lightly doped and has hardly any colour to it at all (When looking through the rod at a white light the most visible colouration is from the antireflective coating on the rod ends, not the dopant in the rod itself). As well, the cavity mirrors are quite high reflectance: the rear mirror being 99.9% R and the front output coupler having about 1% T. YAG has a relatively high gain so it is surprising that the output coupler has such low transmission.

The output power from a YAG laser will rise with pump energy until finally ALL Nd3+ ions have been excited. At that point the rod is saturated with pump energy and no extra amount of input energy will create any extra output. This effect occurs in other types of lasers as well. A rod designed for use with a flashlamp pump will have a heavy concentration of dopant since the peak power of the flashlamp is enormous: orders of magnitude above what any CW lamp can possibly produce. With a CW laser the rod is doped quite lightly so that it reaches saturation quite quickly. When a CW rod is pumped by a flashlamp it becomes saturated quickly and most of the pump power is lost to heat. It would have been a better idea, in the case of the six lamp laser, to sequence the lamps so that you get six closely-spaced pulses. The other possibility for obtaining higher powers from the laser is to use a longer pulse, rather than higher power pulse. The best way to obtain that is using a pulse forming network.

Surplus rods can be judged by their colour. If the rod is almost colourless (In the case of my CW rod most of the colour comes from the AR coating on the rod ends) it is designed for CW use. A rod designed for flashlamp pumping will show a pronounced violet colouring. Note, too, that the dopant concentration can be judged by the colour of the rod when a flash photograph is taken. A rod which appears clear under normal lighting conditions but violet in a flash photo has a relatively low concentration of dopant as opposed to a rod which always appears violet.

TEST #4: Pulse Forming Network With Single Xenon Lamp

In order to experiment with the laser further I intend to configure it once again for single lamp operation. I have two lamps available for experimentation:

#1: 11-1036, 6mm o.d., 45mm between electrodes, 360V max, 90J max
#2: 11-1031, 4mm o.d., 40mm between electrodes, 360V max, 70J max

First, the initial impedance Ko of each lamp must be calculated. From EG&G's literature as well as from Don Klipstein's Web Site the lamp may be characterized as follows:

Ko = 1.28*l/d*(p/450)0.2 where ...

Ko is the lamp impedance in ohm-amperes1/2
l is the distance between electrodes in mm
d is the bore size (i.d.) in mm
p is the fill pressure in torr

According to the above references, most small photoflash type lamps are filled to a pressure of about 80 torr. The first lamp I have available hence has an impedance of Ko=10.19 ohm-A1/2. The second lamp has a Ko=12.08 ohm-A1/2.

Xenon photoflash discharge With no PFN (i.e. using a capacitor alone and no inductor), one would expect a completely underdamped situation however an oscilloscope capture showing current through and voltage across a Xenon lamp employing a 'standard' photoflash capacitor with no external inductance inserted suggests otherwise. In the above test a 350uF/330V photoflash cap was discharged through a photoflash lamp with a 0R33 resistor in series as a shunt to measure current. Clearly the photoflash capacitor has some significant inductance to match it to the flashlamp's relatively high impedance. As expected, the discharge has a pulse width of 1.1mS (Photoflash units are almost always rated at 1/1000th of a second). For further experiments on PFN's, 'regular' electrolytic capacitors with lower self-inductances will be used.

PFN for Nd:YAG laser A PFN itself has the general form to the left. The impedance of this multi-mesh network is simply Zn=(Lt/Ct)1/2. We may now match this network's impedance to that of the lamp. (Refer to the book Solid State Laser Engineering by Koechner, ISBN 3-540-60237-2, for more info on PFN's ... thanks Pyro for the reference!).

If all elements are the same value in this network value then the calculation of L and C values is trivial using the following formulae:

The pulse width is tp = (Lt*Ct)1/2
The total capacitance is Ct = tp/2*Zn
The total inductance is Lt = tp*Zn/2

Of course Energy stored in the circuit is calculated as usual using 0.5*Ct*V2 where V is the voltage the capacitors are charged-to before firing the laser.

For my first test I decided to make a four-mesh PFN. Four capacitors rated at 220uF/450V each are used with 27mH inductors. According to calculations the impedance of this network is 11.08 ohms. This is quite well matched to the smaller xenon lamp which has an impedance of Ko=12.08 ohm-A1/2. At 400 Volts the circuit will store 70.4J, exactly what the lamp is rated for! A match made in ... math. Oscilloscope shots below show the performance of this PFN.

PFN Output
The above oscilloscope shot is an overall view showing the entire discharge. The horizontal scale is 1mS/div. Voltage is scaled at 50V/div and current (on a 1V/div scale) is measured via a 0R1 ohm shunt in series with the xenon tube. The peak current was measured to be 80 A and the pulse width about 4.5 mS. Some ringing does occur during the start of the discharge as detailed below:
PFN Output, Details of Ringing
The ringing which occurs on the voltage waveform during the first 25uS of the discharge is detailed here (The horizontal scale is now 50uS/div). As soon as the lamp has fully ignited (i.e. current begins to rise) the ringing disappears. To protect the lamp, a reverse-biased diode should be placed across it according to EG&G notes.

I will try the PFN in the laser soon (configured for a single lamp) and will, as usual, post results here.

Diode-Pumped Solid State (DPSS) Lasers

Like the above laser, most YAG lasers are pumped by a CW arc lamp or be a pulsed flashlamp but the narrow absorption bands of the material make either pump source inefficient. For the ultimate in efficiency a narrow-band source operating this absorption band is required - in short an IR diode laser.

In the case of DPSS lasers high-power infrared diode lasers (cheap and plentiful by today's standards) are used to optically pump a solid state laser crystal. The pump light may be incident on the sides of the crystal (like the traditional YAG) or on-axis right through the rear mirror as is frequently done with smaller DPSS lasers. These lasers come in a variety of sizes from a 5 mW green laser-poiner to a monster emitting over 5W of CW light! This page will focus on the laser pointer as these are available to most amateurs. As of this writing the pointers are still about $300 USD (in late 2001) but it is expected that the price will drop like any other technology (I remember when red laser pointers were well over $100!). As well, anyone lucky enough to find a 'dead' green laser pointer can always rebuild that (I suspect a 'dead' laser pointer might well mean a simple replacement of the pump laser diode as it is difficult to damage the solid-state crystal except by mechanical means).

Green DPSS The green laser pointer contains much more than a simple diode as a red laser pointer does. Red pointers are a simple diode while green pointers contain a host of components. Because green diodes are not available (yet) the only way to make a green pointer, short of a small gas HeNe tube, is to use a frequency-doubled DPSS. First off, the laser rod is made from Nd:YVO4, not YAG. This material, called 'vanadate', is a more efficient material for DPSS lasers (it has a low threshold and so is suited to use with low-power pump lasers). It has a primary absorption peak at 808nm just barely in the infrared. The pump diode is hence selected to emit at this wavelength for optimal efficiency. The pump light at 808nm is fired at the crystal directly through the rear mirror of the laser. The rear mirror, although totally reflecting at the vanadate's emission wavelength of 1064nm (same as YAG - the Nd atom actually sets the lasing wavelength, YAG or vanadate are simply the host glasses used) are quite transparent to 808nm. This is possible because these are dielectric mirrors made of multiple alternating layers of thin-films and are reflective only at a single designed wavelength while transmitting other wavelengths. HeNe mirrors are also dielectric and you'll note that you can see blue light through an (unpowered) HeNe tube.

So, pump light at 808nm excites the vanadate crystal which then emits light at 1064nm in the infrared. To obtain green light intra-cavity second-harmonic generation is accomplished. To do this a crystal of potassium titanyl phosphate, or KTP, is put directly in the laser cavity between the front of the vanadate crystal and the front mirror (output coupler).

OutputCoupler Via a non-linear optical process the 1064nm light is doubled to 532nm (green). The KTP crystal is inside the cavity itself where power levels are much higher than outside the cavity: this is required for efficient conversion of light. The front mirror or output coupler is another dielectric mirror allowing green light to pass through while reflecting infrared light at 1064nm (and excess 808nm pump light that makes it through the crystal) back into the vanadate to keep it lasing. This is evident from the photo in which the OC appears green. Note that the mirror was fabricated directly onto the surface of the KTP crystal (by thin-film deposition techniques) and was made into the shape of a circle to define the beam (otherwise the beam might have been square since the vanadate and KTP crystals are both blocks).

Nd:YVO4 and KTP crystals

The entire laser is shown here many times larger than life. Components have been labelled for easy identification. The diode laser is not clear here since it is inside the brass mount. It is in a metal can. I did not attempt to remove it so I do not have specs on it. Judging by the way in which it caused havoc with both a scanner and digital camera (we had to find one which filters out IR light) the diode must be reasonably powerful. The entire pointer consumes about 0.16A at 3.0V so the electrical input power is about a half-watt. The rear crystal, glowing violet, is the vanadate while the front one, glowing green, is the KTP.

Nd:YVO4 and KTP crystals

The amazing thing about this laser is the scale. Coming from the world of YAG lasers where the rods are 5cm or more in length you find the vanadate crystal is a square block, not a rod, and is only 2mm in length. In front of the vanadate is the KTP crystal, also a block, about 4mm in length. In my laser the mirrors are fabricated directly onto the faces of the vanadate and KTP crystals. In other lasers (such as the pointer outlined on the DPSS page in SAM's laser FAQ) the mirrors are discrete and are held in mounts which may be adjusted.

Collimating Lenses Once green light is emitted from the laser is is collimated into a beam by a set of focussing lenses. These are required since the short cavity length invariably results in a beam which disperses.

IR Filter Finally the beam is sent through a filter to remove any stray infrared light (both the pump light and the 1064nm light from the Nd itself) and emerges as a green beam.

Quantel SHG YAG
Of course the green laser pointer represents the 'small' end of the spectrum as far as DPSS lasers go. In my lab, I have a large Quantel YAG capable of over 1W of SHG output as seen here in this photo. This laser uses a flashlamp to pump a YAG rod which is then doubled by a KTP crystal. A second laser (due to arrive shortly) is almost identical however is diode-pumped eliminating the flashlamp.