High Light efficiency tests (TEKNIK) - 2.47 umol/j CRI 94.2

Prawn Connery

Prawn Connery

Well-Known Member
I gave a QB324 V1 to TEKNIK to test and he did, but because the board had a few grows on it and was not a fair comparison, I agreed not to publish them. I can say the results were over 20% lower at comparable LED currents.
 
BuddyColas

BuddyColas

Well-Known Member
Prawn Connery said:
I gave a QB324 V1 to TEKNIK to test and he did, but because the board had a few grows on it and was not a fair comparison, I agreed not to publish them. I can say the results were over 20% lower at comparable LED currents.
Click to expand...
Please provide a few details for, "...the results were over 20% lower at comparable LED currents." Inquiring minds want to know.:mrgreen:
 
Stephenj37826

Stephenj37826

Well-Known Member
Randomblame said:
QB324 use Nichia diodes too but its a 2 chip diode with 6v which can run at higher wattage. QB324 is also already 2 years on the market so this alone makes if not a fair comparision.
Click to expand...
The LEDs on the 324 are V1..... I would think at the very least there is a 12%+ efficiency bump with v3f1. V1F1 was super expensive at the time and didn't handle much current. Also at lower currents the 6 volt didn't keep gaining like the parallel 3 volts. In real world usage the 3 volt and 6 volt were very similar with the edge going to the F1.

Also is these numbers including driver loss? Our numbers include driver losses as well.
 
Last edited:
Greengenes707

Greengenes707

Well-Known Member
Randomblame said:
Yeah, I still say the same. Tested on the same test rig(teknik's) with the same parameters, distance and so I don't think there is such a big difference. Its for sure better efficiency with CRI80 but the gap should be lower. Especially when measured from 380-780nm like in this tests.
Maybe we will see some QB tests soon too cause he told me something like that.
Click to expand...
A rig like tekniks...you mean a gonio, the dude has a gonio. There is no difference between what teknik is using, and a sphere for total output and efficacy figures. There is no distance factored in...it is total output.
Sphere's measure 380-780 too...it's been common for way before any talk of Bpar. You keep trying to bring in factors that don't exist. The measurements are under the same standards, sphere or gonio.

Other than the diagram of the beam angle...there is nothing additional to total output being used in those readouts from his gonio that would be different to what a sphere test would show.

High CRI and low nm blue bump chips...again... the nature of the beast for chasing SPD.
 
Prawn Connery

Prawn Connery

Well-Known Member
BuddyColas said:
Please provide a few details for, "...the results were over 20% lower at comparable LED currents." Inquiring minds want to know.:mrgreen:
Click to expand...
Like I said, it wasn't a fair comparison because the board had seen a few grows and was not in pristine condition (not 100% clean), which would have affected its output. I only mentioned it to show other boards had been tested and the results were in line with expectations.

The guy who owns the QB324 V1s is replacing a bunch of 600W HPS with High Lights and so we thought it would be interesting to see how they compared (even after a few grows) to the newer boards. The QB324 V1s use a mix of Nichia CRI80 and 90 LEDs, so in terms of spectrum, they are a bit closer than typical CRI80 boards. It was precisely because we had been using QB324 V1s and liked the results of the extra red (as well as the truer colour rendition when looking at the plants) that we decided to build a high CRI board.

In fact, if HLG hadn't sold out of QB324 V1s when we wanted to buy them, we probably wouldn't have gone ahead and designed our own board!

Stephenj37826 said:
The LEDs on the 324 are V1..... I would think at the very least there is a 12%+ efficiency bump with v3f1. V1F1 was super expensive at the time and didn't handle much current. Also at lower currents the 6 volt didn't keep gaining like the parallel 3 volts. In real world usage the 3 volt and 6 volt were very similar with the edge going to the F1.

Also is these numbers including driver loss? Our numbers include driver losses as well.
Click to expand...
No, the measurements were taken at the board. As you know, different brand drivers can have different efficiencies, and even the same drivers have different efficiencies depending on load.

We used the most efficient CRI90 LEDs we could find, and of course we paid for it. The V3F1s cost 20% more than the NFSW757GT-V3 we were also offered, and the Optisolis and Sunlike were almost double that again. They weren't cheap, but we built them primarily for our own use, so we wanted the best LEDs we could find on the market.
 
Prawn Connery

Prawn Connery

Well-Known Member
Randomblame said:
Yeah, I still say the same. Tested on the same test rig(teknik's) with the same parameters, distance and so I don't think there is such a big difference. Its for sure better efficiency with CRI80 but the gap should be lower. Especially when measured from 380-780nm like in this tests.
Maybe we will see some QB tests soon too cause he told me something like that.
Click to expand...
I was actually happy with the results, because that's about where I thought they would be. I was thinking perhaps even a bit lower (2.4 umol/j) considering the high CRI and mix of LEDs (the Sunlike are not very efficient, and the Optisolis are behind the V3F1). Also the fact there are 450 LEDs, I would expect there to be some losses associated with the bigger board in terms of longer supply tracks adding resistance (especially with the way we've wired the LEDs). All that should add up to slightly less efficiency than a single diode test.

I've also seen (I think you've seen them too?) official Nichia tests of Samsung LM301Bs that did not match Samsung's specs (were about 5% off), so it's not unlikely that manufacturers inflate their figures a little bit, or at the very least use selected components to base their data on. Remember, "lumens per watts" sell LEDs - there are plenty of inflated figures out there. Which is even more reason to get these boards properly tested so you know exactly what you have.

EDIT: Also, just to put things into perspective, here is a Cutter board with 4000K CRI80 Cree LEDs that TEKNIK tested. If you think 2.47 umol/j from a CRI94/95 board with UV-based diodes should be higher, then you would probably expect the same from 4000K CRI80. Like I said, I was happy with our results.

Screen Shot 2019-05-13 at 14.19.50.png
 

Attachments

Last edited:
Prawn Connery

Prawn Connery

Well-Known Member
One thing I forgot to mention is that all the tests TEKNIK did were at full operating temperature. So the boards were powered up and weren't tested until their temperatures and voltages had stabilised. As you know, the hotter the LEDs, the less efficient they are, so these were "real world" tests meant to replicate what you would see in a grow room environment at full operating temperature.

TEKNIK had warned me that his test results were typically lower than similar sphere results for the above reason. The fact is, if you conduct a test in a sphere (fully closed environment) and wait for the LED to fully stabilise, the excess heat cannot escape, leading to an erroneous reading. So some sphere tests are conducted before the LED has fully stablised (heated up), which can lead to higher readings. At least, that was the way it was explained.

I suspect there may have been another factor, but I can't remember what it was :oops:
 
Last edited:
Prawn Connery

Prawn Connery

Well-Known Member
CobKits said:
i dont have a calibrated sphere

any light lab thats doing it for money list how they calibrate their equipment (basically you get a light source that has been calibrated to one at NIST (or whatever the AU equivalent is). I dont see this in tekniks report, not saying his equipment isnt calibrated, just that hes not including it in his reports (in what we see here) and it would be useful

look at the reports on chillled's site for an example of what im talking about

some labs report par range differently as well (again take a look at a multitude of official test reports)
Click to expand...
Hi mate, have a look at the photos in this post: http://www.rollitup.org/t/cutter-cree-strip-tests.987995/#post-14846979

There's a calibration sticker on the back of the VISO. I assume they're calibrated before they are sent out.
 
Randomblame

Randomblame

Well-Known Member
Prawn Connery said:
I was actually happy with the results, because that's about where I thought they would be. I was thinking perhaps even a bit lower (2.4 umol/j) considering the high CRI and mix of LEDs (the Sunlike are not very efficient, and the Optisolis are behind the V3F1). Also the fact there are 450 LEDs, I would expect there to be some losses associated with the bigger board in terms of longer supply tracks adding resistance (especially with the way we've wired the LEDs). All that should add up to slightly less efficiency than a single diode test.

I've also seen (I think you've seen them too?) official Nichia tests of Samsung LM301Bs that did not match Samsung's specs (were about 5% off), so it's not unlikely that manufactures inflate their figures a little bit, or at the very least use selected components to base their data on. Remember, "lumens per watts" sell LEDs - there are plenty of inflated figures out there. Which is even more reason to get these boards properly tested so you know exactly what you have.
Click to expand...

Yeah, 2,4μMol/J at ~100w/2A(my desired drive current) is really good for CRI95. In a real world comparision grow I also expect to see faster growth rates under the high CRI boards. They should also finish the run faster compared to CRI80. Even the yields could be nearly the same or even better.
I still remember @The Dawg 's CRI80 vs CRI90 tests where CRI90 has constantly outperformed CRI80 at the same wattage. Seems the better YPF performance of the spectrum and the far-red part can make up for the lower efficiency and speed things up too.
You know, peeps often only look at the efficiency numbers(lumens per watt) to chose a new white light(or type of LED's). I wanted only to make sure everyone understands the difference and why the efficiency numbers look so low at the same drive current.(150mA per diode). At 4,5A it should be around 2,1 and a QB288v2 is sphere tested with 2,49. That's a big difference in the first moment.
 
Prawn Connery

Prawn Connery

Well-Known Member
Yes, but at 4.5A, all our LEDs are at their maximum rating, according to Nichia. It is safe to max out the LEDs due to the size of the board and heatsink.

What current was the 2.49 umol/j for the QB288 V2? I note that at the maximum board rating of 3A (HLG maximum), those LM301Bs are only at 83% of their maximum (Samsung) rating of 200mA each.

Admittedly, the Nichia's have a lower rating to begin with - 180mA for the V3F1 and 150mA for the Optisolis and Sunlike.

I guess the only way to have a fair comparison is to test the boards on the same equipment under the same conditions.
 
Last edited:
Randomblame

Randomblame

Well-Known Member
Prawn Connery said:
One thing I forgot to mention is that all the tests TEKNIK did were at full operating temperature. So the boards were powered up and weren't tested until their temperatures and voltages had stabilised. As you know, the hotter the LEDs, the less efficient they are, so these were "real world" tests meant to replicate what you would see in a grow room environment at full operating temperature.

TEKNIK had warned me that his test results were typically lower than similar sphere results for the above reason. The fact is, if you conduct a test in a sphere (fully closed environment) and wait for the LED to fully stabilise, the excess heat cannot escape, leading to an erroneous reading. So some sphere tests are conducted before the LED has fully stablised (heated up), which can lead to higher readings. At least, that was the way it was explained.

I suspect there may have been another factor, but I can't remember what it was :oops:
Click to expand...

That's an important point I totally forgot to mention!
Usually there is a 13-15% difference between pulsed measurments at 25°C and stable output at 85°C. At 55°C it's still 7-8%. And in the QBv2 sphere tests they also have mentioned 25°C ambient temps but no stabilized board temps. So I'm pretty sure it was a "cold measuring".
That's probably the most important reason why the gap between this two boards is in reality much smaller and tested on the same teknik's test rig the QBv2 numbers should be at least 7-8% lower. Calculated from 2,49 that's between 2,29-2,315 heated up to ~55°C. So 2,1 vs 2,3 at max current.
That sounds much more like the typical difference found between CRI80 and 90.
 
Randomblame

Randomblame

Well-Known Member
Prawn Connery said:
Hi mate, have a look at the photos in this post: http://www.rollitup.org/t/cutter-cree-strip-tests.987995/#post-14846979

There's a calibration sticker on the back of the VISO. I assume they're calibrated before they are sent out.
Click to expand...
You can even see the date of the last calibration, 17/11/17



4 boards running at 472,9w so around 118,25w each board. So diodes should run with ~150mA's.
That's below is the QBv2 test but shit, I've overseen the test was made with 45 minutes warm up times so forget my calculation above. The measurings were taken with stabilized board temps. Sorry!
 

Attachments

Prawn Connery

Prawn Connery

Well-Known Member
That's input power, so looks like it includes the driver. I guess the only way to accurately compare is to measure side-by-side on the same equipment under the same conditions.
 
Randomblame

Randomblame

Well-Known Member
Prawn Connery said:
That's input power, so looks like it includes the driver. I guess the only way to accurately compare is to measure side-by-side on the same equipment under the same conditions.
Click to expand...
Yepp, that's the only way for a fair comparision test. Maybe teknik will test them too, he mentioned it at least before he was banned.
 
hybridway2

hybridway2

Amare Shill
Prawn Connery said:
Yes. But that's not really the point I was making.

The spectrum of light that reaches earth does change throughout the day, seasons, latitudes and topography etc. What doesn't change is that - no matter the weighting and ratios of each wavelength - the spectral curve of sunlight is relatively flat, with no substantial peaks or troughs compared to most LED and HID lighting.

In other words, wavelengths don't go "missing" or suddenly spike as atmospheric influences alter the spectrum - unlike LEDs and HIDs, which have lots of peaks (overweighting) and troughs (underweighting) in their spectra by design.

With sunlight, ALL wavelengths in the PAR spectrum are covered. And it is the weighting of those wavelengths and their ratios to each other that affects plant growth - especially if you believe plants have been tuned (evolved) to those hourly, daily and seasonal changes to make use of (photosynthesise) different spectra as they become more or less available throughout the day.

For example, are individual photoreceptors rested during some parts of the day to become more active during others when there is more or less red, green and blue light? And should we be trying to emulate this indoors by altering spectra at different times of the lighting schedule?

Or is it enough to simply "fill in the gaps" so that we have a reasonably smooth spectral curve?

Is it also the reason why plants seem to respond better to full spectrum light (with certain weightings of red, green and blue) than to narrow (targetted) spectra such as "blurple" etc?

I mean, mother nature's been around a lot longer than we have. Should we be paying more attention to her?

View attachment 4332215

View attachment 4332186
Click to expand...
My point exactly
 
Randomblame

Randomblame

Well-Known Member
Greengenes707 said:
A rig like tekniks...you mean a gonio, the dude has a gonio. There is no difference between what teknik is using, and a sphere for total output and efficacy figures. There is no distance factored in...it is total output.
Sphere's measure 380-780 too...it's been common for way before any talk of Bpar. You keep trying to bring in factors that don't exist. The measurements are under the same standards, sphere or gonio.

Other than the diagram of the beam angle...there is nothing additional to total output being used in those readouts from his gonio that would be different to what a sphere test would show.

High CRI and low nm blue bump chips...again... the nature of the beast for chasing SPD.
Click to expand...

Yeah, in theory they all are calibrated to the same light source but there is still a difference from sphere to sphere or sphere to gonio.. And when using a gonio to compare 2 lights you need to test them at the same distance to the sensor. In the end the sensor readings are all interpreted by a bunch of algorithms so small difference within a few percent are normal. Like the 7% usually mentioned in datasheets. To make it a fair comp both should be tested with the same sphere or gonio.

But don't understand me wrong! I don't say QB's are bad in any way. They are not.. It's still one of the best products available! I only wonder for the pretty high difference at 150mA and teknik also told me he would expect slightly lower numbers on his gonio because he already tested a few cutter products.
In the end we will only know it if teknik actually tested them. Everything else is just speculation... Differences could be even higher..
 
Randomblame

Randomblame

Well-Known Member
Prawn Connery said:
There's a bhere: https://www.rollitup.org/t/high-lights-high-powered-cri95-flowering-boards-made-in-australia.988805/page-4
Click to expand...
Still makes no sense to me...
And as far as he told me he has no own company today to offer his service and get paid for it. It was planned to do later if it makes sense but he would need to get enough interested LED manufacturers on board.
Maybe he should simply come back as Niktek, lol! But this time always with a bit honey under his tongue to not getting banned immediately.
Shit, man! To have someone who is willing to do free tests with its own calibrated gonio would be such a huge benefit for us DIY'ers!
Whether he sounds friendly or like an asshole really doesn't matters.
But to stay fair, I don't know what was going on behind the scenes... I only know he got a few warnings before they have banned him. Probably 90% because of wrong proud (no "excuse me" out of his mouth) and 10% by prejudice, lol!
 
You must log in or register to reply here.
Top