Welcome to Laser Pointer Forums - discuss green laser pointers, blue laser pointers, and all types of lasers

LPF Donation via Stripe | LPF Donation - Other Methods

Links below open in new window

ArcticMyst Security by Avery

Constant Current Driver Design (Also, i'm back :) )

Joined
Mar 29, 2013
Messages
1,221
Points
63
Hello LPF, long time no see!

It's been a good long while since i've been on here. After switching my focus to multirotors and FPV and then moving to the east coast for school, I seem to have forgotten my roots!

Well, 2+ years later i'm halfway done with my Electrical Engineering degree and looking to get back into the laser scene (oh how it has changed so much since I was last here!)

Basically, I am looking to build a lab-style 445 - 450nm tabletop laser to be used in demonstrations (working with my professor). However, as an exercise in circuit design and tweaking, I want to design my own constant current driver from scratch that allows analog control of the output power. I have completed a few semesters of analog electronics but definitely not enough for this project to be a breeze.


So i come to you guys! There's plenty of information on constant current drivers online, but as we all know, laser diodes can be more sensitive than your average load. I was hoping that some of the driver design veterans out there could point me in the right direction as to what makes a good constant current driver and what kind of specifications are important. I am assuming that you can not just pluck any buck IC from the lot and it will be suitable for driving laser diodes.

I guess the first thing I am unfamiliar with is how you take a voltage converter and then turn it into a constant current source. I am assuming that it utilizes a precision resistor inline with the load and then drives the output such that the voltage across the resistor is equal to some internal reference?


I am planning to use a step-down design (~12V input) with a max output current of around 5A (for those new beefy NUBM44 V2 diodes). Size is not a concern, and I plan to use active cooling for both the diode and the driver for continuous operation. I am looking to learn along the way here!

I am familiar with circuit design tools and circuit simulation, if that's relevant.

Any and all pointers in the right direction would be greatly appreciated!

Thanks guys, it's good to be back.

Best,

Matthew
 





Joined
Jan 29, 2014
Messages
12,031
Points
113
Long time no see, welcome back.

You can't just buy one instead of building it yourself? There is a driver you can buy which can handle more than 5 amps, just search under the term "blackbuck driver" on ebay, there are lots of them listed all the way up to 8 amps. I'd design for overkill and heat sink the driver extremely well with that diode.

Chris
 
Joined
Mar 29, 2013
Messages
1,221
Points
63
You can't just buy one instead of building it yourself? There is a driver you can buy which can handle more than 5 amps, just search under the term "blackbuck driver" on ebay, there are lots of them listed all the way up to 8 amps. I'd design for overkill and heat sink well with that diode.

Chris

I'm using this as an exercise in circuit design. Just buying one would defeat the purpose :)

I saw an open source design that uses a standard buck regulator with a current shunt and current sense amplifier going back to the ref/sense line on the buck IC. It seems like this is the route I am going to take.

Now to find some spice models :D
 
Joined
Jan 29, 2014
Messages
12,031
Points
113
You can also buy a regulator chip and wire it for constant current, but if you want to go the whole route, use a transistor and an accurate resistor to sense the current by the amount of voltage dropped across it and make a loop back to adjust the voltage to keep the current at a set value.

For some ideas, maybe not necessary, as you are already seeing what needs to be done, but if you want to look at chip regulator CC circuits, here's a link http://bristolwatch.com/ccs/LM317.htm and another one for the LM317 at 5 amps set up for voltage control, instead of current. Just giving this to you as a reference for the dissipation this device will handle, when properly heat sinked: http://sribasu.com/hobby-centre/high-current-dc-regulated-power-supply-circuit-with-lm317-5-amp.html

If the chip doesn't already provide it, be sure to find a way to design in some lag or ramp up time for the voltage to the diode when turning it on, you don't want to have too fast of a ramp up, that can stress the laser diode and reduce its life.
 
Last edited:
Joined
Sep 12, 2007
Messages
9,399
Points
113
Find an IC with a low Vref. The datasheet will have some design suggestions. The hard part is the PCB layout and component sourcing, which is exercise enough.
 
Joined
Jan 29, 2014
Messages
12,031
Points
113
Most don't bother for such a simple circuit or when powered by a brick, I expect that is what he will be doing.
 
Joined
Sep 20, 2013
Messages
17,252
Points
113
I got the impression that the OP wants to build a buck driver from scratch. He claims to have a schematic of one that works. That would likely be his best bet as these drivers can be temperamental if not designed perfectly as CC drivers. I have seen many failures from members here attempting a switching regulator driver. Anyone can make a linear driver from a voltage regulator IC.
 
Joined
Mar 29, 2013
Messages
1,221
Points
63
Thanks for the feedback guys. Just one last question.

Is there a guideline for the maximum acceptable ripple in the output current? I am currently testing an implementation using the LT1074 that seems to be performing well in simulation with a 60mA ripple @ 4.5A (currently working on reducing ripple). Is there a number at which performance is considered "good enough"?

EDIT: Also, what is considered a good "ramp up time" for the voltage across the diode? Alaskan mentioned that too quick of a ramp up time can damage the diode. My current design has a ramp time of around 300uS.

Thanks,

Matt
 
Joined
Jan 29, 2014
Messages
12,031
Points
113
I see the ramp up time all over the place with different designs, I've seen 300 us, 500 us, 1 ms, 5 ms and 250 ms. This thread shows some oscope display pix of one members ramp up: https://laserpointerforums.com/threads/astral-driver-boost-buck-sepic.100745/page-15#post-1526523 - I suppose the shortest ramp up time depends upon the laser you are using, maybe see if there is a manufacture specification you can find, but good luck with that. Stands to reason the more powerful the diode, due to the waste heat, the more care should be taken in regard to ramp up.

I'd try for a regulation of 1% to 2% myself.

Here's a screen shot from the thread I linked to showing the ramp up time this member choose:

 
Last edited:
Joined
Sep 20, 2013
Messages
17,252
Points
113
Thanks for the feedback guys. Just one last question.

Is there a guideline for the maximum acceptable ripple in the output current? I am currently testing an implementation using the LT1074 that seems to be performing well in simulation with a 60mA ripple @ 4.5A (currently working on reducing ripple). Is there a number at which performance is considered "good enough"?

EDIT: Also, what is considered a good "ramp up time" for the voltage across the diode? Alaskan mentioned that too quick of a ramp up time can damage the diode. My current design has a ramp time of around 300uS.

Thanks,

Matt
What do you mean by a 60 mA ripple? There is no easy way to measure current with a scope. When talking about ripple, it is always in terms of a voltage.
 
Joined
Sep 20, 2013
Messages
17,252
Points
113
Maybe. Then again I have seen this before and it wasn't just a typo. That is why I asked for clarification.
 
Joined
Mar 29, 2013
Messages
1,221
Points
63
Maybe. Then again I have seen this before and it wasn't just a typo. That is why I asked for clarification.
As you wrote, ripple is always in relation to voltage, not current, but you could work the math to come up with it, but not displayed on an oscope!

Sorry about the confusion. The implementation I have is in LTSpice, not on an oscilloscope, which means I can easily view the (theoretical) ripple current, although this would be difficult to do with a real circuit and scope.

The voltage ripple on the output in LTSpice is only a few mV!
 
Joined
Jan 29, 2014
Messages
12,031
Points
113
I don't know it matters how much ripple for either burning or just to make pretty light anyway, unless the peak exceeds the voltage the device is rated to take. Maybe someone has something more to share on this I am unaware of, there's a life time of learning when it comes to lasers, if you want to.
 
Joined
Sep 20, 2013
Messages
17,252
Points
113
This reason, right here, is exactly why I ask. It wasn't a typo after all.
 
Joined
Sep 20, 2013
Messages
17,252
Points
113
You do want to keep the ripple in the output of your driver as low as possible. Otherwise, the LED drivers would work fine for laser diodes too. But, they don't. And most here already know that.
 
Joined
Sep 12, 2007
Messages
9,399
Points
113
The laser diodes in optical drives write at speeds of up to 16x (maybe more?) with 1x being 36Mb/s, which, according to my math, is a "ramp time" faster than 1ns. These diodes sustain these ramp times literally quadrillions of times without sustaining damage. Why would 5 orders of magnitude slower at 100µs not be slow enough? Am I missing something obvious, or is this just collective guesswork and groupthink?

Also, the laser diode doesn't care about ripple (see above, optical writer drives operate on basically 100% ripple), provided the peak current isn't above some known damage threshold. 2A with an obscene 2A P-P ripple would be less hard on the diode than 4A with no ripple. <5% would be nice I suppose, but it isn't critical for a pointer.

There is no easy way to measure current with a scope.


bnh1fdy.gif



Probe across the current sense resistor in the regulator circuit, Paul. Apply Ohm's law. You should know this by now.


When talking about ripple, it is always in terms of a voltage.



hCWLB.jpg




What ARE you on about?? Every bench PSU has a current ripple rating for instance. Are all those engineers just wrong?
 
Last edited:




Top