All About Lighting

Discussion in 'Lighting' started by jcj77d, Aug 24, 2008.

  1. I think its 18 chips in series (18*3v=54v), and the 16 stacks are wired in parallel. I cut up a 288 board into quarters and each quarter still needs 54v driver to run.
     
    • Winner Winner x 1
  2. #382 ChiefRunningPhist, Feb 18, 2020
    Last edited: Jan 17, 2021
    Lol yes, that would make more sense, and I was baffled by what I originally thought I saw. Lots of fear of thermal runaway but then when I saw that they had put the chips 1 row deep, and 18 parallel (thought I saw)??? Lol it didn't make sense. At first I thought "wow, the chips must all have really close Vf's" which made me think about possible wiring scenarios that I originally thought were problematic (and rightfully so!), but then I finally saw the printed "16P × 18S."
     
    • Like Like x 1
  3. Interesting article. All the Samsung datasheets I've sifted through I've never seen anything about pulsing. I've looked at literally all of them and haven't seen it. Their online calculators use straight DC and hold true to what the datasheet shows for the most part if you use the highest flux bin and average voltage bin.

    Yes I see how pulsing would allow you to better rate the efficiency of the chip stand alone and not have to factor in any hardware scenarios in to the readings. But it is also misleading because that is not real world and they are measuring at the on time its brightness. So pulsed ratings would only lead to worse real world scenarios unless you had chip like a 3535 with a heat slug and you had a direct bond to the aluminum pcb which is doable and had less system inefficiencies.

    I was just pointing out the ways why it's not going to be as high as you think. Which is why I hate chip level ratings. I hate board level ratings too but definitely hate chip level ratings. HLG does a pretty decent job of doing board level and at a given temp and that temp isn't crazy low. It is unclear if that 55C is board temp, solder temp of the chip, or junction temp of the die.

    I don't know what a QB trace loss is. Sounds like HLG is rating full board so you don't have to worry about that. Probably those cheaper import ones use 1oz copper leading to twice the drop and less cooling capacity for the junction temperature. Even if they tell you its not, it probably is.

    Then for wiring it up definitely a system that is paralleled in and out of the boards is worst, paralleled from the driver is better, then finally in series is the best for loss. Lower currents being better. Then people like to remote their supplies adding wire in between. Adding some loss. I agree keeping current low and using decent sized wire can minimize loss but it is system dependent and this is a board rating and does detract and can vary. Then the 10% driver loss on top of that.

    Higher currents reduce efficiency mainly due to heat. This is where most of the problem is.

    Anyways best way to know is how you'll actually be using the light and didn't want people to be mislead but the lower power high efficiency numbers. Also wanted to draw attention to the board level rating is different than your system level rating when comparing to a complete product.
     
    • Informative Informative x 1
  4. #384 ChiefRunningPhist, Feb 19, 2020
    Last edited: Feb 19, 2020
    ^^I can agree with that.

    The reason they pulse is due to temp increase under operation. LEDs have a negative temperature coefficient, this means their Vf drops as temp increases. Its just the nature of the LED process that when current is applied they give off light, but heat up as well. This means in order to generate a true Vf to If curve, they need to control for temperature. Pulsing is their way of controlling for temperature. If temp is maintained, then Vf doesn't drop and they are able to tell the I flowed per V applied, and determine chip power in reference to chip light emmission. I've never known a single manufacturer to use DC, but they could be out there. Samsung could be using straight DC, maybe a quick email might clear it up.

    I agree that some chips are better than others at dissipating heat. Some chips can take more current before hitting a temperature threshold, but regardless the chip design, the final temperature is the figure of importance. Heatsinks and other secondary heat mgmt play a major role as well. At higher currents its just not possible to keep the chips as cool as with lower currents. Temperature is just a byproduct of inefficiency. You have a limited size die and at 65mA your flowing a small amount of electrons, but then as you increase current you're flowing a ton more electrons but through the same small area. The result is heat. Just like if you tried to run a huge current through a tiny wire, it will turn orange and melt. A wire has a resistance, its some amount of Ω's and it doesn't change (well a little bit, it has a positive tempo, but it could be any amount of Ω's depending on circumference and metal composition, ect). Watts = Volts × Amps, and Ω = Volts/Amps. When current is flowed through the wire, the wire resistance acts on the current and produces a voltage drop across the wire. This voltage drop can be multipled by the current being flowed to determine watts of heat wasted. The Ω's in the wire never change, but if you increase the current, your heat wattage goes up. This heat wattage is the byproduct of the increased current due to the fact that the Ω's never changed. LEDs have some ESR on top of their non linear R. I'm not sure which R is more responsible, but as you flow more electrons through a finite space you're going to increase waste heat.

    As far as Cu losses, it doesn't really matter if you're in series or parallel. In parallel you'd size your top and bottom rail traces bigger to accommodate for greater current but each chip trace still gets the same current as it would in series. I agree that China boards are probably 1oz as well.

    The only difference between a chip level and board level will be Cu losses, which really won't represent much of a change. Maybe 0.5%? The only difference between board level and real life, is driver inefficiency, and light losses due to hang height ect. Light losses will vary from setup to setup, but that's why they test with an integrating sphere. It removes the setup to setup variability and gives a base line for people to go off of. Most driver inefficiencies are somewhere around 4%-8%.

    HLG shows 2.6μmol/J and that's with driver inefficiency, Cu losses, and light losses included. An LM79 is a real world test. I'm estimating around 150mA per chip.
    CRF_CV_ EFFICIENCY_1.0.png
    CRF_CC_EFFECIENCY_1.0.png
    CRF_Fv_Graphs_1.0.png
     
    • Like Like x 3
    • Informative Informative x 1
  5. I actually understand some of these words. :laughing:

    Had to look up 1oz vs 2oz copper (for anyone like me who didn't know, it's the quality rating of copper used in PCB boards, higher is better), and it sounds like just the kind of thing they could do to reduce costs without the layman user understanding that he is buying an inferior product.
     
    • Like Like x 3
    • Winner Winner x 1
  6. Ya! :love-m3j:

    1oz Cu describes 1oz of copper spread equally about 1ft2. 2oz of Cu describe 2oz of Cu spread equally about 1ft2. So A 2oz Cu will have a thicker trace that has double the depth or height than a 1oz. You can offset low Cu by increasing trace width. So if I had 1oz and wanted the same resistance as 2oz, I'd have to double my 1oz trace width compared to the trace width of the 2oz.
     
    • Informative Informative x 1
  7. You can't change the copper traces on those boards for parallel. Whatever image someone posted it literally said "at board". You can offset the voltage drop with trace width, thermal spreading not so much. China is on the 1oz standard for mcpcb. You are nuts to them when you try to order 2oz. Making the 1oz trace helps but think of it like dropping a bowling ball in a pool and absorbing that energy. Making it wider helps but if its shallow it doesn't help much. At a certain depth the width helps. Making it super deep is better but there are is a point where its not really worth adding. You have twice the cross section to spread that heat out before you hit the dielectric in to the aluminum. There's some good cree articles on this. In parallel this can matter and add temperature rise from the voltage drop. But mostly I was saying that the 1oz if cooling of the junction of the LED wasn't as good could affect efficiencies... targeted at ripoff quantum boards aka bare circuit boards.
     
    • Like Like x 3
  8. #388 ChiefRunningPhist, Feb 20, 2020
    Last edited: Jan 17, 2021
    You increase trace width to reduce electrical resistance.


    Heatsinks work on surface area and thermal conductivity, so if you wanted to use the extra Cu on the board, you'd want to increase the surface area, the depth won't matter so much.

    Kinda crappy pics, but you can see maybe how thin the layer is comparatively with the rest of the board and the 2 sides, aluminum and the Cu.

    2oz Cu... 2' × 1.5' ...
    0512191406b.jpg 0512191404.jpg
     
    • Like Like x 3
  9. It doubles the distance the heat has to spread before it hits the dialectric layer. Think of it like throwing a bowling ball in a different size pool. Heat can now transfer at a different point in the board. Board still has same heat load but its spread better and it keeps junction temperature down which is the main goal.

    https://www.cree.com/led-components/media/documents/XLampThermalManagement.pdf
     
  10. Hey guys I saw this light and thought it would be great for a pc grow the dimensions are 11cm x 34cm (so it would fit because a full size tower is 40cm x 18cm) and it pulls 30w from the wall lol

     
  11. #391 Hybridway, Dec 3, 2020
    Last edited: Jun 17, 2021
    Some lights. 20201024_232054.jpg 20201024_231328.jpg 20201016_002005.jpg 20201019_182133.jpg 20201022_200029.jpg 20201024_231530.jpg 20201024_231618.jpg 20201024_231733.jpg 20201024_231821.jpg
     
    • Winner Winner x 3
    • Like Like x 2
  12. #392 Patricia Clemons, Jan 13, 2021
    Last edited: Jan 13, 2021
    Hybrid way you've got some of the best rooms I've ever seen in my life.

    I hope to dedicate a small corner to a LED space where I can overwinter my succulents, start a bonsai, and maybe a marijuana plant.

    Only have zero idea where to begin, so many options

    This is interesting, definitely prefer the full spectrum, would thing it had UV Best Quantum Board Grow Lights 2020 | 420 Green Thumb

    Sent from my Pixel 3a using Grasscity Forum mobile app
     
    • Like Like x 3
    • Like Like x 1
    • Winner Winner x 1
  13. WOW It's like the LED museum! Those strips lamps look interesting, what brand is that?
     
    • Like Like x 2
  14. #395 Hybridway, Jan 17, 2021
    Last edited: Jan 18, 2021
    Thank you! Give it a go. Good to ease Into it imo.
    Couldn't have done it without you brother!
    Hi bud! Thank you! All the strips/bars in this room are Amare. Even the skinnier ones in the center isle, using one row diodes is the Bar-8 Prototype from a few yrs ago.
    Last pic has an out of business company that had a pretty bad name amongst the internet grow community. That's a prototype HydroGrow board used as supplemental to the wh/red diode variety but kept to a minimum when mixing because there was x3 SBS's going there.
    -HLG Diablo vs. Amare BAR-6
    -x2, 320w , HLG, 96-Elites vs. x2, Amare Pro-4's with the cobs dimmed to 65-75w, leaving them a total of 275w each.
    -x4, CMH, 315's (1260w) vs. The Amare Bar-8, Canna-Spec using 900w.
    Bout to add up the total weigh ins today.
    Will say these 2 things that I can speak on now.
    . First thing is the HLG, Diablo gave me my first 2#, single plant on record for me & it was only in a 7gall. Fabric pott. That was my BrideZilla #4 (wedding cake dom.)
    Second thing is that in every given case, the patients choose Amare bud over anything else presented. The bag appeal, density, smell, taste was all through the roof. I don't blame them. Was Hella surprised to see the Canna-Spec beat out the x4, cmh's in appeal & density. Only because I've never used cmh only hps+mh. I'd say it was very similar if not right on with mixing hps+mh for that better spectrum but with allot less watts.
    Diablo buds were brighter green, Lil bit smaller & denser. Very frosty too. Just not as big, caught about half the stretch as the Bar-6 & lacked the lavender, pinkish colors. May have had a Lil less terms too. Thing is, you'd never know if you didn't have the two sbs.
    Still, absolutely amazed with 2#'s off one plant using 550w average total off the Diablo.
     
    • Like Like x 2
    • Informative Informative x 1
  15. 2# wow thats really nice! I knew I should have bought some of those diablo boards when they still sold them separately :bang:
     
    • Like Like x 2
    • Funny Funny x 1
  16. Thanks for the insight! Very helpful
     
    • Like Like x 1
    • Disagree Disagree x 1


  17. awesome lights. Do u send em out for free ??
     
  18. That's definitely NOT a list to go by. Nothing good there. Couple decent Budget lights in there though.
     
    • Like Like x 1

Share This Page