Thread: Controlling back lights directly?

Hmm, I think we've scared off the OP.
This is a good discussion and it shows there are many ways to accomplish something.
It's also good info to search for but where is the OP. We need your input if you'd like us to guide you to a solution that's right for you.

Originally Posted by OldSpark

IMO the concern about over-current for LEDs is overrated. It isn't that critical - it's more of a life-span issue.

And drivers like the CAT4101 are for a single string AND high intensity LEDs (ie, up to 1A) which I doubt backlights use. (They are probably multiple strings of 20mA LEDs - not that I am up with the latest in backlighting, but high-current LEDs (CREE etc) are usually NOT suitable for backlighting.)

that cat4101 driver is actually a very capable driver, the only reason i recommended against it is the form factor can be a little daunting to someone who hasn't really dealt with smd components before.

there are actually a lot of drivers like that one-- all of them have a max rating, but use a specific resistor connected to a sensing pin to set the maximum current that is allowed to flow through the layout. so while the driver is capable of outputting 1000mA, you can set it to only output 20mA.

Originally Posted by OldSpark

And keep in mind that even a 20mA LED may handle peaks of 200mA etc - eg. for 1% duty cycles and not exceeding 1mS. I'm sure even previous doubter bes51659 will back me up on that.

name brand led's are capable of handing peaks like that. no-name led's are much less forgiving in all out-of-spec regards. but i don't believe that cheap led's were ever used in these backlights, as the color consistency is very important, and cheap led's do not have that type of consistency. they are probably using cree, nicha, or philips led's for it.

Originally Posted by davekra

I agree, OldSpark, that LED's can handle more current than rated if the duty cycle is appropriate. But if you don't get your duty cycle right you risk blowing the LED's.
In our situations it's just as easy to create a current regulated supply that provides the correct current for 100% duty cycle.
And for me the CAT4101 makes it dead simple. The parts values are not critical and you just dial in the current you want.
Couple that with a PIC and you get a very flexible widget.

completely agreed. i really like the low parts count required for it as well!

Originally Posted by davekra

I would also argue the statement that the CAT4101 is for a single string or high power LED's.
If all the led's are working, multiple strings or a single string are the same. Right? (he humbly asks as he has no actual electrical training)
And since it can be adjusted, it can drive anything up to 1amp.

correct. (lucky for you, i've been building/designing led lighting stuff for my car for the last 2 years. the led section at hidplanet.com is invaluable ) the current is just split between each parallel strings. you gain light density at the cost of current control accuracy. when you have multiple strings, different strings can heat up differently, causing each string to slightly change how much current it pulls.

in the case of high power led's, it is preferable to mount them all on the same heatsink so each parallel string is subjected to similar heat amounts, so that fV(forward voltage--each diode has it, and as it heats up, it lowers) variance is kept to a minimum.

in the case of low-power led's(i believe that is typically anything below 75-100mA), they don't output much heat, so it's not as much of a problem.
you really can voltage regulate them without any problems, especially name-brand led's. it is best to current limit each series string with a resistor in that case.

Originally Posted by davekra

In an ideal world each series group of LED's would be driven separately, like this (see pic)
But the display manufactures don't do it that way (at least the ones I've seen).They simply have multiple series strings paralled together. The best you can do for that is regulate the total amount of current supplied.

most uses of low-power led's don't use any sort of current regulation, many times, they just use voltage regulation. you can really regulate either the voltage or the current-- as long as one part of the equation doesn't change, the other part is constricted enough to not change enough to damage anything.

Originally Posted by davekra

Regardless of the current supplied, if one LED dies it puts the rest of them jeopardy.

what you're referring to is a cascade failure(led heats up, the fV lowers, allowing it to consume more power, which in turn heats it up more, this keeps repeating until the amount of current passing through the led exceeds it's mechanical limits and the led burns up)-- that is another thing that most common in high power led's. it's not to say that low power led's can't have it happen--any diode is susceptible to it, but because their power consumption is so low compared to the large amount of copper pad area they typically are connected to, that helps to radiate any heat they create, it can only become a issue when they are pushed well past their limits.

with low power leds, using voltage regulation would not affect anything if a led died- for instance, i use 12v-2A regulators in my car--one for each led board. most of my arrays consume less then 1A each, more typically in the range of 100-500mA per board. if a led dies in a array, yes, it does make more power available to the other led's, but it is excess added to excess, so while the dead led opened up space, the 12v regulation restricted everything to not allow any of the other strings to consume any more power then they were already consuming.

clear as mud?

Originally Posted by davekra

I would also argue the statement that the CAT4101 is for a single string or high power LED's.
If all the led's are working, multiple strings or a single string are the same. Right?

Alas if one string fails, the rest divide the current and pop! Ergo, no point having a sophisticated LED current controller.
They are intended for single-string high-current LEDs which are more critical wrt to current (LEDs are current devices, not voltage).


PWM is unaffected by the number of (failed) strings.
If resistors or current limiting is to be used, then it should be one resistor per string.
The only "reliable in usual practice" designs that use a common resistor for multiple strings is where there are several strings - eg, if 10 strings and one blows, then the others get 10% higher current. And with low-current (20mA) LEDs, that is usually not a significant problem.

The PWM limiting instead of a resistor or current limiter is used for high voltages where power conservation is desirable. IE - say a 12V string off a 24V or 110V or 360V supply etc, the resistor or current limiter would waste (24-12) or (110-12) or (360-12) compared to the 12 used by the LEDs. (Times the current - I'm just indicating the proportions, ie, 1/2 or ~8x (98/12) or ~30x (348/12) the power consumed by the LEDs.)


If current limiters are required, I'd be tempted to use the LM317 voltage regulators as only one resistor is required and they are self protecting (thermally). Just pick the appropriate type to handle the current and heat (or fit a heatsink). IMO that's probably simpler than two-transistor current limiters etc.
But current limiters are usually overkill, especially since one per string is required.

Originally Posted by OldSpark

Alas if one string fails, the rest divide the current and pop! Ergo, no point having a sophisticated LED current controller.
They are intended for single-string high-current LEDs which are more critical wrt to current (LEDs are current devices, not voltage).

you make it sound like failure is guaranteed with multiple strings-- it's not. and while led's are current devices, part of current is voltage, so voltage regulation does limit the current. current control is more preferable because it more tightly controls the led, but low power led's like what are in the screens talked about here are fine with voltage-only regulation, as the heat that the led's produce is easily accounted for by the copper on the circuit boards.

the only way to have a cascade failure happen is to overheat the P/N junction of the led. it is very difficult to do this on low power led's when taking all the design considerations to mind.

i've been running my led tail lights now for almost 2 years with voltage regulation only... but i designed the boards correctly for the situation..

Originally Posted by OldSpark

The PWM limiting instead of a resistor or current limiter is used for high voltages where power conservation is desirable. IE - say a 12V string off a 24V or 110V or 360V supply etc, the resistor or current limiter would waste (24-12) or (110-12) or (360-12) compared to the 12 used by the LEDs. (Times the current - I'm just indicating the proportions, ie, 1/2 or ~8x (98/12) or ~30x (348/12) the power consumed by the LEDs.)

i don't think you fully understand the drivers being talked about-- the pwm input feature on these drivers is only a on/off switch for the regulator-- it doesn't tell the regulator to do anything but turn on and off.

the advantage to this over the LM series regulators is that it needs less parts to be current regulated and have pwm capabilities. this will accept a raw pwm input, where the LM regulators would require a transistor on the output to utilize the pwm signal.

can the lm-series be current regulated? you bet. i even recommended that method above simply because they are a larger regulator, so they are a little easier to work with...

Originally Posted by soundman98

you make it sound like failure is guaranteed with multiple strings-- it's not.

Sorry if it sounds like that, but what I am saying is that it is essentially useless having a current limiter for multiple strings of LEDs (assuming the current limit is for LED protection) - one resistor per string is better. Hence if one string fails...


As to voltage control being sufficient, I agree - hence why IMO current limiting is overkill.

I've been running my vehicle LEDs without regulation and they still work fine. LED characteristics make resistors a very suitable protection -cum- limiting device, eg, 1 to 3 or 4 LEDs from a 12V - 15V supply. That's evident from implementations (with the same LED/s) using different resistors (eg, 470 or 560 Ohms) with almost indistinguishable behavior.


My PWM comments were in relation to a PWM circuit (like the 555 I posted) being used in lieu of resistors etc. No LM regulators were involved.
I presumed that driver is a current-limiter that can be PWM'd for dimming, and that it is intended for high-current LEDs, not low current 20-50mA LEDs.
If those drivers per string are cheaper than an LM317 or typical current limiter per string, then they might be more desirable. A single PWM should control the lot noting that it has to handle the total LED current for the LM and typical (eg, two transistor) current limiters (as opposed to the lower load (fan out) of the controller chips). But as I said, unless using high-current LEDs, IMO current limiting is overkill. IMO a 555 etc PWM driving multiple resistor-LED strings is quite acceptable.


Incidentally, I concluded that using LM317s current limiters was the simplest, cheapest and best solution for up to ~1A (hence avoiding the T03 package) after a recent reinvestigation. (I was certain a single-transistor current limiter could be built, but I fell back to the usual 2 resistor, 2 transistor version - that latter being good for limiting to multiple-Amps.)


Hopefully that clarifies what I meant.

The only thing I can add is that many seem to think that LED currents are "critical", but for typical LEDs (20mA etc) they aren't. But that's a specification thing (LED datasheets).


I'll avoid the voltage-current discussion. The LED current determines its voltage, and current is the more critical parameter for calculating resistance - eg, for 12V, many ignore the LED voltage drop unless multiple LEDs or white LEDs etc. (Hence typical solutions which can have a or the next preferred-value resistor.) But the current-voltage relationship is a duality - I merely argue from the LED's fundamental design parameter.

Originally Posted by davekra

Hmm, I think we've scared off the OP.
This is a good discussion and it shows there are many ways to accomplish something.
It's also good info to search for but where is the OP. We need your input if you'd like us to guide you to a solution that's right for you.

Nope, haven't scared me off! Sorry I didn't get the emails that there was a reply in this thread.

Firstly, Thanks a ton soundman! That is awesome info of the internals. I'm glad to see the 669 has a nice easily accessible wire harness to get at the LEDs, that seems like the way to go.

Thanks for the discussion on the actual led control. I can't see how I could make the 555 work for what I want to do. Sure it can set a duty cycle via a pot but seems like a PITA to make the duty cycle voltage controlled - where the voltage control is the reference voltage taken from the instrument cluster lighting level. Before I can make a decision on how to drive the panel LEDs I'll have to see how my car controls it's lighting... linear, PWM, common ground, common +12V, etc.. Also I'll have to see exactly how the panels LEDs are controlled as well for the same reasons as listed above and also the drive current req'd. I'll probably base my circuit around a chip (TL494) that I used when I used to desisgn/build PWM powersupplies (on a hobby level only). It has a couple of onboard op-amps, ramp generator and comparitors, and a couple outputs that can handle 250mA each. Seem like everything I need - or nearly - on one chip. If more drive current is required or if the current limiting would be easier done externally then the CAT4101 looks like an easy solution, though I admit I only just skimmed the data sheet.

If you're designing stuff around a TL494 (just looked at the datasheet) you're beyond my capabilities.
What I did fits my needs and is simple enough for me to implement. Playing with the PIC programming was fun too.
Good luck with your project.

p.s. I don't think you mentioned your car, but Subaru's 2006 and older use +12volts at the dimmest and close to 0 for the brightest. 2007 and newer use PWM.

The TL494 is an SMPS chip and totally unsuited. (WRT using LM317s etc instead.)


BTW, if a single LED fails in a current limited string, the other LEDs will not blow (even if the LED fails as a short).

Originally Posted by davekra

If you're designing stuff around a TL494 (just looked at the datasheet) you're beyond my capabilities.
What I did fits my needs and is simple enough for me to implement. Playing with the PIC programming was fun too.
Good luck with your project.

p.s. I don't think you mentioned your car, but Subaru's 2006 and older use +12volts at the dimmest and close to 0 for the brightest. 2007 and newer use PWM.

My car is a 2010 Hyundai Genesis Coupe. I would suspect the lighting is PWM - just have to determine if it is common ground or common 12V like the Subaru you mentioned before I can put pencil to paper. I'll also have to figure out what duty cycles follow the car's instrument lighting to see how the duty cycle should follow the level. Funny you mention what I'm into is beyond you, I was thinking the same thing WRT what you're doing. I agree that a PIC would probably simplify things but I know nothing about them and programming one might be more time consuming than it's worth (for me). However I could see it useful to know for future projects... maybe I'll have a look to see how involved the coding is. Thanks.