Canada Grows to the 4 Plant Limit

CobKits

Well-Known Member
The big improvement with this generation of led is the lower heat waste. Take a look at the Bridgelex data sheets for example, 0.04c/W for the Vero 29c @ 1710ma. At 20w that would be 0.8c delta and probably wouldn't even keep a coffee warm.
that is not a measure of chip efficiency (directly)

it is a measure of resistance of heat flow from chips to substrate

from gen to gen there is a similar amount of heat per watt, better chips just conduct it better (making chips cooler, and brighter)

gen to gen most chip are jumping just a few percent (though the resistance measurements you quote may be jumping more than that, it is only one factor

Vero has great thermal properties but other things that make a chip efficient
-efficiency of dies
-density of dies per sq mm of chip
-current per die (a function of # of dies per chip)
-chip size (area of contact of chip with heatsink)- although thats kind of a component of the thermal resistance measure

your post seems to indicate that a veroC at 0.04 R would be over twice as efficient as a citizen 3618 at 0.095 R (when in reality the citizen is more efficient due to die count and larger footprint and possibly die efficiency). but again we are talking very incrementally more efficient- 1-2%
 
Last edited:

Photon Flinger

Well-Known Member
that is not a measure of chip efficiency (directly)

it is a measure of resistance of heat flow from chips to substrate

from gen to gen there is a similar amount of heat per watt, better chips just conduct it better (making chips cooler, and brighter)

gen to gen most chip are jumping just a few percent (though the resistance measurements you quote may be jumping more than that, it is only one factor

Vero has great thermal properties but other things that make a chip efficient
-efficiency of dies
-density of dies per sq mm of chip
-current per die (a function of # of dies per chip)
-chip size (area of contact of chip with heatsink)- although thats kind of a component of the thermal resistance measure

your post seems to indicate that a veroC at 0.04 R would be over twice as efficient as a citizen 3618 at 0.095 R (when in reality the citizen is more efficient due to die count and larger footprint and possibly die efficiency). but again we are talking very incrementally more efficient- 1-2%

That is correct - it would be twice as efficient in regards to thermal generation for our perspective in that it produces half the heat for using the same electrical power.

Your statement of it being incrementally more efficient is incorrect. Any reduction in heat produced at source using the same electrical power has a significant impact.

Theory is all nice but the real test is going out and doing it. To be blunt, I have and am doing it. Never saw the reason to run an led hot and therefore have never used a bulky heat sink. Have not had an led pop, burn out or fail
 

DIY-HP-LED

Well-Known Member
you emailed me but ill answer here so everyone can see it

1/4" thick flat sheet can do up to 15-20W/chip depending on area.

the cheap 2" extrusions from heat usa prob cool as well. in any case thin metal like baking sheets etch doesnt dissipate heat like thicker metal does.

with a good finned heatsink you can get away with 3/16" thinckness. for flat palte with no fins 1/4" or thicker is best.

finned extrusions win on amount of heat dissipated per wigth of aluminum and arent ridiculously expensive. everybody was using them before pinfins became popular
Pay attention, this guy is an expert. Go to his thread if ya wanna learn
 

DIY-HP-LED

Well-Known Member
15w of heat from running at only 20w? More like the other way around 5w for heat and 15 for light. It is probably even less heat than that.

The big improvement with this generation of led is the lower heat waste. Take a look at the Bridgelex data sheets for example, 0.04c/W for the Vero 29c @ 1710ma. At 20w that would be 0.8c delta and probably wouldn't even keep a coffee warm.
You know more than me about this area and CobKits.com should love ya, if yer ideas catch on, COBs will be flying outta his inventory!;) But the important thing about your calculations is that CPU tape should work fine for the COBs, if ya don't mind soldering them (and I or someone else might figure out a better way). It will make implementing your ideas much more economical.
 
Last edited:

DIY-HP-LED

Well-Known Member
Cheers! Thanks for responding here. I didn't want to drag you into the dirty Canadian forum... :cool:
As you can see you've stimulated discussion and fostered new ideas and that I err on the side of caution when I ramble off the top of my head.
You have done a good thing here. And they say pot makes people stupid, ha!;)
 
Last edited:

DIY-HP-LED

Well-Known Member
hey i like canadians. dirtier the better ;)


naw even at 20W per chip youve got 7-8W of heat to deal with. theyre not that much more dramatically efficient. if you want to get to 70% efficiency you need to be 5-10W/chip
Free free to drop by and put yer 2.2 cents worth in, join any discussion, point out when I fuckup, cause I will till I stop breathing! (and so will everybody else). Glad you showed up Cobby guys like you help, they don't spam. Picking up yer parcel tomorrow and I look forward to playing with yer COBs.
Ever think about moving to Canada? You'd fit right in... If ya get 4 more years of Trump, I'll sponsor ya as a refugee.;)
 
Last edited:

CobKits

Well-Known Member
That is correct - it would be twice as efficient in regards to thermal generation for our perspective in that it produces half the heat for using the same electrical power.
Rj-c doesnt mean that at all, it just tells you what the junction temp will be as a function of case temp and input wattage.

while low thermal resistance contributes to efficacy, thermal resistance is not a measure of efficacy.
Theory is all nice but the real test is going out and doing it. To be blunt, I have and am doing it. Never saw the reason to run an led hot and therefore have never used a bulky heat sink. Have not had an led pop, burn out or fail
doing what? youre defying the laws of physics if youre doubling efficacy over typical cob rigs

every single cob/mid-power board/phosphor light source follows this trend at various currents

upload_2017-9-17_17-35-30.png


and look like this. i put your double efficiency target on the graph you can see that no cob can get there (ive tested almost all of them). lumen numbers from datasheets and simulators demonstrate this as well (except for the broken bridgelux simulator that returns over 300 lm/W at low currents)

upload_2017-9-17_17-42-7.png
the lowest datapoint on that chart is 100mA/~3W. you could run cobs at 30 mA/~1W and still only be 50% more eficient than using 1/50th of the number of cobs at 50W each
 

Photon Flinger

Well-Known Member
Rj-c doesnt mean that at all, it just tells you what the junction temp will be as a function of case temp and input wattage.

while low thermal resistance contributes to efficacy, thermal resistance is not a measure of efficacy.

doing what? youre defying the laws of physics if youre doubling efficacy over typical cob rigs

every single cob/mid-power board/phosphor light source follows this trend at various currents

View attachment 4012164


and look like this. i put your double efficiency target on the graph you can see that no cob can get there (ive tested almost all of them). lumen numbers from datasheets and simulators demonstrate this as well (except for the broken bridgelux simulator that returns over 300 lm/W at low currents)

View attachment 4012166
the lowest datapoint on that chart is 100mA/~3W. you could run cobs at 30 mA/~1W and still only be 50% more eficient than using 1/50th of the number of cobs at 50W each

The problem with your position is that you are comparing an improvement from one aspect as having the same significance in another aspect. Going from 50-60% efficiency for example can be described as both a 10% or 20% improvement.

Rather than use the spreadsheets and what not from others, I determine efficiency for lighting sources based on the real heat values which can be measured fairly accurately. The electrical energy used is going to either light or heat. I know the electrical coming from the wall, and I can determine the amount of heat generated for the value going to heat leaving the remainder of the electrical power being applied for light.

In regards to your comment about the broken Bridgelux simulator, until you can prove it is indeed broken you probably shouldn't be stating such. For one it isn't broken. Bridgelux has far too much on the line to be putting out misinformation. Two, you are a vendor selling a competing product so it just makes you look bad and no different than the other light peddlers.

Anyhow, in summary more cobs/light sources, less heat sink.
 

DIY-HP-LED

Well-Known Member
The problem with your position is that you are comparing an improvement from one aspect as having the same significance in another aspect. Going from 50-60% efficiency for example can be described as both a 10% or 20% improvement.

Rather than use the spreadsheets and what not from others, I determine efficiency for lighting sources based on the real heat values which can be measured fairly accurately. The electrical energy used is going to either light or heat. I know the electrical coming from the wall, and I can determine the amount of heat generated for the value going to heat leaving the remainder of the electrical power being applied for light.

In regards to your comment about the broken Bridgelux simulator, until you can prove it is indeed broken you probably shouldn't be stating such. For one it isn't broken. Bridgelux has far too much on the line to be putting out misinformation. Two, you are a vendor selling a competing product so it just makes you look bad and no different than the other light peddlers.

Anyhow, in summary more cobs/light sources, less heat sink.
I'm curious, do you have an Engineering degree? Work in the industry? Have some source of inside information?
I will remind you of a scientific dictum "extraordinary claims require extraordinary evidence" 300lm/watt is extraordinary, it approaches the theoretical limit of white light efficacy. I'd like to see more evidence before ad hominem attacks on someone's professional integrity.
 
Last edited:

Photon Flinger

Well-Known Member
I'm curious, do you have an Engineering degree? Work in the industry? Have some source of inside information?

Somewhat. Let's just say we have been doing research in this field up north here a lot longer than our poor southern neighbors. Plus our green rush is underway so a lot of folks who have been very quiet are coming out of the woodwork so to speak.

(and no, I am not hostile towards Cobkits in last post, it is just a bun fight over how many cobs is enough - the answer is obviously there is never enough)
 

DIY-HP-LED

Well-Known Member
In Canada, such research must be expensive indeed, with the cost of COBs and an integrating sphere and all.
 

CobKits

Well-Known Member
The problem with your position is that you are comparing an improvement from one aspect as having the same significance in another aspect. Going from 50-60% efficiency for example can be described as both a 10% or 20% improvement.
theres no trickery when you say you are replacing a 400W typical cob righ with 214W. thats (about) twice as efficient which is impossible with current tech. plants dont care if your chip is thermally more efficient, they care about photons. From a grower perspective we use the definition of efficacy which is lumens(or photons) per watt
Two, you are a vendor selling a competing product so it just makes you look bad and no different than the other light peddlers.

Anyhow, in summary more cobs/light sources, less heat sink.
of energy consumed. if thermal resistance was the criteria vero C would have higher efficacy than every other chip out there, but in reality there half a dozen chips with higher efficacy and lower Rj-c.

Rather than use the spreadsheets and what not from others, I determine efficiency for lighting sources based on the real heat values which can be measured fairly accurately.
there are standards for measuring light. there are standards or measuring heat as well (bomb calorimeter). ill entertain your argument if you detail your method.

In regards to your comment about the broken Bridgelux simulator, until you can prove it is indeed broken you probably shouldn't be stating such. For one it isn't broken.
ok heres your proof. over 100% efficiency and doesnt match datasheet:

upload_2017-9-17_20-1-33.png

Two, you are a vendor selling a competing product so it just makes you look bad and no different than the other light peddlers.

Anyhow, in summary more cobs/light sources, less heat sink.
veros are fine chips i got nothing against them, they are a good value and in the same class as citi and luminus. there is no magic bullet with this tech
 
Last edited:

DIY-HP-LED

Well-Known Member
Efficacy and Efficiency
This guy is a live wire and one of this nerds favorite places to visit.
LED thermal management according to an engineer. For those who want to know the difference between efficacy and efficiency this video shows how to calculate efficiency from efficacy. He concludes with the rule of "Burnt thumb". It's a great introduction into an esoteric subject and very useful information if you're into DIY grow lights, have a boo, worth the while watching. Education and entertainment combined, best approach IMHO
 
Last edited:

Photon Flinger

Well-Known Member
theres no trickery when you say you are replacing a 400W typical cob righ with 214W. thats (about) twice as efficient which is impossible with current tech. plants dont care if your chip is thermally more efficient, they care about hotons. From a grower perspective we use the definition of efficacy which is lumens(or photons) per watt


of energy consumed. if thermal resistance was the criteria vero C would have higher efficacy than every other chip out there, but in reallity there half a dozen chips with higher efficacy and lower Rj-c.


there are standards for measuring light. there are standards or measuring heat as well (bomb calorimeter). ill entertain your argument if you detail your method.


ok heres your proof. over 100% efficiency and doesnt match datasheet:

View attachment 4012246



veros are fine chips i got nothing against them, they are a good value and in teh same class as citi and luminus. there is no magic bullet with this tech

How is that over 100% efficiency?

A 100% efficient light example would be 683lm/W using a monochromatic 555nm. That is the problem with using lumens for anything other than performance comparisons within the same set because of the SPD weighting. The digitization efforts to determine QER/LER are only a shot in the dark at best. That is why I go the route of measuring the heat produced and, of course, using a PAR meter to measure the photon count.

So, the spreadsheet is still correct.

I think you should just build one yourself and see how low you need to run the cob to where it doesn't need sinking. That is really what we are trying to find out - I am good with the Vero 29c at 20w-46w hanging off some aluminum sheet.

The actual problem I have now is with winter coming, the light source doesn't provide enough heat for the space. Rather than get some space heaters, I might just go hunt down some cheapo blurples.
 

CobKits

Well-Known Member
(and no, I am not hostile towards Cobkits in last post, it is just a bun fight over how many cobs is enough - the answer is obviously there is never enough)
no offense taken, good discussion

i disagree with your premise tho

there is a such thing as diminishing returns and time value of money.lets say a big rig with 100 cobs at 75W. $10/cob

i can justify doubling the number of cobs to go from 75W/cob to 37W/cob to gain 18% in efficacy (+$750)

i can maybe justify going with 4x as many cobs to go from 75W/cob to 19W/cob to gain almost 30% in efficacy (+$2250)

doubing that already high cost to go 8x as many chips only gets you to 39% more efficient (+$5250)

lets double that again. 16x as many cobs and we are a full 53% more efficient than our original rig at 75W for a cost of +$11250

heres an ananysis of that **which doesnt take into account space limitations/time of drilling/tapping/mounting, or heatsink costs

upload_2017-9-17_20-28-11.png

starting with a baseline of 1.4 GPW youd see that the resulting product value (at $1500/unit) covers the cost of teh cobs in every case

*but* since wattage is static you would need to grow 18% more area in the first case, 30% in second case, and your garden square footage would be over 50% larger in the 3rd case. for the first step from 75 to 38W, we are investing $750 more in chips and getting a 8X return on that

for the largest array at 4.75W, we are investing $11250 and getting 1.63x of that amount back at harvest

most people dont have unlimited space to grow.

at this point youre like "aha cobby, i got you! you just demonstrated that you can keep adding cobs and they pay for themselves in just one grow!!!! Sure... but is that the best solution?

lets go another route. lets just increase the size of our garden by using one of the combinations above by running chips harder to cover more area. At 75W per chip i cant really go 153% of this as im out of the chip spec. let look at 30%. thats running chips at 97.5W

addl chip cost = 0
addl driver cost = $562 @ 0.25/W
chip efficiency at 97.5W = 12% less than 75W = 1.23 gpw
total watts =9750W (vs 7500)

metrics look like this:
upload_2017-9-17_20-50-3.png

yes that is less than the supplementary yields above BUT look at your ROI. spending just $562 yields you almost 9X the return. electric bill will pay you back as well, but this also takes time . so many factors that make you consider the time value of money:

-will you produce the crops you plan to or fall short?
-will the crops be worth the same value in the future as they are now?
what else can go wrong? you get sick and cant work? you lose crops to mold or bugs? youre in the wrong area and get popped?

laying out money now for future gains is ALWAYS a risk. and if youre not the average RIU users bangin 10's with your giant dick upon the stacks of money you have laying arount, it usually costs you money down the line to cough up capital now.

long story short, there are advantages to being more efficient but the LAW of diminishing returns assures that the moe you spend/more risk you take, the more incremental the upside becomes.

i like to think of it this way. somebody asked someone:

I'm curious, do you have an Engineering degree?
i might.... and what i might have learned is when you see a graph of yield vs cost and it has an elbow, the dogleg is where you want to be, because once it goes asymptotic, that is the very definition of diminishing returns. and thats industry agnostic, its fundamental. here is a chemical reaction (conversion vs time) that im all to familiar with from dayjobbin'

upload_2017-9-17_21-5-42.png

that 80-90% conversion you can get in 30 min is often more valuable than the 85-95 at 4x the time - you are losing throughput. "working the dogleg" is fundamental to cost analyses - its why we make those graphs in the first place. while its not as pronounced, our graph has a dogleg too. if i plotted cost vs efficacy it would be even more pronounced

upload_2017-9-17_21-9-20.png
 

CobKits

Well-Known Member
How is that over 100% efficiency?

A 100% efficient light example would be 683lm/W using a monochromatic 555nm.
its over 100% on the QER for 3000K 80 which is different than the lumen-centric wavelength you cite

The digitization efforts to determine QER/LER are only a shot in the dark at best.
they are actually pretty well understood from what i know. have at it.

That is why I go the route of measuring the heat produced and, of course, using a PAR meter to measure the photon count.
thats fine, lets see your method. would love to evaluate

So, the spreadsheet is still correct.
which spreadsheet are you referring to?

I think you should just build one yourself and see how low you need to run the cob to where it doesn't need sinking. That is really what we are trying to find out - I am good with the Vero 29c at 20w-46w hanging off some aluminum sheet.
what is your metric? what is "good with"? what level of thermal droop is acceptable and how are you measuring it? veros have a nice footprint, i could see it dissipating 20W toa cookie sheet, 46W seems like a stretch, would love to see some data on thermal droop based on par/lumen per watt or Tc/Tj measured/extrapolated by voltage measurements.

The actual problem I have now is with winter coming, the light source doesn't provide enough heat for the space. Rather than get some space heaters, I might just go hunt down some cheapo blurples.
my problem as well but i think im going with the DE hps i have laying around. efficiency on par with the cheap blurples but it directly increases leaf temp more, to teh point i can mix it with efficient cobs for an optimum efficacy/leaf temp combo
 

Photon Flinger

Well-Known Member
In Canada, such research must be expensive indeed, with the cost of COBs and an integrating sphere and all.

Not really, most universities have them.

The lighting is actually the easy part. Biology side is where the work is at and where we are very far ahead. The problem is much of that research is private so the results are being kept secret for commercial purposes.

McGill, Guelph, U of Man, the eastern universities all have been doing sponsored research for years. Cannabis and hemp have some wonderful properties that are beneficial for more than what we commonly think.
 

Photon Flinger

Well-Known Member
My suggestion is to determine efficiency, or more appropriately an accurate approximation, based on the heat generated.

The reason for the suggestion is to get back to basics - in this case why do we need a heat sink at all and to see if it is possible to eliminate it.

Light good. Heat bad. If we use more electricity to produce more light we unfortunately produce more heat. There are two increases that impact efficiency. The first is that at higher current, more heat is produced per watt than at lower currents. Then as we increase the power by x watts, the previous value based on current is the factor x is multiplied by to figure out how much heat needs to be managed. Maybe consider two cobs

Cob1 - 1710ma 0.04c/W @ 118.7w so it requires 4.75c of heat differential to be managed
Cob2 - 3420ma 0.05c/w @ 254.6w works out to a 12.75c of heat differential to be managed

While the increase in power is 2.14 times, the heat differential is 2.6 times. It isn't linear growth and I am not going to break my head using calculus to figure it out exactly. It just tells me that if I go the other way in lowering current, more of the electricity as a percentage is going to light rather than heat.

Cob3 - 855ma >0.04c/W (unknown but it would be lower than the 1710ma) @ 56.6w resulting in a 2.26c differential. That is less than half of the Cob1.

2.26c is actually still pretty high. From my testing, 1/8th Al sheet 3"x3" can handle 2.0c differential stabilizing at 32c in 25c ambient with good airflow. It will go lower if put on a bigger sized sheet. At 1.0c, the increase over ambient using the same 1/8th Al 3x3 is only 1-2c which is why I am suggesting to dump the bulky heat sinks entirely.

More measurements need to be done but I think that shows the path I am going down.
 

DIY-HP-LED

Well-Known Member
Not really, most universities have them.

The lighting is actually the easy part. Biology side is where the work is at and where we are very far ahead. The problem is much of that research is private so the results are being kept secret for commercial purposes.

McGill, Guelph, U of Man, the eastern universities all have been doing sponsored research for years. Cannabis and hemp have some wonderful properties that are beneficial for more than what we commonly think.
I think the mere mortals eyes are glazing over on this back and forth, Cobby appears to be pragmatic and knows this stuff. He's doing a Hell of a job as a business man in addressing customer concerns and leading the industry. I'm trying to talk him into moving here;)
 
Top