Current and PZT Locking ECDLs

The saturated absorption signal is fed into a locking circuit to stabilise the laser. The actions of this circuit are described below, (Figure 3.12 aids in the description).

Figure 3.12 : Simplified circuit diagram of the Dual current PZT locking electronics. Inputs/outputs referred to in following text. Diagram reproduced from Dr G. Lancaster3.

The PZT is modulated by a triangular waveform which is scaled within the locking circuit, this allows us to scan across the atomic transition of interest. A sinusoidal dither is applied to the diode current (up to 100kHz) which ramps the frequency of the laser across the peak of the hyperfine feature chosen for locking. The saturated absorption signal from the photodiode is sent to an amplifier tuned to the dither frequency and then to a phase sensitive detector. This generates a dispersive signal which can be viewed on an oscilloscope. Integrating the dispersive signal over a short time base results in a high frequency error signal which is sent to the current output , integrating over a longer time base gives the dc error term which is sent to the PZT output . This processing should provide stable laser locking to one of the

atomic transitions for several hours. An example of the locking signal generated from the cooling transition is given in Figure 3.13. The error, or locking, signal is sown in red, it oscillates about a 0V level. The gain of the amplifier is reduced until only one crossover between the error and 0V level exists in view, at this point the feedback circuit is activated. A more thorough explanation of the locking electronics can be found elsewhere4.

Figure 3.13 : Typical locking trace obtained from the dual PZT/current lock box. Possible lock points are selected by the points where the dispersive (red) signal and 0V ground line (orange) cross. The associated saturated absorption signal (blue) is also shown.

This method is a peak-locking method, as such, it still requires application of the 2

(~12MHz) frequency offset to the red side of the cooling transition to allow Doppler cooling to function. This frequency shift can be achieved by passing the portion of laser output sampled for the saturated absorption setup through an AOM. The AOM was not placed directly in the cooling beam as this would have raised power and spatial-positioning issues when tuning the AOM. The AOM placement is shown in Figure 3.14.

The AOMs used (Isle Optics, LM080 and Gooch and Housego plc., M080) had an obligatory frequency shift of 80MHz with 23MHz tuning either side controlled by a voltage input to the AOM driver. This means it was not possible to lock the laser directly to the cooling transition and use the AOM to provide the ~12MHz frequency shift required. The cross-over resonances provided much more

distinct signals and therefore provided more stable locking points. Thus it was beneficial to lock to a crossover transition and tune the laser from this lock frequency

Figure 3.14 : AOM placement in the saturated absorption beam siphoned from the trapping laser. This allows the sat-abs setup (and hence the locking circuit) to see a well defined crossover transition while the laser outputs a not-so-well defined F=3 F =4 transition.

to the cooling frequency. This was performed by not actually tuning the laser but fooling the saturated absorption setup into perceiving a lasing frequency X 2

MHz distant from the actual output, (X was a frequency chosen to shift the lock point an AOM-manageable distance from the lasing frequency, X 2 80MHz 15MHz). Thus the laser would continue to lase on resonance while the saturated absorption setup would have the frequency-shifted AOM output to lock to. The D2 transitions used for cooling Rb85 are shown in Figure 3.2, from there it can be seen that the, well defined, F = (2,4) crossover transition is 92MHz from the cooling transition and so is within range of our AOM s.

Adjustment of the AOM control voltage for trap size optimisation was achieved through a self-built electronic circuit. This circuit provided a calibrated read- out of the frequency deviation from the base 80MHz shift, up to 25MHz, and passed the appropriate voltage to the Isle Optics AOM drive unit. The circuit is shown in Figure 3.15. The LCD display panel used was a Lascar DPM 700 meter, which provided a read-out accurate to a tenth of a MHz. The potentiometer marked with a red X was the control for altering the AOM shift. Adjusting the remaining two variable resisters while monitoring the AOM signal output (against a reference un- shifted saturated absorption signal) allowed accurate calibration of read-out frequency

shift. Later re-calibration confirmed a good longevity of well-calibrated status, (all components commercially available from RS).

Figure 3.15 : Control and monitor circuit for AOM frequency shift of the cooling laser as part of the saturated absorption locking system. The signal output to the LCD display panel meter was scaled and calibrated to yield an accurate frequency read-out. 18V regulators require heat sinks.

Thus, by locking the cooling laser to the 5S1/2(F= 3) 5P3/2(F = (2,4) line with a ~80MHz +15MHz shift from the AOM we can obtain a laser output direct from source with the -12MHz red-detuning necessary for trapping. The hyperfine beam did require AOM manipulation of the saturated absorption signal, it was locked to a transition which was distinct enough without resorting to crossover resonances.