DFB module: Frequency change via laser diode current

[Joerg]

Hello Folks,

Does anyone know a DFB laser module that is a bit more agile in
frequency pulling via diode current? The 1500nm range would be nice but
we can also go towards 1350nm.

Right now we are trying it with a JDS Uniphase CQF915 module and it
feels like kicking an oak tree. Ideally I'd like to be able to drag it
across a 100pm in a few milliseconds, or at least tens of milliseconds.
This would correspond to a current change of about 25mA. I have designed
a dual loop controller that also uses the TEC to support this FM plus
lots of (slower) overrange and it works nicely. But the DFB seems to
buck and as soon as I make the loop too fast it has difficulties locking.

 

[LVMarc]

Hello Folks,

Does anyone know a DFB laser module that is a bit more agile in
frequency pulling via diode current? The 1500nm range would be nice but
we can also go towards 1350nm.

Right now we are trying it with a JDS Uniphase CQF915 module and it
feels like kicking an oak tree. Ideally I'd like to be able to drag it
across a 100pm in a few milliseconds, or at least tens of milliseconds.
This would correspond to a current change of about 25mA. I have designed
a dual loop controller that also uses the TEC to support this FM plus
lots of (slower) overrange and it works nicely. But the DFB seems to
buck and as soon as I make the loop too fast it has difficulties locking.

Joerg,

good description, pulling against an oak tree.

the laser diode is a highly tuned optical cavity, Q's > 1000. you are
not pulling any ppm but rather ppb! Perhaps an LED wouldbe able to
pull, but mutliplexing two lasers and fsk'ing them would absolutley
work..good luck.

Marc Popek

 

[Joerg]

Joerg,

good description, pulling against an oak tree.

Folks out here do that. But only if they have a monster truck with a
monster winch

the laser diode is a highly tuned optical cavity, Q's > 1000. you are
not pulling any ppm but rather ppb! Perhaps an LED wouldbe able to
pull, but mutliplexing two lasers and fsk'ing them would absolutley
work..good luck.

We need to vary over a range, not just FSK. It's more like a specialized
interferometer application. DFBs can pull about 4-5 picometers per
milliamp in current change. I just need one that will do it fast enough.
Or another kind of laser diode that puts out a nice and narrow
spectral line where we'd mount our own TEC to it for the wide range
tuning part. The TEC tuning speed isn't so critical although it was a
bear to make that work at any reasonable tuning speed without too much
overshoot.

 

[Phil Hobbs]

Folks out here do that. But only if they have a monster truck with a
monster winch



We need to vary over a range, not just FSK. It's more like a specialized
interferometer application. DFBs can pull about 4-5 picometers per
milliamp in current change. I just need one that will do it fast enough.
Or another kind of laser diode that puts out a nice and narrow
spectral line where we'd mount our own TEC to it for the wide range
tuning part. The TEC tuning speed isn't so critical although it was a
bear to make that work at any reasonable tuning speed without too much
overshoot.

You're seeing millisecond delays in _current_tuning_? Wild. Ought to
be much, much faster than that if it's the usual plasma-optic
effect--you must be doing temperature tuning instead. Is there a
smaller, much faster modulation you can measure?

Cheers,

Phil Hobbs

 

[Joerg]

You're seeing millisecond delays in _current_tuning_? Wild. Ought to
be much, much faster than that if it's the usual plasma-optic
effect--you must be doing temperature tuning instead. Is there a
smaller, much faster modulation you can measure?

Well, I was able to make the inner loop (which regulates the current)
oscillate at over 100Hz but only at really tiny excursions. Less than a
picometer. When I disabled that loop and its connection to the other
loop (the one for the TEC) and then stepped the current between 35 and
40mA the reaction was rather sluggish. It did scoot the expected 20pm or
so but that took hundreds of milliseconds. Sure looks like temperature
tuning. Now I am wondering if DFB modules can really do that or whether
we may have to try out ECDL.

Laser stuff is not my home turf. One of the frustrations is the rapid
model changing and corporate takeovers. I talked to an engineer at
Emcore and he said their 1751 can be tuned fast via current. But when
the prototype was done that part was already gone from their site. Have
emailed him, maybe they have something that can work.

 

[[email protected]]

I do not know what your application is but I have worked on similar
problems in the past. The difficulty in what you are doing comes most
in proportion to the accuracy with which you want to hit any given
frequency in on any given time scale and how much amplitude modulation
you willing to accept in the process. Diodes like DFB lasers are
designed to resist frequency shift current modulation. This is so they
will stay within their allotted channel as they are modulated on and
off. Fabry-Perot diodes have much larger FM verse current modulation
tuning coefficient. EC diode lasers are the real champs of large fast
tuning applications but have all the problems associated with
mechanical (usually PZT) devices like long term drift mechanical
resonances hysterieses etc.

FM modulation by current tuning is often troublesome do to the fact
that large and small modulation effects take place on different time
scales from 100s of microseconds to a few nanoseconds.

Knowing exactly how much frequency you can tolerate on what timescale
and if you can tolerate significant AM on your laser will help you
converge to the right solution with the least pain.

Good luck

Steve

 

[Joerg]

I do not know what your application is but I have worked on similar
problems in the past. The difficulty in what you are doing comes most
in proportion to the accuracy with which you want to hit any given
frequency in on any given time scale and how much amplitude modulation
you willing to accept in the process. Diodes like DFB lasers are
designed to resist frequency shift current modulation. This is so they
will stay within their allotted channel as they are modulated on and
off. Fabry-Perot diodes have much larger FM verse current modulation
tuning coefficient. EC diode lasers are the real champs of large fast
tuning applications but have all the problems associated with
mechanical (usually PZT) devices like long term drift mechanical
resonances hysterieses etc.

Drift and non-linearities can be handled here because we have a closed
loop system, no need to steer based on a current versus wavelength
calibration table. This loop works quite well but the laser behaves like
a drag chute, very sluggish in response.

FM modulation by current tuning is often troublesome do to the fact
that large and small modulation effects take place on different time
scales from 100s of microseconds to a few nanoseconds.

Yep, I bet that's part of the problem here.

Knowing exactly how much frequency you can tolerate on what timescale
and if you can tolerate significant AM on your laser will help you
converge to the right solution with the least pain.

We can regulate out AM quite well, even if it was fast. This is partly
due to the advent of blazingly fast opamps from TI and others. I have a
1GHz GBW opamp from Texas in there. Costs only $2-$3 and it truly
impressed me. I was mentally prepared to do it all with RF transistors
like usual but didn't have to.

 

[Jürgen Appel]

Yep, I bet that's part of the problem here.

I do not know about DFB lasers, but with Fabry-Perot diodes a modulation of
the injection current has two effects:

a) The temperature of the diode junction changes. This leads to an
increase/decrease of the optical resonator length due to the temperature
dependance of the refractive index and due to thermal expansion.
b) The charge carrier density in the diode also influences the refractive
index.

For modulation frequencies <1 MHz usually a) dominates whereas for >10MHz b)
dominates.

Unfortunately a) and b) cause frequency shifts in opposite directions so
that the transfer function has a 180° phase shift between 1 and 10 MHz
making a high speed frequency control challenging.

I can imagine that the a) effect is much less pronounced in DFB lasers so
that the a) <-> b) transition frequency is much lower.

Cheers,
Jürgen

 

[Joerg]

Joerg wrote:




I do not know about DFB lasers, but with Fabry-Perot diodes a modulation of
the injection current has two effects:

a) The temperature of the diode junction changes. This leads to an
increase/decrease of the optical resonator length due to the temperature
dependance of the refractive index and due to thermal expansion.
b) The charge carrier density in the diode also influences the refractive
index.

For modulation frequencies <1 MHz usually a) dominates whereas for >10MHz b)
dominates.

Unfortunately a) and b) cause frequency shifts in opposite directions so
that the transfer function has a 180° phase shift between 1 and 10 MHz
making a high speed frequency control challenging.

I can imagine that the a) effect is much less pronounced in DFB lasers so
that the a) <-> b) transition frequency is much lower.

Juergen, that's what I suspected as well. What puzzled me though was
that this cut-off seems to be really low. Somewhere in the sub-10Hz
range. DFB module have a long chamber buried inside, maybe that's why.