What Do You Need to Make Cold Atoms? Part 3: Test and Measurement

The third category in our look at lab apparatus, after vacuum hardware and lasers and optics is the huge collection of electronic gear that we use to control the experiments. I'll borrow the sales term "test and measurement" as a catch-all description, though this is really broader than what you'll usually find in that category.


This category covers all sorts of stuff, from power supplies to data acquisition equipment, but we'll start with the oscilloscopes.

The picture above shows two of the many oscilloscopes that rattle around my lab. These are used for almost everything that involves a voltage measurement, which is almost everything in the lab. We use photodiodes to convert light intensities to voltage signals, which we measure on oscilloscopes. Detector outputs are frequently in the form of short electronic pulses, which again, we check with oscilloscopes. The control signals used to measure and adjust the frequency of the laser are, you guessed it, monitored with oscilloscopes.

The picture above shows the extremes of the scope range in my lab. Right in the center is an ancient analog scope that I took out of the teaching lab stockroom, because nobody uses them any more. They're the best tool for a few jobs, though, and this one has a good X-Y mode that makes it a useful monitor of the frequency lock for the laser.

On the right, angled slightly away from the camera, is the high end of the scope scale, a four-channel color 100 MHz digital scope from Instek. I've got two of these in my lab, and we have at least one more for the upper-level lab class. This will display four different voltage signals as a function of time (hence "four-channel"), and will save data directly to a USB drive, for easy transfer to a computer. I only have two complaints about this oscilloscope: one is that it doesn't have an X-Y mode, the other is that it uses a really dumb data format in its saved waveforms, which requires more processing than it ought to. Still, it's relatively cheap for a four-channel digital scope, and serves as the workhorse data acquisition tool for most of what I've been doing lately.

I've also appropriated a few (ok, maybe five) two-channel digital scopes from the teaching labs to monitor various subsystems. These lack the save option of the nice one, and they're not in color, but they have a couple of nice features, particularly a wider range of offset voltages allowing more "zooming in" on small features than I can manage with the expensive model.

At the top left of the picture above, you can see a signal generator (with a red LED readout), which is another important and often-used component, as we frequently want to vary the control voltage sent to something or another in a periodic way. These are generally stolen borrowed for an extended period from the teaching labs, where they don't get much use, so nobody misses them. These let you select the frequency of oscillation, and whether you want a sine wave, square wave, or (most often) a triangle wave. For really sensitive applications, we have a couple of nicer models, but most of the work is done by cheap ones.

Below the signal generator is a stack of two homemade feedback circuits (well, ok, one feedback circuit and one power supply). This is a huge category of stuff for cold-atom experiments, as the lasers need to be held at a frequency very close to the atomic resonance frequency for the species you want to cool.

There are three things that need to controlled very precisely when you're working with diode lasers: the temperature, the current, and the piezoelectric stack used to position the diffraction grating in the external cavity (as described in the previous post). The first two do a very coarse tuning of the frequency, and I bought commercial temperature and current controllers to do those jobs (from Wavelength Electronics and Thorlabs, respectively). You can make your own for less money on parts, but if you factor in the time you need to spend building and debugging the controllers, it's not that big a savings.

i-7a7f04957a9ce56c614cd37fd104ec49-sm_sat_abs.jpgThe third item that needs to be controlled is the voltage to the piezoelectric stack that positions the diffraction grating. This makes an extremely fine adjustment of the position-- it moves something like one micron for 100 V applied to the piezo-- and is used to keep the frequency of the laser within a few megahertz of the transition frequency. The "mega" in there makes this sound like something huge, but the frequency of the light absorbed by the krypton atoms I'm trying to cool is several hundred terahertz, so it's actually a tiny fraction of the frequency, something like a part per million.

To do this, we use a technique called "saturated absorption spectroscopy" (which turns up in undergrad labs) to compare the frequency of our laser to the frequency the atoms like to absorb. This requires a bit of optical set-up, shown at right, with a weak "probe" beam passing through a glass cell containing the gas of atoms (in the middle right of the picture), and a more intense "pump" beam passing through the same gas in the opposite direction. A complicating factor is that the atoms we care about are not in their ground state, so we need to run a plasma discharge in the glass cell, which accounts for the complicated mount and the tinfoil (to screen out RF from the discharge). The combination of pump and probe beams allows us to get around the Doppler shift of the moving atoms in a room-temperature vapor (which broadens the absorption signal by several hundred megahertz), and stabilize the laser at a point within a few MHz of the transition frequency.

The circuit that does the stabilization is home-made (by me, modified slightly from the one in this PDF paper), and compares a voltage from the saturated absorption signal to a reference level, then sends a voltage to the piezo on the laser to push the frequency in a direction that brings the absorption signal closer to the reference level. The output and input signals are used as the X and Y values sent to the analog scope in the picture at the top of this post, to monitor the state of the laser.

The way we have things set up, we only need the one laser stabilization circuit at the moment. Some laser cooling experiments will need to stabilize up to four or five different laser systems, though, so this sort of circuit is extremely common. It also makes a good summer student project, so many an undergrad at an REU program has built a laser lock box.

Also seen in the optics picture above is a round black thing. This is part of the beam diagnostic system for the laser. One of the main failure modes of a diode laser is to go "multi-mode," meaning that it produces light of more than one frequency. To make sure this isn't happening, we use a "spectrum analyzer," which is a commercial Fabry-Perot interferometer that sweeps across a range of frequencies, and displays the spectrum of the light coming from the laser. If we see more than one peak in the spectrum, we know to bang on the laser.

i-719a86e771a54d02711c7690855c0511-sm_computer.jpgThe saturated absorption signal only works if the laser is already very close to the atomic resonance frequency, so we need a way to check the frequency on a very coarse level. This uses a wavemeter from Bristol Instruments, which interfaces with the least sexy element of my lab, the ~8 year old lab computer, to produce the display seen here. The computer is old, and still running Windows XP, but it has a National Instruments board in it that also allows us to use LabView to record data and generate control signals for the experiment (though that is not currently in use). Aging computer hardware is pretty common in cold-atom physics experiments, because it takes so long to get things working just right that the hardware and software has usually been updated multiple times by the time everything works. Scientific software is preposterously expensive, though, and like in every other sort of software, upgrades tend to break everything that was working before the upgrade, so we tend to cling to old computers as long as they can be made to operate, so as not to need to repeat all the work of getting the damn thing running.

There are a number of other types of electronic gear rattling around the lab that I don't have pictures of, including but not limited to:

  • Power meters: A lot of things require good knowledge of the laser's output power, so we have a couple of different power meters. The best power meters, in my opinion, have analog gauges with a needle indicating the level. You just can't make the same fine adjustments with a digital readout than you can when looking at a needle. If a needle gauge isn't available, I'll often put the voltage output of the meter on, you guessed it, an oscilloscope.
  • Power supplies: Laser cooling and trapping use magnetic fields as well as laser light, and we have a bunch of current supplies to drive the electromagnets used to produce these fields. These range from cheap low-power supplies swiped from the teaching labs to 50-amp monsters obtained secondhand.
  • Vacuum gauges: We've got a bunch of these, to make sure that the pressure in the vacuum system stays where it ought to be. Annoyingly, the more modern ion gauge controllers have safety features that are over-sensitive, and tend to shut the gauges down after a short time.
  • Amplifiers and difference circuits: A lot of the signals we look at are very small-- light scattered from atoms in a beam, say-- and thus need to be bumped up a bit before they can usefully be displayed on a scope. We've got some commercial low-noise amps that we use for this (one can be seen atop the analog scope at the top of this post), but also some homemade circuits that amplify a small signal, or take the difference between two small signals and amplify that.
  • RF supplies: We use a bunch of radio-frequency sources, typically in the 50-200MHz range. Some of these are commercial-- a couple of ham radio sources that we use to drive the RF discharges, an AOM driver used with one of the lasers-- others are cobbled together from miscellaneous parts. I drive one AOM using a voltage-controlled oscillator from MiniCircuits that's run through a couple of attenuators then into a big RF amplifier that I dug out of deep storage. It's a kludge, but it was about $2000 less than buying a dedicated AOM driver, and every dollar counts.

That's a look at the electronic gadgets that we use to get atoms cold. This stuff is pretty common to all cold atom experiments, with some needing more of one thing or another.

Questions? Comments? Critically important systems I forgot to mention?

More like this

Over at Confused at a Higher Level, Melissa offers an alphabetical list of essential supplies for a condensed matter experimentalist at a small college. This is a fun idea for back-to-school time, so I'll steal it, and offer the following alphabetical list of essentials for Atomic, Molecular, and…
Last week's series of posts on the hardware needed for laser cooling and trapping experiments dealt specifically with laser-cooling type experiments. It's possible, though, to make cold atoms without using laser cooling, using a number of techniques I described in two posts back in January. Those…
While the superstar of the particle physics world, the Large Hadron Collider, gets all of the attention (and the glamor shots), there's plenty of interesting science that can be done on the atomic level within an otherwise ordinary laboratory on the campus of an update New York university. Consider…
Derek Lowe has posted an article about X-ray lasers in chemistry, which amused me because of the following bit: Enter the femtosecond X-ray laser. A laser will put out the cleanest X-ray beam that anyone's ever seen, a completely coherent one at an exact (and short) wavelength which should give…

I drive one AOM using a voltage-controlled oscillator from MiniCircuits that's run through a couple of attenuators then into a big RF amplifier that I dug out of deep storage.

Why are you attenuating a signal and then amplifying it again? I'm sorry if this is obvious -- my understanding of AC circuits is hazy, to say the least.

By Grep Agni (not verified) on 13 Aug 2010 #permalink

Why do you want an x-y mode for the laser lock? Isn't the error signal vs time the natural thing to look at?

I was just remembering some of your old laboratory based posts nostalgically. It's great to read this new piece and get a taste of the working physics world. I think it was David Mamet, the playwright, who noted that it was important to establish one's competence in one area to make the rest of one's writing believable. He once used his knowledge of the restaurant supply business, of all things, to make us believe in his characters, their hopes, their motivations and their emotional reality.

Grep Agni: if the VCO is a total piece of shit, running it through attenuators can clean the signal up a bit. If it's good, then attenuators can be needed to bring it into range of the amp.

Chad: I'm surprised you aren't using some sort of direct digital synth locked to gps(cheap)/cdma(expensive) trained 10mhz clock for driving things that need RF. They would require more work to modulate based on analog signals, but you would gain total phase control and crazy levels of frequency control.

By Robert S. (not verified) on 16 Aug 2010 #permalink

As a personal interest, I would like to ask how much investement is required for purchasing necessary equipments to create an atom trap?