Diode Laser Synopsis - University of Connecticut 3/2000


Most of the information on this page is part of the thesis of Steve Gensemer, which can be downloaded here.

Contents:

Introduction

This is an HTML document with links to ASCII and binary files with various uses.

The purpose of this webpage is to enable anyone with a little electronics experience and some understanding of external cavity diode lasers to reproduce the system we have developed. Our system works exremely well for neutral atom cooling and trapping, which is primarily what it is used for at UConn. However, it also has many other applications.

When considering other applications of this system, note that this system is designed to be locked to the D2 or D1 line of Rb, Cs, Li, and maybe K. The external cavity laser could certainly be used, with the addition of an additional reference cavity, to lock to much weaker spectroscopic features, such as iodine molecular lines (Opt. Commun. 132 p. 94 (1996)). The Hansch paper (Opt. Commun. 117 [1995] pp. 541-549) does a very good job of describing the characteristics and possible uses of the external cavity we use. Of course one is always at the whim of the diode laser industry when searching for specific wavelengths.

In general when using this system locked to the same frequency day after day, these lasers require almost no adjustments. It is common to enclose them in a sealed box and never open the box for several months.

Plans for Machining of External Cavity Mount

Material

The drawings here are for the mount described in the Hansch paper. We have made them from aluminum, although TUI (German company that manufactures this) makes it from a composite called "Neusilber" which is similar to invar. The thermal conductivity and heat capacity of invar are worse than those of aluminum, but it has a much lower thermal expansion. Neusilber is reportedly advantageous because it is more elastic, and so has less plastic deformation than aluminum. However, aluminum is easier than neusilber or aluminum to machine and our mounts are very well thermally stabilized so we have stayed with aluminum.

We use 100-pitch screws for the grating adjustments and 0-150V square piezoes, both available from Thorlabs. We have used holographic gratings from Edmund Scientific, 12.5mm square. Be sure to choose the lines/mm correctly; see the Hansch paper. We use aspheric plastic collimating lenses from Thorlabs as well.

The laser diode should be oriented so that the divergence angle is larger in the horizontal dimension - this way more lines on the grating will be filled, which should improve mode selection.

Here are the drawings. They are in both JPG and Canvas format - Canvas is the layout program we used to design them, so if you have Canvas you can edit the drawings easily. There are location holes on some pieces that are only for machining.

Notes:

Debugging problems with External Cavity Mount

Q: My laser was on the atomic resonance in the past, but it isn't now. What do I do if I can't seem to find the atomic resonance with my diode laser?

A: Follow this procedure:

  1. Make sure that your laser current and temperature are the same as they were last time you saw the resonance.
  2. Scan your PZT with a 10Hz ramp wave with as large an amplitude as possible.
  3. Now, while looking at your absorption signal, dial the current up and down between threshhold and the maximum you plan to run this laser.
  4. If there are no signs of life yet, do the same thing and watch for fluorescence in a cell (with a CCD camera) which is getting most of the laser power.
  5. If still no signs, put the laser current where it was when you last saw the fluorescence, and scan the temperature from 5 degrees below your original setpoint to 5 degrees above that point, while watching for fluorescence from the cell.
  6. If still no signs, re-align the vertical adjustment. Stop the PZT scan. Dial your laser current down until only your output beam is only about 1/2 mW. Now, using a power meter or photodiode, monitor the power of this beam. Mark where the vertical adjustment screw is now so you can go back to your starting point. Do not turn the vertical screw more than 1/4 turn in either direction. Very slowly tweak the vertical screw until you find a peak in output power. When the vertical alignment is correct, the output power will be at least double what it is when the vertical alignment is way off. (This is because you are changing the finesse of the laser cavity.) Repeat 1-4.
  7. If still no signs, re-align the horizontal adjustment. Keep the PZT scanning. Turn the laser current to the maximum that you want to use. Set the Temperature to about 5 degrees below room temperature. Do not touch the vertical alignment. Mark where the horizontal alignment is so you can go back to where you started. Do not turn the horizontal adjustment screw more than 1/4 turn in either direction. Watch the fluorescence in the cell and very slowly tweak the horizontal adjustment screw, until you see the fluorescence in the cell. If no signs, turn the laser current down 10mA and try again. Turn it down 10mA more and try again, etc.
  8. If still no signs, make sure you laser is not dead (it still puts out as much power as it is rated to - remember you lose about 30% off the grating.) If your laser is not dead, but has worked in the past, then check to see if other things are wrong - is the PZT still working? (Put in a 1kHz square wave and listen for a 1kHz sound wave) Are you using a dead cell? (try another cell, try this cell with another laser you know is on resonance.) Is your laser still collimated? Is your laser mount full of condensations? Is your temperature controller doing its job? (does the mount feel cold to the touch when you cool it down?)
  9. If still nothing wrong, use an extra diffraction grating to find out if your laser is locked to the cavity or not. Turn off the PZT scan. If you watch the 1st order diffraction off a grating (let it propagate 1m or more) you will see the laser spot jump back and forth between the free-running wavelength of the diode and the cavity-locked wavelength, as you scan the horizontal screw. Use your brain to decipher what the meaning of this jumping is so that you can figure out if your resonance is in the tuning range of the laser. If it is not, you may need to heat or cool the laser more substantially to bring the resonance back into the tuning range of the laser.
  10. If you still can't get the resonance, combine this laser with a laser that you know is on resonance and bounce them both off of the grating - now you can tune your laser until it matches the wavelength of the known laser. This should help in case you are badly misaligned. If you have to do a lot of tweaking on your horizontal adjustment, you should periodically re-set the vertical alignment. Often, a large horizontal adjustment will also slightly change the vertical alignment.
  11. If still no luck, try another diode laser.
  12. If still no luck, ask for help.

Circuit Design Notes

The instructions are designed to help graduate students or undergraduates who are relatively new to this kind of electronics. Please don't feel your intelligence is being insulted - the instructions reflect the numerous mistakes that students have made while constructing the circuits listed here.

For a quick start, just email one of these files to APCircuits and you'll get back a printed circuit board for:

If you want to edit the circuit boards before sending them off, download this file. This contains an old version of the program Qcad which we used to make the boards. In the subdirectory "boards" you will find the files (*.pcb) corresponding to all of the printed circuit boards described below. They can be edited by running Qcad and selecting "pcbedit".

If you want to view the gerber files before you send them off, download this file. It contains the program GC-Prevue, which is freeware. In the subdirectory "boards," you will find files (*.cwk) which you can view with the GC-Prevue command "Restore All". The Gerber files can be exported from GC-Prevue if you need them independently. If you use APCircuits, you can email them the *.CWK file itself and they will send you back the circuit boards. They are quite cheap and good quality. Any other circuit board manufacturer will take the exported gerber files (*.p11), drill files (ncdrill.dat), aperture files (*.apr) and drill rack (drill.rck) and make circuit boards for you.

If you are going to design your own circuit boards from scratch, we recommend the new version of the program QCAD, which is only US$100, and has a schematic drawing front-end. After you generate the gerber files with QCAD, you are urged to view them with GCPrevue to verify that they are really what you want, and to do any last-minute changes to the apertures or drill sizes.

In our experience it is a great advantage to work with commercially produced printed circuit boards because they can be much more easily and quickly assembled, tested and debugged than working with Vectorboard prototype boards.

We recommend assembling these boards one section at a time. For example, the schematics have been broken into pieces that can be easily understood. Assemble the circuit in easily-understood pieces, soldering in all the parts that perform one function of the circuit, and then test them, then move on. You may notice that if you don't understand how the circuit works, you won't be able to test it. Don't just solder everything in place, plug the board in, and plug it into your laser diode! The orientations of diodes, capacitors, ICs and the values of resistors can be confused . . .

If you are going to design your own circuits the way we have, we reccomend that you first plug the circuit into a solderless breadboard and test it thoroughly to get any possible bugs out before you commit it to copper.

If you are only making one of a circuit it may not be worth the trouble of making a whole printed circuit board for it - you can do a very nice job with Vectorboard type 4112-4 prototype board, which has a ground plane to keep noise down. However, thoroughly testing the circuit on a solderless breadboard first is still strongly recommended.

The trick with these layouts is to make boards with a ground plane. A GROUND PLANE IS CRUCIAL TO THE PERFORMANCE OF ALL THE CIRCUITS DESCRIBED IN THIS PACKET! IF YOU MAKE THEM WITHOUT A GROUND PLANE YOU CAN EXPECT LOTS OF NOISE AND CROSSTALK BETWEEN DIFFERENT PARTS OF THE CIRCUITS!

In addition, each of these circuits should be enclosed in its own metal enclosure - die-cast aluminum enclosures are nice to use because they are sturdy and can be easily drilled. Any signals that require low noise should be brought to and from the circuit boards with coax cables - the shield of the coax cable should be soldered to the ground plane, and at the connector end the shield should be connected to the metal enclosure. (RG-316/U cable is good for this because it is thin but can handle high temperatures and so will not be easily damaged by soldering.) The enclosures should be explicitly grounded (not just through signal cables) and power lines should be connected to the metal box at the point they enter the box with 10uF or more electrolytic capacitors. One way to achieve this is to simply set all of your circuit boxes on a grounded piece of sheet metal or optical table.
[Basically the last two points are just RF circuit technique. Even if your circuit has only low-frequency signals, if you want low noise you must use good RF technique because you want to keep _out_ RF noise from the environment.]

If you are new at building analog circuits like this here are a few tips:

To make boards with a ground plane, you can either email the single-sided board layouts to the circuit company (we have used AP Circuits) and ask them to make the ground plane for you, or you can do the following:

In each of the directories for the circuits in this page is a list of files. The have different uses. Here is a list of what you should find in each directory:

If you need to know the aperture sizes and drill sizes so you can use these gerber files with some program other than GC-Prevue,

Diode Current Controller

The files here describe a diode laser current controller that was designed by Jan Hall's group at JILA. The printed circuit board layout was designed by us at UConn.

A paper the JILA group published describes a current supply and its performance, which far exceeds anything we have ever needed. The current supply circuits we have made are actually based on a later version produced by the JILA electronics shop. Based on the variety of current supply designs we have received from JILA, it appears that their electronics shop adds new features to their circuits as often as we fill our coffee cups. Keeping in mind that the students who have to build and test these circuits may not have the experience of the JILA technical staff, we have decided to try to keep these circuits as simple as possible, while retaining the low noise and the high modulation bandwidth of the JILA supply.

We have tried a number of commercial diode laser current supplies and have found that they are generally far noisier than our homemade supplies, in addition to being almost impossible to fix when something goes wrong. Therefore we recommend that you build these current supplies, which can be made for ~ $200 - less if you are making several at the same time.

The current is limited by U80, which compares a set voltage with the monitor voltage and opens Q80 if the monitor voltage exceeds the set voltage, thereby draining current away from the diode. We have found that the current can be much more simply limited by choosing R16 such that the maximum current, roughly 7V/(R16+R1), is less than 50% above the rated operating current of the laser diode, which is basically the most you can crank a laser diode before without killing it. Note that if the max. current is high enough, R16 may need to be heat sinked; Vishay rates this part (VHP4, tolerance .01, to 3W without a heat sink.

For simplicity, our schematic is broken into pieces.

Use 1% resistors on this circuit. R16 is Vishay VHP-4, 5ppm/C temp. coefficient. R1 is MK132 Caddock, Allied PN524-5050 - R20 is 1/2 watt - R23 is a 10-turn precision pot mounted on the enclosure. C22,23 are 10V Tantalums - C3,4,31,102,103 are 20V tantalums - C20 is 35V tantalum - C12,60 are silver mica - C30,105 are electrolytic, all else are monolithic ceramic.

Here are the files for manufacture of the printed circuit boards that these schematics refer to. See circuits notes for an explanation of the PCB and Gerber files.

Lockin Detector/Servolock Circuit

Many thanks to Kurt Gibble of Yale University for this circuit and the locking scheme using AOMs.

Layout files here

This circuit has basically 5 different sections that perform different functions, all related to locking our diode laser mounts to spectroscopic features using lockin detection. It probably only makes sense in the context of the overview of our locking scheme.

This circuit can clearly be used for a variety of different locking schemes - appropriately adapted, it could be used for lockin detection at dither frequencies into the GHz range. (Although we haven't tested it for heterodyne offset locking, it should work for this purpose, provided one has a good enough heterodyne signal - the basic scheme is described in the Hansch paper.) One could also bypass the lockin detector portion of the circuit and lock to the side of a mode of a fabry-perot cavity. This has been done successfully in cases where the spectroscopic features are much fainter than alkali dipole transitions. (Basically one locks the laser to a fabry-perot cavity mode, and in turn locks the fabry-perot cavity to a spectroscopic feature using an additional servo loop.)

Rb saturated absorption signals and the error signal from this lock circuit are here. Look at the readme file first.

The five sections of the circuit work as follows: (Here is the schematic so you can follow along)

Signals that should be routed to and from the board with coax RG-316/U cable are marked on the partmap with an X. Other signals can be routed to the front panel with hookup wire but we recommend twisting pairs of hookup wire for the integrator switch connections to be sure that you don't introduce unwanted high frequency noise because you are looping signals around each other.

Photodiode Circuit

Layout files

This is the photodiode amplifier we use for our saturated absorption. Its gain is determined by the feedback resistor R1. The feedback capacitor C1 deliberately limits the bandwidth to keep noise down, at the expense of introducing a phase shift at the high end of the frequency response - we use a trimmer cap so we can tune this. The parts specified here are chosen to give a bandwidth of ~100kHz.

We have used the Thorlabs FDS100, although this is actually a Hamamatsu photodiode and can be purchased directly from them for less. The photodiode is biased here to -VCC. We run this amplifier off a pair of 12V lead-acid batteries.

The output should be run from the board to the box with RG-316/U. The power leads should be grounded at the box with 1uF caps. Drill a hole in the box to let light into the photodiode - if this hole is small it will help keep out room light and make it easy to put a filter in front of the photodiode. It can fit in a very small box so that it doesn't take up much room on your optical table.

For higher-speed applications one can use a smaller-area photodiode, biased by a battery, - small signals at high frequencies can be amplified with MiniCircuits connectorized amplifiers (ZFL line) - we wire these up on a little piece of prototype circuit board, so there is no special printed circuit board for this.

AOM Driver Circuitry

Modular construction

We built some AOM drivers with all the parts on one board, but found that they are a terrible hassle. RF parts are prone to being killed, so we now wire up each component on a little piece of circuit board inside its own die-cast aluminum box. They can all be bought in connectorized boxes, but we have decided to do it our own way and thus save ourselves a lot of money. Having everything in separate boxes makes diagnosing problems very simple. Follow good RF technique when wiring these up. The parts and circuit boards below should enable you to build your own AOM drivers (as usual, better and cheaper than the commercial ones) to drive AOMs that require < 200MHz.

VCO circuitry

Layout files

This circuit facilitates running a POS-series VCO from Minicircuits. We use them to provide a carrier signal for our AOMs. The circuit only requires a single power input of 28-30 VDC - this is because our RF amplifiers require the same voltage so we can run the VCOs from the same power supply and thus simplify plugging them in.

The supply voltage for the VCO is +15V, set by a Zener diode.

The output of the VCO can be attenuated coarsely by including a load resistor RL. We put a header pair in the holes for RL so we can solder and unsolder RL easily without damaging the circuit board - RL is generally needed because the POS puts out about 9dBm and while our AOMs need 30-33dBm to give their maximum shift, the amplifiers we use (Motorola CA2832) only need -5 to -2 dBm input. Roughly speaking, if you use the CA2832, RL=680ohm will give you about 1.25W of RF power (33dBm), and RL=1.2Kohm will give you about 1W of RF power (30dBm).

The tuning voltage is clamped by Zener diode D1, which is 18V for the POS-100. Jumper J1 selects AC or DC coupling for the modulation input. If DC coupling is selected, then the tuning voltage will be controlled entirely by the voltage at the MOD input. If AC is selected, then the center frequency will be controlled by the trimpot, and the voltage at the MOD input will be AC-coupled in.

The AC-coupled signal can be inverted or non-inverted, depending on the position of J2. This is very handy when you are using an AOM to dither a saturated absorption beam for locking: if you change whether you use an up-shifted or down-shifted AOM beam, you can switch the phase of your dithering by 180 degrees instantly with this jumper!

In our locking scheme we generally run the dither output of the locking circuit into the MOD in of this circuit and use AC coupling. We set the pot for the center frequency we want and we use an RF spectrum analyzer to set the amplitude of the dither so that we are dithering the AOM beam by roughly the linewidth of the feature we are locking to - this gives us the largest possible error signal. If you don't have an RF spectrum analyzer, you can measure the amplitude of the dither at VTune (it will be amplified a little by the opamp) and determine how much it will dither your VCO based on the tuning voltage curve from the VCO specs. However, you will still need a frequency counter to set your VCO center frequency.

Attenuation

We have used bi-phase attenuators - TFAS-line Minicircuits - to attenuate the VCO output before it is amplified. If only a manual control of the shifted beam intensity is required, one can simply wire a potentiomenter to the attenuator.

If, however, one wants linear control of the shifted-beam power with an external voltage, one must servo the shifted-beam power. This is easily done with a photodiode and a simple integrator, whose output controls the bi-phase attenuator. The integrator keeps the photodiode voltage equal to some set voltage it is compared to. Because the gain of the RF amplifiers is generally temperature-dependent, a servoloop is generally required in order to accurately and quickly adjust the shifted-beam power, or in general to stabilize it over long time periods.

Switching

For many applications in quantum optics you will want very fast switching of your AOMs. If you are using a linear broadband amplifier like the CA2832, you will find that you can switch your AOMs at the limit of their specs simply by switching off the input to the amplifier with a PIN diode switch. We have successfully used the YSWA switches from MiniCircuits.

Layout files for YSWA switch

We made this circuit board so we could mount these switches in our own boxes. We put +/-5V voltage regulators on the board so we could use our usual +/-15V supplies to power the switches. The control input takes a TTL signal. The switch is two-way, so it could be used to switch an AOM on/off, to switch a single VCO between two different AOMs, or to switch a single AOM between two different VCOs. The YSWA's off-attenuation is something incredible like >50dB, so this switch is much faster and allows less RF leakage than almost any of the commercial drivers.

Amplifier

Cheap AOMs (<$700 new, <$100 surplus if you can find them!) generally require 1Watt RF power, which can be supplied by a CA2832 Motorola broadband linear amp, which goes up to about 200MHz.

Some AOMs in the < 200MHz range require up to 5W RF power - you'll need to find another amplifier to supply them with that power. The power amplifier modules get pretty pricey, although with more work you can certainly build your own amplifiers with bipolar transistors (you'll want to make your own circuit board for this, and it gets a little tricky.) One problem here is that these are not linear RF amplifiers, they're tuned, and they won't shut off so fast when the input signal is turned off. If you want fast switching with a high power AOM you may need to put a FET in the supply voltage line to shut off the transistor instead of or in addition to shutting off the input signal. Then again you may want to buy a commercial driver. (Then again you may want to just buy low power AOMs!)

Above 200MHz, MiniCircuits parts can certainly be used as outlined above, although you may want to use connectorized versions of them and skip our circuit boards. Again, the amplifiers will get pricier.

Temperature Controller Circuitry

Here is the circuit board layout for the temperature controller from the JILA paper. It is a PID controller and is copied from the Nist design (p. 94). If one only wants to cool (not heat) a mount, one of the power transistors could be omitted. The transistor used should be well heatsinked and mounted on the outside of the enclosure, and it can be powered by a different power supply than the +/-15V that runs the controller circuitry. Know the resistance of your TEC and choose a power supply accordingly. Tuning this circuit is explained well in PID CONTROL AND CONTROLLER TUNING TECHNIQUES.

Generation of Sidebands by Microwave Modulation

We have had very good preliminary results putting microwave sidebands on diode lasers at 3 and 6.8GHz. Other groups have done this for many years. Using a bias-tee (MiniCircuits) to couple the RF signal into the laser diode, and soldering the diode onto the end of the coax cable with very short leads gives fairly efficient RF coupling into the diode, so that with 10 dBm of RF power you can easily get enough sideband light for the repumping of a Rb MOT, even if the laser is in an external cavity or is injection locked.

Companies that supply parts mentioned in this page:

Papers referred to in these documents:

by Steve Gensemer, revision 3/7/2000