Calling All Current Regulation Masters - Unique Driver Needed

[Helios]

Its not for a laser but it is for a HUGE led.

To feed the beast I need a driver that outputs 3.5A at 30-36V!

I plan to use LiFePO4 18650 batteries - 8-16 of them
12pcs X A123 NEW original 18650 Lifepo4 battery element | eBay

I also need the driver to incorporate adjustable current either with a selector switch or better yet a variable resistor so that I dont have to use it at the full 7000 Lumen all the time.

*Bonus Point Feature*
-Integrated charging circuit for the battery pack


If someone was willing to either help me come up with the schematic or build the driver for me I would be more than willing to compensate them with some sort of trade :beer:



EDIT: I am in contact with rhd about this driver so I think I have the help I need

 

[Cyparagon]

Linear might be the most efficient in this case. The 3V dropout of a typical linear regulator is only 10% of the power in this case which beats most switching drivers.

 

[Helios]

That is what rhd suggested as well.

After a bit more research it sounds like the integrated charger is quite an undertaking... But I dont want to have to charge 12 batteries individually! Any ideas?

 

[rhd]

I have one strange thought -

What if between every single battery in series, you had a DPDT switch. In one position, the + and - of two batteries would connect (creating a series setup), and in the other position, the + would connect to the + of each other switch assembly, and the - would connect to the - of each other switch assembly.

Effectively, you'd have 12 switches, and when you switched them all in one direction, you'd get a series config, and in the other direction, a parallel config.

The benefit would be that in parallel config, you could use ONE charger circuit harvested from a decent li-ion charger to charge all the batteries at once. It would take a while, but if you got one of those rapid-chargers, it might not be too horrible. (6 hours?)

 

[Helios]

I have one strange thought -

What if between every single battery in series, you had a DPDT switch. In one position, the + and - of two batteries would connect (creating a series setup), and in the other position, the + would connect to the + of each other switch assembly, and the - would connect to the - of each other switch assembly.

Effectively, you'd have 12 switches, and when you switched them all in one direction, you'd get a series config, and in the other direction, a parallel config.

The benefit would be that in parallel config, you could use ONE charger circuit harvested from a decent li-ion charger to charge all the batteries at once. It would take a while, but if you got one of those rapid-chargers, it might not be too horrible. (6 hours?)

WOW thats some out of the box thinking! I like it.

What would happen if you switched the LED on in parallel config? The low voltage would just stop the LED from lighting? I dont think there would be any damage so theres no chance of an accident.

Ive seen very high current chargers for lifepo4 batteries so that could really help.

 

[rhd]

I don't think anything would happen - though you'd probably have bookend switches that would effectively kill the feed to your driver setup anyway. I guess you'd need 13 switches for 12 batteries. It's a shame you couldn't find a 12P2T switch

- But, I think I have an easier solution.

Buy 1x or 3x of these:
Smart Charger for 4S 14.4V Li-ion / Li-po Battery packs | eBay

Then wire a DC jack to each group of 4 cells in series. It shouldn't matter that each group is also in series with the other two groups of 4. You could either plug in this charger, wait for the charge, and then move it to the next group, or you could probably do all three groups at once with 3 chargers.

If the latter, you *might* want to put a simple toggle switch between the groups of 4, so that you can break their connections with each other, but I'm actually not sure if that is at all necessary.

 

[chefla]

For best efficiency I would strongly recommend using Pulse-width modulation (PWM) for the brigthness adjustment. A simple resistor would get freaking hot and a be a pain to heatsink.
I use PWM for all my LED builds together with a microcontroller for a linear brigthness adjustment. You could also use an arduino or basic stamp which are easier to programm. The programms can be found online. There are LOTS of support forums on the net.
If you don't want to use a uC you could also use a NE555 for PWM. The adjustment via Poti will be not linear then. If that is not a problem, then this design is much easier to accomplish.
I would use the PWM together with a LED buck driver which has an enable pin. Then you can easily hook up your PWM circuit to this driver and enjoy the best battery runtime possible.

 

[rhd]

For best efficiency I would strongly recommend using Pulse-width modulation (PWM) for the brigthness adjustment. A simple resistor would get freaking hot and a be a pain to heatsink.
I use PWM for all my LED builds together with a microcontroller for a linear brigthness adjustment. You could also use an arduino or basic stamp which are easier to programm. The programms can be found online. There are LOTS of support forums on the net.
If you don't want to use a uC you could also use a NE555 for PWM. The adjustment via Poti will be not linear then. If that is not a problem, then this design is much easier to accomplish.
I would use the PWM together with a LED buck driver which has an enable pin. Then you can easily hook up your PWM circuit to this driver and enjoy the best battery runtime possible.

I don't believe Helios was suggesting that he would place the resistor in-line with the diode's current path

 

[chefla]

I have seen strange designs in my days... *Where's that funny smell coming from* .. *Oh look, my IC starts smoking!*

 

[HIMNL9]

A pair of 3083 in parallel will give you all the current you want, with the possibility to regulate it with a simple trimmer or potentiometer (linear drivers).

And for the charging, til you don't change the banks, it will not be a problem, charging all the pack together ..... just use slow charger, and be sure to get all the cells from the same production lot.

 

[Cyparagon]

Effectively, you'd have 12 switches, and when you switched them all in one direction, you'd get a series config, and in the other direction, a parallel config.

That sounds like it'd blow them up. Maybe I misunderstood, but... you can't switch them all at the exact same time, and so at one point you'll have 11 cells in series trying to charge one cell in parallel with them.

 

[rhd]

You'd switch them all, before connecting a charger. But you wouldn't need to literally switch them at the exact same instant. You could switch them one after the other.
Here's what the switch would look like:

 

[chefla]

A pair of 3083 in parallel will give you all the current you want, with the possibility to regulate it with a simple trimmer or potentiometer (linear drivers).

The LT3083 is rated up to 23V only. So in this case that won't work since Helios is looking for an output of 3.5A at 30-36V. I have not found one linear driver which is capable of providing this output voltage. Have you?

 

[Cyparagon]

I see, thanks for the diagram. Well, that'd be a hell of a lot of wiring. I'd rather just take them out and put them in a charger, myself

Which 100W LED are we talking about, anyway?

I have not found one linear driver which is capable of providing this output voltage. Have you?

Our favorite LM series can.

 

[Helios]

There seems to be only one core design. I cant figure out if its made by multiple sources or just one because there is some variation but not much between any of the LEDs I have seen.

http://www.future-lamp.com/download/hpled/RG-DGL-100C-CW_WW_datasheet.pdf - PDF

I ordered the warm white because I prefer the vivid colors you get with warmer color temps.

 

[HIMNL9]

The LT3083 is rated up to 23V only. So in this case that won't work since Helios is looking for an output of 3.5A at 30-36V. I have not found one linear driver which is capable of providing this output voltage. Have you?

Uhm, you're right, sorry ..... i was referring to the high voltage configuration from the datasheet (page 21), but i realized right now that the configuration is for lower the INPUT high voltage, not for have it at the OUTPUT ..... my bad.

It can probably be made, adding an extra mosfet in a different configuration, but this can require a different circuitry ..... must think better about this.

At this point, the better solution can be a simple sink current driver, using a power mosfet and an op-amp, and a very low Rsense for limit the dropout ..... i made in the past some drivers for led banks (20 x 10W in serie for an aquarium) that was working at 200V for 1A ..... have to check the circuit for see if it can be modified for work for 4A or more at lower voltage, but, once you know the min and max input voltages, i don't see too much difficult, in this .....



E£DIT:

There seems to be only one core design. I cant figure out if its made by multiple sources or just one because there is some variation but not much between any of the LEDs I have seen.

http://www.future-lamp.com/download/hpled/RG-DGL-100C-CW_WW_datasheet.pdf - PDF

I ordered the warm white because I prefer the vivid colors you get with warmer color temps.

I had that suspect ..... was just for ask you if you was making a "flashlight" with one of these ones