Workin on an idea here

[Warlock]

The idea is bascially making a dimmer using the renard guidelines PWM control that would control the front side of a led for dimming..It seems (from what I have read thus far) to be smoother to dim by controling the front side voltage rather then using a pulse method as the pulse still have too many voltage fluctuations that cause flickering for anything that is static for longer then 10 seconds.With mainly it affecting the red and blue or green leds.The white seem so far to be unaffected by this or it is not noticable..
SO making a dc or ac dimmer that uses a different approach might be useful as it wil control any led you can toss at it...
Hope this idea appeals to some.Designs will follow shortly..Still working on a few bugs and location of parts..
Joe..

[P. Short]

Huh? What is a front-side voltage?

[Warlock]

Some call the cathod side of the led the front side and the anode side the back side cause of positioning.So if by controling the voltage of the front side ( or cathode) of the led would create smoother dimming levels and no flickering plus you can also control them at a higher rate of speed without any of the long life effects that are associated with doing it at a higher rate of speed.
Hope this helps explain the idea..
Joe..

[phenagan]

Voltage control is NOT the way to go with dimming LEDs. They are non-linear in brightness with respect to voltage, which is why you'll find that commercial LED dimming systems use PWM rather than varying the voltage. Also, white LEDs will color shift when dimmed by voltage whereas they don't with PWM.

Not sure what you're referring to when you're talking about long life effects w/ regard to PWM - as long as you're not over currenting the LEDs, there should be no detriment on their life expectancy from PWM.

[P. Short]

LEDs can be controlled by varying the current. One way to vary the current is by using linear analog techniques, which ends up dissipating a fair bit of heat. Another way to vary the current is by using switching power supply techniques, either buck or buck-boost designs. Either way ends up being a fair bit more expensive than conventional PWM techniques.

[Warlock]

True Phil that was why I was asking how the PWM does the dimming...Any help you can give in this area would be really helpful...
Thanks.
Joe...

[awhaley]

PWM 'dims' by switching things on and off really fast. Your eye has a slower refresh rate than the LEDs, so it averages their intensity over a period of time. The perceived brightness is basicly determined by the duty cycle... what percentage of the time the LED is on.

If it's slow (say... 60 or even 120 times per second) then your eye can see the flickers. The easy and perfectly fine solution is to make sure that the PWM rate is fast enough that nobody will notice. You can get perfectly smooth dimming with precise control over the curve. You should examine RJ's first Lynx design. It's a PWM dimmer (meant to dim both incandescents and LEDs.) It uses Mosfets instead of Triacs, because the latching characteristics of Triacs would screw up the PWM.

The only other solution I can think of would be to 'slow down' the LEDs turn on and turn off time, by adding a capacitor so they behaved a little more like incandescent lights. Getting this right, and making sure that it allowed voltage to get to 0 to unlatch the triacs might be tricky... I've never played with the idea. And it sounds way more expensive and difficult than the PWM solution, which is what pretty much all professional LED dimming arrangements use.

There's nothing mysterious about LED flicker... they're flickering because the power's turning on and off all the time. You've got to make it happen faster if you want to stop seeing it... or go back to incandescents. The differences you may see from one dimmer to another are based on the differences in how they behave at low dimming levels with low current draw. If it's not latching the triac correctly every cycle, the flicker will get worse. But it's going from 'nyah, alright' to unacceptable. The only way to get better than 'nyah alright' is with high speed PWM.

Voltage control, as Phenagan pointed out, is not good aesthetically. You'll never be happy with the curve and the color shift will be rather unattractive. I'm a bit of a color snob, but I actually have never found a white LED I liked anyway... for anything other than strobe and special effect applications. I'm not giving up my incandescent minis anytime soon, even if it will save on my electric bill. :P

Current control via an amplifier arrangement could be interesting... You know, the big boys in dimming are using 'sinewave' dimmers these days which are basically amplifiers and signal generators... maybe it's time one of us develops a version for Christmas lights... I wouldn't bet on it being cost effective, but it would sure be cool... and you can run your lights at any frequency you'd like then.

Art

[Warlock]

Thanks Art..Info was exactly what I wanted...How can you make the pwm faster or is it fast enough already ?????How well do the triacs latch up and trigger with this pwm code and can you run leds directly off the opto and have it work better then regular code ?????
Thanks
Joe...

[scorpia]

How well do the triacs latch up and trigger with this pwm code

If theres one thing i have learnt about led's this year its that they are much better with DC rather than AC. and since triacs are an AC only device i would suggest staying away from them in a new LED only dimmer.

maybe FET's or something similar are a better idea.

Peter

[awhaley]

PWM can be as fast as you make it. When most people on here mention 'the PWM code' they're talking about the PWM code for the Renard... which if I understand correctly, (I haven't built any Renard hardware yet, so I haven't had a reason to read up too much) leaves the Triac trigger signal on for the whole time the lights should be on instead of just pulsing it when they need to come on. This improves the latching for light loads, because if the current wasn't high enough to latch when the triac first came on, a few uS later it might be, so you've got a way better chance of getting something to come on during any given cycle.

But as long as you're talking about AC and Triacs, as Scorpia points out... you've got some inherent weaknesses to the system. Unless you're running your own custom generator (or a wicked capacitor spark gap primary for a tesla coil... :P ) you can't make the AC PWM any faster... the cycle is fixed at 120hz and you can only control what part of that cycle the light is on for.

If you really want to design a PWM dimmer to kick butt at dimming LEDs, you want a microcontroller or a dedicated PWM chip like the TLC5940. I was doing some research a while back about a project along these lines too.

The TLC 5940 based dimmer board would look and work a lot like the grinch, only it would generate dimming pwm signals instead of just on and off. You dump channel levels in serially and latch them and it spits out really high speed pwm which will dim LEDs or Incandescents beautifully...

The catch is you can't use a Triac SSR for it. You need to build a circuit like the DC SSR on the wiki using mosfets, but for 120V DC and using big rectifiers on every power line you run to your new 'MOS-SSR Pack.' The cool thing is that if you keep it triggered by sinking current, the MOSSSR ought to be backwards compatible with the phase angle dimming systems and really shine with your new PWM controller.

I'm describing a system VERY much like RJ's first lynx, as he pointed out when he wandered by a discussion on it... so you should go take a look at it. Now, RJ abandoned that design in favor of phase angle dimming because the PWM system was more expensive... but people who are intent on switching to LEDs may well be willing to pay a little more to build a controller that is really great at it... since they're already paying too much for the lights. :)

Art

[P. Short]

Another point...for Renard (except Ren-C) it is possible to set up the firmware in a PWM mode that is not synchronized to the power line. With this mode it is possible to connect up a DC SSR to drive LEDs directly. The PWM frequency here will end up around 100 Hz if I remember correctly, but could be tweaked higher if desired.

[omzig]

Why couldn't an SCR be used?

[awhaley]

SCRs, like Triacs, latch on once they reach their holding current. SCRs and triacs are very similar devices... with the primary difference being that Triacs function with AC as well as DC.

And Phil, that's good to know that Renard equipment can be used for DC PWM. Very cool stuff. :) Working on my own dimmer project has really made me appreciate everything you've done in the Renard all the more.

Art

[Matt_Edwards]

...f The PWM frequency here will end up around 100 Hz if I remember correctly, but could be tweaked higher if desired.

Isn't this a bit low. I seem to remember that PWM frequency of 1K was the best for LEDs.
KR
M

[P. Short]

Where did you hear that? Although I didn't say it, I was thinking of numbers in the range of perhaps 150 hz to 200 hz. These values should be easy to attain without overloading the processor, while I think that 1 kHz would be a problem.

There are other techniques that could do the same thing with a smaller processor load than the simple PWM methods. One of these is Bit Angle Modulation (BAM). This method requires less processor loading, which allows the base modulation frequency to be increased. There are at least two other techniques that I'm aware of that would keep the base modulation frequency the same but greatly reduce that lower harmonics, thus making the flicker a lot less apparent.

[Matt_Edwards]

Phil,
I don't remember for sure, but let me dig through my papers at home.
KR
M

[scorpia]

If your talking led's it might be worth looking into jec's code.

from memory it has a pretty high refresh rate. and can easily be expanded to more than the 3 channels he runs it on.

Peter

[Matt_Edwards]

Where did you hear that? Although I didn't say it, I was thinking of numbers in the range of perhaps 150 hz to 200 hz. These values should be easy to attain without overloading the processor, while I think that 1 kHz would be a problem.


Reading a Nat semi seminar too quickly. they recommend above 120hz, and then show some examples at 500Hz and 1kHz

I dim my AVRs using 1khz.

The reason I pick up of 120-Hz, is because there are some they say they can see the LED flicker at 100hz.

You are correct about overloading the uC, I tried 1500khz and it was NBG.

sorry if I caused confusion.
KR
M