High Power LED Driver, 9 amps

[Hetore]

Ok to answer the questions:

This is a true current source. It is a buck topology converter. There will be some ripple current but not much and it will be at a very high frequency. To make a converter handle this much current requires some larger components. All these components tend to be square in shape. So a round board is a little harder. I did some quick layout options and the smallest diameter I can go it 1.2" maybe a little bigger. I will work on that and let everyone know. Board coverage is already VERY high.

Efficiency should be somewhere between 80-90%. Its hard to calculate and until I build some and measure it that is as close as I can get. But I sacrificed a little efficiency for size.

I don't think I can get the price any lower. Because of the high current the cost of components goes up. Also building low quantities drives the prices higher and I don't plan on the demand being high enough to justify true production runs.

I think with some component changes we could drive at least 3 SST-90s in series.

Time frame has now been pushed out by needing a new board layout 😉

I will work on the new layout and let you know. The design is finished. Boards I get from a prototype house and take about 3 days to turn. All components are in stock at various suppliers and should take about a week to get here. Then I need to build and verify the design. Everything said and done we are most likely looking some time in Jan 2010.

I don't have any of these LEDs for testing. I'm just going to drive the power into a few power diodes I have that can generate the same Vf at that current for testing. There cheaper if I blow them up.

Hello Sir, would this heat-sink work(not melt the leds) for three SST-90 LEDS?

http://www.thermaltakeusa.com/Product.aspx?C=1152&ID=1586

OR THIS:

http://www.arctic-cooling.com/catalog/product_info.php?cPath=2_&mID=108

Thank you very much
[email protected]

 

[PhotonFanatic]

I certainly hope this come to fruition, and when it does, I would like about five of them. 😀

 

[monkeyboy]

The driver will be able to do this just fine. However in this case I may need to change some other components out because of the lower drive current.
Let me get the first few built and then I will adjust the circuit for your needs. It may cost a few more bucks because I need to order some other parts. But it will drive just fine, the efficiency will be higher than the 9 amp 3.6 volt version as well.

I see, so it's not possible to have both 1-3A low mode and 7-10A high mode on the same board?

 

[fppf]

I see, so it's not possible to have both 1-3A low mode and 7-10A high mode on the same board?

I will need to test how stable the lower drive currents will be. That is a 10:1 deranging of the design. I would not drive below 2 amps at all. This is just a function of how this controller works.

 

[fppf]

Well folks, good news
Preliminary testing went good, the board gets really warm but you can still hold on to it. So I estimate operating temperature in free air at 70 degrees to be about 110-115 degrees. This is when driving 9 amps.

More testing is needed and I will keep everyone updated.

 

[jamesavery22]

I'm in for one as well.

 

[ma_sha1]

I am in for one too

 

[fppf]

Well folks, good news, bad news.

Good news first, the power drive section looks good and robust. I even over drove it by a bit and nothing blew up. Thermal looks good, even in a housing, though I would not put this in a plastic housing. Checking operating temperatures in your housing will be important. Ambient operating temperature should be able to go as high as 50-60 degrees C without overheating the components.

Now for the bad news, really bad news.
The controller chip specs where written by a sales person. The chip is not controlling the way the specs says it should. After double checking my calculations and the app notes I'm not able to get the control I should be getting. It works, but it has poor current regulation over the input voltage range. Current regulation across output voltage is good. There is also some really funky stuff that happens close to shutdown voltage. The chip probable works fine for much lower current systems, but falls down in this application.

So, where do we go from here.
Well, seeing the very high demand I will put the effort into a custom controller. First steps will be to use one of my demo FPGA boards to write and test a custom control system for the current drive section. Then after I have a good control scheme I will source out a micro and other parts to do custom control. This will also give what many have been asking for in the means of dimming and adjustable current levels.

However, this is going to take some time. I will work as fast as possible and keep updates as the work progresses. The price point could also be in flux. I did not want to use a separate micro because of all the support items it needs, which increase part count, cost, and build times.

If it was easy everyone would do it, right?

 

[Fulgeo]

Hey fppf,

I wanted to put my vote in and say a 2 or 3 level driver would rock. I have all the parts I need to fashion my SST-90 build now except for the driver. I am really interested in making a SST-90 flashlight that has a low setting of 3.2A and a high setting near the emitter's top of 9.0A. In you last post I think you hinted that you might have to go back to the drawing board as it were. You going to dust off that idea you had of using a pic chip? Also not to infer the obvious as inspirational but if you have already purchased alot of the components for the earlier design that you can not use now perhaps you could de-tune the design to drive the SST-50? Not sure of the demand but food for thought. Anyhow wanted to tag this tread and follow it. Good stuff!

 

[fppf]

I have not had to totally redesign. The power drives are good, I actually cranked 15 amps through them and they held good, but hot.

Using a micro poses a lot of challenges due to sizing and costs. The controller I wanted had everything on one package and was reasonable priced. With a micro you need regulators, op amps, Mosfet drivers, and a whole slew of other support components. Each one of those will push cost up, and board space was already a problem.

I never buy lots of parts until I know a design is proven to work. So I don't have many wasted parts. I do have a bunch of boards, but I will just need to eat that. I will see if I can adjust the current design to work for the SST-50. The controller might be stable with some adjusted parts.

I'm still working on interfacing with my FPGA board. I use this because its faster to develop on. Then I can move into a micro if I can find one that fits the mold.

 

[HarryN]

Hi - I have been quietly following your progress. Now that the original controller has failed - would you mind telling us which one it was ? If not, it is ok - I can understand the value of your effort.

Thanks

HarryN

 

[fppf]

Hi - I have been quietly following your progress. Now that the original controller has failed - would you mind telling us which one it was ? If not, it is ok - I can understand the value of your effort.

Thanks

HarryN

Sure, in hopes to avoid someone else wasting there time. LM3409

 

[HarryN]

Thanks for the info. While I applaud your effort to pull this off - I am amazed at the extent required. A custom FPGA design for a driver? - wow.

 

[fppf]

It does not need an FPGA.
I'm a systems integrator by trade, I have demo FPGA boards laying around. So the simplest way for me to prove out a control design is to use one of these boards and the many routines I have already developed.

After the design prove out I will move the code into a micro for the final project. The control is pretty simple, which is why I'm very surprised that controller did not work well. I guess it would be ok if your input voltage range was really small and you had external drop out control. It did function and it did drive the circuit, just not to my liking.

 

[blesbok]

Interesting... I also have a design for a bike light controller based on the 3409 (Vin = 15-21V, Vout ~13V @ 1A) and I'm just about to order a few boards. I also have a small micro for voltage/temp monitoring, strobing, etc. Based on your experience, do you see any obvious issues with this scenario?

I also recently used a 3409 eval board in a dive light (Vin = 21-29V, Vout ~19V @1A) and it seems to work well, but to be honest I haven't actually measured the line/load regulation.

 

[fppf]

Its hard to say.
Your trying to use the chip more in lines of the data sheets. I was trying to get high drive currents at low Vf. I think my problem was with the usage of Vf to control the constant off time like the refrence designs depict. I tried using a steady reference voltage and it made the chip behave much better. But it still was not to my liking.

I find if you stay close to the data sheet examples most things work (except for maxum, there sales staff write data sheets). I designed a the circuit to meet all the noted parameters but I think there are a lot of little details they left out which is causing the instability.

 

[fppf]

Folks
Just an update...
I think I found parts for a new control set. I will be ordering some parts for prototype and will go from there.

 

[stub]

:twothumbs

 

[Techjunkie]

subscribed

 

[paetzi]

Any new information about your project?
What´s going on?