Driving and dimming numerous LED strings together

mcgyvr

Joined Oct 15, 2009
5,394
I like that. Only question is if they make them bright enough and properly sized for my application. I don't suppose you know of one that's 14 mm to 16 mm long?
They are available in all sorts of sizes and plenty bright too..
Not to mention all LEDs are already in series/parallel already so you only need 1 constant current source vs 24 if you use discrete LEDs.. Component count drastically reduced..

As to frequency.. 60Hz can be picked up by "some" humans.. At around 75Hz no one will see the flicker..
 

k7elp60

Joined Nov 4, 2008
562
What mcgyvr said is possible as eye perception.
Remember that the ON/OFF cycles are about 120 Hz, but the human eye-brain system can not track much over 20 Hz. When you see a motion picture (a movie) you are not watching motion at all, but a series of stills flashed up in rapid succession, somewhere in the neighborhood of 30 pictures a second. Do you remember all 30 pics in that second? No. You perceive that as 1 second of fluid motion. How much more so, then, will you perceive light flickering at 120 Hz to be a smooth and constant light level?
The above information is from:
http://physics.stackexchange.com/questions/10938/perception-of-a-rapidly-flickering-light.
On another note, I have some high intensity white surface mount LED's, the yellow square is the led, but when powered up the light is round.
 

Alec_t

Joined Sep 17, 2013
15,121
How does it affect power requirements
It should be more power-efficient. Admittedly the boost converter is unlikely to be more than 85% efficient, but on the other hand the present proposal of 2 LEDs per string wastes a lot of power in the resistive current-limiters/volts-droppers.
Keeping the maths simple, assume a LED drops 3V at 50mA and you have 30 LEDs on a 12V supply.
With 15 strings of 2 LEDs each the total power = 15 x 12 x 0.05 = 9W.
If you connected all 30 LEDs in one string and used an 85% efficient converter generating 95V the total power would be 95 x 0.05 / 0.85 = 5.6W.
 

mcgyvr

Joined Oct 15, 2009
5,394
What mcgyvr said is possible as eye perception.
Remember that the ON/OFF cycles are about 120 Hz, but the human eye-brain system can not track much over 20 Hz.
I returned my "LED Christmas lights" as I couldn't stand the mains flicker..
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Flickering:
All due respect, this isn't a motionless sign, this is a fast-moving object. The faster that object moves, the quicker the duty cycle needs to be to compensate. 120 Hz is mighty slow for something moving 60 MPH - crunching the numbers, that's 44 feet for the entire duty cycle, which at 50% would be 22 feet on and 22 feet off. For outdoor, billboard-type signs, I have personally noticed this design flaw at highway speeds. It's difficult to make out and hugely distracting when something is lit up and staggered in movement like that. One of the reasons why I never cared for the LED billboards -- the concept is nice, but the execution has room for improvement.

The science behind how fast a human sees is accurate, but when that human is in motion, it changes the perception.

COB's:
I've tried searching, but I'm not finding a wide variety of sizes. This also gives me less quality control as it's mostly just rigging up someone else's design to a battery and controlling current. I like the idea, but until I am capable of producing my own COB, I will have to rely on whoever fits my need best, which may be hard to find.

High-voltage solution:
I like that a lot! If I want to drive 48 x 50 mA, 3.8 Vf LED emitters in a single string, that comes out to about 48 x 3.8V =182.4 volts required. Bumping it up to the nearest ten, that's (190V x 0.05A) / 0.85% = 11.176W. Doing this for the three sections independently, that's 16 x 3.8 Vf = 60.8V, we can round that to 65V, or (65V x 0.05A) / 0.85% = 3.82W, or 11.47W overall. I like the available redundancy of the three-circuit approach combined with it still using shy of one amp, so I am seriously interested in this approach. For that, what would the circuit look like? Is 65V a good number to bump it up to?

Here's what I'm thinking in schematic form to get started:
White_Test.png
 

mcgyvr

Joined Oct 15, 2009
5,394
You do NOT want to exceed 60VDC (including transients) ever as that moves you up out of the "SELV" voltage ratings and brings you into a whole new level of safety requirements..
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
You do NOT want to exceed 60VDC (including transients) ever as that moves you up out of the "SELV" voltage ratings and brings you into a whole new level of safety requirements..
65V would only be for the white portion of the circuit as that's the highest voltage drop, but duly noted. I can do four sections then (one per actual column of light, or 12 LED emitters), which is 12 x 3.8 Vf = 45.6V, bumping that up we have (50V x 0.05A) / 0.85% = 2.94W, or 11.76W overall, which is still shy of an amp, so I'll take it.

Is 50V acceptable?
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Is 50 less than 60 ? :p
Thought so, just making sure ;)

So the next question becomes how to bump the voltage up to a stable 50V. What is this type of circuit even called, and what kind of components does it use? I can hand-solder some SMD components (like the LED emitters), but if it's too small or too sensitive to heat, I'd likely kill the whole component. 3MM round LED size is about as small as I can hand-solder when it comes to SMD parts.

After that, it should be a simple matter of controlling the current, adding in some (PWM maybe?) dimming, and we should be set!
 

Alec_t

Joined Sep 17, 2013
15,121
Good catch on the 60V safety thing, Mcgyvr.
4 strings @ 50V sounds reasonable. Current draw would be 200mA total (but somewhat less would be advisable in the interest of LED longevity), which any competent PWM dimmer ought to be able to handle.
 

Alec_t

Joined Sep 17, 2013
15,121
Something like this is a bit over the top for your purpose, but searching might find you a lower wattage, cheaper 'DC boost converter'. Probably cheaper to buy than to build.
 

mcgyvr

Joined Oct 15, 2009
5,394
Good catch on the 60V safety thing, Mcgyvr.
4 strings @ 50V sounds reasonable. Current draw would be 200mA total (but somewhat less would be advisable in the interest of LED longevity), which any competent PWM dimmer ought to be able to handle.
My daily job is building devices that run on -48VDC so I'm well aware of that one..
 

mcgyvr

Joined Oct 15, 2009
5,394
I'm almost positive you can find a "boost" LED driver IC with PWM input.. and probably have the schematic all built by the "design tools" on the chip manufacturers sites..
Thats the way to go..
http://www.linear.com/products/step-up_(boost)_led_drivers

maybe this
http://www.linear.com/product/LT3952
maybe this
http://www.linear.com/product/LT3492
lots of choices depending on final LED schematic decisions..

Note I just quickly pulled those 2 at random to show whats out there.. I'm not saying they will or won't work for your application..
 
Last edited:

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Current draw would be 200mA total (but somewhat less would be advisable in the interest of LED longevity)
Regarding the underlined, what's the current per string you see as "advisable"? Would 45mA work? I've read somewhere that 90% light output in an LED is nearly indistinguishable to the human eye than 100%, although I have not been able to test this with my own eyes. Is a 5mA drop sufficient?

lots of choices depending on final LED schematic decisions..
What kind of final schematic decisions? I'm open to anything at this point.

Going to your first link, I narrowed it down by the requirements:
http://www.linear.com/parametric/step-up_(boost)_led_drivers#!gtd_!1034_>=60

That's still 11 to go through though. How do I pick the best one for my needs? I've checked, and some have analog dimming, but they all have PWM dimming. Not sure which is better, or if the PWM relies on an external 555 timer (which is still fine).

Some had three outputs, which would be super nice to have since the other two colors are a 2.4Vf. Crunching the numbers, that's (48 total / 3 strings, or 16-LED strings) x 2.4Vf = 38.4 V, well within spec. Those numbers become 57.6V if you start with two strings of 24 LED emitters each, too close to the 60V limit you mentioned. I might as well use the output if I've got it though, that's a mere 50mA that I'm adding onto it, not that much in the grand scheme of things.
 

Alec_t

Joined Sep 17, 2013
15,121
what's the current per string you see as "advisable"? Would 45mA work?
I tend to err on the cautious side, so personally would go with ~80%. Your mileage may differ. Check first to see what the "50mA" rating really means. Is that continuous or pulsed? With heat-sink or without? What ambient temperature? .....
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Crap! I'm SO glad you made me check the datasheet, I just caught that these are three separate LED units in one housing, which means triple the power requirements!!! In addition, I had selected these based on a few different criteria that, due to the change in design direction, no longer applied! Icky!

Thank you for the wake-up call, even if you didn't mean to give me one!

Phew, now that THAT has been settled, I've adjusted my LED choices. I went to both Mouser and DIgiKey, and kept digging... and digging... and digging... but finally found what I was after! I think, anyways.
For white, I'm using Samsung SMD units:
White - Samsung 65mA 2.8Vf LED emitters

Red and amber are based on Cree Piranha-type units. Surprisingly, these seem like the best "bang for your buck" when it comes to light output for the power usage (although I hope I'm mistaken here. Anyone got anything better?):
Amber - Cree 70mA 2.6Vf LED emitters
Red - Cree 70mA 2.5Vf LED emitters

Since the highest of these is now 2.8Vf, I can now use 16 * 2.8Vf = 44.8 V, or 50V as my specification standard, so that still works. 80% of 65mA to 70mA is anywhere from 52mA to 56 mA -- would running everything at 55mA for consistency work? Or is that too high for the white ones?

(Side note: I'd like the amber to be sequential still, as per my other thread on this subject -- hopefully that doesn't introduce more complexity into the matter, although I can adjust if it does)
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Perfect, 55mA it is then!

How do I narrow down the list of LED drivers? Anything special to look for? Quoting my earlier part about it:

[...]I narrowed [the list] down by the requirements:
http://www.linear.com/parametric/step-up_(boost)_led_drivers#!gtd_!1034_>=60

That's still 11 to go through though. How do I pick the best one for my needs? I've checked, and some have analog dimming, but they all have PWM dimming. Not sure which is better, or if the PWM relies on an external 555 timer (which is still fine).

Some had three outputs, which would be super nice to have [...]
 

Alec_t

Joined Sep 17, 2013
15,121
Your link goes to a blank form for ICs on Linear's site? Are you intending to build the LED drivers rather than buy off-the-shelf ones?
 

Thread Starter

MythicalCoder

Joined Aug 28, 2015
83
Your link goes to a blank form for ICs on Linear's site? Are you intending to build the LED drivers rather than buy off-the-shelf ones?
It seems they're having site issues at the moment. I can't even get the first link to load properly. Darn...

Not sure what you mean by building LED drivers -- I was hoping to get one of those IC's mcgyvr linked to, but I was having trouble narrowing it down to the specific one that would be the best fit.
 
Top