I run 58.8w COBS @1050ma as a rule (clu038’s) I’ve also used clu048’s. I don’t see the point in paying twice the price a COB to drive it lower than it’s smaller sibling 37.8w average from the clu038’s a COB @1050ma.
I have also ran a 480w solstrip rig in a 2x4ft tent. (Around 60w/sq ft) yield was massive but crazy foxtails and shit. I’ve found my personal sweet spot to be around 42w sq ft with good white LED
Not sure if anyone hit this yet and I’m not sure of numbers (par, lumens, umol, etc.), but there is a point where too much light will take you to the point of diminishing returns. Too much of a good thing. Others who know better will elaborate, I’m sure.
We’ve discussed this before, mate. Quite some time ago so I don’t remember the details, but apparently I didn’t do a very good job of explaining why I run my COBs at the values recommended by Citizen.
I’ll make that effort now. Let’s take your CLU038-1206C4 COBs as an example. I’ll reference the Citizen PDF as I go.
The first hint is in the chart of electro optical characteristics. This chart represents the ideal operating conditions of the COBs. It is in the manufacturer’s best interest to present the characteristics of the COBs under optimal operating conditions with respect to efficiency and reliability. As you can see, all measurements of these characteristics were taken at the nominal forward current of 540mA.
The next data point is the graph showing forward current vs. relative luminous flux. The graph clearly shows that relative luminous flux at 540mA is 100% (they chose this point as “100%” for a reason) and if you look where I have highlighted 1050mA you can see that the relative luminous flux is ~170%. This translates directly to a rather significant drop in efficiency. Doubling the forward current does not double the relative luminous flux. Also note, this data was collected at a case temperature of 25C, which will be impossible in real-world use unless you are using liquid cooling or something. At higher (read normal) temperatures, the efficiency loss will be that much worse.
It is easy enough to do the math to discover just how much the efficiency drops at 1050mA forward current. First, we need to look back at our chart of electro optical characteristics.
If we use the same math but substitute higher forward current, we get 34.6V x 1050mA = 36.33W
We already know (from the graph) that at 1050mA forward current, our relative luminous flux is 170% of the value at 540mA, so to calculate the approximate luminous flux at 1050mA forward current.
2827lm x 1.70 = 4805.9lm
Therefore, the efficiency (“efficacy”) of the unit at 1050mA is 4805.9lm / 36.33W = 132.28 lm/W
This all assumes that temperature is kept constant, but we know that is not the case. All other variables kept equal, more current will result in higher case temperature, and higher case temperature results in further efficiency losses:
Finally, I’ll address reliability. It is no accident that all of Citizen’s reliability testing was done at the nominal forward current of 540mA. Like many electronic components, running a COB at nearly double its nominal forward current will absolutely reduce MTBF. By how much, I do not know, but these things tend to be exponential, not linear.
Based on all of this, I decided it was well worth it to spend $85 CAD on five CLU048-1212 COBs and run them at exactly nominal forward current values, rather than spend $52 CAD on five CLU038-1206 COBs and run them at double their nominal forward current. It was a $33 no-brainer for me, based on all the factors considered above, and the minimal price difference. I was building one light, not hundreds, and I wanted to do it right.
Note: Yes, I am aware of the irony that I went to these lengths to understand and design my COB solution, but then rigged it up by zip-tying heat sinks to the side of a blurple and using thermal compound and kapton tape to secure the COBs. Story of my life. Spend hours researching and understanding something, and then slap it together in 30 minutes like McGuyver. So far, so good, however. 
That explains it perfectly. And in sure we have discussed it before mate but I spend about 90% of my time high af so if I ain’t wrote it down I usually forget 
my first light build was pretty similar to yours but 6 x clu048 driven by a meanwell HLG320hc700a. Super efficient af but didn’t warrant the extra spend. (The price difference UK sides is exactly double a chip btw)
I’ve never had an issue with the clu038s being driven too high on a meanwell driver. There’s still a few I’ve ran on cheapo ll2250pf drivers @50w and only one failed so far not in my presence lol.
I do appreciate the irony very much btw
Our approach was actually quite different if you look a bit deeper. Mine is 5 of those COBs driven by a HLG-185H-36A, wired in parallel. It is a 36V supply rated at 5.2 amps, which just happened to be right on the sweet spot for 5 of these COBs (36V @ 1040mA each). The driver runs at maximum efficiency, and was quite affordable. Running parallel wiring allowed me to use lighter gauge wire because each wire only has to carry enough power for one COB. Just so happened that I had a huge roll of solid-core CAT-5 cable. I used 2 strands for positive and 2 strands for negative, which easily handles the ~36W.
I’m trying to figure what sort of wire to use for my rig. My instinct is that this should be ok but could use a little feedback.
What you think? It’s only for an 80W system
I meant more of I spent a lot of time researching. In all fairness at the time I had no idea the longevity of the COBs and heat output running passive. which is why I drove at 700ma. Not saying the build was similar more the principles of the build (Max efficiency given the relative circumstances of each party)
Min draw.
80W isn’t much, but in order to calculate it accurately I’d need volts and amps.
Also, parallel or series?
In my head, that’s what my first light was going to look like. A nice frame - everything done neatly.
Nice job, mate! Some day I am going to re-do mine, get rid of the blurple and make my COB rig nice like that.
By the way, what device are you using there to display amps, voltage, etc?
Voltmeter ammeter. You can pick them up on eBay really cheap. I just got an ip rated housing and cut a slot out to house the screen and hide the shunt and wiring
Very nice. I’m doing that for sure.
And as you can see I had 9xclu048s on that. I know I said 6 earlier. I really need to get some sleep man 
115v and 700mA. Wired in series. The specs say 300v
Yep. 18GA will do it. In fact, 20GA (which is actually smaller if you didn’t know that already) would even be fine.
I highly recommend getting two colours of wire, however!
For future reference: https://www.solar-wind.co.uk/info/dc-cable-wire-sizing-tool-low-voltage-drop-calculator
Much obliged. Thank you.
1-2 degrees C under ambient ;), cause yes i have measured lol
@cogitech now i have a similar mindset in running at listed nominal ratings for efficiencies and heat loads and what not, but at the same i also have very few quams of turning up a light past those nominals to a point where your getting extra light if its only costing a few extra watts in inefficiencies “70% more light but 14% less efficient overall” so sure in your application thats what 25watts wasted if running at 1050mA, but say you wanted to match that light level output wise running at nominal, you would need another 3.5 of the clu048-1212s to match the over driven ones at a cost of what $17? a piece by what you said you paid so say $60 extra to buy those chips ignoring i split one in half lol.
But say you wanted to factor in cost of electricity vs the initial build offset cost of overdriven vs nominal while discounting the general cost that you’d be paying anyways electricity wise.
quick calculation and guesstimating lets say
25w of waste x 18hr a day = 0.45kwh a day x $0.11 per kwh = $0.0495 per day
That extra $60/0.0495 = 1212 days or 3.2 years just to break even vs the upfront cost to be like for like light level wise…
And for a lot of us we tend to upgrade by that point anyways…
So im all for and advocate for running at nominal ratings for efficiency, heat load, and longevity sake, but realistically when factoring in cost offsets and when one typically upgrades, might as well turn up the juice if things can take it or you have the ability to do so ;).
Also i found all the HLG drivers ive had they actually put out a good portion more than there rated amperage if you wanted it by a good 20+%, another something to consider if you wanted to play around.
Making a second post just as a separate point.
something should be said about the intensity of a light source vs the distance and happy zone of plant growth that it can support.
So say you have a spread out strip array, you may have a 12" happy zone for growth but due to the intensity of the initial source that happy zone may only be from 2" to 14" away from a light, whereas say running cobs, or single point light sources, that same happy zone may start at 24" going to 36" away due to being too intense the closer up you are.
Now specific correlations of distance vs intensity and how far away one needs to be or at what light levels our plants are happy all that i cant say because it varies from space to space, setup to setup, and plant to plant.
In your honest opinion SIR…if I build a killer Bridgelux setup, can it match my Gavita De 750 HPS???
I have the Gavita so dialed in, and it covers my 4 x 4 area so damn well, but it throws so much heat and burns $$$ electricity like mad. Im still worried about penetration. I set my light at about 25-28 inches when cranked to 750 or 825 watts, and she will penetrate well to the grow room floor. I trim lower stuff, never grow larf, want to keep it that way.
I trust your judgement/wisdom on this subject…and others too, lol!!!
