Double check my numbers here...

[Scotch089]

ive talked with supra, im sure way too much for him, trying to wrap my head around simple shit... and one has resurfaced. I think I got it... but honestly the numbers are hard to believe. I did a general 'mean' calculation and going from the average of the 3k/6k/670 efficiency to the 5k/670 efficiency, it was 20% more efficient, even with the reds........ so I was at work and decided to... not work on work. but crunch these numbers.

my original sheet for my setup is off (assuming because of the different Voltages, the ABs were put in at 32.6, when they are really 33.6 according to supes chart, examples like that..) so I dono how close those numbers are because calculated draw ended up being 10-12w more than what actual is (124 vs 112) but I came up with:

700mA
x8 cxa 3k ABs are 52.37%, so 12.36Pw from 23.61w (total= 98.88 of 188.88w)
x8 V5 2A XP-L 5700-6k are 53.49%, so 1.126Pw of 2.051w (total=9.008 of 16.408w)
x11 XP-E FR 901s are 48.53%, so 0.771Pw from 1.589w (total=8.482 of 17.479w)

+ the 7% driver losses so I multiplied the total draw of all three colors by 1.14 (two drivers) (I take it back, two drivers takes it up to 119 calc'd watts, at least a little closer)
+ 11.5w for the x4 fans

so 114.92/274.976= 42.79% efficient at a max current of 700mA (1100 umol at 18")

I did this the exact same way for the 350mA and came out to: 58.76/136.754= 42.96%

For going half the power you would expect it to climb more, or is that just me...? but the results hold true for the new 5k/red going in its place.....


700mA
x8 cxb 3070 5k BD are 63.04%, so 14.88Pw from 23.61w (total= 119.07 of 188.88w)
x19 XP-E FR 901s are 48.53%, so 0.771Pw from 1.589w (total= 14.649 of 30.191w)

+ the 7% loss on the WHITES ONLY from an HLG-185H-C700A
+ the 11% loss on the REDS ONLY from an HLG-40H-54A
+11.5w for the fans

comes out to 131.85Pw/247.11w= 53.3% at max current of 700mA, hoping to go higher on the umols at 18" given the less phosphor

same for 350mA= 67.35/124.6= 54% efficient

excited about the slightly less draw, talking 25+ watts, but I am wondering how solid these numbers are.... please @solomontin, @SupraSPL, @Positivity, etc. Correct my mistakes. And I just copied this all off the messy paper I wrote on from work so numbers could be copied wrong

 

[salmonetin]

...too much for solomontin...sorry

...i dont have supra spreadsheets for all your cree models... i cant do my maths...

pd.. CXA 3k AB...(3070?...3590?...) Vf @ 50ºC @ 700 mA????
...the values for % eficiency are for 700 mA??
XPL V5 2A 5700 - 6000k... etc
XPE FR 901s... etc
CXB 5k 3070 BD...etc

...maybe with PCT tool... i never tried... i cant play with a lot of online tools...sorry


saludos

 

[Scotch089]

Lol sorry for the mispell. I knew you name was close to that..

 

[Positivity]

What you looking for scotch? Total effIciency?

Can't really help much there, sorry. Numbers look good, maybe missing the percentage of monos in mix?

I just go about things in a round about way. get the best emitters I can and leave it at that.... 4k with red works well but the strain I ran is new so I'm not sure whats its supposed to look like. Tight nugs..

 

[Scotch089]

Yea total efficiency. My personal goal has been to make myself a panel that hits 50%efficiency. But not to sacrifice spectrum for it. After watching greens super panel kill it that kind of reassured me that even just nw could wreak havoc. So I'm going back in time (like you have) and mimicked the AT and SGS I loved. At this point I have been considering just ordering all the parts and making a whole other panel but don't have another area to use it AND the original layout one. So plan is to clone a keeper and run it under both. Judge which one I like more. After reading Sds' old thread I missed I'm feeling pretty confident in the 5k+660-670

I calculated the monos in and added the parwatts and output to each respectively and then divided it out, so they SHOULD be included. Just super happy their low efficiency (not all that low, actually I messed up on the original post [I put 48 and it's 42 at 700mA]) didn't drag it down that far. I think it was a combo of the ABs, reds and the lower than cob xpls efficiency dragging me down.

 

[alesh]

Yea total efficiency. My personal goal has been to make myself a panel that hits 50%efficiency. But not to sacrifice spectrum for it. After watching greens super panel kill it that kind of reassured me that even just nw could wreak havoc. So I'm going back in time (like you have) and mimicked the AT and SGS I loved. At this point I have been considering just ordering all the parts and making a whole other panel but don't have another area to use it AND the original layout one. So plan is to clone a keeper and run it under both. Judge which one I like more. After reading Sds' old thread I missed I'm feeling pretty confident in the 5k+660-670

I calculated the monos in and added the parwatts and output to each respectively and then divided it out, so they SHOULD be included. Just super happy their low efficiency (not all that low, actually I messed up on the original post [I put 48 and it's 42 at 700mA]) didn't drag it down that far. I think it was a combo of the ABs, reds and the lower than cob xpls efficiency dragging me down.

What drags you down at the lower current is that 11.5W for fans.

If you have 7% driver losses, divide the the draw by 0.93 (or multiply by 1/0.93 - same thing). And if you have multiple drivers with the efficiency of 93%, the losses are always 7% no matter how many drivers you have. In the first example @700mA, the total draw w/ fans should be ~251W.

There's also one more thing to consider:
deep/photo reds produce about 2.2 µmol/J @40% efficiency
3000K 80CRI whites produce 2.2 µmol/J @45% efficiency
cool whites 70CRI produce 2.2 µmol/J @48% efficiency

 

[SupraSPL]

Yep, multiply by 1.075 to add 7% driver losses rather than multiplying by 1.14. You could get your fan power numbers lower by optimizing the heatsink design for efficient cooling (long channels) and using an efficiency level 5 fan PSU that is properly loaded 75-80% of its max current.

Here is what I get for the second setup:

@700mA
X8 CXB3070 5K BD (23.6W) 63% = 119W of 189W.
X19 XP-E FR 901 (1.54W) 42.4% = 12.4W of 29.26W
spectrum 22% blue
131.4/218.26 = 60.2% average efficiency
w driver losses and fans = 246.13
131.4/246.13 = 53.39% system efficiency


@350mA
X8 CXB3070 5K BD (11.42) 66.4% = 60.66W of 91.36W.
X19 XP-E FR 901 (.702W) 48.53% = 6.47W of 13.34W
spectrum 22% blue
67.13/104.7 = 64% average efficiency
w driver losses and fans = 124W
67.13/124.05 = 54% system efficiency

with fans shut down = 112.55W
67.13/112.55 = 59.5% system efficiency


So according to the Cree PDF, dimming the CXB3070 from 700mA to 350mA does not give us a very large boost in efficiency (5.4% more light/W) but the deep reds do (14.5% more light/W)

 

[Scotch089]

So I was right on the second build but messed up doubling up on driver inefficiencies... super awesome. At least I have the general idea. And that means the first build is a fuzz more efficient than I calculated. But @alesh I went out and double checked my killawatt meter after you said 251, that would be wicked but I knew it was up in the 270s. Is this just the 5% tolerance coming into play on the amps? I'm off somewhere because I go from 112-113 to 275.

 

[alesh]

So I was right on the second build but messed up doubling up on driver inefficiencies... super awesome. At least I have the general idea. And that means the first build is a fuzz more efficient than I calculated. But @alesh I went out and double checked my killawatt meter after you said 251, that would be wicked but I knew it was up in the 270s. Is this just the 5% tolerance coming into play on the amps? I'm off somewhere because I go from 112-113 to 275.

View attachment 3446292 View attachment 3446293

The theory doesn't always correspond to reality. Can you please post exact setup you've got? It would be interesting to compare calculated values against reality.

 

[Scotch089]

X2 hlg-185h-c700a
X8 cxa3070 ABS (4 to each driver)
X8 xpl 5700-6000 (4 to each)
X11 xpe 670s (5 and 6 to each)
X4 prolimatech fans (dono the model off the top of my head but I can find out)
X1 fan psu (but I'd have to dig up the specs on it)

On a 7.28x28" sink

 

[Scotch089]

It's a 12V 2A fan psu and pro 140mm fans. I can dig up specs if you need me @alesh

 

[Scotch089]

After talking with you guys I may shave off two of those fans,may have go buy another Psu but it doesn't sound like 4 is really doing much more good than 2

 

[alesh]

X2 hlg-185h-c700a
X8 cxa3070 ABS (4 to each driver)
X8 xpl 5700-6000 (4 to each)
X11 xpe 670s (5 and 6 to each)
X4 prolimatech fans (dono the model off the top of my head but I can find out)
X1 fan psu (but I'd have to dig up the specs on it)

On a 7.28x28" sink

The drivers are not maxed out - lower efficiency. They probably also put out slightly higher current than 700 mA (can you measure it?). I believe that this makes that missing ~20W.

(https://www.rollitup.org/t/diy-led-cree-cxa3070.789575/page-97#post-11010292)

 

[Scotch089]

Yea but anything roughly >55% of load is 93% efficient so we took thst into consideration. they could be running at 735mA and 333mA, I may have crunch numbers and see how much more draw that is. But damn, 25 watts.. I guess that'd take it from 250 to 262, but that's already including fans. But it does make even more sense on the opposite side, 119 to 113w.

O well thank you for all the help alesh, I need to make a spreadsheet where it's plug and play the numbers instead of manually using a calculator slowly but surely. Lol

 

[guod]

with a 50% load on the driver, you will be fucked up by the Crest-Factor. of the input current.
a known issue with modern switching supplys and cheap power-meters.

 

[alesh]

Yea but anything roughly >55% of load is 93% efficient so we took thst into consideration. they could be running at 735mA and 333mA, I may have crunch numbers and see how much more draw that is. But damn, 25 watts.. I guess that'd take it from 250 to 262, but that's already including fans. But it does make even more sense on the opposite side, 119 to 113w.

O well thank you for all the help alesh, I need to make a spreadsheet where it's plug and play the numbers instead of manually using a calculator slowly but surely. Lol

Read carefully Supra's post I linked. There's a difference between 50% load high voltage/low current and vice versa, which the graph in the data sheet does not address.
Also when I'm looking at Prolimatech website, their 140mm fans have ratings ranging from 2.04W to 5.4W.
There are also some margins for error - driver, kill-a-watt (I'll bet it's 5% at least there).

e: guod makes an excellent point.

 

[salmonetin]

...thanks Guod... ...

with a 50% load on the driver, you will be fucked up by the Crest-Factor. of the input current.
a known issue with modern switching supplys and cheap power-meters.

....crest-factor know issue? ...oh my god... any link to read about crest factor issue?...

...Wilson....time to infusion an read...

...pd... mmmmm....driver max out (90-100 % charge???)....... ....but heyyyyy.... they talk about an HLG driver (high efficiency... regulation on vf outoput ...amps output regulation too... lot of driver protections (but good?... )... ...i think 70-80% charge(70-80% output vf from max output Vf of driver) is good for efficiency and good for the driver lifespan...more charge less driver lifespan... 90-100 % charge its a bit extreme to my eyes,,,but its only my inexpert pov... ...on enclosed enviroments (lamp cases) maybe need check temps inside of driver (not on the driver case)... on high charges and long runs...

saludos

 

[Scotch089]

Read carefully Supra's post I linked. There's a difference between 50% load high voltage/low current and vice versa, which the graph in the data sheet does not address.
Also when I'm looking at Prolimatech website, their 140mm fans have ratings ranging from 2.04W to 5.4W.
There are also some margins for error - driver, kill-a-watt (I'll bet it's 5% at least there).

e: guod makes an excellent point.

The ones I am using at 2.4w bh-1 or something like that.

I'm about to read your link and search the crest factor. Thanks for the clarification guys

(I miss you guod)