LED drivers with very high analog dim range?

I'm not sure how they come up with that "analog dimming range", isn't it just a constant current buck? so really the difference between digital and analog dimming is the speed a filter and the feedback?

shouldn't any old buck with feedback from a sense resistor work?

this looks cheaper:

-Lasse

Yes, we can make one (actually, just program one). All you need is a chip with more than 8 bits D2A. Or with external op-amp scaler.

So far all LED designs I was involved in were with regular switcher chips. But this is about dedicated LED chips where you can command 0.5% dim or whatever and then it sets the current accordingly. It has to deal with a lot of noise and other stuff as well, so it's not quite that easy.

I can make one as well but I was thinking about something that can be picked from a distributor's menu and bought :-)

Are you sure you want linear dimming, not logarithmic? A second, similar, LED as a regulator with the right tempco, and a potentiometer, would do a pretty good logarithm-law job of it.

Are you able to consider a combination of current adjustment _and_ PWM to get the range? Or do you really need 3+ orders of dynamic range in setting the current? And do you need linear or geometric steps across that range (how many, either way?) And does it have to be a single chip? Or just cheap?

And finally, any way of closing the loop? Are you observing the optical output or is it open-loop?

Jon

NXP ? PCA9685

Cheers

Joerg a écrit :

Wow, $2/1k! Feeling like a spendthrift?

Linear or not doesn't matter too much right now, just wanted to know whether there a chip with more range. I know how to do it in discrete and with opamps, just wondering if there is a chip already.

Nope, absolutely no PWM. That's how most chips including this LM3409 achieve higher range, by combining modes.

Cheap, and ideally lots of linear steps. The more the merrier.

No, that isn't planned yet but possibly some day.

Thanks but that won't work, can't use PMW-dimming. PWM is the easy way out, that's why most chips do that. And unfortunately a lot of car tail lights.

Well, of course in an application where a bazillion are made per year one would not do that :-)

But then one also can negotiate confidential deals, in Asia that's often called "priced as needed".

Have you tried capacitating the output of your PWM? If you don't need too fast of a response, you could use a big enough cap to provide essentially DC.

Of course, that would be in the form of a current regulator - maybe an inductor would smooth out the current better (LEDs are current-mode, after all), but inductors scare me.

Good Luck! Rich

Just for fun, I took some high-efficiency green LEDs up to the cabin, got really dark adapted, and tried to see how small a current I could distinguish as making light, under ideal conditions.

About 700 pA.

John

How did you read the meter in the darkness?

Years ago, a friend at the Exploratorium was trying to make light from lemon batteries. They had sound and motion. I had a sample of an early high-efficience red LED from HP. I took it to the lab one weekend and left the lights off. I could see it down to the microamp level. (10 V and 10 megohms, or something like that.)

The trick is that it needs two lemon batteries. The LED threshold voltage is below 1 lemon.

I had a variable power supply (with tape over the pilot light) and some gohm resistors. I tweaked the power supply, and toggled the LED connection, until I could just distinguish on from off. Then I turned on a tiny flashlight and read the current.

You can wind a coil, connect it to an LED, and drop a magnet through it, and get a flash. I think you can light an LED from electrostatics, shuffling your feet on the floor.

I have a Tadiran lithium battery, a 1.5M resistor, and a green LED next to my bed on the bookshelf.

ftp://jjlarkin.lmi.net/Tadiran.JPG

It's running about 1 uA and will last the rest of my lifetime. In a really dark house, it's just usable to light your path. But you can hold it in your mouth and bridge the resistor leads with your tongue and adjust the brightness as needed.

But for the bedposts in the cabin, I went for nuclear power.

ftp://jjlarkin.lmi.net/truckee/Tritium.jpg

John

That's an excessive budget. Simply put a fet in series with the LED(s). If you want to take it further, put another fet or tr with base R in parallel with the LEDs as well.

NT

you have obviously tried so there must be more to it than I think there is, but I am still a bit confused at what the problem is.

are you saying that a standard buck topology, like that NCP3065 for example, cannot set the led current with < 0.5% accuracy from the command input?

what voltages/currents are you looking for?

-Lasse

You can, but in order to do that you need a considerable amount of external circuity. That requires space which in these applications often isn't there. I am looking for an IC that has all that right in there. In other words where you simply feed it a control voltage and that lets the light dim over a 500:1 range or ideally even more.

Roughly 20V, derived from 120/240V mains. Current on the order of 1A to