Examining Samsung's 3.03μmol/J Claim

[ChiefRunningPhist]

The Brief:
They state: 3.03μmol/J
This calculates to: 80% efficiency
The datasheet calculates: 67.35% efficiency
What's the skinny: the difference is from the losses on the phosphor



The Claim:




LM301H 6500K flux data:




LM301H 6500K testing parameters:




Digitized LM301H 6500K SPD:





The Breakdown:


The LM301H uses a blue base pump die with a peak of ~451nm. If phosphor conversion efficiency is 100%, which in real life it can approach but hasn't yet, then the max efficiency possible is ~3.77μmol/J.


(451nm) ÷ (119.6nm·J/μmol)
=
3.771μmol/J


The claim of 3.03μmol/J must then mean a total efficiency of 80.35%.


(3.03μmol/J) ÷ (3.771μmol/J)
=
0.8035; 80.35%


According to the digitized spectrum, the maximum lumen output per watt is ~342lm. If 342lm is multipled by the 80.35% efficiency calculated earlier via their 3.03μmol/J claim, a final lumen/w figure of ~275lm/w is achieved. This is greater than what they print in the data sheet.


(343.1lm/w) × (0.8035)
=
274.88lm/w


Looking at the flux characteristics of the best 6500K LM301H listed (SM bin), a maximum of 42lm per chip is printed. This equates roughly to 235lm/w.


(42lm) ÷ (0.065A × 2.75V)
=
234.97 lm/w


Looking at the printed calculation of ~235lm/w, and comparing with the digitized radiometric luminous efficacy of ~342lm/w, a total chip efficiency is calculated at 68.68%.


(234.97lm/w) ÷ (342.1lm/w)
=
0.6868; 68.68%


So which one is it? 68% or 80%?


Well I've since remebered lol Dammit! The discrepancy is due to 1 metric describing the power input vs the energy emitted after the phosphor losses, and the other metric is describing the power input vs the energy before the phosphor losses.

When I assume a CE of 1, I get ~3.3μmols/J. The stated claim is 3.03μmol/J, so I'm assuming the CE for the phosphor film they are using is not 1 but something close to 0.918. I'm going to wrestle a little more with this.

 

[Airwalker16]

Very interesting information here....

 

[PSUAGRO.]

I don't think Sammie is the only one guilty of this? Alesh brought up yrs ago with cree...Citi , also fuzzy #? Or didn't spec out in sphere test, don't remember anymore.

Best bet would be if @Stephenj37826 would chime in on reading partner spec sheet.

Your conflicting math seems correct to me , that's not saying much though.... Ha

Wonder what Teknik thinks on the other forum, I'm locked out since the web move.

 

[diggs99]

I don't think Sammie is the only one guilty of this? Alesh brought up yrs ago with cree...Citi , also fuzzy #? Or didn't spec out in sphere test, don't remember anymore.

Best bet would be if @Stephenj37826 would chime in on reading partner spec sheet.

Your conflicting math seems correct to me , that's not saying much though.... Ha

Wonder what Teknik thinks on the other forum, I'm locked out since the web move.

Ya weve lost a few since the switch, hope you guys can make it back.


Great stuff Chief, as always.

 

[PSUAGRO.]

Haha ya I'm sure you're right, Sammie not the only ones, they're just the ones I noticed claiming over 3.0 without a test report linked.

This type of calculation hasn't really been implemented (that I know of), it makes use of "photon conversion efficiency," but there's always some ambiguity to what the voltage needed to flow 0.065A is/was, and what the exact lumen output is (within the range given by a bin designation). Essentially, it allows for an educated calculation of the photon efficacy when only lumen efficacy is given.

I'm locked out too but I'm not losing any sleep lol some good dudes but I think most came from here and end up bouncing back and forth.

Yes your right on that regard, +3umol/j is quite a claim..... As stated, you won't generally see that system level anyways even if true on sheet. Just made up numbers for the headline/efficiency chasers. Remember the cree lab results? 300 lm/w like 2 yrs ago, lol, just news bait.

Tek was about to ruffle some feathers on his gonio testing. Slight spec bump worth the large increase in price? Good question and China encapsulated leds are rapidly closing in on top dogs.

Tek will get his hands on these. Probably a good chance Nick has/ will run them through the sphere for mega v3 fixture, if the upgrade is worth it? He's all in with Sammie for yrs now .

I'll email him, see what he says.... He's pissed at me though for false promises

 

[ChiefRunningPhist]

@PSUAGRO. @Airwalker16 @diggs99
If you're following this thread, I edited. I was wrong!

 

[ChiefRunningPhist]

So nothing new to report. Multiply QER by efficiency derived from LER and datasheet to achieve a photon efficacy of the chip.

(235lm/w) / (342.1lm/w)
=
0.6869; 68.69%

(0.6869) × (4.49μmol/J)
=
3.09μmol/J

3.03 / 3.09
=
0.98

0.98 is close to 1, I'm chalking up the 0.06 difference to CE for the moment. Thought I was Dick Tracy for a second there lol

 

[Stephenj37826]

I don't think Sammie is the only one guilty of this? Alesh brought up yrs ago with cree...Citi , also fuzzy #? Or didn't spec out in sphere test, don't remember anymore.

Best bet would be if @Stephenj37826 would chime in on reading partner spec sheet.

Your conflicting math seems correct to me , that's not saying much though.... Ha

Wonder what Teknik thinks on the other forum, I'm locked out since the web move.

I had this conversation with them back in the 561C days. I'll see what I can find out. The 3.03 umol/j is the only metric I really care much about of course. Sticking with similar spds I do rough estimates on them using lm/w metrics. If something stands out I get samples and off to the sphere. We have a new board coming in Feb that will be capable of 3.0+ system level at fairly good wattage. We are making several changes at HLG. We should have a nice upgrade to our offerings soon. Man it's been a minute since I've been on these forums. Seems like life moves so fast these days.

 

[PSUAGRO.]

I had this conversation with them back in the 561C days. I'll see what I can find out. The 3.03 umol/j is the only metric I really care much about of course. Sticking with similar spds I do rough estimates on them using lm/w metrics. If something stands out I get samples and off to the sphere. We have a new board coming in Feb that will be capable of 3.0+ system level at fairly good wattage. We are making several changes at HLG. We should have a nice upgrade to our offerings soon. Man it's been a minute since I've been on these forums. Seems like life moves so fast these days.

Good to see you back....... Let us know how these samies turn out in testing

 

[SourDeezz]

I had this conversation with them back in the 561C days. I'll see what I can find out. The 3.03 umol/j is the only metric I really care much about of course. Sticking with similar spds I do rough estimates on them using lm/w metrics. If something stands out I get samples and off to the sphere. We have a new board coming in Feb that will be capable of 3.0+ system level at fairly good wattage. We are making several changes at HLG. We should have a nice upgrade to our offerings soon. Man it's been a minute since I've been on these forums. Seems like life moves so fast these days.

Would these be an upgrade to the HLG550?

 

[Stephenj37826]

Would these be an upgrade to the HLG550?

Absolutely

 

[Stephenj37826]

I don't think Sammie is the only one guilty of this? Alesh brought up yrs ago with cree...Citi , also fuzzy #? Or didn't spec out in sphere test, don't remember anymore.

Best bet would be if @Stephenj37826 would chime in on reading partner spec sheet.

Your conflicting math seems correct to me , that's not saying much though.... Ha

Wonder what Teknik thinks on the other forum, I'm locked out since the web move.

If I remember correctly the 80% numbers are a reference percentage of the theoretical efficiency of phosphorous converted white using a 450nm blue "pump". 65% is more in reference to LER or the lumens a respective spectrum would produce at 1 watt 100% efficiency.

 

[2com]

Absolutely

Hi,
Can I ask what type of upgrade this new offering/board will be; what area? Spectrum, output power (umol), or just more efficient?
Thank you.

 

[Stephenj37826]

Tweaked spectrum, higher efficiency, and overall output increase.

 

[ChiefRunningPhist]

The 80% refers to the efficiency before phosphor losses, ie μmol/J, the 68% refers to the efficiency after phosphor losses, ie mW output to mW input.

If you calculated up the mW output of (X)μmols of 450nm you'd get (Y)mW. The 450nm photons don't stay 450nm, some do but most dont. The majority lose energy on the phosphor to become lower energy photons or different color photons. So the calculated (Y)mW of (X)μmols of base pump is diminished by the phosphor. This is why Samsungs 80% μmol/J efficiency means the same thing as their 68% chip efficiency (though not convinced they're 3+, still thinking 2.7 max but just opinion) 1 metric describes how much energy it takes to create (X)μmols per watt regardless the color (assuming CE = 1), and 1 metric describes the energy coming off the phosphor. This energy is the same as the (Y)mW of base pump minus phosphor losses, because some of the base pump has to downgrade its energy even though the total μmols are conserved (again assuming CE = 1).

When you do a SPD LER you're calculating based off of end mW output, or post phosphor mW output. You assume the relative SPD is in joules/nm, this was confusing me because for every photon emitted from a PC LED, there's a constant power draw needed (base pump efficiency), it doesn't change like the relative SPD shows (all differnt types of photons at different energies). I was getting confused because I knew that the energy going into creating each photon was the same but trying to reconcile calcs that were derived from photons of all energies like seen in the relative SPD. 1 describes pre phosphor chip efficiency and 1 describes post phosphor chip efficiency.

 

[Stephenj37826]

The 80% refers to the efficiency before phosphor losses, ie μmol/J, the 68% refers to the efficiency after phosphor losses, ie mW output to mW input.

If you calculated up the mW output of (X)μmols of 450nm you'd get (Y)mW. The 450nm photons don't stay 450nm, some do but most dont. The majority lose energy on the phosphor to become lower energy photons or different color photons. So the calculated (Y)mW of (X)μmols of base pump is diminished by the phosphor. This is why Samsungs 80% μmol/J efficiency means the same thing as their 68% chip efficiency (though not convinced they're 3+, still thinking 2.7 max but just opinion) 1 metric describes how much energy it takes to create (X)μmols per watt regardless the color (assuming CE = 1), and 1 metric describes the energy coming off the phosphor. This energy is the same as the (Y)mW of base pump minus phosphor losses, because some of the base pump has to downgrade its energy even though the total μmols are conserved (again assuming CE = 1).

When you do a SPD LER you're calculating based off of end mW output, or post phosphor mW output. You assume the relative SPD is in joules/nm, this was confusing me because for every photon emitted from a PC LED, there's a constant power draw needed (base pump efficiency), it doesn't change like the relative SPD shows (all differnt types of photons at different energies). I was getting confused because I knew that the energy going into creating each photon was the same but trying to reconcile calcs that were derived from photons of all energies like seen in the relative SPD. 1 describes pre phosphor chip efficiency and 1 describes post phosphor chip efficiency.

Which coincides with what I said. If 80% is the pre phosphorus conversion then basically you are at 80% of the theoretical limit of blue pumped phosphorus. Quantum dot may be a different story eventually.
Oh they are definitely 3.0+ at low drive current. Our new light will be damn close to that at system level at temperature.

 

[ChiefRunningPhist]

Oh they are definitely 3.0+ at low drive current. Our new light will be damn close to that at system level at temperature.

Good to know, without a sphere I have only the printed documentation to go off of and I'm leery to trust much of it without 3rd party confirmation. Maybe it's warranted, or maybe it's not, idk. Anyways, 3.0+ at system is great! When estimating other chips besides LM301B/H, they seem to be drastically less output, that's what made me question 3.0. Most "high quality" competitors seem to be calculating 2.5 and lower.

EDIT:
Idk what I calculate Nichia NF2_757-F1 at, with a CE of 1, avg lumen measurement of 39.4lm, and at their documented 65mA, I arrive at 3.15μmol/J.

From data sheet:
2.72V × 0.065A = 0.1768W

From data sheet:
[0.5×(36lm + 42.8lm)] / 0.1768W = 222.8lm/W

Digitized Calc:
(222.8lm/W) / (345.5 lm/W) = 0.645

Digitized Calc:
0.645 × 4.88μmol/J = 3.147μmol/J

 

[Stephenj37826]

Good to know, without a sphere I have only the printed documentation to go off of and I'm leery to trust much of it without 3rd party confirmation. Maybe it's warranted, or maybe it's not, idk. Anyways, 3.0+ at system is great! When estimating other chips besides LM301B/H, they seem to be drastically less output, that's what made me question 3.0. Most "high quality" competitors seem to be calculating 2.5 and lower.

EDIT:
Idk what I calculate Nichia NF2_757-F1 at, with a CE of 1, avg lumen measurement of 39.4lm, and at their documented 65mA, I arrive at 3.15μmol/J.

From data sheet:
2.72V × 0.065A = 0.1768W

From data sheet:
[0.5×(36lm + 42.8lm)] / 0.1768W = 222.8lm/W

Digitized Calc:
(222.8lm/W) / (345.5 lm/W) = 0.645

Digitized Calc:
0.645 × 4.88μmol/J = 3.147μmol/J

Nichia has some really good LEDs. Just pricey. We considered them several times and used a few million. Gotta love that S.Korea vs Japan unspoken tech war Haha.

 

[welight]

Good to know, without a sphere I have only the printed documentation to go off of and I'm leery to trust much of it without 3rd party confirmation. Maybe it's warranted, or maybe it's not, idk. Anyways, 3.0+ at system is great! When estimating other chips besides LM301B/H, they seem to be drastically less output, that's what made me question 3.0. Most "high quality" competitors seem to be calculating 2.5 and lower.

EDIT:
Idk what I calculate Nichia NF2_757-F1 at, with a CE of 1, avg lumen measurement of 39.4lm, and at their documented 65mA, I arrive at 3.15μmol/J.

From data sheet:
2.72V × 0.065A = 0.1768W

From data sheet:
[0.5×(36lm + 42.8lm)] / 0.1768W = 222.8lm/W

Digitized Calc:
(222.8lm/W) / (345.5 lm/W) = 0.645

Digitized Calc:
0.645 × 4.88μmol/J = 3.147μmol/J

Did you do this on the new Nichia HT chip?

first samples just arrived, 224LPW and our SSK-272 will use these with G bin Cree Photo Red, we expect sys eff at 3μmol/J at least
Cheers
Mark

 

[ChiefRunningPhist]

Did you do this on the new Nichia HT chip?
View attachment 4441400
first samples just arrived, 224LPW and our SSK-272 will use these with G bin Cree Photo Red, we expect sys eff at 3μmol/J at least
Cheers
Mark

I believe so. I didn't add the HT designation to the .pdf, but attached is the data sheet I was using.