Tuning speed of a laser diode

Thermal settling is incredibly sensitive to the geometry. Thermal conduction slows down quadratically with distance, and common materials differ by factors of 10**3-10**4 in thermal conductivity (Cu gets up to about 400 W/m/K, glass is about 1, plastic is around 0.1, air is about

0.025). If you want fast response, sit the diode laser on a thin-film alumina heater (e.g. certain Vishay resistors work), and control the heater instead of (or in addition to) the TEC. With active control, you can make that settle in a few seconds.

The monitor photodiode makes a really nifty temperature sensor for this use--forward bias it just a little every few ms, and measure the voltage drop. It's typically brazed to the same small metal stud as the laser, leading to thermal delays in the hundreds of milliseconds.

Cheers,

Phil Hobbs

Hello Phil,

The TEC would be really slow because of it's mass. I wanted to do it via the laser diode current. Problem is, often the data sheets have no entries whatsoever for that. Sometimes they do but then it's mostly for very fast FM. An example from one data sheet was 100MHz/mA for 30kHz FM which is an extremely tiny amount of frequency change (but a very fast change). They do not tell you what it would be for, say, 10Hz or 100Hz of FM. One could be brazen and extrapolate from 30kHz towards the "DC" change per mA but that is kind of tough if the thermal properties aren't known. Hence my question.

Yes, that would be a neat way of doing it. However, in this case I'd have feedback about the actual wavelength it puts out.

--
Regards, Joerg

http://www.analogconsultants.com

Oh, I thought you were quoting the _TEC_ current, not the laser current.

The current-tuning behaviour does have some thermal input, because it's usually measured at constant case temperature, but not very much. The main effect is the change in the effective bandgap due to higher level injection.

I don't know if you can get enough temperature swing by just changing the injection current, at least assuming you don't want to blow the facet off the diode. The temperature transient will probably have a small, rapid change due to the thermal resistance of the diode, stud, and thermal interface, followed by a much slower one as the spreader plate temperature changes.

Cheers,

Phil Hobbs

Hello Phil,

Sorry, should have been more clear about it in my post.

I am not too familiar (yet) with what's going on in there. But from what I have read there are sweep times that people measured that were sometimes in a second range, sometimes much less. This was for rather huge swings of more than a nanometer. An added difficulty is that laser diodes seem to be a hot commodity that comes and goes like fresh bread. You read a study, find out the diode they used, call the mfg and then either it's out of production or the phone line is disconnected because the whole place is gone.

That would be ok, the regulation loop would take care of that. As long as the whole tuning process isn't like molasses ;-)

--
Regards, Joerg

http://www.analogconsultants.com

In the work reported in Sloman A.W., Buggs P., Molloy J., and Stewart D. "A microcontroller-based driver to stabilise the temperature of an optical stage to 1mK in the range 4C to 38C, using a Peltier heat pump and a thermistor sensor" Measurement Science and Technology, 7

1653-64 (1996), I measured the thermal time constant of our set-up with three different thermo-electric modules. and the heat capacity of the modules themselves looked to be quite low - less than 2 J/K for the 70W Marlow DT-1089 and probably quite a lot less., which makes the instrinsic time constant of the TEC itself something less than 2.5 seconds, probably quite a lot less.

--
Bill Sloman, Nijmegen

Hello Bill,

Well, yes, that's why I don't want to use the TEC for anything other than to keep the diode at a constant temperature.

--
Regards, Joerg

http://www.analogconsultants.com

Tell me about it. It's worse than hard disc read channel amplifiers. Early DVD drives used self-pulsating laser diodes that had a saturable absorber layer in the cavity. They had a lot of useful properties, but good luck finding one now.

It might be or it might not, depending on the mechanical design.

There's some discussion of all this in my free thermal chapter,

The figure on P.5 is useful here, I think.

Cheers,

Phil Hobbs

Some time ago I worked on a laser diode based spectroscopy instrument. The technique is called two tone FM spectroscopy. The laser diode is modulated with RF and the laser current is ramped so the diode frequency (carrier) is sweep across the absorption line of interest. The sweep frequency was few hundred Hertz but it could have been much higher since the diode frequency is dependant on the injection current. If the sweep frequency is low enough the frequency pulling due to heating might be noticeable.

Regards,

Boris Mohar

Got Knock? - see: Viatrack Printed Circuit Designs (among other things)

void _-void-_ in the obvious place

--
Posted via a free Usenet account from http://www.teranews.com

Joerg, laserdiodes tend not to have that much headroom in regards of current. Meaning no one gets a 30mW diode when 3mW are sufficient. Physicists are a quite different type of folks. While an EE doesn't use a component above 70% of the max rating, physicists tend to be sure that the device also stands 120% continous. Laserdiodes are notoriously sensitive on overcurrent. The lifetime drops rather fast close to the max rating. Don't even consider it.

You can use a Peltier with low mass and change the temperature within few seconds.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

I remember a journal article by Carl Weiman from the late 1980's or early

1990's where for an atomic physics experiment, they were modulating the current through a 852 nm laser diode to tune the laser frequency by 9.2 GHz (0.02 nm) and have the laser frequency stable to about 1 MHz within a few microseconds. To do this they tailored the current pulse to compensate for two thermal relaxation time constants, one of which was about 1 microsecond and the other was about 10 microseconds.

I expect that for a 1550 nm laser diode that the time constants will be or the same order of magnitude. For quantum cascade lasers, which use the same InP substrates as 1550 nm lasers it takes almost 1 millisecond for heat from the active region to reach the submount to which the laser is soldered, so I don't think that the mounting of the laser diode will have much effect on how fast you can current tune the laser wavelength

Regards, Bret Cannon

Hello Phil,

That's why I am even pondering whether or not to design my own TEC and diode current controllers. Although AD offers both and TI offers diode drivers, and both companies have a good reputation for longevity of their chips.

Cool! Nicely written, although I am not planning to do this via the TEC. That would be too slow.

I like Mark Twains statement. He has a way of saying things so that the logic behind it becomes crystal clear.

Does your book contain info about tuning via diode current changes? Might be a good book to have on the shelf.

--
Regards, Joerg

http://www.analogconsultants.com

Hello Boris,

For some diodes there is a MHz/mA spec for FM. However, mostly for rather high FM frequencies where thermal effects may put a big damper on things. But if you got a few hundred Hz of sweep that is encouraging, assuming it had to be somewhat linear of a sweep. Did you go to 0.5nm or more of sweep?

--
Regards, Joerg

http://www.analogconsultants.com

Hello Rene,

Not in this case. I won't go past max rated CW minus fail-safe tolerances and put a hard limit on that where the protection circuitry kicks in. Modern laser diodes aren't like the old cleaved ones anymore. When they are spec'd at 30mW then they are usually made to run at that level for many years. It's all telco stuff where customers expect cast iron reliability. Like an aircraft engine versus a motorcycle engine.

The old types wouldn't even lase at 70%. You only had a tiny window somewhere between 80% and 100%, and woe to those who went to 105% only for a nanosecond. Bzzzt.

Telco stuff can usually be run at the spec power, pretty much forever. Provided you have good protection in case of cold starts etc.

Way too slow in this case :-(

--
Regards, Joerg

http://www.analogconsultants.com

Hello Bret,

9GHz in microseconds sounds encouraging. That would work. Over-shooting the current is something I can't do as it might exceed the ratings for a split usec and send a few hundred bucks into the weeds.

A millisecond would still work, as long as it's not 50msec or more.

--
Regards, Joerg

http://www.analogconsultants.com

It has been a while. We were looking at HF absorption line around 1500nm. I do not remember how wide the line was but we did manage to sweep across the whole thing.

-- Boris

--
Posted via a free Usenet account from http://www.teranews.com

HF - Hydrogen Fluoride - nasty stuff.

-- Boris

--
Posted via a free Usenet account from http://www.teranews.com

I talk about current-tuning, including a lot of neat measurements you can do that way, but not specifically about temperature-tuning via bias current. I've never heard of an application before where that would be a good fit--normally we're trying to optimize the diode current for best laser performance, and adjusting everything else to achieve that goal.

Cheers,

Phil Hobbs

Hello Phil,

Well, in this case we have no choice, we must tune. Anyhow, I think my client should really have your book. What's the best avenue to buy it? IEEE? Amazon?

--
Regards, Joerg

http://www.analogconsultants.com

There's a list of places to get it on my web site,

Cheers,

Phil Hobbs