I've been driving an RGB LED array using a few pre-biased PNP's and a simple current limiting resistor scheme for a while. While it works just fine, it does generate a bit of heat due to the inefficiency. Recently, I've been looking at LED driver IC's in order to increase the efficiency and reduce power consumption. It seems that there are as many driver choices as there are LED's to drive.
I'd like to hear from anyone who has had hands on experience with any particular device, with the hope that I can narrow down my search to slightly less than the umpteen available units.
The circuit is simple, using the ZX to output a PWM signal for brightness control, and a digital output for on/off.
So far I've looked at devices from Linear, Maxim, National Semiconductor, Analog, Fairchild, TI and others. I'm just having a bit of trouble deciding exactly what I do and don't need insofar as features. Obviously, I have little experience in this area.
-Don
LED Drivers
Re: LED Drivers
You might look at the ULN series, e.g. ULN2803A. Although the current capacity of these might be far in excess of what you need they may work well as a general purpose driver.Don_Kirby wrote:Recently, I've been looking at LED driver IC's in order to increase the efficiency and reduce power consumption. It seems that there are as many driver choices as there are LED's to drive.
- Don Kinzer
Thanks Don. I hadn't thought about using darlingtons. You're right though, a bit more than I need.
Yesterday, I happened to be looking at voltage regulators. There are many of them with an 'enable input' that allows a TTL level signal to turn on/off the output voltage. I'm considering using multiple adjustable regulators in constant current mode for each group of LEDs. It would do away with resistor biasing, and if the on/off switching times are fast enough, would allow my PWM signal to vary LED intensity directly. Most datasheets state switching times on the order of 40uS, which should suffice. I'll have to look at the output on the 'scope to know for sure though.
I haven't found the exact part to use just yet, still sifting through the datasheets. Adjustable output, 100ma current, SOT-23-5 or similar, max Vin of 32V or more, min Vout of 2V or less. If I can find the right one, it'll allow me to reduce the board size, as well as distributing the heat over a larger board area compared to my origional design using 1 large regulator.
Any suggestions and/or comments?
-Don
Yesterday, I happened to be looking at voltage regulators. There are many of them with an 'enable input' that allows a TTL level signal to turn on/off the output voltage. I'm considering using multiple adjustable regulators in constant current mode for each group of LEDs. It would do away with resistor biasing, and if the on/off switching times are fast enough, would allow my PWM signal to vary LED intensity directly. Most datasheets state switching times on the order of 40uS, which should suffice. I'll have to look at the output on the 'scope to know for sure though.
I haven't found the exact part to use just yet, still sifting through the datasheets. Adjustable output, 100ma current, SOT-23-5 or similar, max Vin of 32V or more, min Vout of 2V or less. If I can find the right one, it'll allow me to reduce the board size, as well as distributing the heat over a larger board area compared to my origional design using 1 large regulator.
Any suggestions and/or comments?
-Don
Just an update for anyone intersted.
I've looked at white LED driver chips, as well as various contant current schemes. All of them offer higher efficiencies than the simple bias resistor method typically associated with powering an LED. Unfortunately, my space constraints are particularly severe, and the inductors and other components required to impliment a higher efficiency design simply require too much PCB space.
There is a somewhat comfortable middle ground though.
In my particular application, efficiency came second to the heat generated by biasing resistors. I can live with the power usage, as long as the board doesn't let the magic smoke out.
That said, I've chosen to use mutiple votage regulators (SOT-23 @ 100ma) to power my LEDs. I still need to use the biasing resistors, but by using a seperate regulator of an appropriate voltage for each bank of LEDs, the resistors have to dissipate less power, and generate less heat.
All told, my design uses 5 seperate regulators, 1 at 5 volts for the ZX, 3 at lower voltages for the LEDs, and another at 5 volts for ancillary circuitry. The end result is slightly more total power usage, but spread over a much larger board area. The biasing resistors are 1/10th watt 0402 versions, each one dissipating very little power (read: heat).
So, in the end, even with the many available specific LED driver chips available, and considering the advantages of swithing regulators, the solution to my requirements turned out to be very low-tech.
-Don
I've looked at white LED driver chips, as well as various contant current schemes. All of them offer higher efficiencies than the simple bias resistor method typically associated with powering an LED. Unfortunately, my space constraints are particularly severe, and the inductors and other components required to impliment a higher efficiency design simply require too much PCB space.
There is a somewhat comfortable middle ground though.
In my particular application, efficiency came second to the heat generated by biasing resistors. I can live with the power usage, as long as the board doesn't let the magic smoke out.
That said, I've chosen to use mutiple votage regulators (SOT-23 @ 100ma) to power my LEDs. I still need to use the biasing resistors, but by using a seperate regulator of an appropriate voltage for each bank of LEDs, the resistors have to dissipate less power, and generate less heat.
All told, my design uses 5 seperate regulators, 1 at 5 volts for the ZX, 3 at lower voltages for the LEDs, and another at 5 volts for ancillary circuitry. The end result is slightly more total power usage, but spread over a much larger board area. The biasing resistors are 1/10th watt 0402 versions, each one dissipating very little power (read: heat).
So, in the end, even with the many available specific LED driver chips available, and considering the advantages of swithing regulators, the solution to my requirements turned out to be very low-tech.
-Don