Driverless AC COB LED build

Aolelon

Well-Known Member
Are we looking at the same graph? What do the abbreviations on the side and bottom mean?

You are correct, I don't have data yet. I have been asking you how to obtain data to gauge the chip performance that will correlate to the tested chips. Is there a test that I can perform that will quantitatively assess the performance, as to compare it with "HQ" chips, without utilizing a research lab?



Please reinforce these statements with citations or links, no offense, but there should be justification to these statistics.



I based my assumptions upon the lowest performing chip in your graph, as stated in the response. Without data, nor an adequate method of testing, we can only base observations on the information available.

Regardless, I'll post the development of the array as it progresses. I'll provide pictures of the components and the creation process ideologies, unbiased. Any data and tests I perform will be provided, and an analysis of the data per my perspective will be performed. The raw data will be available for individual analysis and consideration. As this project unfolds updates will follow.

Thank you CobKit's for your contribution to my endeavor. Your interest and ideas are appreciated. I look forward future debates that fuel creative innovations.

- ZXC
I highly doubt your chips will be on par with the lowest chip in that graph. I'd be surprised if they were.
 

Randomblame

Well-Known Member
Here is a teardown, mate!


And another one!

You get what you pay for!
I would take twice as much chips from the beginning 'cause you'll need to change broken chips anyway sooner or later.
I bet sooner...!
Listen to what cobkits says, he is one of the most reputable persons here and his intension is to help you.

Elec. conversion is ~80% efficient(at least for the 20w chips) and assuming the cheap 1 or 0,5w diodes are 40% efficient(probably it's even less) the 20w chips are 32% efficient. This means at least 68% heat! With cheap Citi1212 you can already get 50-55% system effiency so at least 23-28% less heat.
10 vs. 2$ per chip is a lot but over time it will take more money, effort and time to repair the system. After a few months at the latest, the first chips will die. I had a 20w floodlight that did not even run for 2 months. Not like these driverless, but same quality level china COB and parts inside.
At least you use brake-away connectors which makes interchanging chips a lot faster...
 
Last edited:

CobKits

Well-Known Member
Are we looking at the same graph? What do the abbreviations on the side and bottom mean?
x axis =power = DC input power to cobs
y axis =PPFD/W = the PPFD measured at a given wattage divided by that wattage to give a relative measure of efficiency

maybe "unitless" isnt the right word. In any case, that sphere isnt calibrated to a known NIST standard so while the measurements are all valid, we dont really know exactly where they fall on a lumen/W or gross PPF basis. But
From the graph regarding the 50W chip efficiency range there was a variation of approximately 39% - 43%.
is certainly not the case- those arent efficiencies. The same chips measured in free space without a sphere read around 10 PPFD/W - its only relative to other measurements in that particular test. To complicate matters, even making measurements on the same setup on different days yield different results, those CREE measurements are exactly the same chip measured on two days with different ambient temperatures. So youre on the right path in regard to controlling chip temp, but at the end of the day you will be limited by the chip you choose.

You are correct, I don't have data yet. I have been asking you how to obtain data to gauge the chip performance that will correlate to the tested chips. Is there a test that I can perform that will quantitatively assess the performance, as to compare it with "HQ" chips, without utilizing a research lab?
yes, see post 15 above. Again, without any lumen maintenance data from the mfr you are measuring a new chip and hoping its output stays relatively constant over its lifetime, which is a big assumption

chip testing is expensive (not just photometics but the lumen maintenance data in particular), do you think the makers of these $2 cobs do any of that? and if so where are their test reports? its analogous to the chinese fixtures which cant offer real specs on their products when pressed.
 
Last edited:

CobKits

Well-Known Member
Please reinforce these statements with citations or links, no offense, but there should be justification to these statistics.
just search this forum there are several threads about these exact same chips and also plenty of threads of people with the experiences of the cheap chinese lights. the conclusion is always the same - you get what you pay for

I based my assumptions upon the lowest performing chip in your graph, as stated in the response. Without data, nor an adequate method of testing, we can only base observations on the information available.
exactly. i wouldnt assume those driverless chips are anywhere near that performance, until demonstrated otherwise.. at this point there is no data so you cant rationalize cost or anything else. I think thats a common trap that people fall into, assuming all LEDs are the same when in reality youd be really hard pressed to find a lighting technology with a wider range of performance (most HIDs, flourescents, etc perform relatively similar - the best LEDs can be more than twice as efficient as the worst ones on the market)

Measuring AC input power accuately will be another challenge in quantifying their efficiency. The watts up AC meters are not all that accurate, esp at the low power levels you need to measure chip efficiency. I'm not aware of any affordable benchtop AC power supplies that would let you do that kind of testing at different current levels. They exist but they are in the thousands of dollars. (as opposed to regulated DC power supplies that can be had for as little as $100)


if they are the same chips as randomblame posted, just off the bat even in the case that the individual diodes were as efficient as the ones in major manufacturer chips, the "driverless" chip is using less than half of the diodes than any other chip in that test, so it will be at a serious disadvantage. running that chip at 50W would be like running the lowest perforning chips in the graph above at over 150W (possible but not recommended). If you want to get in the same efficiency ballpark as HQ chips with those you are going to need to run them at ~10-15W each, and again thats no guarantee they even get to the same level.

as for the drivers a good 200W meanwell driver weighs several pounds. There are a lot of other components that go into making a driver operate safely and efficiently with good longevity, much more than just making AC into DC from a 10 cent circuit.

Something not to forget is the value of reliability. anybody whos had crops suffer due to cheap pumps, timers, gfis, etc, know that it can be a horror show when things go wrong. Which is one reason why people are constantly seeking the easiest most reliable solutions, as time is indeed money. Not trying to dissuade research but starting with chips of dubious quality and then adding water cooling system is adding several layers of complexity over simple air cooled cobs/strips/boards that would probably run reliably for 20 years with no moving parts or maintenance if left unattended. There was a time not so long ago that the best cobs were so expensive that we needed to make efforts to squeeze everythign we could out of them. @ttystikk did his water cooling on CXB3590s years ago when those chips were over $50 each. Now that we can get that same level of performance for 25% of that chip cost most people would just err on the side of more chips (or strips or boards) run softer - which is an easy route to better efficiency as well as reducing load per diode and increasing lifespan. and simple. and redundant as with many more points of light if one fails it may be unnoticable to the plants

if you look thru the forum you will find more LED people having problems with cold temps than high temps. And i understand the concept of using waste heat but anybody air cooling their rig in a building that still requires heat is doing this already. I too had some 2x1 aluminum extrusions welded into waterblocks way back then.... never used them as i never ended up having a need for them and wrote it off as a big overthinking exercise. At this point, its just too easy to get to the efficiency levels we need by reducing current per chip. Temp effects are real but reducing current also does this as well. strips and boards and cobs run soft can get darn close to ambient case temp. maybe not exactly ambient but close enough that you are seeing major diminishing returns when trying to cool using ambient temp solutions like recirc'd water. Trying to get below ambient is another ballgame as chillers are powerhogs and will quickly eradicate any savings on operating cost alone (apart from being complex and expensive capital cost items)
 
Last edited:

CobKits

Well-Known Member
Just get a high quality chip with well known efficacy parameters. Do a side by side comparison of the intensity (PAR) values measured at 24" at the same power levels. Compare the difference.
id go a little closer than that for better results. the design of that chip is quite a bit different than the typical COB. I always used 12" on axis in free space with good results. you need to do it in the dark and make sure the sensor is level, and also make sure the cobs are similar CRI and CCT. it might be difficult to find an actual spectrum measurement on the AC chip, and then even so there might be variance from chip to chip. Even with good cobs not all 80CRI chips are the same (thats a minimum, they will usually be 82 or 83 when measured)
 
Last edited:

wietefras

Well-Known Member
that dozens of people have come before you looking for the silver bullet of high performance for pennies on the dollar, and wasted lots of time and money in the process.
Please reinforce these statements with citations or links, no offense, but there should be justification to these statistics.
After over half a dozen people warned you that these COBs are no good, you still need "justification"?

The forum does have a search function ...
 

1212ham

Well-Known Member
Very nice! Beautiful array. Did you weld the square tubes to make the closed loop? Additionally, did you put thermal glue between the metal plates and the tubes while securing them with the zip ties? I like that configuration, it is similar to what I envisioned. The connecting tubes act as a manifold for the distribution of the cooling fluid, right? I considered using aluminum strips to increase the width of the plate, but I am thinking that the conjoined tubes will increase surface area and fluid turbulence allotting additional thermal conveyance. It will be an added expense, but the total heat capacitance will be increased too. How hot does your system run vs. flow rate? Thanks 1212ham, much appreciation!
Yes, welded.
Yes, dirt cheap Chinese thermal paste. It's quite sufficient with the low heat density of strips. Use the good stuff for cobs!
Yes, tubes on the sides are manifolds, eliminates a lot of plumbing. The hoses connect to opposite corners to balance flow among the tubes.
Very little flow needed, a tiny $6, 12v, 6 watt fountain pump was used.

I water cooled just to experiment and because I thought I had some better ideas to test. It turns out that my ideas were valid, but I don't need the extra heat removal of water cooling, the light is now being run passively and temps in the tent are perfect, around 29C/85F. That may change in the heat of summer.
 

ZxcStaz

Active Member
Hey guys! I have an update on the progress. I can't post too much during the week, but I'll try to carve out some time. My time constraints prohibit me from adequately addressing all of the replies and comments. Thank you, all who have posted for your time, thoughts and efforts, and I appreciate the insights and wisdom. I assure you, I dwell on each post and I know the effort that is invested. I have to thank CobKits and 1212ham especially, you guys are the best! That is said without diminishing any of the other contributions, and please forgive my lack of social graces and I'm sorry for any taken offense. Only a few can tolerate me as I fumble through conversations; at the Mensa I eat quietly.

Well, the chips arrived early and I worked on them for a few hours today. Preliminary testing holds promise. I tested the first three that I randomly picked to see if they would power up, and wow, those things are bright. They all powered up, and I could not discern any flicker. (Even if they did flicker, it would not affect the photosynthetic response.) It looked like all of the LED's were functional. I tested the full spectrum, 4000K and 7000K chips.
20180605_101547_resized.jpg 20180605_105807_resized.jpg 20180605_111107_resized.jpg

This was performed by attaching the leads to a standard 120VAC outlet using a stripped computer cord. I did not yet test the wattage draw. I figure that I'll do that when I have them attached to the cold plates. I threw the chips on a chunk of aluminum just to dissipate the heat generated as not to damage the chips.

20180605_105801_resized.jpg

One annoying aspect was the soldering of the wires to the chip. I decided upon 12AWG stranded wire. After about 5mm was stripped, the wire was coated with solder. This was then soldered onto the chip. I needed to use two soldering irons to make a good connection because the thickness of the wire and the aluminum chip carried the heat so effectively. Even though I had 60/40 rosin core lead based solder, I used flux to pre coat the wires. By the end of the day I only got through 14 chips.

20180605_190342_resized.jpg

I'll keep working on the chips and posting updates as I progress. So far it seems to be working out. I have a bunch of other projects going on. Here is a picture of one of my ladies in early bloom, I hope you don't mind the tangent...

20180604_174140_resized.jpg

Upon close inspection, you might notice the crab claws. This little lady is making some XX pollen for me. She should keep me from spending too much for fem seeds, and rejuvenate my OG Kush line. It's just an aside that I thought you guys would enjoy.
 
Last edited:

CobKits

Well-Known Member
for that current you could use 16 or 18 solid if you can find a breaker small enough to protect it

the IEC sockets that take the AGC fuses are handy as they make them in all sizes

agreed that stranded is a pain in the ass to solder, esp #12!
 

CobKits

Well-Known Member
for that current you could use 16 or 18 solid if you can find a breaker small enough to protect it

the IEC sockets that take the AGC fuses are handy as they make them in all sizes

agreed that stranded is a pain in the ass to solder, esp #12!
 

johnny lee

Active Member
Hey guys! I have an update on the progress. I can't post too much during the week, but I'll try to carve out some time. My time constraints prohibit me from adequately addressing all of the replies and comments. Thank you, all who have posted for your time, thoughts and efforts, and I appreciate the insights and wisdom. I assure you, I dwell on each post and I know the effort that is invested. I have to thank CobKits and 1212ham especially, you guys are the best! That is said without diminishing any of the other contributions, and please forgive my lack of social graces and I'm sorry for any taken offense. Only a few can tolerate me as I fumble through conversations; at the Mensa I eat quietly.

Well, the chips arrived early and I worked on them for a few hours today. Preliminary testing holds promise. I tested the first three that I randomly picked to see if they would power up, and wow, those things are bright. They all powered up, and I could not discern any flicker. (Even if they did flicker, it would not affect the photosynthetic response.) It looked like all of the LED's were functional. I tested the full spectrum, 4000K and 7000K chips.
View attachment 4146737 View attachment 4146739 View attachment 4146741

This was performed by attaching the leads to a standard 120VAC outlet using a stripped computer cord. I did not yet test the wattage draw. I figure that I'll do that when I have them attached to the cold plates. I threw the chips on a chunk of aluminum just to dissipate the heat generated as not to damage the chips.

View attachment 4146734

One annoying aspect was the soldering of the wires to the chip. I decided upon 12AWG stranded wire. After about 5mm was stripped, the wire was coated with solder. This was then soldered onto the chip. I needed to use two soldering irons to make a good connection because the thickness of the wire and the aluminum chip carried the heat so effectively. Even though I had 60/40 rosin core lead based solder, I used flux to pre coat the wires. By the end of the day I only got through 14 chips.

View attachment 4146744

I'll keep working on the chips and posting updates as I progress. So far it seems to be working out. I have a bunch of other projects going on. Here is a picture of one of my ladies in early bloom, I hope you don't mind the tangent...

View attachment 4146747

Upon close inspection, you might notice the crab claws. This little lady is making some XX pollen for me. She should keep me from spending too much for fem
 

1212ham

Well-Known Member
Sounds like a good way to cook a semiconductor. :shock:
Far better to use a bigger soldering iron (or gun),that will get the job done quickly, allowing less heat soak of LED.
 
Last edited:

1212ham

Well-Known Member
The cold-plate/heat-sinks will be self-contained on a standalone recirculating system powered by a 17L/min recirculating pump. The heat will be dissipated to waste in the summer and redistribute to habitation in the colder months. Utilization of a non-electrolytic or polar medium should circumvent corrosion at the expense of higher heat capacitance, yet afford infinite system recharges from cold city water in the summer months.
Can you elaborate? How will the heat be dissipated to waist? Do you intend to drain a percentage of the water coming from the heat sinks and replenish with cold tap water fed to the inlet of the heat sinks?
 
Last edited:

ZxcStaz

Active Member
Can you elaborate? How will the heat be dissipated to waist? Do you intend to drain a percentage of the water coming from the heat sinks and replenish with cold tap water fed to the inlet of the heat sinks?
Hey 1212ham, sorry for the delayed reply. I got sidetracked with building an outside screened-in grow area. It looks pretty awesome now that it is finished, and it houses six of my aero modules for the summer. I can elaborate if there is interest, but that is another topic altogether.

20180610_152748_resized.jpg 20180610_152725_resized.jpg

Nice! Did you peek at my plan book? That is the concept I considered. I contemplated using a DPDT relay to delegate the power. It would be thermostatically controlled. The circulation pump would loop the coolant through the system and carry the heat out of the tent on the normally closed poles. When the temperature reaches the high point, the relay throws from the circulation pump to a main line solenoid that feeds cold city water by energizing the normally open poles. The additional coolant displaces the hot coolant that is overflowed to waste.

In the colder months, this system could be adapted to a fan-driven radiator to move the heat into a living area, or store it, using a different media, for dark time when the temperature drops. I did think about forcing the hot water to a radiant heating array under the floor, but I think a radiator and fan would be more economical. Either way, I have to gauge the total heat of the system before I start configuring the dissipation system.
 

ZxcStaz

Active Member
So, I have made a bit of progress on the lighting project. I have soldered the 70 chips. It took about five hours to cut the leads, tin them, and solder them to the chips. I needed a second set of hands to use two irons. This was needed to provide enough heat for the 12 gauge stranded wire. In the future I'll use 16 AWG stranded, and this is because each chip is only 50 watts.

I purchased the aluminum square tubing and cut it to make the unit. I have to get it welded tomorrow, and that is no small feat. I ran into issues trying to braze it, and I consumed a lot of time, but I learned quite a bit about joining metals with heat. I could post some pics of the failures, put I think I still harbor some PTSD from the whole ordeal. One side observation that might be of value to anyone endeavoring this project: aluminum is a great conductor of heat...hands that do not have thermal protection are prone to cellular degradation with a concurrent activation of pain neurons... Oh, ethanol helps alleviate the nagging reminder of one's oversight.

I drilled the ends of the tubing for water flow. I just used a titanium coated cheap drill bit. I also cut the pipe with a vibratory saw. It's just a cheap Harbor Freight tool, but it worked well. I suggest using ear protection while performing either of these tasks; it's louder than children - and that's hard to do.

So, here are some pics. I have to resume my self-inflicted ataxia to provide a modicum of amnesia, Hopefully I will be able to solder, ah wait, soldier on... :)

20180611_191810_resized.jpg 20180611_191804_resized.jpg
 

nfhiggs

Well-Known Member
So, I have made a bit of progress on the lighting project. I have soldered the 70 chips. It took about five hours to cut the leads, tin them, and solder them to the chips. I needed a second set of hands to use two irons. This was needed to provide enough heat for the 12 gauge stranded wire. In the future I'll use 16 AWG stranded, and this is because each chip is only 50 watts.

I purchased the aluminum square tubing and cut it to make the unit. I have to get it welded tomorrow, and that is no small feat. I ran into issues trying to braze it, and I consumed a lot of time, but I learned quite a bit about joining metals with heat. I could post some pics of the failures, put I think I still harbor some PTSD from the whole ordeal. One side observation that might be of value to anyone endeavoring this project: aluminum is a great conductor of heat...hands that do not have thermal protection are prone to cellular degradation with a concurrent activation of pain neurons... Oh, ethanol helps alleviate the nagging reminder of one's oversight.

I drilled the ends of the tubing for water flow. I just used a titanium coated cheap drill bit. I also cut the pipe with a vibratory saw. It's just a cheap Harbor Freight tool, but it worked well. I suggest using ear protection while performing either of these tasks; it's louder than children - and that's hard to do.

So, here are some pics. I have to resume my self-inflicted ataxia to provide a modicum of amnesia, Hopefully I will be able to solder, ah wait, soldier on... :)

View attachment 4149596 View attachment 4149597
12g wire is massive overkill. 18g is plenty big enough for a 50W load at 120 VAC.
 

ZxcStaz

Active Member
12g wire is massive overkill. 18g is plenty big enough for a 50W load at 120 VAC.
Yeah I know, I have issues. In my own defense, there was an ideology behind it. The copper has a better heat conductance than aluminum, so if I use thicker Cu, more heat will dissipate and it will run the chip cooler. It was a redundant measure such that if the liquid coolant temperature climbed, I would have a fail-safe. It is more expensive, more difficult to work with, and just bloody ugly, but there is less resistance all around. ...I don't know, I'll let you know how it performs. The only detriments that I can foresee are expense and ease of use.
 
Top