Hi guys, I mentioned in another thread that I'd started building a switched mode driver for high flux LEDs, and today I've transferred it to matrix board to test it properly before making some PCB's for lights I have planned. Here's the circuit: It's a buck converter based around a PIC10F204 (8 pin in the picture, but the smt version is 6 pin sot23 IIRC) and follows the pretty standard circuit configuration. In this case it's configured as a 500mA constant current source for 3W LEDs. I chose not to drive at 700mA since the difference in brightness is pretty minimal and the LED doesn't get so hot. The green LED = power good, blue LED is the PWM output. The7805 voltage regulator provides 5V for the 1mA current draw of the logic That'll be changed to something more suitable on the PCB. Driving a red 3W led with a 12V supply yeilds some waveforms like this: Channel 2 (blue) = PIC PWM output, Channel 1 (yellow) represents the inductor current. The switching frequency with the inductors I had around is fairly low for a switching PSU (at around 20kHz nominally), but has the effect of reducing switching losses to an extent. The efficiency is was then measured for a range of supply voltages (max supply voltage is actually 30V due to the 7805): Max efficiency is about 80%. I suspect this is limited by the bipolar transistor I'm using, but I have no plans to swap this for a FET since I got loads of SOT-223 transistors for a great price because they're not RoHS compliant. Add more LEDs in series and the current remains constant at 500mA. Bright! Next I'm working on being able to alter the brightness over either a CAN or RS485 link. Let me know what you think
Its great! A couple of questions though: I've noticed a few PIC based current PSU's. Why PIC's rather than a SMPS IC? I understand how you could sample a voltage regulated SMPS with a PIC, but how do you regulate current without having high wattage resistors in a feedback loop?
Why? A PIC10F204 costs 27p! To reduce the power wasted in the feedback loop, you lower the resistance of the resistor, then either reduce your reference voltage or amplify it to bring it up to something more reasonable. 0.01R resistors caused me some problems, I suspect mainly because the breadboard I was using was not good enough and increased the resistance somewhat. I aim to use 0.1R resistors with either an external amplifier, or one of the op-amps built into some of the PICs.
I came across a "Simple High-Efficiency Adjustable Current Source", 50mA - 1.5A, looked useful. RS have the LT1510 chip - but it's over £10 to get one.
£3.99 + vat at Farnell, which isn't quite so bad... (order code 1273761). Maxim did a few devices, but I don't think they were suitable for 3W LEDs.
Sweeeeeeet. I have no idea what it does, exactly, but you said I'd be better off with one for what I want to do. So erm...cool!
That looks pretty good SteveyG. For the 5v, you could use a zener diode since the current draw is pretty small.
Absolutely awesome mate. I have been trying to make one for about aweek now but keep having snags. Do you have a schematic?
Was there an advantage to that over using COTS constant current drivers? I've had good results using constant current drivers designed for flashlights, then just adding in a simple MOSFET and resistor to allow a microcontroller to PWM it. But I guess your advantage is that it doesn't rely on an external PWM source? Does that small orange pot adjust the PWM duty cycle? Either way, cool project. What are your plans for it? I have one that looks similar (except with COTS parts), here. I usually just stick it against a white wall and go into PARTY MODE Bow chicka wow wow! http://forums.bit-tech.net/showthread.php?t=188453 It's great, and I like it, but the cycle rate (goes through all colors in about 5 seconds) is just too fast, and it's a chinese RGB controller so no easy way to adjust it.
Hi, bit late for a reply, but since I'm here... The idea behind this was as a cheaper more versatile driver than using something like a ZXLD1350 which I could lay out on my own PCBs rather than using pre-built drivers. The orange potentiometer controls the current through the LED by adjusting the setpoint for the current feedback. What I actually ended up doing was using a single bigger PIC with three 'buck' style regulator circuits on the output of three PWM pins for RGB control. The PIC did all the current regulation, so I just needed an external transistor, inductor and diode per channel and also acted as a DMX slave so was about to control the brightness of the LEDs. Since I used a PIC with three hardware PWM channels, I could get the switching frequency much higher and therefore reduce the size of the inductors.
