34063 switcher as high-power LED driver

[flieslikeabeagle]

Some of the guys have been using 1-watt Luxeon LED's for night-flying, and these high power LED's present some issues. They can draw up to 700 mA, at which point using traditional resistors to set current flow becomes problematic: the resistors have to be large and heavy, and lots of power will be wasted heating them.

I'm trying to use an MC34063 switching regulator chip extracted from a 99-cent surplus-store cellphone charger as an LED driver for these LED's. The plan is to modify the circuit to get enough voltage to light three or four of these LED's in series, then modify the current limit resistor of the 34063 to set the LED current where I want it.

I have the voltage where I want it on the prototype (stepped up higher than the 12.6V provided by a fully charged 3S lipo pack), but I'm running into issues setting the current limit. What the heck is the formula for the short-circuit current limit of the 34063 voltage regulator? The peak transistor current is simply (0.3V/Rs), but that does not set the actual short-circuit current from the circuit. Anyone have the right formula?

Some of the datasheets have a formula that relates the short circuit current to t_on/t_off, but I remember having seen a simpler formula in terms of V_in, V_out, and the current set resistor Rs; but I cannot for the life of me find that formula anywhere now. Any help would be much appreciated!

Datasheets: http://www.datasheetarchive.com/search.php?q=GM34063

-Flieslikeabeagle

[Acetronics]

Hi, Beagle

The Luxeon Leds Have to be CURRENT driven and not Voltage driven ... so, you just have to place them in the high side of the voltage feedback divider and the low side will be a resistor which value is Vref / I led ...

Note a Thermistor, to reduce applied power if Led overheat is not a "view of mind" in such situation.

Alain

Alain

[fredhs]

Hello Flieslikeabeagle,

Alain is definitely right about the need for a current source. One charge pump chip especially designed for this application is called "LTC3215"; I have used it quite successfully with 3 NiCads (unfortunalety, the chip is not useful for input voltages > 4.4 V):

http://www.linear.com/pc/productDetail.do?navId=H0,C1,C1003,C1042,C1094,P98 92

The chip does not cost an arm and a leg and does not need a lot of external components. Only drawback (or advantage) is the tiny package...

Make sure you get the correct type of capacitors (see datasheet).

Fred

[Tom in Cincy]

I am using a similar circuit. http://www.rcgroups.com/forums/showpost.php?p=5828469&postcount=8

[flieslikeabeagle]

Alain, Fred, thanks for your replies.

Sorry if I wasn't clearer in my first post. I understand very well that the LED's need to be driven from a current source, and not a voltage source, due to the exponential current/voltage relationship in a semiconductor PN junction. However the 34063 switching regulator has built in current-limit circuitry: if you short-circuit the output, it turns the device into a constant-current source, keeping the pass transistor from burning out due to excess current. Think of it as similar to analog lithium chargers, which at low currents have a constant-voltage output, which current-limits and turns into a constant-current output if the load resistance is reduced beyond some threshold.

So my intention is to use the 34063 to first tweak the open-circuit voltage a bit higher than required by the LED string. When the LED's are connected, they will act almost like a short circuit across the output, due to that low dynamic resistance they have, and so the internal current-limit circuitry in the 34063 will kick in to set the current when the LED's are connected.

I took a quick look at some LED driver chips used in cellphones, etc, and all suffered from a few limitations: most will handle very limited input voltage, and none would handle the amount of current drawn by these 1-watt LED's (which can be up to 700 mA, according to the datasheets I saw a while ago). An additional bonus is that I can find the 34063 and most of the passive components it needs in 99-cent surplus automotive cellphone chargers from the local "99 Cents Only" stores. :)

The high LED current is the reason why I do not want to use plain old current-limiting resistors, because there will be a lot of wasted power, heat, and added weight due to the need for high-power resistors.

-Flieslikeabeagle

[flieslikeabeagle]

I am using a similar circuit. http://www.rcgroups.com/forums/showpost.php?p=5828469&postcount=8
Tom, great minds think alike?? :D

In fact I was one of the guys on that 99-cent UBEC thread that you referred to - Gfcermak started the thread with his Radio Shack 9V switching regulator, but the second post in the thread is from me, discussing the 99-cent cellphone charger as a BEC.

While we both seem to have had the same essential idea, one minor difference between your approach to using the 34063 as an LED driver and mine is that I changed the circuit to the voltage step-up mode, because I want to light three or four of these LED's in a series string, and I want to do this running off a 3S lipo pack in the model, so the input voltage is between about 11 and 12.6 volts. Not quite enough for four of these high-power LED's in series.

What I still cannot find is the equation for choosing the short circuit current limit resistor, the one you called Rsc. The drive transistor current is clipped to (0.3V/Rsc) amps, but that is not the same as the output current from the circuit. One of the datasheets I linked to has an equation that connects the two, but that equation contains another parameter I don't know: the ratio of on/off times of the drive transistor, t_on/t_off.

I guess I'll just have to do some experimenting on the bench. So far I have been using resistors as dummy LED's because I didn't want to risk the health of those $10 LED's until I had the circuit working, but I guess it's time to dive in and hook up the LED's.

-Flieslikeabeagle

[Chippie]

that equation contains another parameter I don't know: the ratio of on/off times of the drive transistor, t_on/t_off.

-Flieslikeabeagle

that info is already there.....its derived from the input voltage, output voltage and the forward voltage drop across the diode, taking into account the saturation voltage of the switching element..for a step up convertor which I guess you are building..

[flieslikeabeagle]

Chippie, I've been around analog electronics for a while, but am new to switching regulators, so I'm missing what is obvious to you...can you point me at the relevant equation? And yes, what I built yesterday was a step-up DC-DC convertor, thinking I'd need the extra volts to light several LED's in series. I thought I had four of the 1-watt LED's in my parts box. However, it turns out I only have three, which led to todays fun and games.

Today I spent some time with a DC power supply, a Fluke 87 DMM, and a 25 ohm, 10W resistor, measuring the three 1-watt LED's I have (a red, a green, and a blue one, all bought from All Electronics).

What I found was about 3.2 volts drop across both the blue and the green LED at 300 mA forward current, and about 2.0 volts drop across the red one at the same current. I measured voltage drops at currents from 100 mA to 500 mA for all three diodes, but without additional heatsinking, the blue and green LED's get too warm at currents much above 300 mA.

These voltage drops are lower than I had been told (I'd been told 4V - 5V per LED). The interesting thing is that I can put a red and a green LED in series and only have about 5.2 volts drop across both at 300 mA. One of the 99-cent cellphone chargers I have puts out 8 volts - which means I don't even have to modify the charger to voltage step-up mode! All I have to do is change Rsc to set the current limit, and I can run two LED's right off it. I might even be able to run three LED's in series (about 8.5 V total at 300 mA) by tweaking the voltage feedback resistors to up the voltage a little. If the 34063 will work with only a volt and a half difference between input and output voltages, that is.

I decided to try modding the charger to light two LED's, one red and one green. I have a shortage of sub-ohm resistors, but I scrounged up a 0.47 ohm resistor, and three 2.2 ohm resistors which I wired in parallel to make a 0.733 ohm resistor. Combined in series, I have about 1.2 ohms, and on a stock, unmodified cellphone charger, using this for Rsc sets the current limit (measured with the Fluke 87) at 373 mA. Close enough for what I want!

I hooked up the LED's, and yup, it works. The 34036 stays cool, the green LED gets a little warm but should be fine on a model airplane with cooling airflow, and the red LED was barely warm due to its lower voltage drop.

Cool, another good use for those cheap 99-cent cellphone chargers!

-Flieslikeabeagle

[Chippie]

Chippie, I've been around analog electronics for a while, but am new to switching regulators, so I'm missing what is obvious to you...can you point me at the relevant equation?
-Flieslikeabeagle


beagle....Sorry for my flippant remarks...to answer your question, if you look on page 13, in the step up convertor column all the calcs for the various parameters are there, just input your values to obtain the answer...you may need to make a few tweaks here and there to get preferred values for components but it should all work..

[flieslikeabeagle]

When I made measurements of the short-circuit current limit set by the 34063 chip, I found the value of this current is dependent on V_in and V_out, as well as on the sense resistor Rs. This is unfortunate, because it results in some variation in LED current with battery voltage. The variation is within acceptable limits for the LED's and illumination requirements, but is "untidy".

Alain posted a method to make the 34063 into a precision current-source, using the same feedback method that is used when making an op-amp current source:

The Luxeon Leds Have to be CURRENT driven and not Voltage driven ... so, you just have to place them in the high side of the voltage feedback divider and the low side will be a resistor which value is Vref / I led ...
Alain
The LED's I'm using are rated for up to 700 mA of current draw, and Vref for the 34063 is 1.25 V, so I was initially not enthusiastic about Alain's suggestion because of the resulting 875 mW of heating in the resistor across Vref.

However after some night-flying experiments with these LED's, it turns out 200 mA of forward current is all I need to have adequate brightness to very easily see my model in the night sky. At 200 mA, only 250 mW of heat will be dissipated in the resistor across Vref, and now Alain's suggestion looks like a great solution.

As soon as I can find some resistors in the right range (nominally 6.25 ohms for 200 mA), I will try out Alain's idea.

I also found out that the particular 34063 I have seems to require about 4 volts minimum difference between V_in and V_out, in voltage step-down mode, which is how the 99-cent cellphone chargers are configured. I cannot tweak the output voltage any higher than 8.3 V or so with an input voltage around 12.4 V. So it works fine to drive 2 LED's in series off a 3S lipo pack, but not 3 LED's.

The night flying last night was done with one green LED lighting up the underside of the right wing, and a red LED lighting up the underside of the left wing. It worked very well, but a third LED to help me see where the tail-feathers are in relation to the wing would be a good addition.

I've already made one step-up circuit, but it required removing all the components from the 99-cent charger and building up my own circuit on a small bit of proto-board. Several guys at the flying field are interested in the circuit and I don't want to have to build several...so I'm keen to use the chargers with minimum mods to the circuitry.

I might try a lower-power LED to light up the tail-feathers, one that can be run off the +5V BEC voltage with a suitable series resistor, since the current levels will be lower. Or I guess I could just parallel it with the two high-power LED's, again with its own series resistor to keep the current through the third LED suitably low.

-Flieslikeabeagle

[flieslikeabeagle]

A couple of pics of the model, lit up with the 1-watt LED's, driven by the 34063 cellphone charger circuit.

-Flieslikeabeagle

[flieslikeabeagle]

I just started a thread on this RC plane (illuminated for night-flight) here:
http://www.rcgroups.com/forums/showthread.php?t=599320#post6417233

-Flieslikeabeagle

[hsen]

Some of the guys have been using 1-watt Luxeon LED's for night-flying, and these high power LED's present some issues. They can draw up to 700 mA, at which point using traditional resistors to set current flow becomes problematic: the resistors have to be large and heavy, and lots of power will be wasted heating them.

I'm trying to use an MC34063 switching regulator chip extracted from a 99-cent surplus-store cellphone charger as an LED driver for these LED's. The plan is to modify the circuit to get enough voltage to light three or four of these LED's in series, then modify the current limit resistor of the 34063 to set the LED current where I want it.

I have the voltage where I want it on the prototype (stepped up higher than the 12.6V provided by a fully charged 3S lipo pack), but I'm running into issues setting the current limit. What the heck is the formula for the short-circuit current limit of the 34063 voltage regulator? The peak transistor current is simply (0.3V/Rs), but that does not set the actual short-circuit current from the circuit. Anyone have the right formula?

Some of the datasheets have a formula that relates the short circuit current to t_on/t_off, but I remember having seen a simpler formula in terms of V_in, V_out, and the current set resistor Rs; but I cannot for the life of me find that formula anywhere now. Any help would be much appreciated!

Datasheets: http://www.datasheetarchive.com/search.php?q=GM34063

-Flieslikeabeagle
This link has a program to calculate values of 34063 Convertidor DC-DC
http://www.neoteo.com/ltspice-iv-simulador-de-circuitos-de-alta-15470.neo