Please explain the Pros and Cons of different LED driver methods

I have been itching to start designing my LED driver for 1 and 3 XM-L LEDs. I found a really good bin of them from a distributor and want to use them since I would get about 4500K color at a high Lumen rate for 700mA of current.

I was searching DealExtreme for drivers and noticed many of the boards had very few chips and some with no inductors at all. Then presto, found one the Chinese manufacture had forgot to scratch the top of the IC off for the picture.
It was an ATMEL TINY microcontroller and what appears to be a diode, some voltage divider resistors probably to power the micro and 3 FETs (7135 35A) is the markings.
So my thought on this is simply this driver is direct drive and the micro is simply using a PWM signal to switch the MOSFETs? :thinking:

But wouldn't that also mean the same "mode" on the flashlight would get dimmer as the single cell battery depleted? :thinking:
So this would not provide for constant current then through the life of the battery???

I think I can hit the ground running once I determine what is considered a "good" driver, but I was hoping to get some input on here.

My thoughts for a 3 XM-L light with the LEDs in series would be I would first need a boost circuit if running off a 2 cell 18650 in series. From the boost circuit would it be as simple as putting a micro controller running a MOSFET with a current sensing resistor to maintain constant current? :thinking:
Or since the boost circuit is maintaining a constant voltage, could I just designate a 99%, 50%, 20% duty cycle PWM signal and hope the current would remain constant due to the constant output voltage from the boost circuit? :thinking:

I have read that PWM allow the color of the LED to remain consistent even at lower outputs like 50% and 20% so what is so bad about PWM????

If it is related to flicker, then why can't the PWM frequency be higher like 120Hz, 200Hz, 30kHz, or 100kHz? :thinking:
Is there a limitation to how high a frequency the XM-L LED can be PWM'd at?

This brings me to the next method using a constant current voltage regulator. Something like this LT3080 configured to deliver a constant current like 3Amp and then tweeking it so that the constant current setting can be adjusted. Is something like this a good method? :thinking:
Datasheet: http://cds.linear.com/docs/Datasheet/3083fa.pdf
What could be the disadvantages other than the LED color shift at lower currents?? :thinking:

The last option I have thought about is using an actual "LED Driver" such as the LT3478
Datasheet: http://cds.linear.com/docs/Datasheet/34781f.pdf
This looks to have the DC to DC converter built in, but this has once again PWM dimming.... so is that bad even though it is at 200kHz or higher?

Or another chip with similar operations, the LT3755
Datasheet: http://cds.linear.com/docs/Datasheet/37551fd.pdf


Thank you for your input.

Clint

somename said:

I have been itching to start designing my LED driver for 1 and 3 XM-L LEDs. I found a really good bin of them from a distributor and want to use them since I would get about 4500K color at a high Lumen rate for 700mA of current.

I was searching DealExtreme for drivers and noticed many of the boards had very few chips and some with no inductors at all. Then presto, found one the Chinese manufacture had forgot to scratch the top of the IC off for the picture.
It was an ATMEL TINY microcontroller and what appears to be a diode, some voltage divider resistors probably to power the micro and 3 FETs (7135 35A) is the markings.

Click to expand...

It's a constant current linear regulator with about a 120mV voltage drop (350mA per chip). If you only need 700mA current, it's going to work quite well with a single 18650.
http://www.micro-bridge.com/data/add/amc7135.pdf
You could also power 3 XML in series from 3 18650 in series (just make sure to limit the supply voltage).

So my thought on this is simply this driver is direct drive and the micro is simply using a PWM signal to switch the MOSFETs? :thinking:

Click to expand...

Those drivers typically use PWM for dimming (with all the 7135 parallel).

But wouldn't that also mean the same "mode" on the flashlight would get dimmer as the single cell battery depleted? :thinking:
So this would not provide for constant current then through the life of the battery???

Click to expand...

That really depends on how much current you want. At 3A, yes regulation won't be good on a single 18650.

My thoughts for a 3 XM-L light with the LEDs in series would be I would first need a boost circuit if running off a 2 cell 18650 in series. From the boost circuit would it be as simple as putting a micro controller running a MOSFET with a current sensing resistor to maintain constant current? :thinking:

Click to expand...

You mean using the MOSFET as variable resistor? Yes, that can work, but getting it to work well (especially at low currents) will be tough. The first version of the TaskLED LFlex driver was doing that (and had some flickering issues).

Or since the boost circuit is maintaining a constant voltage, could I just designate a 99%, 50%, 20% duty cycle PWM signal and hope the current would remain constant due to the constant output voltage from the boost circuit? :thinking:

Click to expand...

Well, if you don't care too much about current accuracy, you can build a poor mans current source by having the voltage-controlled boost driver and a resistor in series with the LEDs. Note that the LEDs have a negative temperature coefficient (meaning the forward voltage goes down, when they heat up). With 3 XML in series, I would want to drop at least 2V on that current-regulating resistor. The boost regulator may not like the PWM dimming - that's going to switch between 0% load and 100% load - and react with large voltage swings.

I have read that PWM allow the color of the LED to remain consistent even at lower outputs like 50% and 20% so what is so bad about PWM????

Click to expand...

Flicker and power consumption (the LED is going to be significantly more efficient at lower currents - at the expense of the color shift).

If it is related to flicker, then why can't the PWM frequency be higher like 120Hz, 200Hz, 30kHz, or 100kHz? :thinking:
Is there a limitation to how high a frequency the XM-L LED can be PWM'd at?

Click to expand...

The LED won't care about the frequency, but there will be a limit on what the driver can do. Some drivers are quite slow and need a fair amount of time to stabilize - so they will only allow a low PWM frequency.

This brings me to the next method using a constant current voltage regulator. Something like this LT3080 configured to deliver a constant current like 3Amp and then tweeking it so that the constant current setting can be adjusted. Is something like this a good method? :thinking:
Datasheet: http://cds.linear.com/docs/Datasheet/3083fa.pdf
What could be the disadvantages other than the LED color shift at lower currents?? :thinking:

Click to expand...

Much higher voltage drop than the AMC7135, so it's not going to work with a single cell. With 2 18650s in series it will be quite inefficient.

The last option I have thought about is using an actual "LED Driver" such as the LT3478
Datasheet: http://cds.linear.com/docs/Datasheet/34781f.pdf
This looks to have the DC to DC converter built in, but this has once again PWM dimming.... so is that bad even though it is at 200kHz or higher?

Or another chip with similar operations, the LT3755
Datasheet: http://cds.linear.com/docs/Datasheet/37551fd.pdf

Click to expand...

Those guys will very likely be quite difficult to work with and get a functioning design (and stable circuit). You don't sound like you have designed a switching power converter before, so you should be prepared for a very large amount of effort to get it working properly.

BringerOfLight said:

It's a constant current linear regulator with about a 120mV voltage drop (350mA per chip). If you only need 700mA current, it's going to work quite well with a single 18650.
http://www.micro-bridge.com/data/add/amc7135.pdf
You could also power 3 XML in series from 3 18650 in series (just make sure to limit the supply voltage).

Click to expand...

I thought those 7135 were MOSFETs because I couldn't find a datasheet on them. I didn't realize they were tiny constant current regulators.

BringerOfLight said:

You mean using the MOSFET as variable resistor? Yes, that can work, but getting it to work well (especially at low currents) will be tough. The first version of the TaskLED LFlex driver was doing that (and had some flickering issues).

Well, if you don't care too much about current accuracy, you can build a poor mans current source by having the voltage-controlled boost driver and a resistor in series with the LEDs. Note that the LEDs have a negative temperature coefficient (meaning the forward voltage goes down, when they heat up). With 3 XML in series, I would want to drop at least 2V on that current-regulating resistor. The boost regulator may not like the PWM dimming - that's going to switch between 0% load and 100% load - and react with large voltage swings.

Click to expand...

I had almost forgot about the problems that can arise with an unloaded boost converter. Thanks for reminding me. I had seen that several years back once when working on a project that was doing variable rate charging on lead-acid batteries up to 20 Amps taking the input voltage from a small wind turbine.

BringerOfLight said:

Those guys will very likely be quite difficult to work with and get a functioning design (and stable circuit). You don't sound like you have designed a switching power converter before, so you should be prepared for a very large amount of effort to get it working properly.

Click to expand...

I actually thought these were 2 of the simpler integrated LED driver ICs I have found. I have built several switching power supplies both buck and boosts. I was just wondering how these compare to the "good" LED drivers out there for 3 Amps output?

Can you give me examples of what are considered some of the really good LED drivers on the market right now? I just figured after looking around at those on deasextreme, that I can build one of better quality (certainly bettery solder quality) even if it is not cheaper. It is the method the high end drivers are using I am trying to figure out.

For something more precise I was thinking about using one of those LT circuits and for something really simple I was considering using an array of LM317 chips such as in Figure 13 of the datasheet. http://www.ti.com/lit/ds/symlink/lm317.pdf But with that design I would have to probably run 3 18650 cells in series for the supply since I believe it can only operate like a buck supply.

Can you provide some information on the high end LED drivers out there for adjustable current outputs to 3 Amps or higher.

Regards,
Clint

For something more precise I was thinking about using one of those LT circuits and for something really simple I was considering using an array of LM317 chips such as in Figure 13 of the datasheet. http://www.ti.com/lit/ds/symlink/lm317.pdf But with that design I would have to probably run 3 18650 cells in series for the supply since I believe it can only operate like a buck supply.

Click to expand...

The LM317 is a linear regulator. The problem with most of those linear regulators is the voltage drop. For constant current usage, they generally need a large sense resistor in series with the LED which results in a large voltage drop.

The AMC7135 doesn't need that and you can easily run them in parallel up to whatever current you need. They also run stable without any capacitors and should be able to do at least 10kHz PWM (if you want PWM), or you could regulate current in 350mA increments by turning some off.

For 2A current, the 7135 would work fine for 3 LEDs in series run from 3 18650s in series. With the LM317, you would want 4 cells in series for decent regulation.

Can you provide some information on the high end LED drivers out there for adjustable current outputs to 3 Amps or higher.

Click to expand...

For a buck driver, you should take a look at the TaskLED b3Flex - I don't think you will be able to build something that's significantly more efficient.
http://taskled.com/techb3flex.html

BringerOfLight said:

For a buck driver, you should take a look at the TaskLED b3Flex - I don't think you will be able to build something that's significantly more efficient.
http://taskled.com/techb3flex.html

Click to expand...

I actually ran across this TaskLED a few hours ago and was looking at the b3Flex, the H6Flex, and the Hyperbuck. Glad to hear you recommend the same board as well.

I really like the B3Flex with the ability to have a status LED and the board looks high quality with the ability to attach to a heat sink. I also noticed several of their board designs use Linear tech chips very similar to those I posted on here. So I may just go with a few of their boards to start with and then later add features to the light on a separate board that doesn't need the heatsinking.

Can the driver current settings being changed after the board is received, or do the settings all have to be programmed from TaskLed before they are sent to you and then you are stuck in that setting?

Regards,
Clint

Hi Clint,

The b3Flex current settings are user selectable. The manual for the user interface is in the technical library on their website.

-Desirider.