dionysus4
Well-Known Member
do u even know where china is on a map?
seriously though, how does that qualify cobs sold in china being binned correctly?
Nike is also made in china, are you suggesting I cant find fake or mislabeled nike shoes in china
Top bin COB comparison
- Thread starter SupraSPL
- Start date
- Status
do u even know where china is on a map?
seriously though, how does that qualify cobs sold in china being binned correctly?
Nike is also made in china, are you suggesting I cant find fake or mislabeled nike shoes in china
Greengenes707
Well-Known Member
Cree's blue die are made in Durham NC. Then shipped to china(malaysia) and the whole package as we know it is assembled/built.
EfficientWatt
Well-Known Member
Hi there.
@Stephenj37826 : When will you have access to that sphere you mentionned ?
@SupraSPL : Have you received, and will you test, GG's CXB3070 3500K BB (non-jerry) ?
Jerry is stocking up on them this week too ... it would be nice to compare, I'd be really really surprised if Jerry's are not legit, and don't beleive it to be the case, but I'm about to order 2500$ of chips within a week, and would love to know for sure before I do...
@Stephenj37826 : When will you have access to that sphere you mentionned ?
@SupraSPL : Have you received, and will you test, GG's CXB3070 3500K BB (non-jerry) ?
Jerry is stocking up on them this week too ... it would be nice to compare, I'd be really really surprised if Jerry's are not legit, and don't beleive it to be the case, but I'm about to order 2500$ of chips within a week, and would love to know for sure before I do...
Stephenj37826
Well-Known Member
After the new year..... It's 30 minutes down the road and funny enough the engineer lives 5 minutes from me lol.Hi there.
@Stephenj37826 : When will you have access to that sphere you mentionned ?
@SupraSPL : Have you received, and will you test, GG's CXB3070 3500K BB (non-jerry) ?
Jerry is stocking up on them this week too ... it would be nice to compare, I'd be really really surprised if Jerry's are not legit, and don't beleive it to be the case, but I'm about to order 2500$ of chips within a week, and would love to know for sure before I do...
robincnn
Well-Known Member
Fair to have some doubts about Jerry since he is not authorized seller . It is fair to doubt anything that comes out of China.
However the data between Vero and CXB is not sufficient to prove doubts about Jerry giving us lower bins. For all we know Cree sucks compared to Vero and we should not blame Jerry for it
The sample size of 2 is not big either.
Although CXB did outperform the Vero 29 in my test, the CXB failed to offer any significantly higher efficiency at lower current compared to Vero 29.
I was going to buy 2 Vero 29's for making next lamp. I will buy 2 3590CD to help with the test.
Digi has CXB3590-0000-000N0HCD35G for $68 each - I can buy 2
Mouser has CXB3590-0000-000N0HCD35G for $89 each - no way
Cutter has DB bin but they are not authorized distributor of cree .. right?
any other authorized suppliers for 3590 3500K CD 36V
KingBrite Kush
However the data between Vero and CXB is not sufficient to prove doubts about Jerry giving us lower bins. For all we know Cree sucks compared to Vero and we should not blame Jerry for it
The sample size of 2 is not big either.
Although CXB did outperform the Vero 29 in my test, the CXB failed to offer any significantly higher efficiency at lower current compared to Vero 29.
I was going to buy 2 Vero 29's for making next lamp. I will buy 2 3590CD to help with the test.
Digi has CXB3590-0000-000N0HCD35G for $68 each - I can buy 2
Mouser has CXB3590-0000-000N0HCD35G for $89 each - no way
Cutter has DB bin but they are not authorized distributor of cree .. right?
any other authorized suppliers for 3590 3500K CD 36V
KingBrite Kush
Last edited:
sounds like a plan! I'm guessing that there is a wider variation in the vero29s, seeing as they are not binned, unfortunately to prove that, it will take a decently large sample set size.Fair to have some doubts about Jerry since he is not authorized seller . It is fair to doubt anything that comes out of China.
However the data between Vero and CXB is not sufficient to prove doubts about Jerry giving us lower bins. For all we know Cree sucks compared to Vero and we should not blame Jerry for it
The sample size of 2 is not big either.
Although CXB did outperform the Vero 29 in my test, the CXB failed to offer any significantly higher efficiency at lower current compared to Vero 29.
HockeyBeard
Well-Known Member
My only thought - What sort of chip layout would you use to evenly blend the different colors?
robincnn
Well-Known Member
I give you numbers and you tell me if difference is worth paying extra cost for implementing binning in Vero 29's
6.3% variation for 10 samples tested at 55C
5.1% variation for 10 samples tested at 85C
Source Vero 29 LM 80 test report Lm-80-Vero29-04-1X dated Dec 2014. Please contact bridgelux for a copy. I cannot share mine as it has my name all across.
Last edited:
sumthin is wrong in the 85 degree set of data. see sample 6, 7, 8, 9 all are way under.
how does a 5% standard deviation compare to the differences between the vero29 and the 3590?
robincnn
Well-Known Member
Fixed it. see the Vero chart in my previous post again.
Now consider Cree 3590
ignoring variation within bins, which actually would it even more worse.
55% variation between AD and DD.
With Cree the variation between just CD and its next higher bin DB itself is 8.3%
What do you think ?
Last edited:
Is it to MY or CN?? MY is S. of me and CN is N. of me and Seagate is on the W.side of town and my
Nike's came from the E. of me, VN.
SLITLOS
Rahz
Well-Known Member
Straightroadelectronics has 3590 3500K CD36V for $60 each.
Fair to have some doubts about Jerry since he is not authorized seller . It is fair to doubt anything that comes out of China.
However the data between Vero and CXB is not sufficient to prove doubts about Jerry giving us lower bins. For all we know Cree sucks compared to Vero and we should not blame Jerry for it
The sample size of 2 is not big either.
Although CXB did outperform the Vero 29 in my test, the CXB failed to offer any significantly higher efficiency at lower current compared to Vero 29.
I was going to buy 2 Vero 29's for making next lamp. I will buy 2 3590CD to help with the test.
Digi has CXB3590-0000-000N0HCD35G for $68 each - I can buy 2
Mouser has CXB3590-0000-000N0HCD35G for $89 each - no way
Cutter has DB bin but they are not authorized distributor of cree .. right?
any other authorized suppliers for 3590 3500K CD 36V
KingBrite Kush
View attachment 3566354
welight
Well-Known Member
Please check the Cree web site where to buy/australia, we have been an authorised Cree distributor since 2005, long before Mouser/Digi. AUTHORISED, We get product shipped to us direct from Cree in HKFair to have some doubts about Jerry since he is not authorized seller . It is fair to doubt anything that comes out of China.
However the data between Vero and CXB is not sufficient to prove doubts about Jerry giving us lower bins. For all we know Cree sucks compared to Vero and we should not blame Jerry for it
The sample size of 2 is not big either.
Although CXB did outperform the Vero 29 in my test, the CXB failed to offer any significantly higher efficiency at lower current compared to Vero 29.
I was going to buy 2 Vero 29's for making next lamp. I will buy 2 3590CD to help with the test.
Digi has CXB3590-0000-000N0HCD35G for $68 each - I can buy 2
Mouser has CXB3590-0000-000N0HCD35G for $89 each - no way
Cutter has DB bin but they are not authorized distributor of cree .. right?
any other authorized suppliers for 3590 3500K CD 36V
KingBrite Kush
View attachment 3566354
REALSTYLES
Well-Known Member
Realstyles for the win
Sent from my SM-G900P using Rollitup mobile app
Sent from my SM-G900P using Rollitup mobile app
SupraSPL
Well-Known Member
OK I have the CXB3070 3500K BB from PLC and a CXB3590 3500K CD from Kingbrite. I decided to scrap my original data and start over with a better methodology. This time each COB was mounted on the exact same heatsink and the heatsink was hung using chain with the holes marked so it hangs exactly the same every time. The measurements were pulsed and taken with a distance of 12" from the LES to the surface of the Apogee sensor. I used 3 multimeters to monitor PAR, voltage and current at the same time. All the COBs are brand new and this is their first use. Each curve represents at least 10 measurements.
At one point I received a sealed CXB3590 50 pack from Kingbrite, but this test was done with a CXB3590 3500K CD that did not come from a sealed pack so it is a "trust" COB. I think these results confirm we are getting top bins as promised from Kingbrite and I have full confidence in Cutter as well.
The CXB3070 3500K BB was provided by Pacific Light Concepts, thank you!
The CXa3070 came from a sealed 100 pack from Arrow. (Very interestingly this COB has been shipped from China to the Netherlands to the US to Australia and back to the US.)
The curves are "compressed" and I am not exactly sure why that is. I think the same goes for temp droop so that explains why measured temp droop always seems to be a bit lower than the spec sheets predict. So these curves cannot be used to extrapolate actual efficiency at low currents as I originally hoped, but they do confirm the continual increase in efficiency even at very low power. So it looks like we need an integrating sphere to get the absolute slopes (or we trust the CREE datasheets).
I plan on adding Vero 29 3000K V2.0, CXB3070 3000K AD, CXB3590 3000K CB, CXB3590 5000K CD 80 CRi to the graph but this is what I have so far. Thankfully, these results agree with the datasheets.
At one point I received a sealed CXB3590 50 pack from Kingbrite, but this test was done with a CXB3590 3500K CD that did not come from a sealed pack so it is a "trust" COB. I think these results confirm we are getting top bins as promised from Kingbrite and I have full confidence in Cutter as well.
The CXB3070 3500K BB was provided by Pacific Light Concepts, thank you!
The CXa3070 came from a sealed 100 pack from Arrow. (Very interestingly this COB has been shipped from China to the Netherlands to the US to Australia and back to the US.)
The curves are "compressed" and I am not exactly sure why that is. I think the same goes for temp droop so that explains why measured temp droop always seems to be a bit lower than the spec sheets predict. So these curves cannot be used to extrapolate actual efficiency at low currents as I originally hoped, but they do confirm the continual increase in efficiency even at very low power. So it looks like we need an integrating sphere to get the absolute slopes (or we trust the CREE datasheets).
I plan on adding Vero 29 3000K V2.0, CXB3070 3000K AD, CXB3590 3000K CB, CXB3590 5000K CD 80 CRi to the graph but this is what I have so far. Thankfully, these results agree with the datasheets.
Last edited:
SupraSPL
Well-Known Member
To go a step further to try and analyze the difference in efficiency between the 3070 and 3590 top bins, since they were mounted on the exact same heatsink I adjusted the power to exactly 25W and exactly 50W and allowed the heatsink temp to stabilize. Then I used a digital probe to try and get an accurate reading, although it was tricky because the probe needs to be preheated and the probe interferes with the convection, so I tried my best to be consistent and get accurate numbers.
CXB3070 3500K BB:
@25W = 131F (55C)
@50W = 180F (82C)
CXB3590 3500K CD:
@25W = 127F (53C)
@50W = 169F (76C)
This temperature test should give us some indication of the actual conversion efficiency, which might come in handy for evaluating COBs that we do not know the Apogee correction factors for. And it could come in handy for the Vero vs Cree showdown.
The heatsink used was 3.5X6" Heatsink USA (PLC CXB3070 pictured)
CXB3070 3500K BB:
@25W = 131F (55C)
@50W = 180F (82C)
CXB3590 3500K CD:
@25W = 127F (53C)
@50W = 169F (76C)
This temperature test should give us some indication of the actual conversion efficiency, which might come in handy for evaluating COBs that we do not know the Apogee correction factors for. And it could come in handy for the Vero vs Cree showdown.
The heatsink used was 3.5X6" Heatsink USA (PLC CXB3070 pictured)
Last edited:
SupraSPL
Well-Known Member
They will blend, but not evenly, so my plan is to use COBs that we know can stand on their own for flowering (2700K 3000K 3500K 4000K) and then hopefully benefit from whatever blending does occur
Last edited:
EfficientWatt
Well-Known Member
Thanks Supra !
Looks like everything's back in it's place, so to speak.
Looks like everything's back in it's place, so to speak.
OK I have the CXB3070 3500K BB from PLC and a CXB3590 3500K CD from Kingbrite. I decided to scrap my original data and start over with a better methodology. This time each COB was mounted on the exact same heatsink and the heatsink was hung using chain with the holes marked so it hangs exactly the same every time. The measurements were pulsed and taken with a distance of 12" from the LES to the surface of the Apogee sensor. I used 3 multimeters to monitor PAR, voltage and current at the same time. All the COBs are brand new and this is their first use. Each curve represents at least 10 measurements.
At one point I received a sealed CXB3590 50 pack from Kingbrite, but this test was done with a CXB3590 3500K CD that did not come from a sealed pack so it is a "trust" COB. I think these results confirm we are getting top bins as promised from Kingbrite and I have full confidence in Cutter as well.
View attachment 3566917
The CXB3070 3500K BB was provided by Pacific Light Concepts, thank you!
View attachment 3566899
The CXa3070 came from a sealed 100 pack from Arrow. (Very interestingly this COB has been shipped from China to the Netherlands to the US to Australia and back to the US.)
View attachment 3566918
The curves are "compressed" and I am not exactly sure why that is. I think the same goes for temp droop so that explains why measured temp droop always seems to be a bit lower than the spec sheets predict. So these curves cannot be used to extrapolate actual efficiency at low currents as I originally hoped, but they do confirm the continual increase in efficiency even at very low power. So it looks like we need an integrating sphere to get the absolute slopes (or we trust the CREE datasheets).
I plan on adding Vero 29 3000K V2.0, CXB3070 3000K AD, CXB3590 3000K CB, CXB3590 5000K CD 80 CRi to the graph but this is what I have so far. Thankfully, these results agree with the datasheets.
View attachment 3566900
robincnn
Well-Known Member
At 50 watts Cree PCT says about 6.5% more output with 3590 CD than 3070 BB
At 50 watts our spreadsheet says about 10.3% more output with 3590 CD than 3070 BB
At 50 watts your data says about 6.7% more output with 3590 CD than 3070 BB
So I agree Jerry is legit. Thanks for the test @SupraSPL
Looking forward to your Vero 29 3000K test. I am curious if it offers similar or better performance then 3070 BB
In my Vero 29 3500K and 3590 3500K test i saw vero looked more cool/blue and 3590 was more warm/red
Now i do not need to buy 3590 and can build Vero 3000K lamp as i originally planned.
I assume you talking about this compression i highlighted in black.
Your results look very accurate. Even the datasheet current vs flux looks slightly compressed.
I think case temperatures rises very fast at higher currents and assume junction temperature would increase even faster.
I think the small rise in Tj at higher current is giving some compression even though we trying to take pulsed reading.
from vero 29 datasheet
The efficiency vs current looks very linear. In real world scenarios(not pulsed) the case temp will be higher for higher current. So even more compression.
Which means we gain more and more efficiency as the drive current is reduced.
At 50 watts our spreadsheet says about 10.3% more output with 3590 CD than 3070 BB
At 50 watts your data says about 6.7% more output with 3590 CD than 3070 BB
So I agree Jerry is legit. Thanks for the test @SupraSPL
Looking forward to your Vero 29 3000K test. I am curious if it offers similar or better performance then 3070 BB
In my Vero 29 3500K and 3590 3500K test i saw vero looked more cool/blue and 3590 was more warm/red
Now i do not need to buy 3590 and can build Vero 3000K lamp as i originally planned.
I assume you talking about this compression i highlighted in black.
Your results look very accurate. Even the datasheet current vs flux looks slightly compressed.
I think case temperatures rises very fast at higher currents and assume junction temperature would increase even faster.
I think the small rise in Tj at higher current is giving some compression even though we trying to take pulsed reading.
from vero 29 datasheet
The efficiency vs current looks very linear. In real world scenarios(not pulsed) the case temp will be higher for higher current. So even more compression.
Which means we gain more and more efficiency as the drive current is reduced.
Last edited:
BuddyColas
Well-Known Member
Good on ya Supra! I know it’s not easy being a “COB Truther!” It’s good to see you just took the measurements and let the photons fall where they may.
Looking forward to the heavy weight match when the Vero 29 enters the ring…and to see if it only loses by just 4%. Considering the price difference, an average 4% loss in efficiency (particularly at lower power levels where I like to operate) would actually be a big victory for the Vero 29.
Thanks. I’m glad cooler heads prevailed.
Last edited:
- Status