Thanks for bringing that up -- I, too, was questioning whether the LED itself was really going to be capable of a 10000:1 range. I'm sure that on a 1st-order analysis you just vary the current and assume that the current to photon flux number is constant -- but I have trouble believing that the photon flux is really a constant factor of current over a
10000:1 range of current!-- www.wescottdesign.com
Drive behind a car with LED tail lights during a very dark night and move your head/eye left-right-left rapidly. As if looking for deer, which we have to out here). Some people aren't bothered by it, others like me find the stroboscopic effect hardcore annoying. Same if two or more dimmed LED sources beat.
It's ok up there, I've been involved in aircraft LED design as well. Aircraft rarely have a passenger car with LED tail lights in front of them :-)
Yep, got to calibrate.
The light intensity as perceived by a human eye is not linear.
-- Regards, Joerg http://www.analogconsultants.com/
d
cy
to
tlot
sked
s ovel
s sst
e tI just looked at a cree datasheet, and the curve for current vs. flux is does have a small bend
and from 25 to 150C output drops 30% at constant current
-Lasse
Are you aware of all the science on this? I've got gobbles of it. Use a fast enough PWM and it becomes very hard to notice. I've done a lot of testing on this, as well, to verify what I've read about. You are right if the PWM is down around 70 Hz -- unmoving you don't see it but even slight changes of eye motion and it becomes easy to notice. But go up a few orders to the small kHz range and PWM inside of that and I haven't been able to find an issue. Most folks are good below 1kHz rates.
Been there, done that.
But they are _do_ care a great deal about even the slightest hue differences. And I thought, at least in part, you were discussing that as well. Besides, that wasn't the only application. I also dealt with drive-by signage. That's fast. Custom stuff to be sure. But the buyers were paying many, many millions of dollars for what they bought. And they cared, as well.
I have to let you know, I was able to develop a unique method to do that which Seimens and Osram hadn't conceived of and were shocked to receive as a product. It cut their calibration costs by two orders of magnitude. You might be interested.
You are telling me this? Shocks me that you'd even imagine for a second I'm not fully aware. I've written a small course on the subject, if you'd care to read it.
Jon
It's initially roughly linear for (non-white) LEDs.
See, for example, fig. 3.11 here:
Note that, aside from the nonlinearity, there is roughly a 5:1 or 10:1 variation in expected luminous intensity at 1% of the full current! Presumably that's 3-sigma statistical variation. But if you have three colors matched at full current, they're likely to be WAY off in an unpredictable way at 1% current. A good reason to use PWM if you can.
The chromaticity shifts quite a bit with current (another reason why PWM is to be preferred if you can use it). Then self-heating will result in a reduction in light output as the chip heats. Some LEDs will change 3:1 or 4:1 in luminous intensity at a constant current over their operating temperature range! Then there's aging..
Clearly Joerg's proposed application is extremely unusual and for a specialized niche so he's just complaining on general principles. ;-)
He might want to look at the specialized chips designed for video LED screen applications (eg. stadium screens). This is a decently large business so the chips exist. They have trims built in to adjust gamma. Not sure if they would be directly applicable though.
t
and
le
tency
tor
l to
er
hat
e glot
Cs
rked
ain't the just because they run the pwm at some far too low frequency? if pwm can do full bandwidth audio it should be possible to do dc
-Lasse
When you go into the kilohoitzes you lose dimming range. Because the switcher core frequency is usually only in the low hundreds of kHz.
Thing is, up there PWM can be very ok. Down here on earth is where it's often not ok because some folks see the stroboscopic effect. Once you see that stuff any hue inaccuracy won't make a difference, because then it's already perceived as junk. Like LED tail lights often are rather dismal designs.
Yes, that can be very intersting here once we get to a product (that's entirely up in the air so far).
No, that was in answer to Tim's comment, not yours. But I am sure he knows that as well, it's just to confirm that I am aware of that.
-- Regards, Joerg http://www.analogconsultants.com/
Self-heating is not much different between analog and PWM. In fact, it can even be worse with PWM.
Nope. It certainly is not my main turf but there is a another guy in the group who is significantly older than me and has decades under his belt in this business.
Good idea. Do you know who might make those chips?
-- Regards, Joerg http://www.analogconsultants.com/
that
ing
something like this?
-Lasse
Yes. For car tail lights there is no reason to do that, or to do PWM in the first place, because they only have two dim levels. But I guess they want to be cheap.
That is entirely different. Audio amps "drink" power from already prepared power rails. They may not be well regulated but the voltages are about right. With LEDs that is generally not the case, you have to do that in the same IC. And for some reason many switcher chips have rather poor duty cycle limitations to boot. My discrete solutions never did.
-- Regards, Joerg http://www.analogconsultants.com/
I doubt that I can design a chip for Joerg... sounds possibly like we're competitors... since I'm heavily into LED video... and I'm winning ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
[...]
Unless I am reading this wrong it is 12bit PWM (just a hard switch) and
6-bit linear current adust over a small range. That won't be too useful for lighting purposes.-- Regards, Joerg http://www.analogconsultants.com/
maybe not since it is made for led video a screen, wasn't that waht was asked just above?
guess you could put all the outputs in parallel and use it for lighting looks like it is using 30MHz, so ~7.3KHz pwm at 12bit
-Lasse
Not really, I thought that Spehro meant "real" stadium screens. This size:
Ok, yes, but only 60mA.
-- Regards, Joerg http://www.analogconsultants.com/
It _will_ be worse, definitely. Maybe not significantly, but it will be worse.
Not offhand, unfortunately, or I would have suggested the maker. This is one, but I think the ones I've seen are from another maker (I don't think Asian, though the maker I was talking to was in the Antipodes and the factory in the Middle Kingdom):
Best regards, Spehro Pefhany
-- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
If update speed isn't an issue, you can use a charge ramp system via a cap with a analog sample and hold. There is a name for this, I just can't think of it at the moment;)
I did that years ago, it give me resolution beyond your dreams how ever, it wasn't too fast when you wanted to change the output value.
Just think of how many clock cycles you can accumulate in 1 sec time in a modern uC. That reference can be used to process the Sample&Hold update and reset the cap on the output for the next cycle.
Just a thought and a real smelly one at that.
Jamie
Ah, thanks, now I understand. The chip that Lasse mentioned and the one you mentioned above is what they use by the bazillions in displays. Unfortunately that won't work in this case because we want to send an amp or more through each LED string when set to max brightness.
These massively parallel LED chips can only support a few ten mA and they rely on PWM dimming. Analog (current set) dimming is only for vernier purposes and not done switch-mode, it's just a controlled transistor in there. And the analog dim granularity is only 7 bits in this one (6 bit in the other).
-- Regards, Joerg http://www.analogconsultants.com/
How can you win with 'Lie Emitting Democrat' videos? ;-)
-- You can't have a sense of humor, if you have no sense.
and
simple
frequency
for
to
switcher
that
doing
lot
=20
worked
level
issues
least
the
=20
believing=20
I suspect that will have to be divided as about 100:1 in PWM and 100:1 in current control, simultaneously. Challenging all by itself.
?-)
I think it is for screens like that, think about it, lets say;
1000x1000 pixels *3 for RGB * 120mA * ~2Vled = 720kW that's not your average big screen tv
per output, and it says 120mA at Vcc > 3.6V
-Lasse
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.