High End COBs vs $2 COBs...

HideousPenguinBoy

Well-Known Member
I'm absolutely fine with moving to new tech, but I'm trying to find efficiency numbers for all these cheaper COBs because I pay almost $.30/kWhr. Paying an extra couple bucks to save $$$ over the course of a year is absolutely worth it. Is there a chart comparison for these? Currently I'm running CXB3590s at 1050mA and dropping down to 700mA shortly.
 

muleface

Well-Known Member
I'm absolutely fine with moving to new tech, but I'm trying to find efficiency numbers for all these cheaper COBs because I pay almost $.30/kWhr. Paying an extra couple bucks to save $$$ over the course of a year is absolutely worth it. Is there a chart comparison for these? Currently I'm running CXB3590s at 1050mA and dropping down to 700mA shortly.
holy crap! .30/kwh! I average .10 a kwh.

In the winter it gets down to .06
 

CobKits

Well-Known Member
I'm absolutely fine with moving to new tech, but I'm trying to find efficiency numbers for all these cheaper COBs because I pay almost $.30/kWhr. Paying an extra couple bucks to save $$$ over the course of a year is absolutely worth it. Is there a chart comparison for these? Currently I'm running CXB3590s at 1050mA and dropping down to 700mA shortly.
the math is easy. a 1000W DE pulls 1100+ at the wall. a rig with cree at 50-55W or luminus/veroC/clu1825 at 70-75 W should replace a DE with roughly 700 cob watts or 750 wall watts

in this case your savings is 350W, or 0.35KWh (per hour)

at 1212 there are 4380 hours in a year

annual savings = annual hours x kwh savings x rate

4380 x 0.35 x 0.3 = $460

thats straight electrical cost of lights alone and does not consider the cooling you wont need for the 350 less heat watts you are producing, which could bump that savings by 10% or more. also the bubl and reflector changes which are at least $30-50 year usng the cheapest bulbs you can find and stretching out over a year

so lets look of the cost of that cree rig to do the above

say you want to cut your cob wattage in half to 700 mA. lets say a 700W (1400 mA) cree rig is somewhere around
$35 x 14 chips= $490
700W in CC drivers = $200
heatsinks and holders etc (if you go pin fin) = $ 260
________________________________
$950

assuming a bargain gavita/epap/whatever and a couple bulb changes over a few years is $450

you paid $500 more to save $460/yr in electricity- seems like a fair investment that will keep on returning year over year

the trick is calculating diminishing returns.

using the "cut current in half and bump efficacy by 10%" rule of thumb, the same cree rig (in PAR/lumens) at 700 mA would look something like say 600 cob watts/675 at wall:

$35 x 24 chips= $840
700W in CC drivers = $200 (same- but with CC drivers you dont have the clearest upgrade path unless you run strings in parallel which has risks with a non-voltage limitable high voltage CC driver)
heatsinks and holders etc (if you go pin fin) = $300 (downsizing heatsink size to small pinfins or star extrusions)
________________________________
$1340

in this case youre spending almost $400 more to save another 0.075 KWh

4380 x 0.075 x 0.3 = $99/yr -> 4 year payback - not as great as the case above
 

CobKits

Well-Known Member
two of the big cost of that second rig are the chips and the heatsinks- so how do we save? we can also shave our driver cost if were frugal in selection

1. chips. Crees are workhorses but they are expensive. If you look at real world plots of chips run at low currents/wattages (say 40W or less) -they really all do converge to within a few percent of each other. so you can either use a chip that performs the same in that range and costs less (luminus, $17), or a chip that underperforms by 1-2% and either deal with it or use a few more to make up the slack ($10 citi gen 5)

2. drivers. im an advocate of low voltage cosntant voltage high current drivers for two reasons: firstly they are expandable and you can easily start with 4 cobs on an HLG185 driver and add 2 or 4 or 6 more and get an efficiency bump each time. Secondly, by way of adjustable current and voltage, if sized correctly you can easily jam 225W out of one, where with a CC driver if youre getting 200 output watts you filled it well. that extra 10% is still operating at the 93+% efficiency level, but you reduces your driver cost/watt by 10%

3. heatsinks. the lower you go with current per chip the less heatsinks you need. in the case of properly sized linear extrusions. If youre running 4 chips per bar at 50W each or 8 chips per bar at 25W each the bar doesnt care, the heat load and cost per watt are the same and these bars make a lot of sense for low current/low cooling loads. at ultra low currents,super cheap extrusions or straight bar stoick is acceptable. in the case of the PCB boards were seeing some people run the PCB bare when its majorly underdriven
 

CobKits

Well-Known Member
more points for penny pinchers:
-larger drivers are more efficient than smaller drivers. HLG 480 and 600 can get up to 96% efficiency at 220V input
-in almost all cases you will see a 1-2% bump in all meanwells when going from 110V to 220V input, even with the little guys
-while 220V may not be as 'standard' as 110V (in the US at least)- the extra cost of running a few circuits gives you a quick return on investment and also reduces wire size significantly, which can add up in bigger builds. At my 'real job' I have a lot of motors- they are all 460V and anything up to 10HP (7500W) is covered easily with simple common 12GA wire. cheap, easy to pull, smaller conduit, etc etc.)
 

nfhiggs

Well-Known Member
I'm absolutely fine with moving to new tech, but I'm trying to find efficiency numbers for all these cheaper COBs because I pay almost $.30/kWhr. Paying an extra couple bucks to save $$$ over the course of a year is absolutely worth it. Is there a chart comparison for these? Currently I'm running CXB3590s at 1050mA and dropping down to 700mA shortly.
Ouch! Yeah at that price I would look for the most efficient one I could find.
 

nfhiggs

Well-Known Member
::cries uncontrollably:: City living. I can eat 30 different cultures' foods at midnight, but rent/electricity I pay out the butt.
I recall someone a while back saying their "base" allowance was only 400 kW/h per month, and the price tripled over that. I couldn't even run one of my computers on that!
 

muleface

Well-Known Member
@CobKits

hey, i got some 2828 getian cobs. They sent me some holders for them, but they are really weird. They clip from the bottom, so the cob doesn't actually make contact with the heat sink, do you know of any holders that would work with these?
 

CobKits

Well-Known Member
@CobKits

hey, i got some 2828 getian cobs. They sent me some holders for them, but they are really weird. They clip from the bottom, so the cob doesn't actually make contact with the heat sink, do you know of any holders that would work with these?
same bjbs and ideals that the clu048 citis use
 

frica

Well-Known Member
The perceived shortcomings of the "trash cobs" are greatly exaggerated:

1. Yes, they are less efficient - 100-110 L/W or so at full power compared to 150 L/W or so for the "good" cobs. But if you run them at 30-50% you can get into the 120-130 L/W range - not fantastic but comparable to MH/HPS efficiency. So the whole "twice the watts and heat" is just an exaggeration. They will run fine on "regular" sized heat sinks.

2. The supposed reliability issues are also greatly exaggerated thanks to one youtube video. I've purchased 20 of these and have not had one issue with any of them - NONE. They work. They have been working continuously for 6 months now.

3. They come in warm white, they are not all "blurple".

Cheap cobs are a good way to get your feet wet, and experiment a bit to find out what works and what doesn't, instead of simply following the herd. The great thing about the 2 dollar cost is you can easily replace them later with more efficient cobs and not feel like you're throwing away a ton of money.
The problem with 100-110 lm/w is that it means they're quite clearly outperformed by LED bulbs.

I can get philips 13W bulbs for under 5 euros here, they're 115 lm/w with the globe on. 130+ with the globe removed.
No need to get a heatsink, no need to get a driver.
Just a ceiling socket and a piece of wood to screw it in.

Also the system cost of a China cob isn't even that much cheaper because unless you insist on buying Cree the COBs are a minor part of the whole price.
 

CobKits

Well-Known Member
you forgot to mention the bulbs are safer, they are ul/ce listed with normal sockets for normal people. you can build out a whole 120-130 lm/W array at $0.30 USD/W, and make it as small or as big as you want. every time i go onto the big box stores these are better and cheaper than the last time i was there

last year they were $7 for 100 lm/W (with globe on)
now they are $5 for 110-115 lm/W

and if they break (ive never seen a bad one) they are warranteed and a local return in many cases

while these are generally cost-driven by federal efficiency standards, at some point there is no point for the diode manufacturers to turn out low-flux chips when its probably cheaper to turn out high flux chips and use one less per board. ill bet they dont bin that much lower than the top shelf mid powers out there
 
Top