LED Power Supply

OwenS

I'm currently in the process of designing a Propeller based lighting instrument, based upon a trio of RGB LEDs (1W per channel per LED, for 9W power total). The design is overall powered from an unregulated 18V DC supply, and my intention is currently to power the Propeller and other sensitive electronics from a linear regulator, and use the Propeller to provide the LED power supply. The design means the power load can vary wildly and rapidly; therefore the power supply will have to be relatively resilient and will also have to be fast acting.

The LEDs are 3.3V and common anode; my question is how best to supply them with the power they require and regulate it. I'd prefer a software solution; particularly one which can give me statistics back to report to the user (Along with housing temperature information).

Philldapill

Are all the LED's in parallel? i.e. - are all the reds in parallel with reds, blues with blues, greens with greens? OR, are you controlling each individual color of each LED seperately? If they are seperate, TI makes some pretty good LED controller chips that are I2C, and use PWM to control brightness. They are a "set it and forget it" type of chip.

OwenS

While theres no need for them to be in parallel, it's pointless to dim them individually. In any case, TI's most powerful LED controller only does 500mA - my LEDs are 300mA each, so I would need 9 of them.

Philldapill

OwenS, if your LEDs don't share a common cathode or anode for each color, you can use this circuit(attached) to control the color of each individual color of the LEDs. The circuit consists of 3 LED's chained together in series, and 3 groups of these series chains. It might be better to use an N-channel MOSFET in place of the NPN BJTs due to current handling ability. The circuit shown should be for EACH of the three colors, so duplicate it three times for R, G, and B.

Again, this is assuming there are 6 connections on each LED module, so that each color doesn't share any common connections with the others. However, if the LEDs share a common cathode, well, you'll have problems.

With this setup, you should be able to drive the whole set with 3 I/O pins, using PWM to vary the intensity of each color. Each color would be driven from a seperate IO pin.

OwenS

As I said in my initial post, they're common anode.

Timothy D. Swieter

The solution Philldapill proposes works with common anode. I agree that common cathode is difficult to work with.

Using the TI chips is also possible even with the larger current draw of the 1W LEDs. There are apps notes on TI's site for taking the output of their IC and using a MOSFET for controlling higher currents. However you could just connect the Prop to the mosfet and do something similar if the Propeller has free cogs/code space and I/O in your application.

At the moment I am prototyping a product that has 8 groups of red, green, blue LED controlling circuits. The brains of the device is the Propeller. There is 24 outputs, one to each LED. The Propeller will do all the PWM or BAM for controlling the intensity of the LEDs. The Propeller is switching a n-channel MOSFET so that each channel can control up to 7.5A. The product, which I call Illuminator-CA (where the CA stands for common anode) has a switching regulator to the electronics portion. The voltage that is feed into the device is routed directly to the LED voltage. The device will accept DMX512-A for control or have built in patterns. Eventually there will be an interface where users can design their own patterns and store them on the device. I attached an image of the prototype, but the final product will change slightly. I am going to be selling the prototypes, though at a cheaper price than the final product. Let me know if this device is of any use and we can work out a deal.

So - did you get your question answered OwenS? If not, let us know what you are looking for still and we will try our best to up.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.
www.brilldea.com - Prop Blade, LED Painter, RGB LEDs, 3.0" LCD Composite video display, eProto for SunSPOT
www.tdswieter.com

Philldapill

OwenS, if Timothy's response didn't fully clarify, you can always modify that diagram I attached. What you would do, is provide a transistor for EACH LED component(each LED and each color for it). Instead of having three LEDs in series, you would have individual LEDs in that chain. Also, since it is common anode, you would put the resistor between the LED and the transistor, connecting the anode of the LEDs to the +18V. Then, connect all the transistors together via the 1k resistors as shown. You will need to increase the resistor value, however, since it will need to drop more voltage.

Nick Mueller

> The Propeller is switching a n-channel MOSFET so that each channel can control up to 7.5A.

It looks like you are going directly to the FET's gate (with an R inbetween). Could you please tell me the type of MOSFET that does 7.5A with an Ugs of 3.3V?

TIA,
Nick

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!

The DIY Digital-Readout for mills, lathes etc.:
YADRO

Timothy D. Swieter

Nick - For the Illuminator-CA I am using the Fairchild FDS6910 www.fairchildsemi.com/pf/FD/FDS6910.html. The FDS6910 is actually a dual n-channel MOSFET. I am still learning some about using the MOSFETs since I haven't done much designing with them. So far the IC and circuit looks to be working, but I have only powered the system on and let is cycle for a couple hours without much stressing. I probably need to get into the details with the scope to watch a couple signals and do some more calculations.

In my circuit the Propeller output is taken straight to the gate. Theres is a pull down resistor on this I/O to ensure the MOSFET is off and not floating. There is no resistor between the Propeller output and the MOSFET. I know the circuit seems so simple, I keep thinking I might be missing something. What do you think?

Sorry OwenS, I don't mean to take your thread off topic. Jump in if you have more questions and need help.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.
www.brilldea.com - Prop Blade, LED Painter, RGB LEDs, 3.0" LCD Composite video display, eProto for SunSPOT
www.tdswieter.com

Nick Mueller

If you look at the datasheet, figure 2, "On-resistance variation with drain current and gate voltage", you'll set that you wont get happy with a Ugs (or Vgs, it's the same) of below 4V.
Typically, you have an 10R .. 20R going to the gate. This reduces switching speed a bit (RFI; scope for overshots) and limits the current from the prop while switching on/off. You are charging/discharging the gate. 1nF typically according to the data sheet.
Better use a gate driver.


Also, I don't know how you drive LEDs with constant voltage and without an resistor. But maybe your LEDs have something built in.


Nick

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!

The DIY Digital-Readout for mills, lathes etc.:
YADRO

OwenS

The two problems I have with the transistor and fixed resistor based circuit is that, firstly, it's an 18V unregulated supply and 18v is only nominal, and secondly, at 300mA current per LED, each resistor is going to have to drop 15V, and have a power dissipation of 5.4W. 9W waste heat from the LED array is bad enough; another 48W from the resistors is just unacceptable, and thats before you've added all the other heat generating components (Propeller PSU, SMPS for servos, transformer)

The other big problem is that Incandescent light bulbs are beating us in the efficiency game here!

Post Edited (OwenS) : 7/1/2009 3:35:40 PM GMT

Phil Pilgrim (PhiPi)

To get any efficiency at all from that arrangement, you will need a buck regualtor (DC-DC converter) to regulate your voltage down to a sane level. Properly designed, these can operate at 90+% efficiency. Then, the circuits suggested above will be more realistic to consider.

-Phil

OwenS

I suppose the best course of action would be to build a Propeller controlled (I still have quite a few free cogs!) buck converter with 5V output to power the LEDs. 5v is useful because it is also the voltage required by the servos in the design. The MOSFET drivers used will be powered from the 18V rail (They can take it). The whole system will be protected by a 2A self resetting polymer on the 18V rail at it's entry. The aperatus will, of course, include a cooling fan. The Propeller will be programmed to regularly sample one or more temperature sensors in order to detect any components overheating.

Can anyone suggest a method of voltage protection? I was thinking a zener diode between the 18V and 0V rails, in order to trip the fuse?

Nick Mueller

Or (maybe this is a stupid answer) a supressor-diode. They can stand high currents for a short time (until the fuse blows).

Nick

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!

The DIY Digital-Readout for mills, lathes etc.:
YADRO

OwenS

In the time between our posts I looked it up and discovered the crowbar circuit [noparse]:)[/noparse]

Timothy D. Swieter

@Nick - Thanks for the review and feedback. With the MOSFET circuit I am describing the LED do include the current limiting resistor. It is the other circuits, the ones using the TLC5940 that are current controlled devices. For the next revision of the Illuminator-CA I can add a series resistor from the Propeller to the MOSFET gate, I will have to try this out on the prototype and see what effect it has.

Thanks for pointing out figure 2 in the data sheet. I reviewed that for a few minutes and I believe I understand it better. If I am driving the gate with 3.3V, then the on-resistance will be between 1.5 and 2.2 and probably closer to 1.5. I take this to mean that the MOSFET will dissipate more heat with higher current loads because the resistance isn't as low. I could therefore say the specification of the Illuminator is less than the actually MOSFET current driving capabilities to account for the higher on resistance.

Maybe in another thread or PM you can tell me more about gate drivers. I don't fully understand why it would be needed - again, I am just now researching and designing with MOSFETs so it could be something obvious I am not understanding.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.
www.brilldea.com - Prop Blade, LED Painter, RGB LEDs, 3.0" LCD Composite video display, eProto for SunSPOT
www.tdswieter.com

Nick Mueller

> If I am driving the gate with 3.3V, then the on-resistance will be between 1.5 and 2.2 and probably closer to 1.5.

Typically.
The problem with MOSFETs is, that they all need at least about 4V on the gate to fully open. That's why I asked. Now if you drive them with 3.3V, you can't charge the gate fast enough. So there's a certain time while switching when the resistance is high. Power loss, heat.
The other thing is, that you should not switch on/off too fast. Or you will get RFI with a lot of harmonics and the FCC (or how they are called) ringing at your door bell.
With drivers (comparators would work too), its easier to get a high gate voltage, and supply the necessary currents.

Oh, and sorry, it is Figure 4, not 2. -> A Vgs of 8V would be desirerable.

Nick
Nick

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!

The DIY Digital-Readout for mills, lathes etc.:
YADRO

OwenS

The efficiency is gonna be a lot less than you expect without proper drivers. MOSFETs are highly capacitive, and the Propeller is going to have trouble discharging/charging the capacitance quickly.