cheap chinese COB - 2017

Nope, not such a big difference.
I'm not sure you reach 1gpw with hard driven generic COB's but 0,8gpw sure.
With 20 of this next generation 70w COB's mentioned by the OP above(44$) and running them soft at 600-700mA per COB you could build a 450w light, which should be able to reach the 1gpw or more zone. Maybe a 600w hps replacement..

DaveInCave said:

I would say these are very important advantages, since we are not producing light for lighting's sake, we are trying to produce lights for plants to use.
If the spectrum is better, wouldn't that compensate for lower efficiency?


I'm a huge fan of such projects and experiments
I'm definitely not going to base my commercial grow operations on Chinese cobs
Like you, I also like to try new things, maybe I will give it a go and report back if I find something interesting.

It would be nice to able to build a light cheaply, that will also be efficient and produce good yields.
I think if I can make a 800W light for $100, and get 1<g/w yield I will consider it a win, even if it's slightly less efficient than HPS, and brand names LEDs.

Click to expand...

drivers alone will put you over $100 for 800W

caretak3r said:

drivers alone will put you over $100 for 800W

Click to expand...

Yeah, that's why I thought to try the ones with the on-board driver, of course like you mentioned they have their own problems which might not make them very cost-effective.

I was thinking to buy 6 of those 150w cobs for $66, heatsinks and fans for another 35, so 0.8 g/w might even be good enough, it will still yield more than a 450W brand name led with 1.2 gpw (assuming the Chinese ones don't fry within a week), for $100 it's not so bad.
I think I'll try it as some point when I can afford to play a little bit.

I would love it if the parameter we used for comparing lights would be gram (of yield) per dollar (which will include initial costs and running costs of a given light.)
It will give such a better comparison than gram per watt, lumens per watt, PAR etc.
Sounds like one hell of an excel sheet

DaveInCave said:

I would love it if the parameter we used for comparing lights would be gram (of yield) per dollar (which will include initial costs and running costs of a given light.)
It will give such a better comparison than gram per watt, lumens per watt, PAR etc.
Sounds like one hell of an excel sheet

Click to expand...

never going to happen - too many variables (grow abilities, genetics, etc)

Btw, I have made some CAD of water-cooled aluminum fixtures for DIY lights.
At some point I'm planning to get a CNC and make some prototypes, maybe I will share the process here.

caretak3r said:

never going to happen - too many variables (grow abilities, genetics, etc)

Click to expand...

One can dream

DaveInCave said:

Yeah, that's why I thought to try the ones with the on-board driver, of course like you mentioned they have their own problems which might not make them very cost-effective.

I was thinking to buy 6 of those 150w cobs for $66, heatsinks and fans for another 35, so 0.8 g/w might even be good enough, it will still yield more than a 450W brand name led with 1.2 gpw (assuming the Chinese ones don't fry within a week), for $100 it's not so bad.
I think I'll try it as some point when I can afford to play a little bit.

I would love it if the parameter we used for comparing lights would be gram (of yield) per dollar (which will include initial costs and running costs of a given light.)
It will give such a better comparison than gram per watt, lumens per watt, PAR etc.
Sounds like one hell of an excel sheet

Click to expand...

In most cases higher upfront costs will pay back for themself.
The more efficient a light works the more it yields...
Upfront costs would only matter if you use the light only one times. But the longer you use it the more you get back.
800w with 100$ upfront cost = 800g each 3 month's.
800w with 1200$ upfront costs are 1200g each 3 month's.
With 400g more yield each 3 month's, how long you need to get back the 700$ more in upfront costs?
And what makes you sure that these cheap diodes will not fail also within a few weeks, there are more or less the same crappy led's.
The only reason for me to build such a light would be if I have not have enough money to build the one with best efficacy.
In this case I would build a small light to make the money for a really good one.

Randomblame said:

In most cases higher upfront costs will pay back for themself.
The more efficient a light works the more it yields...
Upfront costs would only matter if you use the light only one times. But the longer you use it the more you get back.
800w with 100$ upfront cost = 800g each 3 month's.
800w with 1200$ upfront costs are 1200g each 3 month's.
With 400g more yield each 3 month's, how long you need to get back the 700$ more in upfront costs?
And what makes you sure that these cheap diodes will not fail also within a few weeks, there are more or less the same crappy led's.
The only reason for me to build such a light would be if I have not have enough money to build the one with best efficacy.
In this case I would build a small light to make the money for a really good one.

Click to expand...

You bring up excellent points! I like your thinking and I completely agree.
That's why I only consider this as a little side project and not as a basis for an actual grow.

I did some more testing last night. Again, i observed that @270ma these no-names were performing equivalent to my 4000K Vero18 gen6 (both no-name samples were tested this time) according to my cheap lux meter. I also tested with a 700ma driver (previous test was with 270ma). This may be a bad assumption, but we're going to assume the drivers are roughly equivalent in efficiency because i didn't get multimeter readings to get true power numbers. @270ma the, pair was pulling 17W from the wall, @~700ma, they were pulling 37W from the wall on the killawatt. So, at the wall, we were pulling 2.17X the power. The lumen output @700ma was only 1.6X of the readings I got @270ma. I did not sub in the Vero18 so I'm not sure how the chips compare at this amperage yet, but I suspect they perform MUCH better in comparison at the higher amperage (which should be expected really).

Next up, I did some unscientific readings/comparisons with my QB board, and my theory is that it would take 8+ no-names running at 270ma to match the lumen output of the QB @500ma. Please take this with a huge grain of salt, but it's kind of the benchmark I'm using --- can I run lots of no-names at super low amperage and come any where close to a QB..... When my next batch come in, I'll probably build a panel of 8 and see how they do in a true comparison. I'm not claiming to match efficiency, just to see how close I could get.... currently, in a best case scenario, I think I could only match efficiency of the QB when it's running at max current.

Lastly, I asked for pricing in larger lots... the best price I got was:
$ / qty
$1.65 / 100
$1.55 / 500
$1.45 / 1000

This pricing came from the second factory I ordered from, I have not tested their samples yet but I suspect it to be the exact same product.

I'm not advocating anyone jump on these, as I have no idea about longevity.... it's a fun exercise for me, so I'm just sharing.

This sounds pretty interesting but I think a few other things also matters.

How is the light quality in comparision to a good CRI80/3000k COB like a Vero18? Most of the times the generic warmwhites looks more greenish-yellow than warmwhite, so more CRI60-70 as 80.
Are in cool- and day-white the same differences, what do you think?

I know, plants don't really care the difference and photons counts and one could add monos to improve that..
I'm curious to see how your a low current mulit chip unit would perform against other products.

Randomblame said:

This sounds pretty interesting but I think a few other things also matters.

How is the light quality in comparision to a good CRI80/3000k COB like a Vero18? Most of the times the generic warmwhites looks more greenish-yellow than warmwhite, so more CRI60-70 as 80.
Are in cool- and day-white the same differences, what do you think?

I know, plants don't really care the difference and photons counts and one could add monos to improve that..
I'm curious to see how your a low current mulit chip unit would perform against other products.

Click to expand...

yep, totally valid points. They CLAIM 85+ CRI, but who knows. I'm testing the warm-white version and they don't look green to me, but I'll need to take some comparison shots. I have some 4K and 3.5K Veros I could shoot against.

worse case...........decent camera lighting

For the price they sound like gold

caretak3r said:

yep, totally valid points. They CLAIM 85+ CRI, but who knows. I'm testing the warm-white version and they don't look green to me, but I'll need to take some comparison shots. I have some 4K and 3.5K Veros I could shoot against.

Click to expand...

Yeah, maybe the phosphor layer also improved in quality.
Also it looks like they use their new thermoplastics instead of a aluminum as substrat on their 220/110v COB's, is it also used on this series?
Thermo-plastic or ceramic would be interested to know.
I don't think thermoplastics are as efficient as aluminum or copper in terms of heat conduction but I could be easily wrong with my assumption. After all, Cree has been using ceramic for a long time.

Maybe I will try to get some of this 70w/3.000k COB's from another seller who did shipping to europe.
With a generic 10 or 20w driver one could use them at least in the kitchen to make it ass kickin' bright in there... LOL!
Perhaps for a herbal shelf od so..

for those that like to experiment, I rediscovered Getian cobs (Cobby has tested them and he had good things to say about them). I got their sample price sheet and it looks pretty good.

caretak3r said:

for those that like to experiment, I rediscovered Getian cobs (Cobby has tested them and he had good things to say about them). I got their sample price sheet and it looks pretty good.

Click to expand...

Now you have my full attention.
They seems good for a "low current" build with many COB's and ⅛" alu-sheets.
I was thinking of a 12 COB's at 350mA configuration with HLG-185H-C700 (6S2P) with Citi 1212's / 3k / CRI90.
With this getians I could halve the chip costs with little loss of actual yield.
Do you have it from Aliexpress?
I'll find them, unless you have a link to your hand ...
They look like they're quite comparable to the quality brands and I think I've seen a couple of Cobby's Speadsheets where they gave equivalent results.

BTW,
I love experimenting...

Randomblame said:

Now you have my full attention.
They seems good for a "low current" build with many COB's and ⅛" alu-sheets.
I was thinking of a 12 COB's at 350mA configuration with HLG-185H-C700 (6S2P) with Citi 1212's / 3k / CRI90.
With this getians I could halve the chip costs with little loss of actual yield.
Do you have it from Aliexpress?
I'll find them, unless you have a link to your hand ...
They look like they're quite comparable to the quality brands and I think I've seen a couple of Cobby's Speadsheets where they gave equivalent results.

BTW,
I love experimenting...

Click to expand...

alibaba, talk to Frank
https://getiangroup.en.alibaba.com

those are sample prices, not including shipping/etc... Not sure what price you could get in quantity.

and, I agree - I think this is likely a quality chip comparable to the other big names we know and love.

Nice find..if the spec are true..basically..with 100 watts 3 amps driving 10 units gt2828 cobs you can reach almost 190-200 lumens per watt at 4000k...cost about 55 dollars / 50 euros.... without drivers etc...holders included....shipping cost should be a simple envelope....maybe 10 dollars....What a wonderful times..my first interior was with a 100 watts incandescent in 1988....too much streching

pulled out the cheap cobs to compare them to my qb288 clone. It appears the cobs are a little warmer than the samsungs (I assume these are probably lm561B or Bplus @ 3000K). I compared lumen output with cheap meter and at 50W draw (measured at the wall), they actually measured about identical @ 12". 6 X $2 Cobs versus 1 X QB288 clone. None of this means a whole lot - I'm sure they both can grow weed

sorry if i missed it ..are the cheap cobs you are using ..driver on board?