A two-frequency acousto-optic modulator driver to improve the beam pointing stability during intensity ramps

0.61 MB·English

by  B. Fröhlich

A two-frequency acousto-optic modulator driver to improve
the beam pointing stability during intensity ramps
B. Fr¨ohlich, T. Lahaye, B. Kaltenha¨user, H. Ku¨bler, S. Mu¨ller, T. Koch, M. Fattori and T. Pfau
5. Physikalisches Institut Universit¨at Stuttgart, Pfaffenwaldring 57, 70569 Stuttgart, Germany
(Dated: February 2, 2008)
We report on a scheme to improvethe pointing stability of the first order beam diffracted by an
acousto-optic modulator (AOM). Due to thermal effects inside the crystal, the angular position of
7 thebeamcanchangebyasmuchas1mradwhentheradio-frequencypowerintheAOMisreduced
0 to decrease the first order beam intensity. This is done for example to perform forced evaporative
0 cooling in ultracold atom experiments using far-off-resonant optical traps. We solve this problem
2 by driving the AOM with two radio-frequencies f1 and f2. The power of f2 is adjusted relative to
n the power of f1 to keep the total power constant. Using this, the beam displacement is decreased
bya factor of twenty. The method issimple to implement in existing experimental setups,without
a
J any modification of theoptics.

6
1
I. INTRODUCTION
CCD
amera

]
s
c
i
t
p
o teles
ope
attenuator
. An important applicationofacousto-opticmodulators
s ×1
c (AOMs) is the control of laser beam intensities. The 3 AOM
beam blo
k
i power of the sound wave traveling inside the acousto-
s
y optic crystal determines the amount of light that is FIG. 1: (Color online) Setup for measuring the beam dis-
h diffractedoutofanincominglaserbeam. However,ther- placement of the AOMusing a TeO2 crystal. The size of the
p maleffectsleadtoadisplacementofthediffractedbeams laser beam is reduced with a telescope before it enters the
[ AOM.AbeamblockaftertheAOMstopsalllightexceptthe
when the power of the radio-frequencydriving the AOM used beam, which is attenuated and monitored with a CCD
1 is changed. The position stability is a critical parameter
camera. The distance between the AOM and the camera is
v in many applications using AOMs, especially for dipole
1.4 m.
3 traps formed by strongly focused, far off-resonant laser
8 beams [1]. Such traps areplaying a majorrole in atomic
1
physics nowadays, as they allow for the realization of
1 II. EXPERIMENTAL SETUP
new experiments, forexample the Bose-Einsteinconden-
0
7 sation(BEC)ofatomicspeciesthatcannotbecondensed
0 in magnetic traps such as cesium or chromium [2], or We test the two-frequency method with two AOM
/ the all-optical generation of a BEC [3]. Particularly in models that use different acousto-optic crystals to mod-
s
c crossed optical dipole traps, where two beams have to ulate the light. The setup for measuring the beam dis-
i be overlapped on a 10 µm scale, a small change of the placement of the first AOM using a tellurium dioxide
s
y beam position can have a dramatic effect on the trap (TeO2) crystal (Crystal Technology 3110-199) is shown
h characteristics (frequency and depth), thus causing se- in figure 1. We use anytterbium fiber laser(IPG, model
p vere problems [4]. One way to circumvent them, is to YLR-20-1064-LP-SF) at 1064 nm, with 10 W output
v: use a single-mode optical fibre after the AOM, but this power. The1/e2 beamradiusisreducedwithatelescope
Xi cannotbe donefor highpowerlasers,suchasCO2 oryt- frominitially2.1mmto0.7mmbeforegoingthroughthe
terbium fibre lasers. In this paper we report on a simple AOM.AftertheAOMabeamblockstopsalllightexcept
r scheme, adaptable to any AOM, which strongly reduces theusedbeam,whichisattenuatedandmonitoredwitha
a
the beam displacement. The method is based on driv- CCD camera. We fit the images with a 2D-gaussianand
ing the AOMwith two differentradio-frequenciesf and record the peak position of the beam profile. The setup
1
f ,andadjusting theirrelativepowersP andP sothat forthesecondAOMusingagermanium(Ge)crystal(In-
2 1 2
the total RF power P = P +P in the AOM is kept traAction Corp. AGM-406B1) is slightly different. We
1 2
constant [5]. This article is organized as follows: After use a CO2 laser (Coherent GEM100L) at 10.6 µm, with
describingtheexperimentalsetupwithwhichwemeasure 21W ofpowergoingthroughthe AOM.At adistance of
thebeamdisplacement,wepresentourmeasurementsfor about3mwemeasurethebeamprofileinonedimension
AOMs in the 1 µm and the 10 µm wavelength range. In witha movable pinhole infrontofa powermeter. We fit
an appendix we show the details of the electronic circuit the profile with a gaussianand recordthe peak position.
we use to adjust P relative to P with a single control Figure 2 shows the modified AOM driver one has to
2 1
voltage. use for the two-frequency method. To control laser in-
2
(a)
f1 f2

ontrol VCO atten. AOM AOM f1}1st order
voltage Uin 2f1−f2
x
(b) f2−f1
f1
+

voonlttargoel Uin VCO atten. AOM y⊙ z f1,f2 0f1th−ofr2der
f2
VCO atten.
FIG. 3: (Color online) Schematic of the AOM driven by two
Uout
voltage frequencies. The image shows a picture of the laser beam
adjustment diffractedbytheTeO2AOM.Ontherighthandsideoftheim-
age the frequency shifts corresponding to the diffracted light
FIG. 2: (Color online) (a) Normal setup for driving an are indicated (f1 =99 MHz, f2 =123 MHz).
AOMwithvariableRFpower. Avoltagecontrolledoscillator
(VCO) generates the radio-frequency f1 (blue line), which is
attenuated to a value given by the control voltage Uin. The displacementweoptimizetheanglebetweentheacoustic
signalisthenamplifiedbeforegoingtotheAOM.(b)Forthe
wave and the incident laser beam to have the maximum
two-frequencyAOMdriverweaddanextraVCOandattenu-
powerinthefirstorderoff . Withfullpoweratthisfre-
ator. Theadditional VCOgenerates thesecond RFsignal f2 1
quency andnone at f , we achievediffractionefficiencies
(red line), whose power is adjusted relative to f1 to keep the 2
up to 90%.
total power in the AOM constant. This adjustment is done
by modifying the control voltage Uin with an electronic cir-
cuit(shownindetailintheappendix). FortheTeO2AOMwe
use the following Mini-Circuits components: VCO POS-150, III. MEASUREMENTS
attenuatorPAS-3,combinerZMSC-2-1,amplifierZHL-1-2W.
With the setups described above we measure the po-
sition of the first order beam of f at different RF pow-
1
tensities with an AOM, one has to change the RF power ers for the two AOMs, with and without the second fre-
driving it. This can be done by attenuating a RF signal quency. In figure 4 we plot the angular movement as
comingfromavoltagecontrolledoscillator(VCO)before a function of the laser power in the first order beam.
amplifyingittoitsfinalvalue(figure2(a)). Theamount Figure 4 (a) shows the displacement perpendicular to
of light that is diffracted out of the incoming beam is the plane of diffraction y for the TeO AOM. The dis-
2
thendeterminedbythecontrolvoltageUin. Forthe two- placement in the plane of diffraction x (not shown in
frequency driver we add a second VCO and attenuator the figure)has the same dependence as perpendicular to
(figure 2 (b)) with frequency f2. The two frequencies f1 it, but is smaller by a factor of three. Adding the sec-
andf2 arechosencloseenoughinordertobewellwithin ond frequency keeps the beam position almost constant
the bandwidth of the AOM [6], but far enough to give (below 0.03 mrad), whereas without, a beam displace-
a sufficient separation of the two first order beams. We ment of up to 0.6 mrad occurs. A big improvement is
use f1 = 99 MHz (resp. 30 MHz) and f2 = 123 MHz also evident for the Ge AOM (figure 4 (b)), the angular
(resp. 50 MHz) for the TeO2 (resp. Ge) AOM. The movement is reduced by a factor of ten. The fact that
power of the frequency f2 generatedby the second VCO we are not able to compensate the displacement as well
is adjusted relative to the power of f1 in order to keep as with the TeO2 AOM is due to the higher RF power
the total power in the AOM constant. To do this with the AOM is driven with. For maximum diffraction effi-
a single control voltage, Uin is modified by an electronic ciency the Ge AOM needs 30 W RF power, whereas the
circuit (see appendix) before it is applied to the second TeO AOMneedsonly2W.AnotherTeO AOMthatwe
2 2
attenuator. We adjust the transfer function Uout(Uin) of tested (A-A Opto-Electronics deflector, model MTS80-
the circuit to have a constant total RF power after the A3-1064Ac) uses a sheer mode acoustic wave and needs
signals are added and amplified, the latter condition be- only0.5WRFpowerformaximumdiffractionefficiency.
ing crucial to take into account the amplifier saturation. Its beam movement is significantly smaller than for the
Laser light going through an AOM driven by two fre- other AOMs, only up to 0.1 mrad, but still larger than
quencies is diffracted in many different beams as can be with the two-frequency method [8].
seen in figure 3. The image was taken with the TeO To supplement those steady state measurements, we
2
AOM at about equal power of both RF signals. Besides have also checked for the TeO crystal that the suppres-
2
the zeroth order beam, the first order of both frequen- sion of the beam movement remains good, when the RF
cies,aswellassecondandeventhirdorderbeams,which power is dynamically ramped down over a timescale of a
correspond to multiple absorption and stimulated emis- few seconds, as is done for forced evaporative cooling of
sionofphonons[7],canbeseen. Formeasuringthebeam ultracold atoms.
3
The two-frequency method helps also to stabilize the
(a)
laser power P in the first order when switching the RF
d℄
a powerrapidlyascanbeseeninfigure5,whichshowsthe
mr 0.6 r time dependence of P(t) for the TeO AOM. Without
2
[
t r the second frequency it takes about 10 seconds until the
en 0.4 steady state value is reached, when switching the laser
m
e r powerabruptlyfrom10to100%. Thebeamdisplacement
v
o 0.2 takes place over the same time scale. Only a very small
m r
transient effect in the first second after switching can be
r
ula 0 b b b b b b rb seen, when using the two-frequency method.
g
n 0 20 40 60 80 100 In conclusion we have demonstrated a simple method
a
to improvethe pointing stability of a beam diffractedby
laser power in (cid:28)rst order [%℄ anAOMwhentheintensityisrampeddown. Thesalient
(b) advantage of this technique lies in the fact that only the
d℄ 1 RF driver has to be modified, without any modifications
a r
mr 0.8 of the optics.
[ r
nt 0.6
e
em 0.4 r Acknowledgments
ov 0.2 r
m
ar 0 b b b rb We thank C. S. Adams for useful discussions and
ul b W. Mo¨hrle for the design of the digital control box. We
ng 0 20 40 60 80 100 gratefully acknowledge the support of the German Sci-
a
ence Foundation (DFG) (SFB/TR 21) and the Landess-
laser power in (cid:28)rst order [%℄
tiftung Baden-Wu¨rttemberg. T. L. acknowledges sup-
FIG. 4: (Color online) (a) Measured angular movement of port from the European Marie Curie Grant MEIF-CT-
thefirst orderbeam perpendicular totheplaneof diffraction 2006-038959.
(y)with(bluecircles) andwithoutthesecondfrequency(red
squares) for the TeO2 AOM. The movement is plotted as a
function of the relative laser power in the first order with
APPENDIX: VOLTAGE ADJUSTMENT CIRCUIT
respecttoitsmaximumvalue. (b)Samemeasurementforthe
Ge AOM, measured in thediffraction plane x.
Inthisappendixwepresentasimplewaytorealizethe
voltageadjustmentneededforthetwo-frequencymethod
(figure2(b)). Theelectroniccircuitshowninfigure6(a)
modifies the control voltage U , so that the total RF
in
power stays constant in the AOM. We measured the re-
100 quired calibration curve U as a function of U , which
out in
90 thecircuitapproximatesbyastepwiselinearfunction. To
80 do this, we use an inverting amplifier whose gain at low
%℄ 70 voltagesisgivenby−RR111++RR122. ParalleltoR1 andR2 are
[ 60 otherresistors(R ,R ,...) inserieswith Zenerdiodes. If
y 3 4
t 50 U is larger than the Zener voltage of one of the diodes
si in
n 40 it gets conducting and the gain is increased. For ex-
e
int 2300 a−mple(Ri1f1+4.R31V2) 6 .UTinh6us,6e.a3chVttimheegUaineixsceinedcrseaasZedenteor
10 (R1+R2)k(R3+R4) in
voltage of one of the diodes the gain increases. The am-
0
plified voltage U′ is then inverted to U′′ before in the
−2 0 2 4 6 8 10
last step the voltage U is added. The potentiometer
off
time [s℄ R allows for an extra gain in the last step. We use
16
large potentiometers for all resistors to have a big flexi-
FIG.5: (Coloronline)Timedependenceofthelaserintensity
bility for the transfer function. In figure 6 (b) the mea-
when switching the RF power rapidly. Without the second
frequency (green) it takes nearly 10 seconds for the intensity sured transfer function is plotted. With this we are able
to stabilize to its steady state value. With the second fre- to keep the total RF power after amplification constant
quency(blue)thereisonlyaverysmalltransienteffectinthe within10%,whichisenoughtostronglyreducethebeam
first second. displacement. For the setup using the Ge AOM we use
a more complex control box, which digitizes U with an
in
analog-to-digital converter and then generates the out-
4
(a) (b)
12
R47101 R501k2
Uin 4R710 1R002k −+OP27U′ −11kR195UVofR6f.187k 108 rs rs rs rs rs rs rs
4.3V 4R730 5R04k k3 R6.188k V] rs
6.3V 4R750 5R06k 10R1 R101k4 R6.185k R221k6 −+OP27 Uout U[out46 rs rs rs
rs
8.2V 4R770 5R08k −+OP27U′′ 2 rs
10V 4R790 R501k0 0 rs
0 2 4 6 8 10 12 14
Uin[V]
FIG. 6: (Color online) (a) Schematic of the electronic circuit for adjusting the control voltage. The gain of the first inverting
amplifier depends on the voltage Uin due to the Zener diodes. The amplified voltage is inverted again before a variable offset
Uoff is added in thelast step. (b) Measured transfer function of the circuit.
put voltage U according to a conversion table written in an EPROM.
out
[1] R.Grimm,M.Weidemu¨ller,andYu.B.Ovchinnikov,Adv. sity,2003.
At. Mol. Opt.Phys.42, 95 (2000). [6] WeobtaintheAOMbandwidthbymeasuringthereflected
[2] T. Weber, J. Herbig, M. Mark, H.-C. N¨agerl and R. power as a function of the radio frequency using a direc-
Grimm, Science 299, 232 (2003); A. Griesmaier, J. tional coupler (Mini-Circuits ZDC-10-1).
Werner, S. Hensler, J. Stuhler and T. Pfau, Phys. Rev. [7] D. L. Hecht, IEEE Trans. Sonics Ultrasonics SU-24
Lett. 94, 160401 (2005). (1977).
[3] M. D. Barrett, J. A. Sauer, and M. S. Chapman, Phys. [8] One drawback of the acousto-optic deflector is that the
Rev.Lett. 87, 010404 (2001). sound velocity for the sheer mode in TeO2 is signifi-
[4] R. Dumke,M. Johanning, E. Gomez, J. D. Weinstein, K. cantly smaller (by a factor of five) than for the longitu-
M. Jones and P. D. Lett, NewJ. Phys. 8, 64 (2006). dinal mode, leading to longer rise times. Using the two-
[5] This method is mentioned briefly in M. E. Gehm, Prepa- frequency method with a longitudinal mode AOM allows
ration ofan optically-trapped degenerate Fermi gasof 6Li: one to keep fast rise times.
finding the route to degeneracy, PhDthesis,DukeUniver-