REALSTYLES
Well-Known Member
I just made a light that is 69-70% efficient and if I was to dim it to 250ma it would be 77% efficient.
Sent from my SM-G900P using Rollitup mobile app
Sent from my SM-G900P using Rollitup mobile app
DiY LED - Cree CXA3070
- Thread starter SupraSPL
- Start date
ttystikk
Well-Known Member
I'm impressed. Is it very expensive to get high PPfd numbers at the 70% efficiency level?
REALSTYLES
Well-Known Member
It's all about the bin. I just happened to get the DD bin cost the same as the others just the highest you can get and no Jerry doesn't have anymore. I asked for more and he said no he can't get them lol, but he is getting some 6500k 36v DB soon
EfficientWatt
Well-Known Member
Using supra's numbers right ? that is highly likely over estimated by ~10% ... so more like 63-64% and 70% at 250mA ...
I really think we should use "proper" numbers, or Supra should redo tests or something ...
I really think we should use "proper" numbers, or Supra should redo tests or something ...
REALSTYLES
Well-Known Member
We've gone over it a few times and I've used the Cree pct and at 700ma it's 200 lm/w so the calculation is right
Edit: Supra and I went over it lol
SupraSPL
Well-Known Member
These are the CXB numbers I use. Hard to figure the Tc but not too far off.
The tricky part is figuring low current data for CXB3590. I extrapolated from CXA3590 to CXB3590 using the efficiency difference between 25W and 50W and transferred that to the CXB3590 sheet. So they are not unreasonable estimates.
That is how I get these numbers for the top bins. The DB is very impressive the DD is a big step up from that.
You can check for current droop with a lux meter and a pair of DMMs. They really do put out more lux/W as you decrease current even to very low levels
The tricky part is figuring low current data for CXB3590. I extrapolated from CXA3590 to CXB3590 using the efficiency difference between 25W and 50W and transferred that to the CXB3590 sheet. So they are not unreasonable estimates.
That is how I get these numbers for the top bins. The DB is very impressive the DD is a big step up from that.
You can check for current droop with a lux meter and a pair of DMMs. They really do put out more lux/W as you decrease current even to very low levels
Last edited:
EfficientWatt
Well-Known Member
Hi sup,
No intention to diminish your work, big up to you Supra !
But (As far as my understaning goes) :
- You are using CXA curves to guess CXB data. : Fair enough, but we don't know that as many dies are used, and that the efficiency gains are the same, if you would be so kind as to share all your data ... I think others will find that obviously you are not as consistent with data as ou might have been in the past .. Some correlations & curves don't match, and I beleive you know that.
- Why don't you adjust those curves to the PCT DATA ...I mean, if CREE says that a COB (CXB) has a certain lumen output at a given current and tc, how can you honestly think they are underestimating by close to 10% ? (especially when confronting it with past data .... from past PCT/Datasheets given by the same company ... ) they might be a good company and all ..but come on ... who would undervalue their own product that much while having so precise and top notch tech ... just a Q, but a relevant one, I beleive...
Take care and love, with your garden and loved ones,
EW out !
No intention to diminish your work, big up to you Supra !
But (As far as my understaning goes) :
- You are using CXA curves to guess CXB data. : Fair enough, but we don't know that as many dies are used, and that the efficiency gains are the same, if you would be so kind as to share all your data ... I think others will find that obviously you are not as consistent with data as ou might have been in the past .. Some correlations & curves don't match, and I beleive you know that.
- Why don't you adjust those curves to the PCT DATA ...I mean, if CREE says that a COB (CXB) has a certain lumen output at a given current and tc, how can you honestly think they are underestimating by close to 10% ? (especially when confronting it with past data .... from past PCT/Datasheets given by the same company ... ) they might be a good company and all ..but come on ... who would undervalue their own product that much while having so precise and top notch tech ... just a Q, but a relevant one, I beleive...
Take care and love, with your garden and loved ones,
EW out !
SupraSPL
Well-Known Member
All good @EfficientWatt I make plenty of mistakes and I genuinely appreciate people checking my work and voicing any concerns. I do share all of my data and it is posted above. I will run another current droop test on the CXB3590 and post the results to demonstrate and quantify the increase in lux/W at low currents. Anyone with a CXB3590, dimmable driver, pair of multimeters and lux meter can reproduce the current and temp droop tests.
Reasons why the Cree PCT and my spreadsheets disagree:
The PCT is using minimum figures (example, it says CD [12000]) and my spreadsheets are now using typical figures (because the Vero kind of forced us to with its wide gap from min to typ).
The PCT is incorrectly listing Tj (junction temp) when it is actually going by Tc (case temp). You guys know I have been moaning and groaning about the potential for confusion and mistakes when Tc became popular in the spec sheets, but I understand what they are trying to do.
Cree PCT does not provide data for the CXB3590 at low currents and neither does the spec sheet. The CXA3590 PDF does provide low current data in the 23W range (posted above). It is true that there might be slight differences in temp/current droop between CXA3590 and CXB3590, not something that would cause a 10% error, maybe 1%.
Generally speaking Cree does not seem to support or advise designing systems to run at these very low currents. Commercial systems are more likely to be designed to run hotter and at the nominal current suggested in the PDF. Cree figures if you want low power to your COB, you use a smaller COB. I thought they might embrace high efficiency applications because running at low currents with very high efficiency is what Cree does best, the Vero beats them at running hard and hot so might as well play to your strengths.
Reasons why the Cree PCT and my spreadsheets disagree:
The PCT is using minimum figures (example, it says CD [12000]) and my spreadsheets are now using typical figures (because the Vero kind of forced us to with its wide gap from min to typ).
The PCT is incorrectly listing Tj (junction temp) when it is actually going by Tc (case temp). You guys know I have been moaning and groaning about the potential for confusion and mistakes when Tc became popular in the spec sheets, but I understand what they are trying to do.
Cree PCT does not provide data for the CXB3590 at low currents and neither does the spec sheet. The CXA3590 PDF does provide low current data in the 23W range (posted above). It is true that there might be slight differences in temp/current droop between CXA3590 and CXB3590, not something that would cause a 10% error, maybe 1%.
Generally speaking Cree does not seem to support or advise designing systems to run at these very low currents. Commercial systems are more likely to be designed to run hotter and at the nominal current suggested in the PDF. Cree figures if you want low power to your COB, you use a smaller COB. I thought they might embrace high efficiency applications because running at low currents with very high efficiency is what Cree does best, the Vero beats them at running hard and hot so might as well play to your strengths.
Last edited:
SupraSPL
Well-Known Member
This is the CXB3590 CD, adjusted to minimum figures. You can see that the spreadsheet and PCT agree much more closely (set the PCT to Tsp 30C, which approximates a Tj in range of 50C). In practice, the difference between Tsp and Tj will be lower at low currents and higher at high currents.
Last edited:
BM9AGS
Well-Known Member
Do we have an ETA of this monster?
MonsterDrank
Well-Known Member
alesh
Well-Known Member
ttystikk
Well-Known Member
I'm working on this now.sommeday we'll have that COB chip powered by a super efficient flux capacitor.
BM9AGS
Well-Known Member
I'll attach the cable to the lightening rod at the clock for the 1.21 gigawatts
alesh
Well-Known Member
I was wondering the same. I'm getting about 150lm/W at about 4.1A and 40.5V, Ta = 25°C. I can't really see the relevance of Ta but there you go. Do they mean that the chip has the same temp as ambient? Or are they saying that they're measuring in 25°C ambient which in fact tells us nothing about the COB's temperature?
Does anyone know how much they charge for a piece?
BM9AGS
Well-Known Member
They have to be referring to the chip itself. Because that effects it's efficiency and life expectancy. Which 25c is pretty efficient cooling from large heatsinks to water cooled performance
lednewbie123
Member
Ahoy supra. Ive been following this and some of your other led threads for quite some time. I bought a couple mars hydro lights (the 5w 144 reflector series) for a first time led test run and Im impressed enough to take the plunge with a full changeover to cree. Still a little confused on a few things though and looking for some quick help or to be pointed in the direction of posts its already been answered. Been reading ALOT but still a little confused about these few specifics.
cxa or cxb? My understanding from a few posts I seen suggest cxb is better. Is that correct?
3070 or 3590? When i first looked into this 3070 was the big thing but now I see more recently its turning into 3590. Is one easier to find in better bins than the other?
I understand bins like "z4" "ab" but im not understanding the last letter. Ive seen what looks to me like the same item but have "h" "f" or "g" at the end is there a specific one that should be used? Some examples of what i mean "N0HAB30G" "N0HAB30H" "N00AB30H".
EDIT- I looked over the data sheet some more and found the answer for the last letter being different. Now that brings a new question what is better 2/3/4 step? Found a post saying 2 step would be ideal. Is that correct? And Im still missing the answer to the letter "h" or number "0" before the bin. Example "N00" or "N0H". EDIT- Again lol. Found the H or 0 specifies the cri.
Places to look for ordering? Ive seen mouser and digikey mentioned alot so Ive been looking there. But is there some other option for getting the higher bins? I can spend around $2000
cxa or cxb? My understanding from a few posts I seen suggest cxb is better. Is that correct?
3070 or 3590? When i first looked into this 3070 was the big thing but now I see more recently its turning into 3590. Is one easier to find in better bins than the other?
I understand bins like "z4" "ab" but im not understanding the last letter. Ive seen what looks to me like the same item but have "h" "f" or "g" at the end is there a specific one that should be used? Some examples of what i mean "N0HAB30G" "N0HAB30H" "N00AB30H".
EDIT- I looked over the data sheet some more and found the answer for the last letter being different. Now that brings a new question what is better 2/3/4 step? Found a post saying 2 step would be ideal. Is that correct? And Im still missing the answer to the letter "h" or number "0" before the bin. Example "N00" or "N0H". EDIT- Again lol. Found the H or 0 specifies the cri.
Places to look for ordering? Ive seen mouser and digikey mentioned alot so Ive been looking there. But is there some other option for getting the higher bins? I can spend around $2000
Last edited:
badtorro
Well-Known Member
you prolly want to read some internet, I strongly suggest https://www.icmag.com/ic/showthread.php?t=304279 golden source of info in one place
