CXM22 DIY Build for 2x3 - Advice Wanted

I am just about to buy the components to build a 4 COB light for a 2'x3' area. Full grow, seed to harvest.
I am planning on using 4 Luminus CXM22 3500s near the 4 corners, mounted on 140x70 round pin heatsinks using 2020 aluminum T-Slot for the frame.
The COBs use 51.5 forward voltage with a max of 2.2A.
I am trying to decide if it is better to use 2 drivers (HLB-185H-C1400B) or just 1 (HLB-320H-C1400B). I believe the output achieved will be the same, but is one way more efficient as far as power consumption at the wall?

I was initially thinking that with 2 drivers I could turn 1 off (2 COBs) during the first 3 weeks or so and then turn on the other when the plants start getting big. Powered off means no energy consumption.
Then I was thinking I could save a few bucks by just getting 1 driver, keep all 4 COBs active but dim them down to 10% - 20%.

The purchase difference is only about $20 with 2 drivers and the extra potentiometer (dimmer), so energy consumption seems to be a better concern.

Any advantages/disadvantages to 1 driver vs 2?
I know if I have 2, if ones dies I still have some light.

At 1100mA these COBs give 9215 Lumens. So 4 would give me 36860L ~ 179L/W.

I saw someplace that I needed ~ 250W in a 6 sq ft space to produce 600 to 1000 Par.
So i figured the 4 CXM22 near the corners spaced at ~ 15" x 10" would give a well balanced, full coverage within the 600 to 1000 PAR range.

Any recommendations? Perhaps I should be thinking something total different.

These are the costs I'm looking at:

Frame ~ $75 ish
COBs plus mounts and reflectors - 4@$29.25 ea = $117
Heatsinks - $19 ea = 4@$76
Drivers - 1@$89 or 2@$50=$100
Potentiometers (dimmers) - 1@$6 or 2@$6=$12
Miscellaneous wire and stuff ~$20
Shipping ~ $25
Total ~ $408 or $425
It seems high for a DIY but I think I'm going to be getting way more usable light than I was getting from a $100 ViparSpectra VT300.

Thanks in advance for any advice.

Id suggest both options are hugely overpowered, will you be expanding your growspace at some point? Id suggest 1 driver, using CV instead instead of Cc: 50v cobs are hard to match with cc drivers as they are generally geared towards 36v cobs. One hlg240-54, A version if you can live with dimming on the driver itself, save you a bit of cash. Thats 40w per square feet which seems enough for me. But on that space you might go for a 240 w quantum board kit, should be more efficient, about the same price and,less work and a nice fit for your space. If you insist on cobs id ikip the reflectors (assuming reflective walls) and go for 3500k 90cri.

Rocket Soul said:

Id suggest both options are hugely overpowered, will you be expanding your growspace at some point? Id suggest 1 driver, using CV instead instead of Cc: 50v cobs are hard to match with cc drivers as they are generally geared towards 36v cobs. One hlg240-54, A version if you can live with dimming on the driver itself, save you a bit of cash. Thats 40w per square feet which seems enough for me. But on that space you might go for a 240 w quantum board kit, should be more efficient, about the same price and,less work and a nice fit for your space. If you insist on cobs id ikip the reflectors (assuming reflective walls) and go for 3500k 90cri.

Click to expand...

Thanks for the reply.
Right now I'm on a countertop without reflective walls, so using COBs with reflectors would help me focus the light down to the plants and I think the 4 would give me pretty even coverage for that space. See my theoretical pic Using the 2020 T-Slot allows me to slide the lights in and out as needed. Debating between using the T-slot vs the pre-made frames that Rapid LED sells, (34-canopy-rail). Sorry, rollitup won't let me post a link because I'm a newbie to the site.
I assume you would connect the COBs in parallel using that CV driver. I'm a bit concerned with the CV vs the CC because from what I've learned the CV allows the current to fluctuate which changes the light output. I think the hlg240-54 has to much current for those COBs. The COBs are rated at 2.2A and that driver in A series is dimmable from 2.23A~4.45A (based on the spec sheet) which would mean that even if it does work, the COBs would be running full out even and if the dimmer gets turned up just a bit the COBs could fry, at least the way I understand how the drivers work.
I was going with the 80cri because they offer more light output and I didn't feel that the benefit of the 90cri color consistancy outweighed the extra Lumens of the 80s.

The picture shows coverage that I expect to get, theoretically of course.

• 2 gray circles are the plants.

• 4 Purple circles are the 140mm heatsinks with the CXM22s attached.

• 4 Blue circles show the light coverage, which I can adjust with hanging height or reflectors.

JustChillinLPA said:

Thanks for the reply.
Right now I'm on a countertop without reflective walls, so using COBs with reflectors would help me focus the light down to the plants and I think the 4 would give me pretty even coverage for that space. See my theoretical pic Using the 2020 T-Slot allows me to slide the lights in and out as needed. Debating between using the T-slot vs the pre-made frames that Rapid LED sells, (34-canopy-rail). Sorry, rollitup won't let me post a link because I'm a newbie to the site.
I assume you would connect the COBs in parallel using that CV driver. I'm a bit concerned with the CV vs the CC because from what I've learned the CV allows the current to fluctuate which changes the light output. I think the hlg240-54 has to much current for those COBs. The COBs are rated at 2.2A and that driver in A series is dimmable from 2.23A~4.45A (based on the spec sheet) which would mean that even if it does work, the COBs would be running full out even and if the dimmer gets turned up just a bit the COBs could fry, at least the way I understand how the drivers work.
I was going with the 80cri because they offer more light output and I didn't feel that the benefit of the 90cri color consistancy outweighed the extra Lumens of the 80s.

The picture shows coverage that I expect to get, theoretically of course.

• 2 gray circles are the plants.

• 4 Purple circles are the 140mm heatsinks with the CXM22s attached.

• 4 Blue circles show the light coverage, which I can adjust with hanging height or reflectors.

Click to expand...

With CV drivers you divide that current in 4 for 4 cobs. Even if 2 fails at the same time you should still be ok. Also with A version of driver you can adjust down the Voltage a bit so you could limit how much the cobs can draw.

Running cc drivers far fron full load makes them slightly less efficient. Also CV drivers use less voltage so less danger in zapping yourself.

The issues you quote with parallel connection is very marginal with cobs, its more of an issue with single diodes. Just remember you need the same heatsink on each cob to balance the heat in the same way.

Remember that a cob itself has a lot of parallel connections.

Where are you located?? US / Eu or somewhere else?
For US hit up @CobKits
For EU hit up @diyled
Both have one stop led shops for growers and are more knowledgeable than me, very helpfull, good prices and on this forum.
Still recommend quantum boards if youre in the US

JustChillinLPA said:

The COBs use 51.5 forward voltage with a max of 2.2A.

Click to expand...

not sure if you mean that how its written
the cobs may have a nominal/arbitrarily stated Fv of 51.5, but its well over 54V at 2.2A

JustChillinLPA said:

I am trying to decide if it is better to use 2 drivers (HLB-185H-C1400B) or just 1 (HLB-320H-C1400B). I believe the output achieved will be the same, but is one way more efficient as far as power consumption at the wall?

Click to expand...

see data sheet, 320 is slightly more efficient. 2x 185s are more powerful though (well over 400W), 320 is around 360W

JustChillinLPA said:

I was initially thinking that with 2 drivers I could turn 1 off (2 COBs) during the first 3 weeks or so and then turn on the other when the plants start getting big. Powered off means no energy consumption.
Then I was thinking I could save a few bucks by just getting 1 driver, keep all 4 COBs active but dim them down to 10% - 20%.

Click to expand...

if you got 4 cobs hanging out, dimming them all down is far more efficient and will give better coverage, than turning 2 off

JustChillinLPA said:

At 1100mA these COBs give 9215 Lumens. So 4 would give me 36860L ~ 179L/W.

Click to expand...

at chip level, yes. more like 165-170 after driver losses

JustChillinLPA said:

I saw someplace that I needed ~ 250W in a 6 sq ft space to produce 600 to 1000 Par.

Click to expand...

600 to 1000 is a huge range. 250W in a 6ft space is close to 1000 with a medium efficient cob rig

Rocket Soul said:

Running cc drivers far fron full load makes them slightly less efficient

Click to expand...

down to 50% isnt so bad, though it tails off greatly after that

tell ya what, in that space, your going to love the Luminus cobs, Im running 4 in a 3x3 and its plenty.

Here is my light rig.

4luminus, 2x3500-90cri and 2x4000-80 cri

I have 2 on hlg-185-1400b each, so I can control the spectrum independently, and even totally shut off each spectrum thru my Sonoff 4Chan pro.

After i did it this way, The only benefit of the double spectrum is if the light is high, to get the mix even of the 2 spectrums. need the higher blue in the 4000k but need the higher red in the 3500K, so that was my thinking, because I cant get both with just 1 COB. When i rec=build, going to use 4000K and add reds thru supplemental fashion.

the 4000K max out at 80watts each on the 1400B driver,
the 3500K max out at 88watts each on the 1400B driver,

so i get about 335 watts from all 4 turned max,

at 30 inches, i get 62,000 lux or about 1040 PAR in the center, on 335watts.
at 30 inches, 4000K alone, 470 par, 160 watts
at 30 inches, 3500K alone, 500par, 176 watts

Hoped that helped ya,

Rocket Soul said:

With CV drivers you divide that current in 4 for 4 cobs. Even if 2 fails at the same time you should still be ok. Also with A version of driver you can adjust down the Voltage a bit so you could limit how much the cobs can draw.

Click to expand...

I forgot that with Parallel circuits each COB divides the Total available current, whereas series divides the total available voltage. So with the HLG-240H-54 having 2.23A~4.45A I would divide that by my 4 COBs and be able to run them each at ~ 558mA ~ 1113mA which means that max in this scenario would be just over 50% of the COB's max of 2200mA, with 1100 being the typical FC and what the spec sheet shows as 100%. So if I want to be able to run them a bit brighter I would need step it up to a HLG-320H-54 which is rated at 2.97A~5.97A which would give me 743mA~1488mA which is similar to the CC drivers that I was looking at.

Rocket Soul said:

Still recommend quantum boards if youre in the US

Click to expand...

I'm in the US, Even though I saw that QBs might be a bit more efficient,I think for the cost, the COBs allow me to spread the light out more evenly.

CobKits said:

if you got 4 cobs hanging out, dimming them all down is far more efficient and will give better coverage, than turning 2 off

Click to expand...

I'm not sure that I understand the thought process behind this. Looking at the Efficiency vs Load charts (115V) for the 320H efficiency drops down to about 87% at 30% load, 84@20, and 74@10 where the 185H shows 90@30, 88@20, and 78@10. So wouldn't the 185 be more efficient if run at the same current? Even more by turning 1 driver and 2 COBs off I would have to run that driver at a higher power which makes it even more efficient. I also noticed that the 185H seems to remain constant at around 93% efficiency (115V) @ 50% and above while the 320H doesn't hit 93% efficiency until 70-80% power and then drops to 92%@90-100%.
My COBs will also be pretty adjustable so I can move the 2 that are on directly over the plants and then adjust when I have all 4 on.
Sorry if I come off argumentative, I really do appreciate the help, I just want to make sure that I'm understanding things correctly.

InTheValley said:

at 30 inches, i get 62,000 lux or about 1040 PAR in the center, on 335watts.
at 30 inches, 4000K alone, 470 par, 160 watts
at 30 inches, 3500K alone, 500par, 176 watts

Click to expand...

It's great to see someone that has done something so similar to what I am planning. When you say "1040 PAR in the center" you mean total center of the canopy where all 4 COBs are overlapping? So directly under each COB it would be slightly less, lets say 900 PAR and then around the edges your dropping down to 800 or maybe 700?
I was pondering the 4000k vs 3500K vs 3000K all in 80CRI, But the more I looked at value for the color output I kept coming back to see more value in the lower Kelvin, red range for flowering. I just watched a youtube video called "Grow lighting Masterclass with Dr Bruce Bugbee - Grow Light Spectrum Discussion" it's on the Micro Channel and it appears that by using the 3000K or 3500K you could lower the lights height and/or increase intensity a bit during veg to accommodate the lower % of blue spectrum to help keep the plants shorter while still having the higher % of red for the flowering benefit. So I think I am going to stick with 4 - 3500Ks, although after watching that video and looking at the color spectrum chart for the CXM22s, I'm really wondering if 3000K is better for full cycle, flower is where they really seem to need the extra PAR. I'm just afraid that I'm gonna sacrifice too much blue by being greedy for the Red.
I see that you your COBs locations are fixed, do you like that or would you make them adjustable if you did a remake?
Hard to tell, are you using reflectors? I'm pretty sure that I want/need them to keep the light down and not on the walls (no reflection material). Trying to decide between the Ledil Angelina W 43° silver and the Ledil Angelina XW 90-91° white. I was originally thinking that the W, which should focus more light directly down made sense, but I have the plants on a counter so my total height between the top of my hydro setup to the lights is less than 50". Taking that into consideration I think I might be better off with the XW which I think will giver a broader pattern at the closer height. Thoughts?

Have any of you verified that the power consumption at the wall does actually reduce when the driver is dimmed? If so is the reduction similar to the amount dimmed, ie 60% dimmed translates to 60% less power consumed at the wall?

As I'm completing this I had a summary thought and please correct me if I am wrong ... for the most driver efficiency they should run closer to max, while the COBs tend to have more efficiency when run around 50% max current?

I'm being extra thorough this time around, because as much research as I though I did before buying a ViparSpectra vt300 it just doesn't seem to have enough light for that space, at least during flower. Maybe 1 plant would be OK, not enough red. This light is gonna cost almost 4 times as much so I wanna get it right.

Thank You again for everyone's help.

JustChillinLPA said:

I forgot that with Parallel circuits each COB divides the Total available current, whereas series divides the total available voltage.

Click to expand...

sort of. first part is right, but series doesnt 'divide available voltage.

if you have a series CC driver which is 1400 mA, and your cobs run at 36.5V, they will use 73, 109.5, or 146V if you use 2,3, or 4 cobs. lets say your driver had a max voltage of 156V, you wouldnt be using all the avaialble current in any case

JustChillinLPA said:

I'm not sure that I understand the thought process behind this.

Click to expand...

forget the drivers, i meant on a cob basis (not that youre not right, thats just not what i was talking about

JustChillinLPA said:

Have any of you verified that the power consumption at the wall does actually reduce when the driver is dimmed? If so is the reduction similar to the amount dimmed, ie 60% dimmed translates to 60% less power consumed at the wall?

Click to expand...

that is well known. no its not exact, again, driver efficiency

JustChillinLPA said:

while the COBs tend to have more efficiency when run around 50% max current?

Click to expand...

40% is more efficient than 50%
30% is more efficient than 40%
20% is more efficient than that

etc

InTheValley said:

Here is my light rig.

4luminus, 2x3500-90cri and 2x4000-80 cri

I have 2 on hlg-185-1400b each, so I can control the spectrum independently, and even totally shut off each spectrum thru my Sonoff 4Chan pro.

After i did it this way, The only benefit of the double spectrum is if the light is high, to get the mix even of the 2 spectrums. need the higher blue in the 4000k but need the higher red in the 3500K, so that was my thinking, because I cant get both with just 1 COB. When i rec=build, going to use 4000K and add reds thru supplemental fashion.

the 4000K max out at 80watts each on the 1400B driver,
the 3500K max out at 88watts each on the 1400B driver,

so i get about 335 watts from all 4 turned max,

at 30 inches, i get 62,000 lux or about 1040 PAR in the center, on 335watts.
at 30 inches, 4000K alone, 470 par, 160 watts
at 30 inches, 3500K alone, 500par, 176 watts

Hoped that helped ya,



View attachment 4203525

Click to expand...

How did you measure ? Any light Meter in particular ??

JustChillinLPA said:

I am just about to buy the components to build a 4 COB light for a 2'x3' area. Full grow, seed to harvest.
I am planning on using 4 Luminus CXM22 3500s near the 4 corners, mounted on 140x70 round pin heatsinks using 2020 aluminum T-Slot for the frame.
The COBs use 51.5 forward voltage with a max of 2.2A.
I am trying to decide if it is better to use 2 drivers (HLB-185H-C1400B) or just 1 (HLB-320H-C1400B). I believe the output achieved will be the same, but is one way more efficient as far as power consumption at the wall?

I was initially thinking that with 2 drivers I could turn 1 off (2 COBs) during the first 3 weeks or so and then turn on the other when the plants start getting big. Powered off means no energy consumption.
Then I was thinking I could save a few bucks by just getting 1 driver, keep all 4 COBs active but dim them down to 10% - 20%.

The purchase difference is only about $20 with 2 drivers and the extra potentiometer (dimmer), so energy consumption seems to be a better concern.

Any advantages/disadvantages to 1 driver vs 2?
I know if I have 2, if ones dies I still have some light.

At 1100mA these COBs give 9215 Lumens. So 4 would give me 36860L ~ 179L/W.

I saw someplace that I needed ~ 250W in a 6 sq ft space to produce 600 to 1000 Par.
So i figured the 4 CXM22 near the corners spaced at ~ 15" x 10" would give a well balanced, full coverage within the 600 to 1000 PAR range.

Any recommendations? Perhaps I should be thinking something total different.

These are the costs I'm looking at:

Frame ~ $75 ish
COBs plus mounts and reflectors - 4@$29.25 ea = $117
Heatsinks - $19 ea = 4@$76
Drivers - 1@$89 or 2@$50=$100
Potentiometers (dimmers) - 1@$6 or 2@$6=$12
Miscellaneous wire and stuff ~$20
Shipping ~ $25
Total ~ $408 or $425
It seems high for a DIY but I think I'm going to be getting way more usable light than I was getting from a $100 ViparSpectra VT300.

Thanks in advance for any advice.

Click to expand...

You'll be better off with a single driver. As far as energy savings, remember your COBs will get more efficient the lower you run them. So 4 COBs at 25 watts each is way brighter than 2 COBs at 50 watts each. A good rule of thumb with quality COBs is 37.5 Watts/Square Foot. So 225 Watts would be perfect for your space. You might want to consider getting an HLG-240H-C1050B. It'll run your COBs at 54 Watts each (216 total) and bring your total down to $382. At 54W you can run them 12" above your canopy. Happy growing!

Black Sail said:

You'll be better off with a single driver. As far as energy savings, remember your COBs will get more efficient the lower you run them. So 4 COBs at 25 watts each is way brighter than 2 COBs at 50 watts each. A good rule of thumb with quality COBs is 37.5 Watts/Square Foot. So 225 Watts would be perfect for your space. You might want to consider getting an HLG-240H-C1050B. It'll run your COBs at 54 Watts each (216 total) and bring your total down to $382. At 54W you can run them 12" above your canopy. Happy growing!

Click to expand...

Sorry, I just saw your reply. I think I really want to go with a 1400mA driver so that I have a bit more power available for light hungry plants/strains. I finally ordered the T-Slot for the frame yesterday, but I'm still trying to wrap my head around the efficiency of 1 driver over 2. Looking at the Spec sheets, input only the HLB-320H-C1400B uses 3.5A@115V = 402.5W while the HLB-185H-C1400B uses 2A@115V = 230W Xs 2 drivers = 460W. It seems obvious that the 320 consumes less power, but the 2 185s also have more overall V available for output which explains the higher consumption.
Then I looked at the Output specs, putting them in a spreadsheet and it looks to me that that 2 - 185s will be more efficient (understanding that they will cost a bit more to purchase upfront). Am I missing an additional power factor that would explain why the 1 - 320 should be more efficient than 2 - 185s? Perhaps there is another chart in the spec sheets that I should be looking at.

I attached my spreadsheet showing what I found. Can anyone explain why I can't figure out how 1 - 320 is better than 2 - 185s?

Assuming my chart is correct, For seed to veg. I definitely see now that it is better to run all 4 COBs dimmed down to 29% than to just run 2 COBs at 55%. Watts used is about equal but light output and overall efficiency is better with all 4 running on 2 - 185s, but if I run all 4 at 29% on 1 - 320 the Watts used looks to be higher. Again, I could be missing something?

54 watts each cob (4) at 12 inches will roast the leafs. These cobs are no joke,

Argentocobs said:

How did you measure ? Any light Meter in particular ??

Click to expand...

i use this one, and drilled a hole in the lid, then calibrated it with scotch tape till I get the PAR value = to lux conversion factor. ( 65 ) Sure, its not exact, and its not $126, but it gets you in the general idea of PAR and intensity to dial things in better then using nothing at all, lol..



https://www.amazon.com/Dr-Meter-LX1330B-Digital-Illuminance-Light/dp/B005A0ETXY/ref=sr_1_1_sspa?ie=UTF8&qid=1539193145&sr=8-1-spons&keywords=lux+meter&psc=1

InTheValley said:

i use this one, and drilled a hole in the lid, then calibrated it with scotch tape till I get the PAR value = to lux conversion factor. ( 65 ) Sure, its not exact, and its not $126, but it gets you in the general idea of PAR and intensity to dial things in better then using nothing at all, lol..

Click to expand...

I've read that the cheaper Lux meters don't work well with blurple because they don't see the reds & blues very well. So running the white COBs, do you feel that you are getting a fairly decent full reading? I really like that $34 price if all I have to do is a little math to get me in the PAR ballpark. Are you taking the Lux reading and multiplying by .65 to get PAR? Where did you get the 65 conversion factor? I think I saw someplace that there are different conversion factors for different types of light. I don't understand your calibration,wouldn't you need a control to somehow already know what the PAR is. Can you please explain your calibration process in a bit more detail. Thanks

Hey Justchillin, I get 65 because i use a mix of 4000k 80cri and 3500K90cri, where 4000K80 is usually a 70factor and 3500k90cri is 60, so i use in-between for 65. so, if my lux is 40,000-- I DIVIDE that by 65, gives me about 615 PAR. SO, i drill a 3/8 inch hole in the Lid of my lux meter, ( direct center)then, put the lux meter under lights to find your Lux, write it down,do the conversion, write that down, then use scotch ( clear tape) by layers, till you hit that conversion factor of PAR. Take 1 piece of tape, stick it to your finger tip, to get the oils to transfer to the tape off your finger print. it "blurs" the tape. Then just keep layering over the hole, till U hit the PAR figure. Thats it.. its not exact, but like i said, its a cheap way to get in the ballpark of the PAR.

JustChillinLPA said:

<snip>
I attached my spreadsheet showing what I found. Again, I could be missing something?

Click to expand...

Yes, you are missing the efficiency numbers on the datasheet. For the 185 it shows 94%, for the 320 it's 94%. There you have it, end of story. No need to do any spreadsheet contortions. They are equally efficient drivers.

InTheValley said:

Hey Justchillin, I get 65 because i use a mix of 4000k 80cri and 3500K90cri, where 4000K80 is usually a 70factor and 3500k90cri is 60, so i use in-between for 65. so, if my lux is 40,000-- I DIVIDE that by 65, gives me about 615 PAR. SO, i drill a 3/8 inch hole in the Lid of my lux meter, ( direct center)then, put the lux meter under lights to find your Lux, write it down,do the conversion, write that down, then use scotch ( clear tape) by layers, till you hit that conversion factor of PAR. Take 1 piece of tape, stick it to your finger tip, to get the oils to transfer to the tape off your finger print. it "blurs" the tape. Then just keep layering over the hole, till U hit the PAR figure. Thats it.. its not exact, but like i said, its a cheap way to get in the ballpark of the PAR.

Click to expand...

OK now I understand your 65 factor, I assume that you pulled the 60 and 70 numbers from a chart someplace?
I still don't understand the calibration. Isn't the meter suppose to zero out with the cap on it, no light thus 0 LUX.
Wouldn't you just be able to take a reading under your lights and then divide by the correct factor to get your PAR?
So you take a reading and you get a lux of 40,000 (or where did the 40,000 come from) and divide that by your factor of 65 to get the 615. Drilling a hole in the cap limits the amount of light hitting the sensor, and the tape is acting as a magnifier? This is the part I'm just not understand why/what is happening.
Have you tried any other meters. I saw a couple others that say they have a wider spectral sensitivity to catch more of the lower blues and higher reds but they don't show a graph. A reviewer posted a picture of the instructions for the one you have and the graph shows that most of the sensitivity is focused between 500 - 650 nm, so I was wondering if it is capturing all the blue in the 450 area on these COBs and the reds that venture to 700 and above.
I've never used one of these meters so perhaps I'm missing something.

shimz said:

Yes, you are missing the efficiency numbers on the datasheet. For the 185 it shows 94%, for the 320 it's 94%. There you have it, end of story. No need to do any spreadsheet contortions. They are equally efficient drivers.

Click to expand...

Not that I don't appreciate everyone's input, but I believe that you are actually missing something.
The 94% number is the maximum potential efficiency of those drivers. Once you add loads and start adjusting the dimmer the efficiency changes. For instance, if you look at the spec sheet for the 185/1400 on page 6, there is a graph called efficiency vs load. Based on this graph you can only get 94% efficiency if you are running 233V or 277V at 100% load (dimmer all the way up). In the US with 115V you can only achieve 93% efficiency. On the flip side if you dim the driver down to 10% your efficiency drops drastically to about 79%.

The COBs are pretty much the opposite of the drivers where they are more efficient the lower you run them.

While what you say is true, there would be very little difference between 2 drivers at 94% efficiency driven at 80% each and one big 94% efficient driver at 99% load, but if there was a difference, the edge would go to the 99% loaded driver. Assuming 230VAC, as it looks like efficiency actually takes a little hit at high loads when input is 115VAC.

Edit: Also, I would be willing to bet that there is an efficiency vs. Tcase that they're not showing us. Assuming the bigger driver would run hotter, that might just wash out any load efficiency gains.