[School of Physics - Optics Group]

Diode laser update news

29 March 2004

Current response: Warning: our diode laser web pages are very very out of date. We have learned a great deal and are now in the process of laying out boards for a far superior ECDL electronics system. It will be simpler, cheaper, and better.

One thing we have learned is that current modulation is limited by the diode itself, to about 10kHz:

This figure shows the frequency modulation of our ECDL when the current is modulated at relatively low frequencies. Note that the dips at 700Hz and 25kHz are caused by the (unlocked) laser drifting off the atomic resonance and hand-adjusted back onto resonance. The phase lag is about 30 degrees at 13.5kHz.

17 April 2002

Current modulation: The laser protection circuit is shown with two series inductors of 100uH and a parallel 1uF capacitor. These limit possible current modulation (see item 3 under 15 August, below). The bandwidth depends on dynamic resistance of the laser diode. If the diode resistance is 10 ohms, the maximum frequency will be about 10kHz. If the diode dynamic resistance is 1 ohm, then that frequency drops to just 1kHz. Modern diodes aim for low dynamic resistance; from scanning the web, I expect a cutoff frequency of somewhere between 1kHz and 10kHz. If you want to try proportional feedback to the diode current, remove the inductors and capacitor. You might instead add a resistor in parallel to both inductors, e.g. 22 ohms for 50ohm input impedance, or use smaller inductors, or remove the inductors. Do a little circuit analysis first -- the phase shift at the cutoff frequency becomes rather interesting.

12 April 2002

Minor electronics fault: In CLAUDIUS, the lock signal offset voltage is derived from two LM399 voltage references. Huge overkill (they are quite expensive) - you can just use the supply rails. If you want to use the LM399's, they only provide a range of +/- 7V. Replace R28 (10k) with a 5k and you will have a full range of offsets.

01 February 2002

Minor electronics fault: In CLAUDIUS, the comparator chip used to produce the trigger output at ramp zero-crossings operates outside its rated supply voltages. Current diagrams show a uA710 (LM710) comparator, which has abs max +14V/-7V supply, but is supplied with standard box +/-15V. This may be the cause of poor zero-crossing detection and is poor from a reliability viewpoint
Solution: Swap chip with a LM111J (not LM111J-8, this is an 8-pin DIL) which is pin compatible but has much wider supply range. Preliminary search shows this part may be hard to find, as chip fits perfectly well in 8-pin DIL package and old 14-pin compatability packaging is a bit daft. The LM311 is a widely available 8-pin DIP equivalent, but would require an adaptor.

15 August

At last we have two lasers with beatnote below 1MHz. Indeed, well below — 650kHz combined beatnote or roughly 450kHz each. How did we do it? There were three major issues.
  1. Using Zeeman dither on the vapour cells rather than dithering the injection current. It turned out to be trivial to drive a couple of the coils at ~55kHz and ~70kHz, and obtain 3 to 4 volt dispersion curves with only 50 to 100mV noise. The coil driver is identical to the output driver of our temperature controller. It works just fine with the L165 high-power opamp, but we are now using cheaper ordinary LF356 opamps, +/- 12V supplies, 43ohm sense resistor.
  2. The stack. The stack tunes the laser at 200MHz per volt, so 5mV of noise on the stack (i.e. output of the high voltage amp) corresponds to 1MHz laser frequency noise. 5mV on the high voltage corresponds to just 0.5mV of noise on the signal into the HV amplifier, i.e. from Claudius.
    We are now using a WIDE/NORMAL switch and an additional offset control. To set up the laser initially, we have things arranged for a wide scan (switch to WIDE). The Claudius PZT output (JP10) is connected to the HV amp and then to the stack. The PZT disc is not connected to anything. In NORMAL mode, the Claudius PZT output (JP10) is connected to the HV amp and then to the PZT *disc*, and the stack is connected to a potentiometer across the 150V supply rails. This allows us to set a frequency offset over a wide range via the stack, and use the disc for short scans and for locking. The stack offset is filtered with a capacitor and resistor.
  3. Current injection. We have tried and tried to use proportional feedback to the diode injection current but it's essentially useless. I think this is probably because varying the injection current affects both the laser intensity and frequency, and our frequency reference (sat abs) does not properly discriminate between the two. Proportional feedback to the piezo disc is not wonderful (bandwidth limited, mechanical resonances) but better. You might like to experiment with using sat abs with a reference beam subtracted. Our electronics is designed for doing that, but at this point we can't be bothered.
Many trivial things have been fixed. The 26 Sept Claudius circuits on the web have these changes, including the coil driver circuit, but not the WIDE/NORMAL switch as yet.

25 July

After considerable effort, we have come to the conclusion that proportional feedback is almost useless. This is counter-intuitive and will be the subject of further work. In the interim, the basic CLAUDIUS frequency feedback servo design has been streamlined, and Luke Maguire has dramatically improved the circuit diagram. The lockin board also had serious problems which have been fixed; a revised circuit diagram will be added later today. We also hope to have a beat signal between two optimised lasers very soon.

Remaining problems: the dither signal is ac-coupled to the lockin board, and to the current injection, but both have different equilibrium potentials. This must be fixed to remove offsets on the proportional feedback.

And of course, we really would like to see the proportional feedback doing something useful.

We also suspect that the springs in the Ultima mount could be damped to reduce high frequency mechanical noise.

50/100Hz

Some of our lasers have a problem with 100Hz, as observed by looking at the frequency of the laser output, which has a little jump every 10ms, followed by mechanical ring-down at about 3kHz (see above). Some lasers don't have this, so we think it's a problem with grounding. We will provide some guidelines on this when we've fixed the unhappy lasers.

DC locking

We have investigated a new DC locking technique described in:
S.E. Park, H.S. Lee, T.Y. Kwon and H. Cho: "Dispersion-like signals in velocity-selective saturated-absorption spectroscopy", Opt. Comm. 192 p49--55 (2001).
Early indications are that it's not so simple, and we are (15 August 2001) happy with our AC locking using Zeeman dither. Luke also has good results using DC locking with the Wieman/Hansch polarisation spectroscopy arrangement.

Diode laser electronics
Created: 7 June 2001
Last modified:
Maintainer: Robert Scholten, School of Physics, Email:  r.scholten at physics.unimelb.edu.au
Authorised by: Robert Scholten, School of Physics, Email:  r.scholten at physics.unimelb.edu.au
This page, its contents and style, are the responsibility of the author and do not represent the views, policies or opinions of The University of Melbourne.