SunLike 3500K 95+ CRI midpower led from Seoul Semiconductors tested

Joshk · 2019-11-20T02:17:07.000Z

That’s some cool looking aluminum extrusion. Good job

led4power · 2019-11-20T12:54:26.000Z

SunLike:

If someone need, I just finalized draw pcb for 3030 sunlike leds it is for 1-12pcs (12.5x99mm with 2.5mm holes) and 1-6pcs 3030 leds (22x22mm with 3mm hole) 3S4P or 2S4P or 8P and 2S3P or 3P Also you can cut it if you need

One suggestion, it's much better practice to connect LEDs in series first, and connect those series strings in parallel.

Reason for this is LEDs Vf variance, if you connect multiple LEDs in parallel, current distribution could be unequal, LED with lowest Vf will draw highest current, and because of that if will become hotter, and Vf would drop even more, current would rise even more.. etc. I've seen this problem in LED T8 tube light, shadows and bright spots were visible because of unequal current distribution and measurements confirmed that.

With 3 (or more) LEDs connected in series first, all strings would have total Vf closer to average (more leds - more equal average total Vf among strings because of probability laws).

SunLike · 2019-11-20T13:23:19.000Z

I understand this situation, but:

1. seoul semico recommended to use their 3030 leds to 150mA, I recomend to use it on 85mA and I will use it with 0.5w
2. quality of leds is really good and it is almost with the same voltage.
3. when 1 led will die (shit happens) when those series strings in parallel, string will not work. If here, for example, 1 of 4 pcs will die, another 3pcs will work only 1.333 times more than 4pcs
4. in my situation it is possible to use it 3S4P and 2S4P and 1S8P

I changed price to 3030 leds on my web-site to 0.5$ and power too (now it is 0.5w, was 0.6w)

But seoul semico say that it is 150mA and 200mA max current. It is not good power to this leds, because the size of crystal not 1w. I mean it will die really fast with 150mA.
———

I just draw mixing 3:1, 2:1, 1:1

The_Driver1 · 2019-11-20T14:33:48.000Z

SunLike:

I understand this situation, but:

seoul semico recommended to use their 3030 leds to 150mA, I recomend to use it on 85mA and I will use it with 0.5w

quality of leds is really good and it is almost with the same voltage.

when 1 led will die (shit happens) when those series strings in parallel, string will not work. If here, for example, 1 of 4 pcs will die, another 3pcs will work only 1.333 times more than 4pcs

in my situation it is possible to use it 3S4P and 2S4P and 1S8P ------ I changed price to 3030 leds on my web-site to 0.5$ and power too (now it is 0.5w, was 0.6w) But seoul semico say that it is 150mA and 200mA max current. It is not good power to this leds, because the size of crystal not 1w. I mean it will die really fast with 150mA.

Even "really good" LEDs (how do you know this?) have a variance in Vf from sample to sample.

When one of four LEDs dies, the other three get more current. This increases the chances that another dies (more current => more heat => Vf drecreases => more current => ...). This can continue until all four are dead.

SunLike · 2019-11-20T14:47:55.000Z

It is if leds works with almost 100% power. I will use only 0.5w or smaller.

Joshk · 2019-11-20T16:16:15.000Z

I agree with LED4Power’s suggestion and The_Driver’s opinion. It would be worthwhile to change the PCB layout. At the price point and quality your business targets, you don’t want customers noticing one LED is brighter than the others. It will make them nervous and could result in some warranty claims. I would take their advice if it were my PCB.

SunLike · 2019-11-20T18:28:45.000Z

I have my own approach. If you will find at least a 20% difference in brightness, I agree to give you everything what you will choose on my web-site for $ 200

Ok. I am going to do another board. about 12.5x99mm, what will work from any 18-22v power supply from notebook. I will do their 4P3S

If you know size what will be better, let me know. For example 15x99mm or etc

I am going to use NSI50350ADT4G
or NSI50350AST3G

it is 350mA

SunLike · 2019-11-21T20:30:40.000Z

Hello, peoples of the flat Earth!

Well done! about 6 hours (for this board)

120-240-360-480mA or etc with NSI constant Current Regulators (yes, it is not cheap)

from 18v to 25v

also you can give 12v if instead of 1 LED, you will set a jumper (wire)

also you can use led driver if you will set a jumper (wire) except CCR

12x99x1.6mm m3 holes

also you can cut up to 1.5+mm from the both sides

50%G (Grower)

G≈$

Joshk · 2019-11-21T20:52:22.000Z

It looks good. The CC regulators are not even necessary for good performance.

Backing up to the previous design, you see how the current is given the choice to take the path (the LED) of least resistance at every level? That’s not good.

But your new design is immune, even without the expensive regulators.

SunLike · 2019-11-21T20:57:15.000Z

it is not critical. because there is only ~300mA

And I am going to not parallel this boards, because I need 36-54v. I will be S, so, current will go from left to right or etc.

Also you can do not use CCRs. It is maximum universal. Also I have 5+h video. I make it when I draw this board =) It is real 6h to do this board

Joshk · 2019-11-21T21:14:57.000Z

On a different topic, I was curious about this:
50%G (Grower)
G≈$

Is that a price? I haven’t seen percent used like that before.

SunLike · 2019-11-21T21:17:17.000Z

Joshk:

On a different topic, I was curious about this:
50%G (Grower)
G≈$

Is that a price? I haven’t seen percent used like that before.

Yes. It is 50perCents =)

Also it is real money in my village =)

Joshk · 2019-11-21T21:28:36.000Z

So 50%G is about $25?

SunLike · 2019-11-21T21:30:58.000Z

Joshk:

So 50%G is about $25?

No =) it is 50% from 1$. It is 0.5$ or $0.50 if you like that style

Joshk · 2019-11-21T21:34:06.000Z

Interesting. Thanks.

Joshk · 2019-11-21T21:50:28.000Z

I know you don’t agree the time spent changing the PCB was necessary technically. But perhaps look at it from the business angle. This new design eliminates a possible criticism of your products. Competitors can not claim you made a mistake.

SunLike · 2019-11-21T21:58:58.000Z

I do not think, that 3S4P was with mistakes in design. But 4P3S was not so easy designs pcb. v1.0 was for 1 CCR 350mA, but it was impossible, 1.1 was for 4x4pcs sod123 CCRs. I do not think that 4P3S better than 3S4P. But if you like it more, now you can choose. Tomorrow I will add order to pcb manufacturer for 150-200pcs.

led4power · 2019-11-21T22:29:03.000Z

Those CC regulators from On semiconductors have pretty bad characteristics, current is far from constant, large voltage drop etc., it's better to use Diodes Inc AL5809, they are very stable and accurate.

SunLike · 2019-11-21T22:50:46.000Z

led4power:

Those CC regulators from On semiconductors have pretty bad characteristics, current is far from constant, large voltage drop etc., it’s better to use Diodes Inc AL5809, they are very stable and accurate.

AL5809 also have CCRs in SOD-123 body up to 50mA - page 13 .
And when I draw soldering footprint, I used NSI and AL5809 datasheet.

любой каприз за ваши деньги =)

led4power · 2019-11-22T09:54:42.000Z

You can use version in PowerDI-123 package dimensions are almost the same as SOD-123, but power dissipation capability is better, so there are versions up to 150mA - less $/mA.