Flying searchlight V2

[HugeOne]

Well after successfully flying my FPV airplane at night with the help of a headlight. (see: https://www.candlepowerforums.com/threads/260667)

I am back on the drawing board for the next one, lighter, simpler. better and off course, more lumens. I want to base my next lamp on 6 SST-90 along with Ledil Lyli-spot reflectors. I want to place the emitter in rows of 3 touching each others. In this configuration, the reflectors won't fit, so I want to cut them so the center one is just a portion. Effectively making an oval reflector to obtain a wide spread in horizontal, but not so much vertical.

So now, to begin with, I'm looking for a source of bare emitters, top bin preferably. I found led-tech.de that carry them so far...

 

[CKOD]

Whoa, 200+ watts is one heck of a heat load, definitely watercooled again?

And Ive had good luck with Avnet for SST-90's Ive been using the 5700k top bin LEDs. Going to put them down on a copper plate with a dugout for the electrical connection on the back? If you do, get a chipquik kit and use the low temp solder, and solder those badboys down after you use lead free solder to solder on the wires.

Going to be using a driver for this? or direct drive? the der Wichtel 9A buck driver is pretty nice, built a light for a friend with one. Compact too.

Definitely subscribing to see the results of this.

 

[HugeOne]

That a nice little driver there, but it's not dimmable. I need to variable output not to be blinded on landing. I designed another linear driver with 0-100% instead of Off/Lo/Hi. Problem is that the Vf of 3 SST-90 is high for 3S li-po, once the battery is half discharge I won't be able to drive more than 5A or so. I could use 4S li-po, but then the FETs will deal with major heat.

The SST-50 have a lower Vf and it may be a good alternative to stack 12 of them in a 3cm x 3cm square with a custom made reflector around that.

Of course, I'll recycle my watercooling system. My idea is to solder the led's heat pad to a 5mm square copper pipe in which coolant will flow. That chipquick thing is interesting, how does it work? mix it with the solder paste?

 

[aEx155]

That chipquick thing is interesting, how does it work? mix it with the solder paste?

Chip Quik is essentially solder with a really low melting point (since it contains bismuth). I believe CKOD recommended it since it wouldn't put as much stress on the LED when soldering as it wouldn't have to get as hot.

Just from doing some reading, here are some things I can suggest:

If you're running everything off a 3S Lipo pack, you could try 2s3p for the LEDs to get a lower Vf while still having 6 SST-90s. Of course, that'd mean you'd need a higher current driver...but hey, it's a thought.

I'd also think about doing more than just using a straight tube along the backs of the LEDs; when computers use water cooling, the blocks that go on the various chips have a certain amount of turbulence-inducing walls to increase the time the coolant spends in the block and the amount of time it can transfer heat away. If the tube is straight through, it might not be effective in transferring heat away.

If my calculations are correct (just using lumen numbers from datasheets), a single SST-90 at ~9A will put out as much light as 9 XP-G at 1A, so even with just 2 SST-90s you should end up with more light than your previous 12 XP-Gs.

How much more lumens are you trying to achieve? You could certainly achieve double the output with just 4 LEDs, and the hexagonal shape of the reflectors and stars lends itself to nice diamond shape without having to cut reflectors. And only having 4 LEDs means less heat to deal with and an easier job for your driver.

Hope that helps.

 

[HugeOne]

My xp-g are driven at 2A, so 3 SST-90 will be needed. Problem with my light right now is the spot lens that overexpose the image's center. I could replace the current carclo spots with ovals to solve it. I need a wide beam to cover the field of view of the cam with less height not to waste light in the sky. In order to get a good lighting I need to push the xp-g at their limits. Since their PCB's are thermal epoxied to the water cooler the thermal path is not optimal. One must had an air bubble behind it as it blew up the other day.

I need at least 6-7K lumens for night flight.
That's the reasons why I wish to change to big luminus, so they are not pushed hard and I can solder the thermal path right to the water cooler. I'll probably drive them at 5-6A most of the time, yet I could have a 9-10A boost mode for the sake of pure epic light :rock:

 

[CKOD]

I need at least 6-7K lumens for night flight.
That's the reasons why I wish to change to big luminus, so they are not pushed hard and I can solder the thermal path right to the water cooler. I'll probably drive them at 5-6A most of the time, yet I could have a 9-10A boost mode for the sake of pure epic light :rock:

I was thinking the same...use a dc-dc converter either of your making, or a premade POL unit to get the voltage to just above where your linear regulator needs it to be.

And the chipquick is a kit for desoldering SMDs, it comes with some flux, and the important part, the low melting point solder. I was suggesting just using that solder to attach the LEDs after you solder leads to them. That way you can pre-tin the LED, pre-tin your copper pipe, clean them, flux them both, and solder the LEDs down. You could have wires soldered on your LED already using lead free solder which would make it easier to not have your leads fall off as the solder on the pipe melts

 

[aEx155]

I need at least 6-7K lumens for night flight.
That's the reasons why I wish to change to big luminus, so they are not pushed hard and I can solder the thermal path right to the water cooler. I'll probably drive them at 5-6A most of the time, yet I could have a 9-10A boost mode for the sake of pure epic light :rock:

You could try mounting the LEDs in a shallow-angle V configuration, with 2-3 LEDs on each side. Take your square tubing, cut it at an angle, solder it back together after rotating one side 180 degrees (like cutting a miter on a piece of wood) and you have a V to solder LEDs onto. The advantage of this is you don't have to cut any reflectors and you can still spread the light out sideways more. Even better would be to use a curved pipe, but it's hard to find something like that with flat sides.

Going off my previous calculations, just 2 SST-90s at 9A would be ~6200 lumens, and 4 at 9A should theoretically be over 12,000 lumens. The same 4 at 5A should net you 7000 lumens. 2s2p for 7.4v, 10A with 4 LEDs seems reasonable to make a driver for, and it won't work your batteries as hard.

 

[HugeOne]

use a dc-dc converter either of your making, or a premade POL unit to get the voltage to just above where your linear regulator needs it to be.

I want to stay away from switching device, don't want to end up with lines in the video or reduced control range.

I'll definitely order that chipquick... tinning the pipe/fluxing the leds and then heat the pipe on skillet and fix the led in the pool, sound like a plan.

just 2 SST-90s at 9A would be ~6200 lumens

From the datasheet, it says 2250lumens @9A?

 

[aEx155]

From the datasheet, it says 2250lumens @9A?

The only rating in the datasheet that I saw was 1200 lumens at 3.15A, and the only place I saw "2250" was in the first bullet under Features: "Extremely high optical output: Over 2,250 lumens from a single chip (white)"

The "Relative Luminous Flux vs. Forward Current" graph shows a direct relationship between current and lumens, so I just did some calculations: If 100% is 3.15A, 9A should be ~275%, so 1200*2.57 = 3,300 lumens.

 

[MikeAusC]

The only rating in the datasheet that I saw was 1200 lumens at 3.15A, and the only place I saw "2250" was in the first bullet under Features: "Extremely high optical output: Over 2,250 lumens from a single chip (white)"

The "Relative Luminous Flux vs. Forward Current" graph shows a direct relationship between current and lumens, so I just did some calculations: If 100% is 3.15A, 9A should be ~275%, so 1200*2.57 = 3,300 lumens.

Lumens-per-amp reduces as you increase current. The "Relative Luminous Flux vs. Forward Current" graph is not a straight line.

The SSR-90 datasheet shows the lumens at 9 amps at 3.55 volts (32 watts), through a W65S WN SST-90 as 235% of 3.15 amp = 2820 lumens.

If you put 9 amps at 3.3 volts (30 watts) through 9 R5 Cool White XP-Gs you would get 139x2.5x9 = 3128 lumens

 

[aEx155]

Lumens-per-amp reduces as you increase current. The "Relative Luminous Flux vs. Forward Current" graph is not a straight line.

The SSR-90 datasheet shows the lumens at 9 amps at 3.55 volts (32 watts), through a W65S WN SST-90 as 235% of 3.15 amp = 2820 lumens.

If you put 9 amps at 3.3 volts (30 watts) through 9 R5 Cool White XP-Gs you would get 139x2.5x9 = 3128 lumens

The relationship is essentially linear, especially in the datasheet from Luminus:

 

[MikeAusC]

. . . but the straight-line approximation doesn't go through zero, so it's pointless doing a proportional extrapolation.

 

[HugeOne]

I read about the chip quick and it has a melting point of 136F, too low, any overheating and they'll fall off. I need someting at least above boiling point of water. Maybe by mixing it with lead free solder by twisting the two solder leads together... and then testing the proper ratio for my desired temp.

On the other hand, if the leds can be soldered with lead free, soldering them with leaded paste should not be too hard.

https://www.candlepowerforums.com/posts/3098815&postcount=336

That guy tested the SST-90 on his bench and he got 925lumens at 3.2A and 2136lumens at 9A. Look like it's not top bin but close. He predict 3000lumens at maybe 15-16A

 

[CKOD]

I read about the chip quick and it has a melting point of 136F, too low, any overheating and they'll fall off. I need someting at least above boiling point of water. Maybe by mixing it with lead free solder by twisting the two solder leads together... and then testing the proper ratio for my desired temp.

On the other hand, if the leds can be soldered with lead free, soldering them with leaded paste should not be too hard.

https://www.candlepowerforums.com/posts/3098815&postcount=336

That guy tested the SST-90 on his bench and he got 925lumens at 3.2A and 2136lumens at 9A. Look like it's not top bin but close. He predict 3000lumens at maybe 15-16A

Whoops, for some reason Ive been thinking 136C, not F. Yeah, falling off at 60C would be bad. Still, if you want to solder it on a hot plate after you put leads on, you could use Sn100 solder (pure tin) for the leads and Sn/Pb 63/37 for attaching the LED to the copper pipe. The pure tin melts at 230C, and the tin/lead at 183, a 60C window to hit shouldnt be that hard. Now that I think of it, Ive actually soldered 2 pieces of tin solder together with some 63/37 without melting the tin solder, so I know it should work. Sn100 might be hard to find, since there is newer and better out, but there is plenty of lead free stuff thats nowhere near as good as the real stuff, so should work for this purpose.



Sorry for the confusion.

 

[HugeOne]

Still, if you want to solder it on a hot plate after you put leads on, you could use Sn100 solder (pure tin) for the leads and Sn/Pb 63/37 for attaching the LED to the copper pipe

Or solder the leds on the pipe first, fill the pipe with water and then solder the leads.

My plan so far is to stick the leds on the pipe with solder paste, then heat the pipe on a skillet, monitoring temp with a probe inside the pipe. First, I'll do some practice run soldering IC on the pipe.

 

[HugeOne]

Here's my concept so far, 12 SST-50 on two row, SST-90 just have to high Vf to match them to li-po for my linear driver. FETs will be fixed to an aluminium plate on the back of the pipes.