3 XM-Ls, 18 AMC7135, 1 MCU and a 12V Deep Cycle Battery

Hi all,

I'm finally getting around to doing something with 6 XM-L T6 emitters I have spare from when I dismantled an outdoor light I built for camping. Funny enough I'm rebuilding the light, but given my new-found ability to use eagle, I'm design a PCB purpose built for the task.

I want to use as much as I have already in my 'spares bin' and given I have around 300 AMC7135 chips handy, I wan't to use them. I also have a handful of 7135 KD drivers which I plan to bastardize and steal a few bits and pieces from.

The whole thing will be powered from a 12V deep cycle battery I have. The plan is to connect 3 groups of 6 parallel AMC7135 chips in series, with each group of six chips powering a single emitter. I want to control all of this using a single ATiny13 MCU (so I can use the various firmwares floating around these parts to set up the levels perfectly).

Now maybe because it's late, or maybe because I am just dumb, but I'm really struggling to work out the best way to set the circuit up. The way I see it, the 3 groups of 6 AMC chips will see approx. 4V each if connected in series, which is perfect. But what is the best way to incorporate the MCU into all this? I was thinking a voltage divider circuit and have the MCU running in parallel with the LED circuit, but I'm pretty sure there is some law of electronics that is going to blow the MCU up if I do that. Then again if you look at your typical KD circuits, the MCU is effectively running in parallel with the 7135 chips on those boards; so maybe there is nothing wrong with this set up?

My second problem is that I'm not even 100% sure how to connect the 6xAMC and 3 LEDs in series. I'm so conditioned to hook the 7135 chips in parallel that I can't conceptualize how to connect them in series! Can it even be done? Maybe I need to do some doodling...

I'm going to hit the sack and hopefully wake up enlightened. If not, maybe someone here can help kick-start my brain


- Matt

But am interested to see how this turns out. Perhaps you should PM Pilot PTK, he’d be able to tell you what you’re asking.

Good luck!

Not sure how to connect series, but you may be able to get away with them in parallel if the total Vf of the series emitters is close to the battery’s voltage. I know that it has been show that the 7135’s can handle higher voltages under such condition, but I don’t recall how high you can go.

It might be worth trying this, a while back I succesfully ran two LEDs (in series) off a stack of 8 7135 chips driven by 2 18650s (in series) for over a week straight with no problems.

The current-carrying part of the 7135 doesn't care what the total system voltage is, only the relative difference between the input voltage and the LED's (in this case, combined) forward voltage. System voltage is irrelevant as the 7135 doesn't have any connection to the positive side of the input voltage.

The voltage applied to the 7135's Vdd pin depends on the working voltage of the MCU, so if you do one of the workarounds to drop the MCU's voltage to something it can survive, that takes care of the Vdd voltage issue at the same time. In something where you have the space for extra components (i.e., not a flashlight) best solution for the MCU voltage thing is probably a 5v voltage regulator (7805 or the like).

as long as the Vf is close to Vf, the 7135 chips should be okay, at least up to 3S I recall (don’t know if RBD ever did a 4S set up). To run the MCU without frying it, you can power it off of the +ve pad of one of the LEDs.

I think comfychair has nailed it. I also found this:


I’m basically doing this but without using a KD driver – I’m reworking the board to all be integrated into one solution. The AMC7135 chips regulate the current to ground, so my system would only (in theory) need 6 AMC7135 chips in total:

V+ >> LED >> LED >> LED >> 6x7135 >> GND

The LEDs are all technically operating in direct drive, but the 7135 chips regulate the total string current to 2100mA.

However as pointed out previously the 7135 chips care about the relative difference between VIN and the combined LED Vf. If each LED is dropping 3.24V at 2.1A, then that means by the time the 7135 chips come into play the relative difference between VIN and Vf is 2.24V. At 2.1A that means the AMC7135 chips have to burn off 4.7W as pure heat. That’s bad. I can fix this with a series resistor to drop the voltage a little bit. To ensure sufficient overhead I’ll go with a 0.5Ohm resistor? This leaves the 7135 chips with around 0.8V overhead and 1.7W to burn off, with the resistor burning off the other 3W. Obviously I'd have to spec the resistor proeperly.

The last thing I don’t fully understand about the Poorman setup (and this would be basic electrical theory) is what voltage is the MCU on that 8x7135 driver is seeing? As per diagram 1, the MCU is technically in series with the first 2 LEDs, and in parallel with the last LED. Using the combined Vf of the first 2 LEDs, that leaves around 5.52V going across the MCU (when the LED string current is 2.1A). When the LEDs are running in low mode (lets say Vf of the 2 LEDs is 5.4) the MCU will see 6.6V. The problem is the Tiny13V MCUs only operate up to 5.5V…

In the Poorman method, the P7's sudfficiently drop the voltage for everything to run OK. With the XM-L method, the resistor would have to be placed before the MCU to ensure that the required VIN is not exceeded.

Does this all sound right lol? Or am I waffling?

- Matt

You're waffling. :p Problem solved with a 7805 regulator. If you're doing your own PCB for this it should be dead simple and there's no worries about what mode it's in or what resistor will work over the full range of the battery's input voltage. It'll always only ever send 5v to the MCU no matter what.

The key to picking the right LED configuration is to figure out how low you want to drain the battery, and then set up the LEDs so that the combined total Vf will be slightly below the lowest input voltage, so it's regulated through the full discharge curve.

I have 7135s burning off more voltage than that, running a 6v MTG2 from 8.4v input and they handle it just fine, and they're still on a normal 105C PCB with no additional heatsinking. Keeping them cooler is always better if you can do it, but they will probably be fine if you can't.

Expect problems… when the AMC chips are turned off, very little current flows through the LEDs. At no current the Vf of an LED can be very small (I have see less than 1.4V drop across them).

So 12V-(1.4V*3) means the Vout pin can see around 8V (actually even more since the charged battery will be at around 14.4V, not 12V). AMC chip is rated at 6V…

You cannot depend upon the Vf of the LEDs to produce a reliable voltage drop… they are not zener diodes.

Yes Comfy lol. I will use a 7805 for simplicity sake. Makes sense to do it anyway! I can grab one for $1 on my way home from work, though this is the TO-220 case which sucks but I'll MacGyver it onto my board.

Texas...do you have any recommendations as to how I could set this circuit up? Or do you think it's a bit of a lost cause with the AMC7135 chips?

Will the more-than-6v-spec thing kill the 7135s even when there's no current flowing? And just cause I'm a dummy, that's the 6v difference between source voltage and forward voltage, right, and nothing to do with the circuit's input voltage directly?

Yes, it can easily kill the chip.

The 6V number is the volts between the chip ground pin and any other pin.

Ah. So... they would be fine with these working voltages in a clickie-switch application where input power is completely broken, as opposed to the soft-off type that just removes Vdd signal and leaves +/- connected at all times? :shy:

connect 3 LED series and it it wont be problem.

for maximum input voltage for 7135 ?? There is no Maximum Drain-Source voltage of internal FET marked in datasheet. I think 7V is supply voltage to chip. not the internal FET.


Ah. So... they would be fine with these working voltages in a clickie-switch application where input power is completely broken, as opposed to the soft-off type that just removes Vdd signal and leaves +/- connected at all times? :shy:


Well hold up - this is an outdoor light connected to a Deep Cycle battery, but the whole thing is still turned on and off via a switch. This switch WILL completely remove power from the circuit when the light is off. It’s no different electrically to our flashlights – just a different case.

Now the whole deep cycle battery voltage thingy...when the car is running it's up around 13.8V to 14.4V. When the car is off and the battery is 'resting' it's 12.6V and fully charged. That’s at no load either. The battery is dead at 11.7V, though ideally you don't want to discharge a deep cycle battery past 50% to prolong it's life. I have no intention of running this light with the engine running, so my system calculations should be worked out on 13v max (to add some head room).

According to the XM-L datasheet, LED turn on is at about 2.65V. That’s an LED string voltage of 7.95V, leaving 5.35V to be dropped by the 7135 chips*.

If I still use the 0.5ohm series resistor to take some of the strain off the 7135 chips, use the 7805 regulator to provide power to the MCU, and heat sink everything properly, this should work.

Of course, as usual I could be missing something fundamental…

- Matt

* That’s absolute worst case – for starters I used 13V as VIN, and secondly I have no intention of running the LEDs that low. Minimum string current will be around 500mA, which equates to a string voltage of approximately 8.4. So realistically the voltage dropped across the regulators will be closer to my original 4.2V.

Ok, so I threw this together quickly. I still need to work on component layouts a bit, and neaten up some tracks, thicken a few up a bit, add thermal vias to the back plane for the 3 LEDs, and last but not least add some screw holes for a nice tight fit to the heat sink. The size of the PCB is exactly the same size as the face of the heat sink this will be screwed to. I don't plan on using any optics.

The 2 large resistors or 0.25ohm 3W resistors in series with the LED string to drop the voltage difference for the AMC7135 chips.

D1 and C1 are hangovers from other circuits I've done. D1 is a reverse polarity protection diode - I'm not sure if this is required now that power is coming from a voltage regulator...

Everything else is pretty self explanatory I think.

Anything I need to specifically pay attention to? Anything I've got horribly wrong?

- Matt

I love the board name :slight_smile: Personally I’d rethink the “no optics” thing unless you’re going to sit the LEDs behind some translucent plastic. I’ve built a bunch of work/task/camping lights as well as dome lights for my car and I’ve found bare LEDs to be pretty unpleasant with lots of glare. Even some super wide angle optics or some opaque plastic make the light a whole pleasant. That’s somewhat subjective however and it might not bother you in the slightest, but worth bearing in mind nonetheless.

White plastic pill bottles work great as diffusers. Unbreakable, come in all different sizes, even easy to cut up if you need to make a 'lens' instead of a bottle shape.