looking for LED-efficacy-data as complete as possible

What about these fucken noobs that show up and demand all the data laid out for them in an easy to read and compare format.

I have an idea, why dont you make this wonderful and magical document comparing scores of diodes that will need updating weekly to stay current.

Please share when it's done thanks in advance.

klx said:

What about these fucken noobs that show up and demand all the data laid out for them in an easy to read and compare format.

I have an idea, why dont you make this wonderful and magical document comparing scores of diodes that will need updating weekly to stay current.

Please share when it's done thanks in advance.

Click to expand...

I wish someone would do that for us!

Sadly every time we design an LED fixture we need to go through every single type of diode we are considering using and calculate the umol/j to work out how efficient they will be which determines which diodes we end up using. The problem is that the manufacturer's data is not always accurate. That's one of the reasons we use Nichia diodes because they are very conservative with their flux bin data and often the diodes perform better when we test them than they are on paper. We can't say the same for other manufacturers. We are fortunate to have access to a goniometer so we can test individual diodes against manufacturer claims.

Even if you calculate the efficiency on paper it doesn't mean the diodes you have in your hand (or on your board) are actually that efficient. Chinese manufacturers always claim to have "top bin" Samsung diodes when we know that none of them really do. The best diodes go to those who have the best relationship with Samsung and in most cases pay for the privilege. There are only so many "top bin" diodes to go around so there is no way everyone can have them. That's what makes this exercise somewhat futile. The only way for the OP to actually know how efficient his Quantum-board knock-offs are is to test them in a sphere or goniometer. In reality he will have no idea what flux and voltage bin diodes will be on his boards. He can ask the manufacturer but he may not get an honest answer. They may say "SE" bin but there is only one way to prove it. Even within SE bins there is up to a 7.5% fluctuation in flux with at least four different voltage bins that have up to an additional 14% difference in efficiency.

A 2.8V 73lm 5000K CRI80 LM281B+SE chip may look great on paper but in reality it doesn't exist. And that theoretical maximum is still only 174lm/w which is well behind the top 3030 diodes such as Nichia's 757 series that are now 228lm/w — on average! The top bins are even better. The CRI70 5000K diodes we will be upgrading to are over 71% efficient with an output of 3.27 umol/j. I only know this because we have done the hard work of calculating and physical testing.

PJ Diaz said:

That diode isn't intended for horticulture applications, so you aren't going to derive umol readings from it. You can get luminous flux ranges however. It's basically a linear curve, dependent on current. You would have to do the math.

View attachment 5194623
View attachment 5194624

Click to expand...

The forward current is almost linear, but the forward voltage varies a bit more. Once you multiply forward current by forward voltage to give overall wattage you can see how much more efficient the diode is at lower currents in terms of lm/w or umol/j.



100% luminous flux at 150mA and (let's say) 3V (A2/A3 voltage bin). That's 100% at 0.45W

Let's assume 52% luminous flux at 75mA, which is now 2.875V. That's 50% at 0.215W which is about 9% more efficient.

38% luminous flux at 50mA is now around 2.75V for 0.1375W which is almost 25% more efficient.

These are rough figures based on eyeing the above graphs.

Grow Lights Australia said:

I wish someone would do that for us!

Sadly every time we design an LED fixture we need to go through every single type of diode we are considering using and calculate the umol/j to work out how efficient they will be which determines which diodes we end up using. The problem is that the manufacturer's data is not always accurate. That's one of the reasons we use Nichia diodes because they are very conservative with their flux bin data and often the diodes perform better when we test them than they are on paper. We can't say the same for other manufacturers. We are fortunate to have access to a goniometer so we can test individual diodes against manufacturer claims.

Even if you calculate the efficiency on paper it doesn't mean the diodes you have in your hand (or on your board) are actually that efficient. Chinese manufacturers always claim to have "top bin" Samsung diodes when we know that none of them really do. The best diodes go to those who have the best relationship with Samsung and in most cases pay for the privilege. There are only so many "top bin" diodes to go around so there is no way everyone can have them. That's what makes this exercise somewhat futile. The only way for the OP to actually know how efficient his Quantum-board knock-offs are is to test them in a sphere or goniometer. In reality he will have no idea what flux and voltage bin diodes will be on his boards. He can ask the manufacturer but he may not get an honest answer. They may say "SE" bin but there is only one way to prove it. Even within SE bins there is up to a 7.5% fluctuation in flux with at least four different voltage bins that have up to an additional 14% difference in efficiency.

A 2.8V 73lm 5000K CRI80 LM281B+SE chip may look great on paper but in reality it doesn't exist. And that theoretical maximum is still only 174lm/w which is well behind the top 3030 diodes such as Nichia's 757 series that are now 228lm/w — on average! The top bins are even better. The CRI70 5000K diodes we will be upgrading to are over 71% efficient with an output of 3.27 umol/j. I only know this because we have done the hard work of calculating and physical testing.

Click to expand...

I tip my hat to you mate, it's a lot of work with diodes changing ad infinitum.

[QUOTE="klx, post: 17084579, member: 510035" it's a lot of work with diodes changing ad infinitum.
[/QUOTE]
yeah this
a Sisyphean work

Grow Lights Australia said:

I wish someone would do that for us!

Sadly every time we design an LED fixture we need to go through every single type of diode we are considering using and calculate the umol/j to work out how efficient they will be which determines which diodes we end up using. The problem is that the manufacturer's data is not always accurate. That's one of the reasons we use Nichia diodes because they are very conservative with their flux bin data and often the diodes perform better when we test them than they are on paper. We can't say the same for other manufacturers. We are fortunate to have access to a goniometer so we can test individual diodes against manufacturer claims.

Even if you calculate the efficiency on paper it doesn't mean the diodes you have in your hand (or on your board) are actually that efficient. Chinese manufacturers always claim to have "top bin" Samsung diodes when we know that none of them really do. The best diodes go to those who have the best relationship with Samsung and in most cases pay for the privilege. There are only so many "top bin" diodes to go around so there is no way everyone can have them. That's what makes this exercise somewhat futile. The only way for the OP to actually know how efficient his Quantum-board knock-offs are is to test them in a sphere or goniometer. In reality he will have no idea what flux and voltage bin diodes will be on his boards. He can ask the manufacturer but he may not get an honest answer. They may say "SE" bin but there is only one way to prove it. Even within SE bins there is up to a 7.5% fluctuation in flux with at least four different voltage bins that have up to an additional 14% difference in efficiency.

A 2.8V 73lm 5000K CRI80 LM281B+SE chip may look great on paper but in reality it doesn't exist. And that theoretical maximum is still only 174lm/w which is well behind the top 3030 diodes such as Nichia's 757 series that are now 228lm/w — on average! The top bins are even better. The CRI70 5000K diodes we will be upgrading to are over 71% efficient with an output of 3.27 umol/j. I only know this because we have done the hard work of calculating and physical testing.

Click to expand...

exellent post...

New generation LEDs coming out

They take 6500K or "mint green" to get a higher PPF base white, which also gives them more opportunity to use the most efficient 660nm monochromatics more, to balance out the higher blue/green.
They're not the first to do that....

i spent a little time with the Samsung calculator, that thing is actually pretty cool, you can calculate the lumen for each diode and then convert it into umol/J. Thing is, most of the times you dont get info on the rank/bin.

And then there are the red diodes that are not from Samsung most of the time.

mannitu77 said:

i spent a little time with the Samsung calculator, that thing is actually pretty cool, you can calculate the lumen for each diode and then convert it into umol/J. Thing is, most of the times you dont get info on the rank/bin.

And then there are the red diodes that are not from Samsung most of the time.

Click to expand...

There is also the Samsung LED Component Calculator

thats what i meant...but the guys that sell the panel most of the times dont tell you what bin. LM281B+ at 50ma can have 200 or just 150 lumen per watt für example.
I found nothing for epistar reds or even Osram reds so far.

Red Osram are no problem.
Epistar doesn't make Led at all, only crystals, and you'll never know which ones are in your cases.
I think your main question was 100% answered by Grow Lights Australia.
You can say the same thing more simply.
The difference in the Led parameters of different bins within the same brand can be greater than the difference between Leds from different brands of the same price category. You will hardly know which bin is on your board.
As for the Chinese QBs, there is no sorting by bins at all. You can easily verify this if you turn off the power and see at what different speeds the individual chips go out. Are you going to apply exact calculations to them?

Has said:

As for the Chinese QBs, there is no sorting by bins at all. You can easily verify this if you turn off the power and see at what different speeds the individual chips go out. Are you going to apply exact calculations to them?

Click to expand...

yes i did...they are stated to be "SE-Rank"...and i have no reason to believe theyre not, its a mediocre rank so why should the company lie, they could have just said its top rank. For the bin, i found 4 different bins with slighty different Forward voltages, i just applied the 3rd best, or 2nd worst . So i think i can say i didnt overrank my calculation. The difference in bins and ranks are huge indeed. There are LM281B+ that do 200-210 lumen per watt at 65ma and 301s that dont do 200 at all.

From the best case szenario, best rank/bin 281B+ vs best case 301b i found about 11% difference in efficiency.

I thought i have epistar LEDs on my board, are these called crystals instead of LEDs or what do you mean? Theres a website from epistar with datasheet, but theres nothing with efficiency/efficacy in it, only basic things about size and typical and max currents.


edit: but for the Epistar, someone can suppose they raise overall efficacy, not bring it down, right? that wouldnt make any sense XD...so if the boards efficacy is 2,7umol/J for example, only calculating with 100% Lm281B+ diodes, you can be sure that the 12% reds will increase overall efficacy slightly, no matter how shitty theyre?

mannitu77 said:

I thought i have epistar LEDs on my board, are these called crystals instead of LEDs or what do you mean? Theres a website from epistar with datasheet, but theres nothing with efficiency/efficacy in it, only basic things about size and typical and max currents.

Click to expand...

Epistar are really a die manuf, package manufacturers buy their die and then sell as 'EPISTAR' leds. Epi do make nice RED die, however there is a world of difference between Power RED and 2835 Red, and this speaks to the general dumming down of grow panels, china copied the Quantum and over time cost downed it with lower performing products, lets face it the average amazon punter just does not know or really care

Grow Lights Australia said:

I know it's not the answer you want, but all the info you need is in the datasheets. Easy to find, easy to compare. The problem is when every company claims to have "top bin" Samsungs how do you know who really has them? The datasheets only tell you the theoretical highest bin. They don't tell you the production yield which determines what bins everyone gets.

FWIW we don't use Samsung diodes because there are better diodes out there.

Click to expand...

What do you mean better diodes, Please elaborate. As far as i know LM301H now Evo are the best diodes available.

EL_HERB said:

What do you mean better diodes, Please elaborate. As far as i know LM301H now Evo are the best diodes available.

Click to expand...

Why do you believe they are "the best"? How are they better than Nichia's dual-chip NF2W757H-V2F1? Because Samsung says so?

EL_HERB said:

What do you mean better diodes, Please elaborate. As far as i know LM301H now Evo are the best diodes available.

Click to expand...

I will explain myself but I must warn you that this will be a very long and somewhat technical post.

We'll start with the Samsung data and ask a simple question. Why does it not add up?


The forward voltage is listed as 2.7V at 65mA which is 0.1755W. Now divide 0.56 umol/s by 0.1755W and you get 3.19 umol/j so these figures don't add up. 0.2W is really the class of LED and not the actual wattage which is why that is there.

Perhaps we should look at the LM301H Evo datasheet.


The typical voltage at 65mA is around 2.775V, which is 0.180375W and 3.10 umol/j.

You could eye the data and reach a different conclusion but it is still only published data which does not determine which bins you will get when you actually order.

Now let us compare the above with Nichia's NF2W757HT-F1 which has been superseded by the NF2W757H-V2F1 but for which we have a lot of real-world data.

Typical voltage is 2.72V at 65mA = 0.1768W. Typical lumens for CRI80 5000K is 38.1 which is 215.5lm/W. We know from our SPD data that the theoretical maximum lm/W for this diode is 328.5lm/W and its Quantum Efficiency of Radiation is 4.61 umol/j so this diode is typically 65.6% efficient for 3.02 umol/j (215.5/328.5 x 4.61).

Remember that this is a TYPICAL diode which means it is an average. There are higher flux and lower voltage bins available from Nichia but this is typically what you can expect.


Now have a look at the data for the R70 (CRI70) 5000K which is 39.6lm. I will do the maths for you. It is 224lm/W for 67% efficiency and 3.03 umol/j which is similar to the CRI80 5000K Nichia.

However, when we ordered this diode from Nichia what we actually got was a much higher flux bin than typical. What we got was P12d43 which has an average 41.1lm for 235lm/W and 68.5% efficiency for 3.15 umol/j.

These tests were done in May 2020 which is two years ago and now the new NF2W757H-V2F1 is around 4% more efficient than the older HT-F1.

We started using these diodes (and still do) in our High Light 420 Gen1 and Gen2 LED panels almost two years ago which was a year before Samsung released its Evo.

We can also look at Samsung's past claims about luminous efficiency including their LM301B which was described as being 220lm/W. But when off-the-shelf diodes were tested, they only came in at 204lm/W.

Samsung has been caught out embellishing their specs in the past and there is no reason to doubt they are still doing the same. The main difference between Samsung and Nichia is that Samsung usually quote their best figures whereas Nichia normally quote their average figures. Nichia also tends to over-deliver when you buy their diodes while Samsung can be pot-luck unless you have a good relationship with them and/or pay a premium for higher bins. The disclaimer is that we also pay a premium for top Nichia bins.

This data is almost four years old now so I don't mind sharing it. You can guess who "Company A" is because you only have to look at the diode to know it is an LM301B. The Nichia diode is an even older model, the NF2W757G-V3F1.



Now maybe Nichia is also not telling the truth but the performance of the NF2W757G-V23F1 is equal to 2.67 umol/j for CRI90 3000K according to their datasheet but when we tested the diodes they came in at 2.75 umol/j which is about 3% higher than advertised. We have found this with nearly all the Nichia diodes we have tested. They are all at typical or higher performance specs.

We could also talk about the spectrum because the EVO does have a deep blue pump in the 435-440nm range which is good but perhaps that is best left for another discussion as there are other ways to get 400-450nm light.

I just wanted to show you that you cannot always rely on advertised specs when buying LEDs off different companies.

Here is an interesting read on Samsung's anti-sulfurisation. I don't know how accurate it is because they seem to lump LM301B in with LM301H so perhaps they tested 301B instead of 301H. But anti-sulfurisation is not really that important when you conformal coat LEDs so it provides no real advantage to us. It may seem, like I am knocking Samsung when they are still quite an innovative company but we are realistic and have the luxury of being able to test diodes and not rely on advertising to determine which LEDs we use.

Grow Lights Australia said:

We could also talk about the spectrum because the EVO does have a deep blue pump in the 435-440nm range which is good but perhaps that is best left for another discussion as there are other ways to get 400-450nm light.

Click to expand...

i know you know all this but ill reply anyway...

yer the 435-440nm pump is good in some respects but that pump along with the phosphor coating the evo uses shifts the whole spectrum down the nm range (the phosphor peak on the evo is at 580nm and the peak on the standard 301h is around 600nm) so the evo has less red in its spectrum, obviously when you use the evo on a pcb with a the addition of some good 660nm mono's this adds back in some nm of red light and also raises the efficiency. but i dont think the resulting spectrum is ideal.

id like to see the evo used 50/50 along side a 301h so you get both the 435-440nm pump and the 450nm pump, half the intensity of each pump but a now broader range of nm. that would be a good efficient start to add mono's too to finish of the spectrum...

or just buy some growlightsaustralia boards and be done

Chlorophyll A has an absorption peak around 430nm but there is still plenty of absorption from 400-420nm if you fill in the entire PAR range. Chlorophyll B peaks in the cyan range which is really lacking in the EVO even compared to typical 450nm blue pump phosphors which already have a deep cyan gap. @grotbags is right that the EVO misses out on a lot of red around 620-630nm where the second Chl B peak is so it doesn't really target Chlorophyll B (PS1) receptors.




Here is a 3000K Nichia Vitasolis that is 2.75 umol/j for comparison.


Vitasolis