Best Buck for Red, 7 diode build

Hey all

I need a cheap, simple, driver circuit that works with 3.7 Vin and will drive LPC826 diodes. Off the shelf IC's and components preferred. I have already blown my budget several times over.

Background:

I am building a 7 diode focusable handheld with an integrated beam expander and PWM brightness control. After nearly 1 whole year (I literally thought of this concept the same day I found LPF), the design is finished with one major exception and I am beginning to assemble. This is a prototype build, using cheap lenses, diodes and drivers. Proof of concept until I can go BIG with 445's, high efficiency drivers and low divergence lenses.

Literallly the only thing I have left is the driver circuit. I planned on using linear driver but I can't find one that will work with my battery. I am using a huge 38120 battery LiFePO4 that operates at 3.7V. I need to use this battery because it is
the only one capable of supplying the current needed for the next build with 445's. The battery is capable of supplying 30A continuous! :evil:
Results for BATTERIES

I already have built and tested the PWM module (555 timer). The PWM drives a power MOSFET which controls current to the 7 drivers. The problem is 3.7V is too low for any linear driver I can find. I am guessing that this means I need a Buck driver; the Target ampacity is 550 milliamps or so.

Any help is appreciated and I know there are tons of drivers out there but I wanted to hear some of your thoughts on this based on my design requirements.

Thanks for the help. I am very excited to get this beast completed.

J

Here's the problem I see -

To power 7 LDs with one driver and an input of 3.7V, you're obviously going to be wiring them in parallel. With a constant current driver, if one of those 7 parallel LDs fails (quite possible with 7 diodes TO potentially fail) then the current that diode is consuming gets distributed to the other 6. That increased current has potential to then kill the other 6 (or even just one of them) and the problem snowballs.

If you can't increase the supply voltage, then I would be inclined to use 7 different drivers.

If it was me doing this, I would lay out 7 of the AMC7135 chips on a PCB, wired to provide 7 different channels of regulated 350mA output, from the same input power source. The dropout on that regulator is tiny. As long as you can live with 350mA, this is probably the best way to have something that "resembles" one driver, driving your 7 LDs.

Absolutely what RHD said. OR you can run them in series off a simple linear CC driver, with an input voltage of 25V 350mA (24V min).

Thanks for catching my slip up, RHD! Post edited.

Sigurthr said:

Absolutely what RHD said. OR you can run them in series off a simple linear CC driver, with an input voltage of 25V 3A (24V 2.45A minimum).

Click to expand...

A string of series LDs will provide the same full current output to each diode. Your suggested setup would only need ~400mA (or whatever you're driving the LDs at) as an input current.

In series, you'd want to have an input voltage of (7x Vf) + Driver Dropout, and an input current of whatever you want to driver the LDs at (so say around 400 or 500mA for LPCs)

Thanks guys

I was planning on 7 separate drivers in parallel, all sharing a positive bus that is switched by the MOSFET.
The problem with series is the diodes must be electrically isolated. They are all case neg and since I can't electrically isolate cases without thermally isolating them as well, the neg. pins will be at the same potential and parallel wiring it is! I will check the AMC driver but I wanted 550 ma. I would rathar go with a buck driver than less current.

Yeah, total brain fart there, RHD, thanks for catching it. I don't notice when the old neurons misfire anymore....

I'm very curious to see what this project becomes.

I played around in eMachineShop's software, and came up with a design for a 16 laser diode heatsink that would fit in this host:
http://laserpointerforums.com/f66/ginormous-bug-eye-100w-led-build-72191.html

In my head, I thought of running 3x cells in series, powering a driver that used 8 channels to drive 2 diodes in series per channel.

The big question in my head was how well the beams would line up, and how well they would stay aligned at a distance (and also what the heck I would do with a build that used up 16 pricey 445 diodes...)

I didn't go forward with that project

Yeah I was also curious about allignment. So I made it adjustable with the diode mount independant of the lens mount. That way I can move the diode relative to the lens which can adjust beam angle. You can see what I am doing in the emachine shop thread; I have posted those 2 parts there. The diode mount is bolted to the lens mount with 4 boltd. The hole in the diode mt is big enough for some.play relative to the bolt so there is some adjustment.

RHD how about modifying the Free Opensource Boost driver to work in buck mode? Is that a potential solution? Or should I find some other buck IC.

Jubathoph said:

RHD how about modifying the Free Opensource Boost driver to work in buck mode? Is that a potential solution? Or should I find some other buck IC.

Click to expand...

I know the IC can theoretically do boost buck, but would require two inductors, and wouldn't be a good approach.

Try the X-Drive from Lazeerer.

What do you think about something like this

LM3405 - 1.6MHz 1A Constant Current Buck Regulator for Powering LEDs

I could use the application notes to design the circuit but I have no experience with PCB layout or printing or surface mount stuff. I am up for the challenge though...especially since I have several months until enachineshop makes my parts. It would def be in keeping with the "DIY spirit" of the build.

Any thoughts?

Jubathoph said:

What do you think about something like this

LM3405 - 1.6MHz 1A Constant Current Buck Regulator for Powering LEDs

I could use the application notes to design the circuit but I have no experience with PCB layout or printing or surface mount stuff. I am up for the challenge though...especially since I have several months until enachineshop makes my parts. It would def be in keeping with the "DIY spirit" of the build.

Any thoughts?

Click to expand...

I think the X-Drive is a much more prudent route to go.

Still, if you're limited to 1x cell, then obviously space is an issue. Thus, I cannot imagine fitting 7x drivers into a host that didn't have room for simply adding a second cell....

One cell is not an option. I already have the headway 38120 battery (linked in OP) and the host designed and procured to fit it. That battery is essential for phase two which will incorporate (7) 445 diodes driven at 1.8A each. I figure this will draw around 20-25 amps from the battery which is why I need the 30A continuous current rating and also why I need the LiFePO4 battery chemistry.

In my final review you will see that this a very compact build. The whole thing is 2"x 11"

BTW the battery is a Beast. 5" long by 1.5" wide.

Do you have room for 7 different drivers?

Yep they will be heaysinked to thishttp://www.thorlabs.com/thorProduct.cfm?partNumber=SM2PL

Hate to say it, but I think you should go with 7x flexdrives.

They'll buck for you when you have your red diodes installed, and boost for you when you swap in the 445s.

Short of that, if you don't want the up-front cost of flexdrives, and want simply a buck driver, go with the X-Drive.

That's my advice at least. Cheers