driving LEDs, different drivers different light quality?

I finally finished my under cabinet LED light installation, I've posted here in the past about it. I'm using two different drivers, one I bought on ebay from China, it supports 16 LEDs at 700ma. I'm driving two 18" strips with 8 3.7v 700 ma 3K LEDS on each, connected in series. I inserted my VOM in the circuit and did measured 700ma of current. These lights produce a crisp with light with a shift to the blue end of the spectrum.

Using the exact same LEDS I'm driving the rest in banks of 10 in series powered from a simple transformer, bridge rectifier and LM317 regulator set to regulate current. Calculating the current of 700ma with the 1.2v internal drop in need a 1.6 ohm resistor, I created one using a number of resistors in parallel. I'm powering 4 banks from two regulators. These lights are considerably warmer in color even though they are the same LEDS I'm using with the other driver. Inserting my VOM into the circuit I only see 200ma of current, though calculations indicate it should be running 700ma.

I'm getting about 3.6v across the LEDs, so voltage is fine, it would appear I'm under driving them current wise. They are not as bright, but the light quality is more what I'm after. I guess the question is is it the current that controls the color temperature of the LEDs? Would I be better off driving them with more current in terms of lifespan? I would think under driving them would be better.

Changing the drive levels will definitely change the tint. Higher drive shifts toward blue, just as you observed. I'm surprised it's that noticeable though.

You should check the voltage across your LM317's, from input to output. Check both AC and DC. You need a minimum of 2-2.5V for the '317 to work correctly. If the DC minus some fraction of the AC is less than 2.5V, you don't have enough supply voltage for the number of LEDs in your string. I don't know exactly what the fraction is, but it's between 0.4 and 0.7.

Another easy thing to try is to short one LED in the string. If the others get brighter, you have too little supply voltage (or too many LEDs).

Another possibility is that you have too much supply voltage or too little heatsinking on the '317. If it overheats it will drop the output current to protect itself. If this is the case, the '317 would be so hot a wet finger would sizzle when touching it.

From 200mA to 700mA there should be very little difference in tint, barely noticeable if at all.

I say that, but if you have very poor heatsinking on the LEDs, i.e. if your LEDs at 700mA are running really really hot, then yes they will shift in color.


Transformer, bridge rectifier, LM317 ....... Capacitor?


Semiman

SemiMan said:

Transformer, bridge rectifier, LM317 ....... Capacitor?
Semiman

Click to expand...

Yeah, missed that. VERY important.

Thanks for the input...

Oh, yes, forgot to mention the 2200uf capacitor....

Yes, if I short out one LED the rest do get brighter. Also the Lm317s do have heat sinks and are almost cool to the touch, odds are one could support all the LEDs but opted for two just to be on the safe side. If I leave the lights on all day, the 18" units run about 14 hours a day, the aluminum channel I have them mounted to is moderately warm - I can hold it for a bit before it feels too hot. The other banks are cool to the touch, but they are on for much less time though they have been on for about 8 hours at this point and are still cool.

The transformer I'm using is a 50v ct. unit. i'm only using half of it - 25v tap simply because I can only find capacitors rated at 50v locally. all the lights on the transformer are using the same half in case you are wondering. with the full wave rectifier and capacitor I'm getting about 37v dc out which works out well with 10 LEDs per set even with a bit of loading.

For me there is one simple explanation - you have bought a very cheap Chinese led driver (export item only ). Which is switched-mode power supply with unfiltered output - means that your leds are not driven by a clear DC, but with high frequency noise from the driver . That's why when you are comparing them with properly powered ones you can see difference in light.

Szemhazai said:

For me there is one simple explanation - you have bought a very cheap Chinese led driver (export item only ). Which is switched-mode power supply with unfiltered output - means that your leds are not driven by a clear DC, but with high frequency noise from the driver . That's why when you are comparing them with properly powered ones you can see difference in light.

Click to expand...

Well I agree with you on the quality of driver, your comment w.r.t. the quality of the light is complete fantasy and wrong.

Not going to comment on a cheap driver pushing the tint way off, not my area of expertise.

But I think the fact that shorting one LED makes the others brighter points strongly to having too many LEDs in the string for the supply voltage. I'd love to know what the AC and DC voltages across the regulator are.

OK, did some measurements.

In is 33.5v
ground to LM317 out 30.6v
across LM317 2.39v
negligible AC at LM317

I'm using 10 LEDS rated at 3.6v be bank so it appears I'm under voltage a bit. They are running cool, but I would not object to a little more light output.

One less LED should bring you into regulation.

Negligible AC doesn't sound right. You should have at least a couple volts of ripple with 2200 uF and 750 mA. Oh, but at 200 mA, you'd have under a volt!

ok, will try that.

Should I redesign the soppy to get more current? what are the benefits of doing that?

Remember that DC on the input to the regulator has lots of AC on it ... The output does not. Measure AC at input to regulator. You may find its out of regulation more than thought.

The problem isn't current, it's voltage. You don't have enough to run 10 LEDs.

SemiMan said:

Remember that DC on the input to the regulator has lots of AC on it ... The output does not. Measure AC at input to regulator. You may find its out of regulation more than thought.

Click to expand...

If the input of the reg has AC and the output doesn't, then it's not coming out of regulation (but I think it is). At 200 mA there would only be 0.7V peak-to-peak ripple on the input, probably less on the output. I don't know exactly what the ratio is for that waveform, but that's less than 0.3Vrms. I would imagine that's what OP referred to as 'negligible'.

I did drop out 1 led from each set, light output went up as expected. Still runs cool, so that is good. For the heck of it I tossed together another supply to replace the cheap China made driver. It's running with similar specs to the others I built, mainly 200 ma current. The lights now run MUCH cooler, but the light quality is still not matching the rest. Has running it with the I'm assuming noisy supply permanently changed the LEDs?

Got to say I'm learning a lot from the posts, thanks. Can anyone point me to a good basic primer on driving LEDs? I'd like to learn more and better practices. A few folks have seen my setup and are asking how to do it and I don't want to give bad advice.

matt832 said:

driving LEDs, different drivers ... shift to the blue end of the spectrum.

Click to expand...

yes, this can happen. Usually it is merely observable, not really noticeable. The human eye is much more sensitive to small wavelength changes in the greenish-blue part of the spectrum.
Driving at different currents can result in as much as a 4nm shift in wavelength.

Some time ago, GE developed the VIO LED, which was primarily intended to allow dimming in LED lamps without any color shift (it accomplished this by using a violet emitter with blue phosphor, the phosphor had the same spectral emission, regardless of a shift in the excitation wavelength). The VIO never caught on and was discontinued.

SemiMan said:

Well I agree with you on the quality of driver, your comment w.r.t. the quality of the light is complete fantasy and wrong.

Click to expand...

No it isn't, maybe I didn't write it clear enough - with those cheap drivers 700mA is an average value - which means that you can experience effects similar to low frequency PWM dimming i.e. color shift.

More here :
http://www.digikey.com/en/articles/techzone/2014/feb/led-color-shift-under-pwm-dimming

Almost all small drivers China or not do that and it does not impact color one iota of noticeable difference. DUV is so small over a 50% current range it is completely unnoticeable at the same die temp.

Blue pumped white is not AlInGaAp with huge temp effects as per that article.

Anders current changes over a nominal range make almost zip difference on wavelength .... Temperature ... Big