DiY LEDs - How to Power Them

Boatguy

Well-Known Member
I also want to remote mount the drivers, I've never mounted them to fixtures before. But this'll be three DC out power cords running up and over to each fixture, right? No way around that?
Get yourself a length of 18/6 stranded cable off amazon. 18 is plenty heavy for the dc voltage and current you'll be running there. Add some pigtails to the fixtures and drivers for easy setup and disassembly.

Original 648 datasheet here
 
Last edited:

2com

Well-Known Member
A question re the 320-54AB: it only comes with onboard adjustable current? No adjustable voltage? Seems hard to get a clear answer theu the data sheets.
Correct. No adjustable Vo. No dim-to-off function.
Also, when you say wont light up, no light?
Won't light up. I'm sure I would've have Io pot turned all the way up. If I have time later I'll check it out again.
but what happens in practice always takes precedent over theory.
It's refreshing to hear someone else say this.
Their open voltage is supposed to be 60V with loaded voltage range of 27-57V. HLG uses inventronics drivers of same spec for the diablos (or at least used to). To the best of my knowledge the xlg- should run that board.
It's 27-56V per the data sheet. Yea, they've run a few different drivers from meanwell and inventronics. The xlg-H model should run them, yea. If/when I get some funds I'll likely order a few to try out. I have four qb648 with no drivers for them, wish I had six.
Im not a super fan of the xlgs though, id probably go for a 54A driver and turn up the voltage a bit.
Xlgs get quite hot and have not always reacted how we expected them to. Also: we seen to have more draw on the AC side of our drivers than what we expected, im suspecting the xlgs as the culprit. Will be checking next turnover, got some nice new sonoffs which should be able to measure better than what we got now.
Ok, that's not good to hear. And 54A drivers have no external dimming leads (unless there's a couple wattage/models that do, there are some one-offs and inconsistencies with "features" in their lines from what I've seen). And I don't think the xlgs have CV options (ie "54x") so you must mean a different series when you say this?
Three hlg-240s would cost over $360, compared to $210 for xlg. That's pretty significant...
Thanks.


Get yourself a length of 18/6 stranded cable off amazon. 18 is plenty heavy for the dc voltage and current you'll be running there. Add some pigtails to the fixtures and drivers for easy setup and disassembly.

Original 648 datasheet here
Thanks for the suggestion.

@Lou66 No. Have a nice weekend though.
 

Rocket Soul

Well-Known Member
Correct. No adjustable Vo. No dim-to-off function.

Won't light up. I'm sure I would've have Io pot turned all the way up. If I have time later I'll check it out again.

It's refreshing to hear someone else say this.

It's 27-56V per the data sheet. Yea, they've run a few different drivers from meanwell and inventronics. The xlg-H model should run them, yea. If/when I get some funds I'll likely order a few to try out. I have four qb648 with no drivers for them, wish I had six.

Ok, that's not good to hear. And 54A drivers have no external dimming leads (unless there's a couple wattage/models that do, there are some one-offs and inconsistencies with "features" in their lines from what I've seen). And I don't think the xlgs have CV options (ie "54x") so you must mean a different series when you say this?
Three hlg-240s would cost over $360, compared to $210 for xlg. That's pretty significant...
Thanks.



Thanks for the suggestion.

@Lou66 No. Have a nice weekend though.
Try those diablos on minimum dim, might work. Trying one on full dim should not work so quite normal.
 

shimbob

Well-Known Member
I happen to be using an XLG-240-h-ab at the moment, it's really versatile when swapping lights to play around with various boards.
It looks like two QB288 V2 rspec boards in parallel would settle right around 49.(something)V at 2.4A per board for 120W each.

I run 10GA wire from my driver to the lights, no worries about voltage drop and only "a few pennies" more.
 

2com

Well-Known Member
I happen to be using an XLG-240-h-ab at the moment, it's really versatile when swapping lights to play around with various boards.
It looks like two QB288 V2 rspec boards in parallel would settle right around 49.(something)V at 2.4A per board for 120W each.

I run 10GA wire from my driver to the lights, no worries about voltage drop and only "a few pennies" more.
Thanks for sharing your experience with them.
One thing I'm unsure about now is forward voltage. I thought the forward voltage, or the 56V requirement for the qb648 - for example, was the minimum voltage required to even power the board(s). Is that not correct? Because it's been my experience, but my experience has been almost entirely with constant voltage drivers. Is that not how it works?

10awg!? That's huge. How long is the run?
I already have several "cords", a couple 16awg, sever 18awg. They range from 10'-12'.
 

shimbob

Well-Known Member
Thanks for sharing your experience with them.
One thing I'm unsure about now is forward voltage. I thought the forward voltage, or the 56V requirement for the qb648 - for example, was the minimum voltage required to even power the board(s). Is that not correct? Because it's been my experience, but my experience has been almost entirely with constant voltage drivers. Is that not how it works?

10awg!? That's huge. How long is the run?
I already have several "cords", a couple 16awg, sever 18awg. They range from 10'-12'.
Under most normal operations, the voltage an LED requires is determined by the current you're feeding it. Usually led vendors publish specs that list the max voltage required at the max amps that they recommend. In the case of the qb648, the (pretty crappy documentation relative to other specs Samsung has put out) specs say the qb648 goes up to 58.8V at 2.2A. But an LED will happily run at a lower voltage when you feed it less current.

10ga only because its what I happened to have on hand after building a 700W light with a 15' wire run, got real anal about voltage loss and went big. Current run is 6', maybe.
 

Boatguy

Well-Known Member
Under most normal operations, the voltage an LED requires is determined by the current you're feeding it. Usually led vendors publish specs that list the max voltage required at the max amps that they recommend. In the case of the qb648, the (pretty crappy documentation relative to other specs Samsung has put out) specs say the qb648 goes up to 58.8V at 2.2A. But an LED will happily run at a lower voltage when you feed it less current.

10ga only because its what I happened to have on hand after building a 700W light with a 15' wire run, got real anal about voltage loss and went big. Current run is 6', maybe.
Download wiresizer on your phone. 56vdc at 2.5a over 20' round trip in 18g at 3% voltage drop
I thought the forward voltage, or the 56V requirement for the qb648 - for example, was the minimum voltage required to even power the board(s). Is that not correct?
1708988536113.png
I posted this link earlier.. 58v is the max.. 51 the minimum
 

2com

Well-Known Member
Try those diablos on minimum dim, might work. Trying one on full dim should not work so quite normal.
I just took two of the qb648, got the new-in-box hlg-320h-54ab driver all connected in parallel with them, with Io pot. turned all the way down (duno why, why not?), timberled cased potentiometer connected and turned all the way down ('cause I remembered you suggesting that), triple checked connections, and turned on the power. They...turned...on. I was confused, I still am, a little bit. I had my DMM connected in parallel. I'll come back with some numbers if I get another chance tonight. I'm going to get one of the other 320h-54ab drivers from the other room and try with it too, just to check.

Just popping in to say that, and thanks.
 

Lou66

Well-Known Member
Check the Datasheet of the driver. It does not dim to 0 but only to 10 %. To turn it off you have to interrupt the live.
The ELG series dims to 0.
 

2com

Well-Known Member
Check the Datasheet of the driver. It does not dim to 0 but only to 10 %. To turn it off you have to interrupt the live.
The ELG series dims to 0.
Oh I know the 320h-54ab doesn't dim to off. That's one of the things I don't like about it. I think I mentioned that in the "cons" in a earlier post, but maybe I didn't.
The XLG dims to off as well.
By the way, appreciate you offering suggestions. Thanks.

----------

Anyways. I just tried the other 320h-54ab, it also worked. I can't imagine what I would've done (pots, dimmers, etc.) wrong the first time for them to not turn on. Unless it was some wiring issue other than reverse polarity... I duno.
Now I'm gonna be wondering about the other pair of qb648s, and the other 320h-54ab driver, so I'll likely end up testing those items as well, later.
 

Rocket Soul

Well-Known Member
Constant voltage is a thing on the driver side, not the same thing on the chip side
Constant voltage means your driver will out put current at a constant voltage, not how the chip reacts to voltage and current at the output side.

It basicly means that when you buy a chip (cob or whatever) it will be somewhere between min and max and usually somewhere close to average, if all compared at current. Generally cobs and strips will be within a couple of percent of the average value.

Theres really a long discussion behind this regarding this, generally all values in the spec sheet comes with a little * of "+-7%". Or sometimes as much as 10%; anything you buy which is within these tolerance levels would generally be deemed as within spec.

I would be very surprised if there isnt some number cruncher at every company looking at these numbers and evaluating how to "game" out the datasheet with two things in mind; how to present the product in the best possible light and how to avoid liability by over specing your official numbers and getting sued by someone.

But here in this case, i think the spelled out min and max in part cause its more important and in part cause theres a higher theoretical variance. FWIW count on the mid number but if your running them in series the add a few volts to the total just in case the chips are a bit over mid range voltage.
 

Boatguy

Well-Known Member
Constant voltage is a thing on the driver side, not the same thing on the chip side
Constant voltage means your driver will out put current at a constant voltage, not how the chip reacts to voltage and current at the output side.

It basicly means that when you buy a chip (cob or whatever) it will be somewhere between min and max and usually somewhere close to average, if all compared at current. Generally cobs and strips will be within a couple of percent of the average value.

Theres really a long discussion behind this regarding this, generally all values in the spec sheet comes with a little * of "+-7%". Or sometimes as much as 10%; anything you buy which is within these tolerance levels would generally be deemed as within spec.

I would be very surprised if there isnt some number cruncher at every company looking at these numbers and evaluating how to "game" out the datasheet with two things in mind; how to present the product in the best possible light and how to avoid liability by over specing your official numbers and getting sued by someone.

But here in this case, i think the spelled out min and max in part cause its more important and in part cause theres a higher theoretical variance. FWIW count on the mid number but if your running them in series the add a few volts to the total just in case the chips are a bit over mid range voltage.
That is an old datasheet. Maybe something has changed since they stopped having samsung make the boards. The diablo 750 they are pushing to 180w per panel which is 50% higher than the max according to the info i posted.
I have my 2 board fixture running off a hlg 320-2800 and have run it max but typically dont because of the amount of heat it produces and how far it needs to be away from the plants.
 

Rocket Soul

Well-Known Member
That is an old datasheet. Maybe something has changed since they stopped having samsung make the boards. The diablo 750 they are pushing to 180w per panel which is 50% higher than the max according to the info i posted.
I have my 2 board fixture running off a hlg 320-2800 and have run it max but typically dont because of the amount of heat it produces and how far it needs to be away from the plants.
Sorry, wasnt talking about the hlg datasheet, just how samsung and mfters makes their datasheet: i really doubt its just based on measuring output and rather figure out what number to put to make it look as good as possible while minimizing any liability risks of being outside spec.

Edit: i thought diablos could handle almost 300w if you have enough heatsinking; regular qbs handle 150w and a diablo is more than double the diodes.

Also outoff my own interest; what kind of lightlevels/hanging height do you get with your diablos pushing that driver to the max? Thinking about picking some up
 

Boatguy

Well-Known Member
Sorry, wasnt talking about the hlg datasheet, just how samsung and mfters makes their datasheet: i really doubt its just based on measuring output and rather figure out what number to put to make it look as good as possible while minimizing any liability risks of being outside spec.

Edit: i thought diablos could handle almost 300w if you have enough heatsinking; regular qbs handle 150w and a diablo is more than double the diodes.

Also outoff my own interest; what kind of lightlevels/hanging height do you get with your diablos pushing that driver to the max? Thinking about picking some up
At 240 for the pair i can get them about 12" from the canopy. At max it is closer to 24+.

They have a pretty small footprint and mine are spaced with just enough room for the driver so overlap is an issue.
IMG_2064.jpeg

It is a 3x3 and they are pretty perfect in there. Not many 3x options out there for led lighting, unless you go big and run them low
 
Last edited:

KitnerPush

Active Member
Well according to HLG reds's are 6x the current of whites. So St test-current on whites of 65mA reds will be 390mA. So at test current of white total system would be 2.73A @ 56V (actually probably lower with low voltage bins) that's 152.88W. Thats running at half the test current for red.

If we run at test current of red 700mA, then whites will run at 116.66mA. You're probably going to get some deminishing returns running them that high? Soul?? Anyways, system would then be 295W.

anyways, this is the derating curve:

Screenshot_2024-04-10-22-31-27-70_e2d5b3f32b79de1d45acd1fad96fbb0f.jpg
Looks like you're good up to 80°C at 116mA, or right on the hair...probably pushing it.

At 2800mA at 56V, you're at 77.77mA which means you shouldn't get higher than 85°C.

Wish we had more people who knew the intricates of the diodes... It would be nice to know how each 10mA increase affects the lifetime of the diode??? Anyone???
 

Rocket Soul

Well-Known Member
Well according to HLG reds's are 6x the current of whites. So St test-current on whites of 65mA reds will be 390mA. So at test current of white total system would be 2.73A @ 56V (actually probably lower with low voltage bins) that's 152.88W. Thats running at half the test current for red.

If we run at test current of red 700mA, then whites will run at 116.66mA. You're probably going to get some deminishing returns running them that high? Soul?? Anyways, system would then be 295W.

anyways, this is the derating curve:

View attachment 5385122
Looks like you're good up to 80°C at 116mA, or right on the hair...probably pushing it.

At 2800mA at 56V, you're at 77.77mA which means you shouldn't get higher than 85°C.

Wish we had more people who knew the intricates of the diodes... It would be nice to know how each 10mA increase affects the lifetime of the diode??? Anyone???
For high lifetime try to keep your leds at under 50C at junction, they will probably outlast your life. But keep in mind that lifetime tests graphs dont seem to match up by reallife testing by growers: Migro made some tests a few years ago of various growlights, kept them.on for a year or so and the performance reduction seemed to be a fair bit over what was expected from the datasheet. The test were made in his shop so probably even worse in a more harsh grow environment with higher humidity.


The 6x current is something ive seen pop up from time to time: looking at the pcb it would mean theres 3 reds in series in each string (didnt know, thought it was only 1!) and the numbers actually seems to confirm it (18x2.73+ 3x2=56ish V). It would mean that 18 strings of white feeds into 3 strings of reds on each side of the board. My only gripe with this is that if i could choose 3 monos i wouldnt go for 3 of the same kind, id love to drop a 400nm diode into that string
 
Top