Voltage mode LED Driver IC's are often best?

[D.A.(Tony)Stewart]

LT3755 is good value but switch rating 40V max so external FET needed.

But for the same price, this one is better and goes up to 100V in and 100V output.. so no ext FET needed.
\[\oplus\] ===== \[\oplus\]

Datasheet shows LED driver 94% Efficient 30W White LED Headlamp Driver 81V @370mA not bad!!

 

[mtwieg]

LT3755 is $3.7 for 5000 pieces on digikey.

-thats a bit too expensive.

Okay now you're just shifting the goalposts. If you compare it against general PWM controllers of course you'll find cheaper parts (also linear tech is always a bit expensive...), but then you don't have important things like built in brightness control or open output protection. If you don't need those features then why do you desire LED control chips in the first place?

edit: Also I found a pretty interesting part, the LM3423. It's current mode but uses constant off time modulation to eliminate subharmonic oscillation without slope compensation (something I wish much more pwm controllers used...). And it seems to be very capable in every way a LED controller could be.
https://www.ti.com/lit/ds/symlink/lm3421.pdf

 

[D.A.(Tony)Stewart]

I consider that if you are driving 48V@350mA worth of LEDs or 17 Watts worth at 1~$2/W = $17~34
Then magnetics, AL Clad board etc, shielding, EMI filtering , you afford a few bucks for a driver and get the bells & whistles too.

LT3756EUD stock Newark $3.6@100, stock Arrow $4.31@1
Then for automotive pricing cut that in half.

Choose a SMPS IC that raises the running F to 1MHz has a better chance of reducing magnetics cost. to <$1 for 20W converter.

But probably right with time, that could be cost reduced easily to $0.10/W total.

Yes the LM3423 also looks like a good %95 efficient solution, which is avail. for <$2 @5k, perhaps sub harmonics are not an issue if shielded and > 10KHz for flicker free use in motion. Actually, one could use a very cheap & dirty dimmable boost driver to drive LED's in continuous mode.

 

[grizedale]

Thanks, incidentally, LT3756 is what Bosch use in their automotive led drivers.

The LT3755 And lt3756 datasheets dont tell exactly how much internal slope compensation is added

-as we know, at high duty cycle , a lot of slope compensation is needed.

The high side current monitor looks good in lt3756 etc because it can stop overcurrents to the output due to staircasing....


Also Automotive load dump can whack up the input voltage drastically and again cause huge sudden overcurrents, again needing a good current sense to stop it.

The lt3756 has the current monitor in a great position (in front of the output cap) , but still it is expensive.

 

[mtwieg]

The LT3755 And lt3756 datasheets dont tell exactly how much internal slope compensation is added

Yes and this is a problem. Linear tech never specifies their internal slope compensation. Every time I use one of their current mode control chips I send an email asking them to provide that info, never get a clear response.

Linear is a strange company. They're brilliant engineers and they love to show it. They love putting exotic, complex features in their products, and in general those features work very well. And often they're happy to explain a lot of it in their documentation. But often their documentation isn't completely exhaustive, which can make skeptical engineers cautious in situations like this. So you kind of have to put a little faith in them. I've only had one experience (out of dozens) where one of their pwm controllers didn't work out, but in that case my application was somewhat bizarre and it wasn't too surprising.

That being said, I'm pretty sure they add a great deal of slope compensation so that it will be fine for pretty much any design. A few of their example schematics show boost designs with output voltages of 40-50V with an input range of 8-40V. If you're still unsure, there's always LTspice. Presumably their models have the real slope compensation in them. Heck, by doing analysis on simulations you could probably figure out the value of the slope compensation.

Also Automotive load dump can whack up the input voltage drastically and again cause huge sudden overcurrents, again needing a good current sense to stop it.

Well this is mainly an issue with loop compensation. If the line drops, then the converter will want to draw more current no matter what. Current mode should have less overshoot and faster settling time than voltage mode.

 

[D.A.(Tony)Stewart]

There may be a reason for their silence. But you can research the slippery slope of equalization back to 1989 or most recent citations for best results.

Why are bugs female? (ha)