LED drivers with very high analog dim range?

Yeah, I remember that. The bed post warning beacons?

I could imagine that for a blinking light but for a steady light this would require a total darkness that you can't have in urban area, I assume.

Well, this is for very high efficiency stuff :-)

No, those are tritium.

John

The MAX7219 (and a few other similar parts in the line up) set their PEAK current with an external resistor. It can be tweaked for very low light levels. It also has PWM.

So, designing some avionics display that needs to be read wearing NV?

Of course, it probably isn't cheap enough for you.

...

Heck, even I know that Maxim parts are discontinued the week after you design them into something...

okay. So up to 5A drive (given that PDF you mentioned),

Jon

Well, I'd feel confident to design it (together with an IC professional who knows the fabs and processes). But going the custom chip route carries a huge NRE. And it wouldn't make sense if there is already a chip. My problem is that searching is quite difficult because manufacturers web sites and their search algorithm (if existing) are usually the pits.

Efficiency only needs to be high at high power levels, not when dimmed way down, of course.

what more than inductor,diode and sense resistor? or you want no external parts at all?

-Lasse

Yes, that one is too expensive but it seems it also would not be able to do large analog dimming ranges. And not much power. Table 11 shows it to already go non-linear over a 4:1 ratio so I guess it'll probably become iffy before you'd reach 100:1. This chips is made for meters where dimming is usually shallow, and sometimes it's just setting one particular brightness for the selected display and that's it.

Is it that high? I remember buying a wafer brokered out at about $5k. 8". While ago. Probably at 12", now. Must be folks out brokering capacity like mad, these days. I can't imagine how many of these you'd get from a wafer. You'd need to test, package and then retest, though.

True.

Yes, of course. I'm still stuck on thinking about the 5A and you probably wanting to go well below 1mA if you could get it, and with precision I bet, too. I remember how hard it was to set up a 0.1% precision current control for electron beam emitters used in IC mask generation and have it not drift much from day to day.

Anyway, best of luck. Just for grins, here's another thought:

If the wavelength you are looking for matches up, perhaps you can source some phosphor responding in the blue with a red phosphorescence and taus in the area of, say, 10ms or 100ms or longer. Then PWM the blue LED. The long tau relative to the PWM cycle will simply filter the hell out of the PWM pulsing, in the red anyway. Block the blue or use enough phosphor to totally absorb it.

:)

Jon

You need various other things: Differential amplification of the sense resistor voltage (Kelvin connection), some filtering, then either a gain variable amplifier or a 2nd amplifier to add voltage to the feedback node, bypassing and supply voltages for all this, and so on. It does add up. I have designed stuff like that.

I was only talking about the one line-item. And yes, there is a lot more that needs doing. Of course, I knew before that this wasn't likely to meet your $2 in 1k piece qty -- which obviously equals $2k. You can't get an asic packaged and tested at that price. I could only wish.

Stability was very expensive in the application I mentioned. And even a slight shift would yield either a much brighter beam or a much weaker one, plus if brighter far faster wear (very expensive) since they were usually running these at the wear-out-knee point to get the faster writing times.

Hehe. Well, I just figured I'd do an "out of the box" moment for you. Just for the fun of it.

Jon

something like this then?

-Lasse

Oh, that's just a ballpark max cost figure for a quantity that is popular in distributors' price lists. Qties would be a lot higher. Still, developing an IC is something that is a major investment, quite similar to buying a sizeable house.

That can require a good regulator with a super-hard max current bypass limiter. IOW one that throttles power no matter what if some pain threshold is exceeded. Done it once when the Dollar number of the attached optics stuff exceeded that of a fancy new car :-)

Sometimes out of the box musings result in a "Shazam, that's it!" moment :-)

Quote "The LED current is set by selecting an appropriate value for the sense resistor value and using the EN input to provide simple on-off control or for PWM brightness control using a suitable externally generated PWM signal."

Only PWM, like the other one :-(

Analog

ly

gital

keting

sn't

and

o deal

t easy.

ften

. In

the

in

dd

umbers/6265/6265.pdf

think you forgot to read the part about iref for analog dimming, top page 9 .. If VIREF is less than 1 V then the 100% current level is defined as: ILED(max) =3D VIREF / (10 =D7 RSL ) (3) This feature provides direct analog dimming using a voltage from

-Lasse

-Lasse

Analog

cheaper:

easy.

then?

Yes, but they also mention that it is meant to even out brightness deviations. The current error is given at +/-5% which is huge, and normally that also means a large offset. Which isn't stated because if the feature is only used to even out LED tolerances it doesn't matter.

A chip with a large analog dimming range must normally contain very precise or trimmed opamps inside.

Even for PWM dimming the ratio in this chip seems to be very limited. Quote, page 9: "During PWM dimming, the A6265 switches the LED current between 100% and typically 8% of the full current."

e. Analog

irly

digital

arketing

isn't

al and

P3065.PDF

her

nd 0.5%

to deal

hat easy.

or

often

re. In

ts the

se

gain

add

_Numbers/6265/6265.pdf

but it is hard to tell from the datasheet if that 5% is really there when you use and external input for iref, or it is just that the internal

-Lasse

log

ng

I dont see how youre going to get better efficiency with _analog_ control than a series pnp transistor. Add a very low emitter resistor and you have voltage control. What other component would give you control with less minimum Vdrop?

An opamp could take whatever Vin range you want to control the tr.

NT