Base pump, WV, μmol/J, phosphor CE, LED effeciency

[ChiefRunningPhist]

How do your eyes interpret a leaf as green? A green photon must be reflected off the leaf and be received by your eye's sensor.
Unless green photons positively affect the secondary metabolites, for the sake of photosynthesis green is very ineffective/inefficient. And I do not care about how deeply green is absorbed.

It doesnt mean its all being reflected, and I've seen reds on MJ before too. The science doesn't lie. Read the paper, or perhaps just the title.

Looks like you are correct when talking about the action vs absorption though. I'd have to agree with you that action is the function I'm after. I've never really seen any grow lights advertise that, I was under the impression they were compounding from reading the papers.

In photosynthesis photons are converted to electrons. After that, it depends upon the electron transport chain and the photo chemical reactions.

I'm decently versed in the chemistry, not debating that.

 

[alesh]

The luminosity function is described as being "normalized" about 555nm. And looks like a normal distribution. Albeit you'd have to tweak the equation to model it more or less but it looks like a normal distribution, no?
View attachment 4376307
View attachment 4376309 View attachment 4376311

Looks similar to a normal distribution function but isn't one. If you look closely, you'll see it's not symmetric. You'll introduce some inaccuracy to your calculations if you treat it as a normal distribution function.

A mono LED can be modeled relatively easy I think?? At least if its symmetrical? Though I hear ya, if I were to find an easily modeled equation of the Luminosity function, adding and subtracting known functions to create the SPD function is a huge pain in the ass lol

See above. Some mono LEDs are closer to a normal distribution function some are quite different.

I guess I'd ask for more clarification as to why I can't use peak base pump, (caveat of assuming the base pump SPD is symmetrical)?

Energy of a photon is inversely proportional to its wvl. It's not linear. Therefore, you'll introduce some inaccuracy if you calculate with peak wvl only. Depending on the specific LED, this inaccuracy might be negligible - in case of LEDs with a narrow FWHM.
If you use peak wvl only for a broader spectrum, such as a phosphor converted LED you're inadvertently throwing a huge inaccuracy in your calculation. It's essentially useless.

If the phosphor CE is at best 1, then the qty that is produced at base = total qty. The resulting energy difference in photon downgrading is expressed as heat on the phosphor film, it would mean low CCT chips were warmer than high CCT chips, I think this is observed in reality. When you measure μmol/s in a sphere, you're counting the total emmision. If it's understood that there is no loss in qty of photons, only loss in energy of photons, then you should be able to approximate base pump chip effeciency based on the base pump peak WV and total μmol/s measured. I care less so about the additional losses on the phosphor film, more so about raw performance, Ie what I put in in terms of watts to what I get out in terms of photons.

Manufacturers treat phosphor LEDs as a whole product. If you get a LED specified at certain (photon/radiant/luminous) flux, the phosphor losses should be included.

 

[alesh]

@alesh have you digitized the photoreceptor absorption & action spectra? That's something I'm trying to do, then I can compare lamp SPDs and their output to the total % utilized by the plant.. Create a new metric to judge grow lights from.. Like a UVI but weighted to photoreceptor absorption peak WVs instead of just 295nm like UVI.

I believe I did digitize them. However I don't like using those as they're of dubious relevance. I prefer well defined SI units.

You might want to look into some posts by @stardustsailor though. Sadly he seems to be MIA. Check out his lights too. Craftsmanship on those is excellent.

 

[alesh]

In my estimation, it seems ~22.5% of base pump radiance is converted to heat via phosphor film for this luxeon 5700k, even though photon qty is essentially maintained. I need to correct and use a different term than electrical effeciency. Effeciency is energy out over energy in, so the effeciency metric I've been describing is not exactly total effeciency because it's not accounting for phosphor film energy losses, it's only describing the electrical effeciency per base pump photons created.

I used the luxeon color line to come up with the 22% figure.
View attachment 4376442
According to the data sheet a cool white L1SPCW90002000000 pushes 1.57μmol/J at typical/nominal.

According to the data sheet the deep red L1SPDRD0002000000 pushes 2.65μmol/J at typical/nominal.

View attachment 4376448
Looks like peak base WV are ~455nm for the cool white, & ~664nm for the deep red.

455nm/119.6 = 3.80μmol/J
664nm/119.6 = 5.55μmol/J

455nm cool white:
(1.57μmol/J) ÷ (3.8μmol/J)
=
0.413; or 41.3%

663nm deep red:
(2.65μmol/J) ÷ (5.55μmol/J)
=
0.477; or 47.7%

When comparing these effeciencies to a different and trusted source, I get..

"
455nm cool white = 32.0% effecient
664nm deep red = 47.9% effecient "

It seems to me that the 0.4% difference (47.7% ÷ 47.9% = 99.6%) on the 664 nm deep red was due to the SPD of the 664nm not being symmetrical (amoung other roundings etc), and the 445nm cool white had a 22.5% reduction in effeciency (32.0% ÷ 41.3% = 77.5%), or a 22.5% reduction in [(mW of total radiant output) vs (mW/μmol of base pump)], and due to the losses on the phosphor film. For this 5700k example it looks like ~22.5% reduction in total effeciency due to phosphor film losses.

This could explain the discrepancy I've been having with Samsung saying 3.0+ μmol/J ( I calculate 79% effecient in base pump of 455nm) despite the effeciency from the trusted source giving a much lower effeciency (~70% effecient). Aha! I thought they were just lying through their teeth..

Maybe, phosphor film losses scale linearly with CCT & CRI, but idk, would be useful info, most SPDs of like CCTs & CRIs are pretty similar company to company.

In conclusion, it seems you can calculate a monochromatic LED chips total electrical effeciency, via its measured μmol/s output and its peak WV, but, using this technique for white light requires an additional phosphor loss be added to the the electrical effeciency per base pump photon created (I'm dubbing it the "base pump EE") to finally arrive at a "total electrical effeciency".

@alesh I couldn't message you, says you're locked up like fort Knox lol, hmu sometime..

This whole calculation is wrong.
You might use peak wvl for deep red or royal blue mono with just some minor inaccuracy but you can't use it for cool white spectum as it includes a lot of other wvls. Cool white spectra, depending on the manufacturer of course, generally produce around 4.5 µmol/J.

BTW if you look at the SPDs from the data sheet, you can see that royal blue and cool white peak isn't exactly aligned. It's not exactly the same chip.
Also the phosphor converted white LEDs have their output specified with the phosphor layer already. The losses are counted in.

BTW2 Some other manufacturers do specify their LEDs radiometric output. Then it's simply η= Φe / (U*I)

 

[eatled]

Read the paper, or perhaps just the title.

I've read many papers on green. But green only applies if green is specifically needed for that species.
Green is an expensive LED color in cost per µmol. Green does well for photometric due to the 683x luminosity boost.
Green does poorly in absorption and action. You are still paying for those reflected green photons.
Each photon, regardless of wavelength, is converted to one electron.
For photosynthesis, wavelength does not matter, so why not use the more efficient and beneficial wavelengths and forget the inefficient green.

 

[ChiefRunningPhist]

Energy of a photon is inversely proportional to its wvl. It's not linear. Therefore, you'll introduce some inaccuracy if you calculate with peak wvl only. Depending on the specific LED, this inaccuracy might be negligible - in case of LEDs with a narrow FWHM.
If you use peak wvl only for a broader spectrum, such as a phosphor converted LED you're inadvertently throwing a huge inaccuracy in your calculation. It's essentially useless.

No this is incorrect. The energy per photon per nm scales linearly. Its why a constant can be used as a conversion factor, ie 119.6.

See above. Some mono LEDs are closer to a normal distribution function some are quite different.

In my experience, using this technique with unphosphored chips results within a max of ~3% deviation. Most most times less than 1% deviation. I understand the principle behind the concern but due to the narrow emmision range and due to the small difference in energy per photon/nm the differences are neglible as you say.

This whole calculation is wrong.
You might use peak wvl for deep red or royal blue mono with just some minor inaccuracy but you can't use it for cool white spectum as it includes a lot of other wvls. Cool white spectra, depending on the manufacturer of course, generally produce around 4.5 µmol/J.

BTW if you look at the SPDs from the data sheet, you can see that royal blue and cool white peak isn't exactly aligned. It's not exactly the same chip.
Also the phosphor converted white LEDs have their output specified with the phosphor layer already. The losses are counted in.

BTW2 Some other manufacturers do specify their LEDs radiometric output. Then it's simply η= Φe / (U*I)

No, I'm pretty sure it's correct. Please re-read.

I'm describing base pump EE, not total chip effeciency. If there is a phosphor loss then this phosphor loss would further reduce base pump EE. Base pump EE = the total chip effeciency when no phosphor is applied. When phosphor is applied then chip effeciency = base pump EE - (phosphor loss). Like @eatled said, we only care about total photon output, so the excess energy left on the phosphor film is reducing the total chip effeciency but when CE =1 does nothing to reduce base pump EE or total photon output. Base pump EE is what I care about when it comes to effeciency of plant utilization.

 

[ChiefRunningPhist]

I've read many papers on green. But green only applies if green is specifically needed for that species.
Green is an expensive LED color in cost per µmol. Green does well for photometric due to the 683x luminosity boost.
Green does poorly in absorption and action. You are still paying for those reflected green photons.
Each photon, regardless of wavelength, is converted to one electron.
For photosynthesis, wavelength does not matter, so why not use the more efficient and beneficial wavelengths and forget the inefficient green.

https://chuckersparadise.com/threads/spd.433/post-23673

^^^just showing that I understand how it works.

Ya I don't think you're giving it enough credit. But to each his own. Imo intial cost really shouldn't come into play if trying to build the optimum light. Most these manufacturers run their lights out at $1k+ already for a 4x4. If the results are there, the improved tech and prices will follow. The reason blue is so much more effecient is due to the white light market. It pushed blue to be better and better. Now reds are starting to match effeciencies because of the horti industry. You can see the current film skips ~480nm conversion, so they need to develop a better film for the 480nm range, not sure if Nichia has done that with their optisollis, Id have to look at their effeciency, but imo its all market driven. I can post more studies showing added benefit to including green if you want but makes no difference to me what you do.

 

[eatled]

I don't think you're giving it enough credit

Even bacteria don't like green. Since 1882.

 

[ChiefRunningPhist]

I'm trying to make an indoor grow environment that is economically feasible for growing vegetables.
Initial cost is a big factor. If adding a feature increases cost, it must show an ROI not contribute to the debit column.
So for now that leaves me with Samsung and Bridgelux strips.
I prefer Bridgelux because they run cooler and need no heatsink.
I have some Bridgelux Gen3 Slim on order. I am going to test them thermally. If they pan out I will order 100 of the 2700K 90 CRI.
I've been experimenting with strips for about 3 years now.
I do make some of my own with LEDs that are not available on strips.
Luxeon 2835 White Horticulture and Fresh Focus will be my next builds.
I have a bunch of blank 560 mm strips to build my own.

I do have some green strips made with Lumiled's Rebel but I used those only for Xmas decorations for the past two years.


View attachment 4376907

FWIW, in my experience 2835 luxeon c-line is more effecient than their horti lineup lol. idk why, was curious to me too..

I looked at the rebels for the 480nm gap, if I remember, you have to buy lower fluxing bins to get the peak WV you want. It seems the lower flux correletates not to lower performance but to a shift in the peak WV.

I was that way. I hear you on initial cost, but if green adds morphology to the plant then you're really not optimizing the veggies per the R:B SPD. Because the LED last so long, their cost per day used is ~0. So for me, I'm more concerned about the quality of the product being made and cost of operation, than what it cost me initially. It will just take a bit longer to get my return.

If it costs $100 to make a R:B & $125 to make ideal, I'm only extending the time it takes to get even by 20%, though assuming there's added beneficial morphology from the ideal, then my only loss is really the 1/5 extra time. In either scenario I'll have to wait to get my return, so imo, I might as well be making a better product while I wait. Once the wait is over the choice will be clear.

Cool strips! Are those pre-stenciled? Just add solder paste and chips? Or what exactly am I looking at?

 

[ChiefRunningPhist]

Even bacteria don't like green. Since 1882.
View attachment 4376912

That could be a beneficial thing.

 

[ChiefRunningPhist]

I use the solder paste in a syringe. I only build two at a time and it doesn't me long to apply the paste.
I used Ki-CAD to create the Gerbers and had PCBWay fab them with an ENIG finish.
The copper pour is for thermal conductivity to the screw holes. Can be mounted to a heatsink if necessary.

View attachment 4376930

Very nice.

What was MOQ from pcbway? I'm using EAGLE, but not preferential of one program over the other, just the first program I found.

 

[eatled]

If it costs $100 to make a R:B & $125 to make ideal

I've been on the red white and blue band wagon for a couple of years. The problem has been R and B are expensive.
The new Luxeon 2835 Sun Plus are now available at a reasonable cost.
I'm thinking to build three 48V strips one each with R,W, or B with the Luxeon 2835 Sun Plus where W is the PNK horticulture white or Samsung LM301B
Strip cost, assembled, would be about $25 each with 96 LEDs. That's a big bump over a $6 EB Gen3.
Then a board with an ATtiny and multiple dimmable (analog and PWM) drivers using the LM3414HV.
This board would have a USB connector to set the intensities from a laptop. Maybe Wi-fi.

 

[eatled]

What was MOQ from pcbway?

I purchased 20 strip for $150 including DHL shipping. 2 oz, ENIG
You could get 5 for $21 ($4.20 ea) with standard everything, 1 oz., HASL
Shipping would be $17.
I usually do 3 or 4 different PCBs at a time to keep shipping cost down per board.

 

[ChiefRunningPhist]

I've been on the red white and blue band wagon for a couple of years. The problem has been R and B are expensive.
The new Luxeon 2835 Sun Plus are now available at a reasonable cost.
I'm thinking to build three 48V strips one each with R,W, or B with the Luxeon 2835 Sun Plus where W is the PNK horticulture white or Samsung LM301B
Then a board with an ATtiny and multiple dimmable (analog and PWM) drivers using the LM3414HV.
This board would have a USB connector to set the intensities from a laptop. Maybe Wi-fi.

I'm making my own drivers because I'm not satisfied with the effeciencies of the low wattage drivers currently available, or their reduced performance when dimmed. It's honestly what's taking me so long. Imo it would be a waste to buy the high effeciency chips but only supply them via a 90% PS.

The lm3414 buck converter will only further reduce the effeciency of the multiple drivers. Id leave it out and just use the AC/DC drivers. You can actually get PWM-able small wattage drivers. Whatever the effeciency of the base PS you're using to run the lm3414 will be reduced by an even greater amount by using an lm3414 between LED string and base PS. The additional effeciency loss would be dependent on the lm3414 load, and output V of the PS.

I'm using an esp32 which is Wi-Fi connectable as well as usb connectable. Check it out. I'm implementing both tuning methods, manually or via phone w Wi-Fi. I really like the ATtiny but the esp32 can be coded with dynamic dimming profiles as well (I'm assuming thats why you want to use the ATtiny instead of manual POT knobs, for sunrise sunset ect..).

 

[ChiefRunningPhist]

I purchased 20 strip for $150 including DHL shipping. 2 oz, ENIG
You could get 5 for $21 ($4.20 ea) with standard everything, 1 oz., HASL
Shipping would be $17.
I usually do 3 or 4 different PCBs at a time to keep shipping cost down per board.

That's really reasonable. Ive been scouting some pcb fabricators but leery to send my gerbers because I'm afraid of my shit getting ripped off lickety split! Do you know where they are based from? I guess I could just Google it too lol

 

[ChiefRunningPhist]

There is a reason low wattage drivers are inefficient.
Do you have a lot of switching power supply design experience?
It's going to be tough to beat Mean Well.
I do not think it would be possible to design your own with better efficient and lower cost than an off the shelf Mean Well.
I have 18 years experience running my own manufacturing.
My costs were low because I had my own Zevatech pick and place. An entire manufacturing line in my house.
I do not believe I could beat Mean Well. Not even close.

MW is not even close to the top effeciency right now. My PS might not be cheaper, but like I said I'm not worrying about cost, just worrying about being the best.

And FYI I've designed multiple SMPSs.. I'm currently using one as my lab bench PS lol

switching frequency tweaked to the output current.

Lol can you explain that? Can you explain how varying the switching frequency will change the brightness? What R value are you using between FS and GND? What power rating for Rfs? How are you programming the ATtiny? Did you make an ATtiny programming shield?

 

[ChiefRunningPhist]

How will you know if the photomorphogenesis benefits?
You need a $500,000 mass spectrometer to test.
Each experiment takes months to complete.
It could take months just to identify each cannabinoid and other secondary metabolites e.g. terpenoids.

Its kinda what I been screaming about for the last few pages.. The studies have already been done..

 

[ChiefRunningPhist]

I don't think so. The only study that I know of for narrow band photomorphogenesis was for Basil

Yep they've been done. I offered to post but you couldn't have been bothered so go find em yourself.

And what kind of efficiencies have you gotten? Or expecting? At what cost? Electricity does not cost very much.
An HLG-600H-48 with a 7 year warranty is 96% efficient, 150 W 94%. So you want to beat that?

So if you increase the efficiency 3% on a 1333 watt fixture (1 kW/day). How much do you save running 18 hrs a day?
0.3¢ per day.
If the extra cost is $20 for 1333 watts of drivers it will take 18 years for an ROI at 10¢/kW.

Lol why not grow with HPS then?

Ultra high efficiency. I have a buddy who actually is a rocket scientist and is phenomenal at designing ultra high reliability / efficiency.
He was my partner in a venture 12 years ago. We built a 6 watt 1U 1000 MIP server. The power supply design took for ever to design.
The server won Editors' Choice Best of Show Award at the 2007 Computer Telephony Expo.
He is now working in CA for a firm that does contract jobs for DoD and NASA.

Lol. Good for your buddy. Seriously, good for him. I've some similar friends, and now that youve mentioned NASA lol, actually developed the batteries for them. Good to have smart friends huh? Amazing the shoulders you rub at an engineering university, and I don't count myself out though either tbh.

Watch yourself. That pissed me off. Some things you should keep to yourself.
Don't tell me how little I know, dazzle me with your brilliance.

Really could care less if it pissed you off. You don't know what you're talking about but really trying hard to flex like you do and it's getting tiresome.

You would not be the first. And failed. Look at Steve Jobs and the NeXt Cube.

Ya that Steve jobs... What a failure... Lol if you don't like the truth I've given you about your PS design I'm sorry. I was not trying to offend you, I was only trying to set you straight because it was obvious you were confused. I don't think I did this in a demeaning way. I'm still waiting for my other questions to be answered, if it seems that you're compitent in the basic understanding of those questions and what they'll mean for your design then I'll be more open to hear what you have to say regarding SMPS design ect... Ready...Set... Google! Lol until then I just view it as a grumpy wannabe trying to mine me for my knowledge...

 

[diggs99]

 

[ChiefRunningPhist]

^^^This dude.. Smh..

If anyone wants ANY graph digitized it's super easy. I just digitized the energy weighted yield photon flux graph and imported into excel in 10 seconds.

Check out this link, you upload a jpeg of the graph you want and then you pick a few known points for the webapp to base the plot from and then click digitize.

https://apps.automeris.io/wpd/

Select which color line you want to digitize, the energy weighted curve was blue and that's why you see a blue filled square by the foreground color box because I wanted to digitize the blue line. The blue line on the graph looks red in this pic but its just because all the digitized data points after it ran were colored red...



Data from digitizing, I plotted to see what it looked like..