High power laser driver - where to start?

[nikodem]

Hi all!

This is not a first laser (diode) driver, that I will be designing, but definetely the biggest. Before that I've been working with small laser diodes (mA of driving current) that I've been driving from a DC current source. Now the laser diode has up to 120 W (pulsed), so I need to supply a lot of current (up to 60 A I think). Obviously I would like to make it as efficient as possible, so DC current source on a FET is a bit problematic. I have no idea where to start, but I have lot of questions. Some of them are:

• Are there any white papers / application notes for such stuff?
• Can I power the diode with PWM?
• It it better to regulate the current or emitted power?

Thank you in advance for your assistance!

[Slh]

It really depends what the application is. Things like how short are your pulses, what's your repetition rate and what's an acceptable rise/fall time? Also, do you need to maintain a low current at all times to keep the diode warm and near the leading threshold?

I don't think that the diodes will appreciate simple PWM.
If you're happy with ms rise times then you might be able to do it with a current controlled dcdc. There are some dcdc chips with adjustable current limits.

If you want it to be faster then you might need to do a linear current source. For higher efficiency you set your DC voltage to just be high enough for the voltage drop on the current source.

Obviously you're well into class 4 laser territory here so remember, don't book into the laser with your remaining eye and all other safety  rules...

[nikodem]

It really depends what the application is. Things like how short are your pulses, what's your repetition rate and what's an acceptable rise/fall time? Also, do you need to maintain a low current at all times to keep the diode warm and near the leading threshold?

I don't think that the diodes will appreciate simple PWM.
If you're happy with ms rise times then you might be able to do it with a current controlled dcdc. There are some dcdc chips with adjustable current limits.

If you want it to be faster then you might need to do a linear current source. For higher efficiency you set your DC voltage to just be high enough for the voltage drop on the current source.

Obviously you're well into class 4 laser territory here so remember, don't book into the laser with your remaining eye and all other safety  rules...

This will be a medical laser, so it's not only class 4, but also really able to burn your fingers and start a fire in the lab 

But back to electronics:

I don't want to pulse the laser. The shortest on-time will be 5 ms, so from lasers point of view it's CW. the rise/fall times are not specified, but I'm guessing that around 100 us would be nice to have. The idea about keeping the diode just under the lasing threshold is a nice idea, but this can be achived by just adding a normal current source and summing the currents, right?

Current controlled DC/DC sounds like the way to go. Are there any controller chips for that, or do I have to discover everything on my own 

Regards!
Nikodem

[jbb]

A medical laser you say?

That opens up a can of worms in terms of safety certification. I have no idea what the detailed requirements might be, but I speculate they will include things like failing to a safe state, and maybe using qualified parts?

I would suggest that a good way to reduce certification pain is to look at the requirements now, identify which subsystems of the equipment will be most critical, and try to make those subsystems as simple as possible.

What sort of duty cycle do you need? If it’s not too high, perhaps a linear driver is suitable if fed with a DC-DC voltage regulator which outputs a voltage just a little higher than the laser + cabling + linear driver needs?

Do you have access to both pins if the laser diode, or do you have special requirements like cathode must be grounded?

[nctnico]

For current controller DC-DC chips look at those for LED lighting. More or less the same problem. Likely there will be some feedback loop involved from the laser's output level to have a constant light output level. Using LEDs instead of the actual laser might be a good idea to do initial testing of the circuit.

[nikodem]

Safety requirements and certification stuff are well known here and there is no problem with that. What I do want to know is how to approach the electronics.

DC/DC converter and a linear driver is a reasonable approach. A LED driver sound also a reasonable approach - I'll investigate it a bit closer. For testing we use rectifier bridges connected together to mimimic the internal paralel/serial connection of the laser (we don't know how it's built, but with voltage drop at ~20 V there definitely are several diodes in series).

Regards!

[Dan_Ohm]

Your laser project sounds very interesting! 
I'm starting to work on a laser project and at some point will need to turn on/off the beam maybe around 5K - 10K Hz or more, but I assume the laser can't do it so will need to find another way to achieve it.

[nikodem]

Your laser project sounds very interesting! 
I'm starting to work on a laser project and at some point will need to turn on/off the beam maybe around 5K - 10K Hz or more, but I assume the laser can't do it so will need to find another way to achieve it.

The laser should have no problem, with switching that fast, as long, as your electronics will be fast enough... Of course it depends on the laser type etc. What kind of laser is that?

[jmelson]

I'm starting to work on a laser project and at some point will need to turn on/off the beam maybe around 5K - 10K Hz or more, but I assume the laser can't do it so will need to find another way to achieve it.

Admittedly, what I have is not high power.  But, I have a laser photoplotter I made 20+ years ago.  It uses a 5 mW 670 nm laser diode.
The pixels are 5 us long, so the laser needs to turn on/off at rates up to 100 KHz.  It does this easily.  I just use an NPN transistor and a 100 Ohm trimpot to set the on intensity. 

Jon

[poorchava]

At 60A, with pulses short enough, a very low rds mosfet and carefully regulated, low voltage this may be doable as long as the resistor is almost perfectly non-inductive. 60A pulses will store a significant energy in even a minor inductance and will then ret released as inductive spoke that may d up your laser and/mosfet. If the current needs to be adjustable, then this will not be a reasonable solution. 

A constant current source that drives a mosfet or IGBT in linear region would almost certainly exceed soa of any mosfet or IGBT. The only solution I think would be to parallel numerous power BJTs, but those are quite obsolete for any normal use nowadays.

A switch mode source would have to be really, really fast, think MHz range to provide sharp enough rise/fall times, especially at such high current. Normal fets that can handle such current (few in parallel) will require a ton of gate charge and consequently a very high current have driver to switch fast enough. That is if you manage to handle the inductive kickback from literally any piece of conductor in the current path.

Maybe GaN would be a solution? I have no practical experience with it though.

Maybe you could also run some form of forward converter with transformer and primary switches operating in current mode? It would mean that you would have to somehow handle the negative part of the switching cycle so as to not f up your laser diode, but that way you could avoid 60A of  current through a semiconductor in linear region.

[nikodem]

I don't need such short pulses as jmelson, for me a pulse od 5 ms is minimum, what I need. I've found such driver:

https://www.analog.com/en/products/lt3744.html

Does it make sens to use it for driving a LD or is there a catch here..? I know, that 20 A is max, and should be sufficient for some uses.