12V LED dimmer/timing circuit advice required, please!

You already have the 12V supply, so I would keep it as part of the project. For one thing, this prevents you from having to deal with the mains. There are probably several LEDs in series, plus a resistor. I don't think they would use 12V per LED since the drop across the resistor would be high.

What you want to do is pulse width modulate the power to the LED. That is, let duty cycle control the brightness.

Most cheesy PWM circuits consist of a sawtooth generator which you can make out of a 555, and a comparator.

Bill's circuit doesn't reduce the voltage (much) what it mainly does is reduce the on-time of the LEDs (by blinking them at a rate that's too fast to see) this is a very power-efficient way of dimming lights.

an irrigation controller perhaps ? this is way more complicated than Bill's PWM dimmer. it may be easier to hack a few electronic timer-plugs and use their timers to control on-off and brightness.

OTOH it would be easy to wire a light-sensor a variation of bill's dimmer circuit so that when it sees "dark" it dims the lights and just use the timer plug to turn them off when you want off.

One thing to consider when chosing between a switched (pulse width controlled) dimmer, and some linear system, is that in pulse width modulated LED supplies you get a sort of stroboscopic effect when your eye scans accross the LEDs (you see more then one, some distance apart). If you do not like this, then use DC.

As you are not very experienced, maybe it would be simpler to use the serial port. For example to write to an antique UART, and use the 8 data lines to for example drive a resistor R2R network to make an analog voltage to drive the brightness, for example drive a 2N3055 emittor follower.

That will save you from programming a PIC. Here is a simple serial communications program for Linux:

okay. It's one less headache, I guess.

Nor me, but the transformer claims it's chucking out that and these are somewhat superbright LEDs. I'll get a multitester on the DB and measure the voltage at the parallel end, see if it really is that much.

okay.. I recall some diagram that had a strobe-like effect to simulate dimness, although it looked more like a square wave (I could probably pop some caps across the output points to smooth out the wave).

Ta for the help!

Given they're not going to be near anything else strobing (overhead lights @50Hz, monitors/TVs etc) I wasn't too bothered about the strobing effect, but I know what you mean. Would upping the PWC frequency to something much higher (maybe in 200Hz range) disguise the strobe effect?

snagged, untarred and read the README... mmm.. nice! Thanks for that, I may toy with a few proof-of-concept experiments and that util.

Can't go wrong with a 555... where would we be?

(don't say "taking a pair of scissors to a 556"...)

I read that reducing the power output (strobing) is the more efficient method than splitting it and dissapating the unwanted power elsewhere, but as we're talking a total of 1W for my lights I didn't know if it was worth it. I'll have to do some costings.. perhaps I should have done that first.

(/me ponders a blog...)

mm... never thought of that. Will investigate, ta.

Essentially I wanted to have the lights at 100% during evening/dusk so some LDR-based circuit could trigger the "on" cycle, but then wanted them to dim to "comfort light" after a pre-determined point (20% from midnight onwards?), so what was an illuminating effect doesn't become too dazzling to sleepyheads. Bit of a strange requirement, I know.

okay, that's given me some pause for thought - ta for that info!

I can just see one preparing a ham sandwich with a strobe light. The blood from the cut-off fingers would be something to see.

Yes. Build the Bowden circuit - it runs at 200 Hz so there is no need to up the frequency. You were also concerned about the square wave output - that is not a problem, at all. Get the dimming portion of what you want working first, sort out the timing part later.

Be sure to check that the voltage coming from the supply included with the lights is 12 volts DC.

Ed

This is my recommendation, but I wouldn't use a 2N3053 and 2N2955 - I'd use something more like a PN2222 and a TIP32, because they're much more recent designs, with lots of beta and stuff. If you want to switch the low side, lose the 3053 and just drive a TIP31. They're less than a buck at digi-key.

And yes, 200 Hz is high enough to make the flicker go away. :-)

Have Fun! Rich

No, a rectangular pulse is what you want. Your eye will average the intensity. The 200Hz suggested frequency is plenty fast.

From: Rich Grise

okay.. so just replace 3053->2222 and 2955->TIP32?

Roger that.

I don't know why, but I'd prefer to have the switch at the high side - something about having floating grounds worry me.

And yes, my idiotic eyesight never actually READ the site carefully and noticed that! Major DOH! Excuse my noobness/impatience; it's strangely nice to be on the other side of the clueCurtain once in a while!

Ta!

Nope - they're paralleled up.. and just checked the voltage. 13.55V at the supply end!