Pulsed Lighting = more efficiency?

[heckler73]

SDS referred to a NASA document in his Soft-start thread which caught my attention.
It was regarding the use of LED pulses (at 5000umols/m^2.s) vs a continuous stream at 50umol/m^2.s



I haven't sat down to do any calcs, but my intuition tells me this would be more efficient all around.

Hypotheis: Using a frequency of 6.5kHz and 1% duty cycle, allows higher currents to be applied with less concerns about thermal energies.

BUT...there is the issue of decay times. When "heat" is generated, it will have a dissipation time determined by the materials involved. However, if one is already using a system which handles dissipation for continuous currents, then it should have little issue (if any) in moving those phonons around from the pulses.

I suppose I can test this using some 3W LEDs I have kicking around--to see if it is technologically feasible in the first place,but if anyone has any experience with this idea, I'd appreciate the input.

 

[churchhaze]

I've always been skeptical of this theory. There used to be a lot of people trying this back around 2005 when people were trying to grow with traffic lights, and I always figured all experiments must have shown conclusive results of failure. I hear it brushed over every so often, and it never amounts to anything substantial.. Almost brushed under the carpet and forgotten about.

Straight off the bat, one of the reasons this doesn't make sense to do with leds is because of the efficiency advantage you get from running at a lower current. If you're pulsing the output at a higher current, but for shorter duty cycles, you will be driving the led at a very inefficient operating point. For this reason, even if you did manage to gain a boost in photosynthesis, it would likely be strongly outweighed by the loss in efficiency by running at the higher current. This point alone is damning when the theory is looked at from a practical perspective, even if the theory itself might be true.

I'd love to see evidence that shows otherwise, but I'm almost positive that average power output is more important than peak output (when it comes to photosynthesis). I think sds started an experiment on this and either changed his mind, or ...?

 

[heckler73]

I've always been skeptical of this theory.

That's normal...so am I (skeptical, that is...not normal), which is why I want to test it

Straight off the bat, one of the reasons this doesn't make sense to do with leds is because of the efficiency advantage you get from running at a lower current. If you're pulsing the output at a higher current, but for shorter duty cycles, you will be driving the led at a very inefficient operating point.

Are you sure about that? Semiconductors are odd characters in the world of electronics, being governed by Boltzmann statistics, fundamentally.
What's the point of running higher currents with active cooling systems, then?
NASA seems to see it as an apples-to-apples comparison. 50umol/m^2.s vs 1%of 5000umols (= 50umols)
The difference will be thermal...When the thermal excitation of majority carriers is mitigated, more optical energy is allowed to be created per unit time, being the sub-hypothesis I am operating from. We can see that effect in action just from regular operation. It starts out bright but dulls over the day. The idea here is to use high-injection levels to generate excess minority carriers, then let the LED diffuse that energy in the off-cycle which will continue to transmit photons with an exponential decay, minimizing waste heat build-up. This is where the crux of the efficiency question I am trying to answer lays hidden.

Some further experimental data for consideration c/o citizen science:

The LED shown below is a 08LCHW3 InGaN white LED with a 2MHz small-signal bandwidth. This LED is spec'd for a maximum DC current of 30 mA and a maximum pulsed current of 100mA (at 100µsec pulse width and 1% DS(DS=Duty Cycle)). The peak pulse current as measured by the voltage across Rm is about 1.1Amp, more than 10 times higher than the actual rated pulse spec. The oscilloscope traces below show the light output pulse using a high-speed photodiode circuit. The pulse period is 230 µsec and the pulse width is ~ 400ns corresponding to a low DS of 0.2%. The peak optical output power was measured to be ~ 30x that of the LED operating at 20mA DC. After several hours of operation, the LED showed no sign of degradation in peak optical output.

http://www.jensign.com/PulsedLED/index.html

As an aside, here's a funny note at the end:

As another benchmark, we can ask how close to the human eye must this pulsed LED be placed so that the peak illuminance Ev (lumen/m^2) will be the same as that of the sun at the zenith on a clear day? According to the RCA EO Handbook, the solar illuminance Ev is 1.2E5 lumen/m^2. (In radiant units, for the solar spectrum this corresponds to ~ 1.3kW/m^2 or 130mW/cm^2). The illuminance of a source is simply the luminous intensity divided by the distance squared Ev = Iv/R^2. Therefore the 1.1A pulsed white LED must be placed 3 cm or ~ 1" from the human eye to have a peak brightness equal to that of the sun. Again, the average brightness of the LED will be much lower.

Anyway, this is just my mind exploring a tangent. If no one has experimented with this, then it's just more reason to do it. Maybe it's crap? I don't know...I need to see it fail in action before dismissing it, unless I am staring at the wrong equations (also possible).

 

[churchhaze]

I have no doubt that this trick can fool the human brain into thinking something is on continuously when it's only on for a short period, but the human brain is based directly on electrical signals, whereas plants have a "hormone/enzyme lag". Eyes are also made to sense light, while chloroplast are designed to collect as much energy as possible. I've worked on plenty of designs that involved scanning leds one at a time to make it appear to a human like multiple are on at once. This often makes it possible to use the sourcing/sinking of a microcontrollers output since significantly less average drive current can be used to make all the leds (say for example in a 4 digit 7 segment display) look equally bright.

About the "Are you sure" part, yes.. You can see from looking at the charts on the datasheets that efficiency is significantly improved when operating at lower currents, and so far I haven't seen an LED where that isn't the case. The datasheets indicate a quantifiable amount of current droop and temperature droop which many members here have run through pixel counting tools to find the luminous efficacy of source at various currents. There are a ton of spreadsheets laying around put together by supra, that along with luminous efficacy of radiation for a given spectral distribution curve, very good estimates of efficiencies at different currents and case temperatures have been made and put into very convenient spread sheets for just about every led discussed here. You can basically check these spread sheets to see what efficiency is for different cobs at different currents, and if you don't trust the spread sheets, you can check the datasheets and count the pixels yourself.

 

[churchhaze]

Here is an example of how the spread sheets are laid out for each cob (credit to @SupraSPL )

he's basically made one of these for everything by now. It says what the efficiency is at given currents, and how much the startup cost is per radiant par watts. The increase is significant, between 34.34% and 51.67% in the current range listed in the table.

 

[Greengenes707]

SDS referred to a NASA document in his Soft-start thread which caught my attention.
It was regarding the use of LED pulses (at 5000umols/m^2.s) vs a continuous stream at 50umol/m^2.s

View attachment 3387263

I haven't sat down to do any calcs, but my intuition tells me this would be more efficient all around.

Hypotheis: Using a frequency of 6.5kHz and 1% duty cycle, allows higher currents to be applied with less concerns about thermal energies.
BUT...there is the issue of decay times. When "heat" is generated, it will have a dissipation time determined by the materials involved. However, if one is already using a system which handles dissipation for continuous currents, then it should have little issue (if any) in moving those phonons around from the pulses.

I suppose I can test this using some 3W LEDs I have kicking around--to see if it is technologically feasible in the first place,but if anyone has any experience with this idea, I'd appreciate the input.

So in the test, they are supplying 50µmols to each? One is continuously on to accumulate to that. While the other is higher intensity when supplied, but because it is pulsed, is only is supplying 50µmols as well when time(integral) is factored in?

I would expect equal performance. Plants don't use an instantaneous measurement of light...they use the energy accumulated over the photoperiod(assuming no oversaturation). The same concept as adding time to your light cycle with a 13/11 cycle...increased DLI, or energy captured...or allow a lower intensity source more time to make up for it's lack in output.

Indoors DLI is not on our minds because nothing changes with our lighting usually...if we know the instantaneous reading...we know the DLI. But in an environment, like outdoors, where light is constantly changing, DLI becomes the necessary measurement.

If they were able to get the same amount of photosynthesis from less quanta that would be growing more efficiently...also breaking laws of nature/physics. Photosynthesis is a quantum process.

That test has rearrange how the light is given to the plants. It's still the same amount of light, and takes the same amount of power to produce it.

 

[Doer]

Kinda like turning the lights on and off on a factory floor. Might make the workers irritable.

And Heck, the point is, these LEDs of any kind, are born immaculate perhaps, but current beyond a 1/10 of a milliamp is on the inefficiency curve. From there it is all trade offs.

I had the same sort of head scratch on this. It isn't the heat causing it, The inefficiency curve is causing the heat. The heat makes matters worse. It is a form of electron tunneling, maybe.

So regardless of the heat transported, you'd like to keep the junction temp close to 25c, for other reasons.

It seem the question is at what low current can you get what high light output? 700ma is better than 1400ma just because that is where the efficiency is to justify the cost....in some of our minds. Others, may run Max amps knowing they are not necessarily being the most efficient but they have other trade offs in mind.

I too wanted to run the CXB3950s at 2/3 power but after being shown the error of my thinking I backed that off to 1/3 power.

I have been made to see my light. What is the sense of running the most efficient COB in the least efficient way? <dooh>

So, by pulsing very fast could you run less current and get more light?

 

[Positivity]

As LEDS get more efficient you can run them harder. Easier to dim down then dim up..

Personally I do it mostly for fun and learning. Commercial should be right in the efficiency sweet spot

 

[SupraSPL]

Are you sure about that? Semiconductors are odd characters in the world of electronics, being governed by Boltzmann statistics, fundamentally.
What's the point of running higher currents with active cooling systems, then?

As far as current droop goes, I did test it to see if the effect goes all the way down to very low currents and it does. You can verify this with a lux meter and kill-a-watt or for more accuracy a pair of multimeters.

When we size the heatsinks many of us go by surface area/dissipation W rule of thumb. Our COBs are often run at low current but we pack on lot of power in the canopy so the heatsink has to be appropriately sized whether active or passive cooled. When it comes down to it, current droop losses are much larger than temp droop losses so we have learned to invest in that direction. COBs take advantage of that strategy very well. Temp droop can be as low as .25% when run soft and 2% when run hard, so there is not much to be gained there.

But you could always try pulsing the lights at low current to see if it yields more g/W than a steady current?

 

[heckler73]

So in the test, they are supplying 50µmols to each? One is continuously on to accumulate to that. While the other is higher intensity when supplied, but because it is pulsed, is only is supplying 50µmols as well when time(integral) is factored in?

That is how I interpret it also.

I would expect nothing less than equal performance. Plants don't use an instantaneous measurement of light...they use the energy accumulated over the photoperiod(assuming no oversaturation).

And that seems to be what NASA is saying as well.

That test has done nothing but rearrange how the light is given to the plants. It's still the same amount of light, and takes the same amount of power to produce it.

That part is where I make a screwface. In constant current, one needs to constantly deal with thermal waste and actively getting rid of that to maintain the efficiency. In a pulse, that thermal energy has time to dissipate before the next "feed". Consider transistors...How is it they can operate at high pulsed-currents without worrying about heatsinking?

Do you see what I am getting at here? If the plant doesn't care, then take advantage of the bulk reduction in thermal management.

Here's what CREE recommends after they played around with 1kHz pulses:



CREE says I should do it in the follow-up note.
I agree... At 1kHz, your eyes aren't going to see a difference, and it doesn't seem the plant will either. Besides, it will be fun to see how far a Cheapo Chinese LED can be tortured, right?

Perhaps a little more background to my madness would be useful here.
Ladies and gentlemen,
The Haynes- Shockley experiment:

 

[churchhaze]

That is how I interpret it also.


And that seems to be what NASA is saying as well.

What NASA is saying is that given the same total uMol/m^2, the rate of photosynthesis will be the same. That is not the same thing as saying it costs the same amount to produce those same amount of total uMol/m^2. A lamp running more efficiently (given SPD doesn't change) will produce more photons per watt per second, and thus it will cost less to produce the same amount of total photons.

Nobody here seems to have a problem maintaining continuous current with a case temperature around 30-40C without going through great lengths to do so. Even at 85C, the efficiency loss insignificant to compared to the loss of running at a higher current.

 

[Devildenis69]

Hi guys

and what's about pulsating it at a regular current ?
seems ok for the led, few things to manage for the driver ...

on another board, a member with quiet a good biological background, figured out the "no-light time" which doesn't disturb the plant ... I'm gonna try to find this

 

[heckler73]

In the meantime, as I mull over the details, have a gander at this little gem I stumbled across:

http://leds.hrt.msu.edu/assets/Meeting/Optimal-spectrum.pdf

 

[SupraSPL]

Good data! (with the caveat that the LEDs the are referring to on page 18 of the PDF were mostly generics and were already out of date).

 

[heckler73]

In the process of finding that pdf, I was reading discussions on LinkedIn (of all places) where there was questioning going on about pulsing. I am not the only one thinking about this, at least.
What I did see was a good point of rationale for doing this (other than my thermal hypothesis). I believe it was a commercial GH who is using high-current pulsing to get more canopy penetration. This makes sense, too. If the intensity is increased, more light will penetrate, and because it is pulsed, it can be dropped to the canopy without concern of saturation. I suppose that raises the question of the necessity for reflectors and lensing, then.

PS I ripped the data from a CXA2590 chart and found output peaks at ~3.69 A and ~200% for Tc=298K, so that would be an upper bound on that COB, and it falls in line with CREE's 300% figure for over-driving.




It's all just speculative voodoo from my end, though. I need to set some bulbs on fire!
SCIENCE !!!!!111@2!

 

[heckler73]

I found an open-access paper from China that did some experiments with pulsing.
It has some interesting results, and compared to the handful or so of papers I've read tonight, it's not bad all around, despite the less than fluent English.
It has some answers for me to ponder, at least, so I thought I'd share.

http://www.sciencedirect.com/science/article/pii/S1875389211005049

It's too bad I can't share this other one I found from China. It's bloody hilarious. I don't know how some of these things make it past peer-review.
The title is:
Backlight LED Pulse Drive Method and Luminous Efficiency
by Huaijiang and Hong

The closing line caused me to howl:
Because the authors understanding is limited, some theoretical depth of study is not enough.

 

[Doer]

Who are the peers is the constant question.

 

[Doer]

Get ready for a joke.

You've been here 5 minutes. Buld the device! No need to plan and plan!!

 

[Doer]

All this makes sense somehow on a quantum level.
The question is about how to make an experiment?

And it occurs to me that you don't have to flash all the lights so you never get to no light.

You could flash as a supplement to see about results of adding intense flashing to get higher molar flux without ruining COBs, wasting juice, or depriving/overdriving the factories to distraction.

And iac, on a quantum level I wonder what continuity has to do with it anyway? It could be rather instant. One photon=one event of synthesis. It may not matter about the photon before the event, or after. A Quanta, as they say.

 

[Doer]

That first slide deck is a gem for sure

Just that one picture of a warm white seedling being 2x the size of the cool white seedling is course changer for me

I was planning to go with cooler COBs for veg.