Need (lots of) help on DIY light

[bradsab]

Yes, higher voltage means lower current = lighter wire (assuming you have proper switchmode drivers, definitely NOT with linear drivers). Nominally, 12s gives twice the voltage and half the current of 6s. The drain on each cell is identical, so battery life should be identical. The wire losses are lower, so you'd think that's an improvement...

But it ain't necessarily so. It turns out that most drivers are most efficient if the input and output voltages are close. This is really important in linear drivers like the AMC7135 based ones, where efficiency = Vout/Vin. It's less important in switching drivers, but still nothing to sneer at. Doubling the input voltage decreases wire losses, but increases driver losses. Since wire is cheap compared to drivers, the rule of thumb is to optimize for the driver, which for most drivers means keeping Vin/Vout as close to 1 as possible, without violating min/max input voltages.

There are times when you want to break the rule of thumb. To know when to do that requires thorough understanding of the tradeoffs. That means knowing not only the wire losses, but the driver efficiencies at the appropriate operating points. Since this is information you probably don't have access to, I'd go with the rule of thumb. Do you have an option for 4s3p configuration? That's the config that follows the rule, for most drivers.

In a week or so I might be able to get you two prototype 7A drivers. See this link if you're interested High current (SST-90) driver - CandlePowerForums It would work best with 4s configuration.

D

One of the reasons I'm looking at these higher voltages is the attempt to drive one SST-90 in dive light mode, and then two SST-90s in video mode. If I run two LEDs in a series, the Vf doubles, doesn't it? I've been following two threads on CPF developing constant current drivers, and my head design has room for two of said drivers. BUT...

Your driver adds a new twist. It's small enough I might be able to get three of them in the head. It depends on the heat sink area and overall ht. Would you have a footprint drawing?

A real handicap is I just don't understand enough about drivers to know what to do. I'd like to drive each SST-90 with 3.7 - 3.8v on high. Per this thread White LED lumen testing - Page 11 - CandlePowerForums that would give me 1580 - 1965 lumens; X2 for the video floods. Will a linear driver hold the Vf constant?

The other issue is I'm running out of time. I don't get to dive very often, and I have a trip coming up the first week of March. I've been putting off the build trying to decide what to do. I was going to build the canister today, but now I'm not sure how big to make it. I can cut down my current design and make 4s2p, but then I don't have the option to direct drive if the drivers don't come thru in time.

I attached some drawings of the light and canister so you have a visual of what I'm playing with. I'm having a lot of fun with this, just hope it gets done in time.

 

[bradsab]

I may be complicating this more by trying to use the toggle switches I have. They're rated at 15A @ 120VAC, but there's hardly room for them. If I could use mini toggles rated at 5A @ 120VAC or 28VDC, I could use a 3 or 4 pole switch and switch one of the drivers output from the spot LED to one of the flood LEDs. That way I would only need two drivers. Don't think I'll have enough room in there for 3 drivers, but with the smaller switch, I could share a driver between two SST-90s. Will this switch safely carry two SST-90s? Digi-Key - CKN1499-ND (Manufacturer - 7303TZQE)

 

[DIWdiver]

...Will this switch safely carry two SST-90s? Digi-Key - CKN1499-ND (Manufacturer - 7303TZQE)

Well, that's a tough question. As a professional, I would say 2x9 = 18A. 3x5=15A. Switch is not rated for that. However, as a hacker, it's tempting to say it's awfully close, so yeah it would probably work but the switch wouldn't have the rated 1 bazillion operations.

But maybe that's not the right question. Two of these switches could handle high/low/off and dive/flood, within specs, if wired properly. And it would only take two drivers. Actually, you could use less expensive switches too. One 7301TZQE and one 7203TZQE would save you a buck or so.

In the pic L1 is your dive light, L2 and L3 are your floods. SW1 (not labeled) selects dive/flood, SW2 selects hi/off/low. SW2 carries low current, so could be almost any switch that fits. This would work for the driver I'm about to be offering. For other drivers the diagram might change a little, but I would think it could be worked out. I could help with that, and I'm sure others would as well.

D

 

[DIWdiver]

One of the reasons I'm looking at these higher voltages is the attempt to drive one SST-90 in dive light mode, and then two SST-90s in video mode. If I run two LEDs in a series, the Vf doubles, doesn't it? I've been following two threads on CPF developing constant current drivers, and my head design has room for two of said drivers. BUT...

Your driver adds a new twist. It's small enough I might be able to get three of them in the head. It depends on the heat sink area and overall ht. Would you have a footprint drawing?

A real handicap is I just don't understand enough about drivers to know what to do. I'd like to drive each SST-90 with 3.7 - 3.8v on high. Per this thread White LED lumen testing - Page 11 - CandlePowerForums that would give me 1580 - 1965 lumens; X2 for the video floods. Will a linear driver hold the Vf constant?

The other issue is I'm running out of time. I don't get to dive very often, and I have a trip coming up the first week of March. I've been putting off the build trying to decide what to do. I was going to build the canister today, but now I'm not sure how big to make it. I can cut down my current design and make 4s2p, but then I don't have the option to direct drive if the drivers don't come thru in time.

I attached some drawings of the light and canister so you have a visual of what I'm playing with. I'm having a lot of fun with this, just hope it gets done in time.

Wow! Cool 3-D CAD work!

Design looks very nice. Maybe you can build my next light.

The fins would help a little in air, but are probably unnecessary in water.

Before going further, I want to say that LEDs are really current devices, not voltage devices. You should aim for a particular current, not a particular voltage. At 9A, you should get the specified lumens, no matter what the voltage is (and it will vary a little from one LED to the next).

As far as the configuration, can you build a canister for 4s3p or 4s4p and put dummy cells in if necessary to get what you finally settle on? That way you could change on the fly and adapt to driver changes if need be.

Yes, having two LEDs in series does double the Vf. Some drivers can handle this and some can't, at least not well.

The nomenclature of 'true constant current' is not as definitive as the originator intended. Both switchmode and linear drivers can be constant current. As far as I'm concerned there are four kinds of drivers, in order of decreasing desirability: Switch mode constant current, linear constant current, resistive, and other <expletive deleted>. Direct drive is actually a subclass of resistive, as the currents are determined primarily by resistance of wires, cells, and LEDs.

The difference between switching and linear constant current drivers is what happens to the efficiency as the input/output voltage ratio changes. With a switch mode driver, the efficiency can remain relatively constant over a large range. With a linear driver, the efficiency is fixed at Vout/Vin. Both should hold the current constant, when operated within specified limits.

The individual building the 'true constant current' driver is (at least as far as we can tell) building a really high quality switch mode driver that's adaptable to various input and output voltages with high efficiency. If that's what you need and you can wait for it and afford it, it's probably the best thing going.

As far as the footprint of my driver, I don't have any drawings yet. The prototype is 1.25" square, 0.32" thick, and requires a flat surface to attach to, either with one screw or thermal epoxy. The only difference in the production version is that it will be 1" diameter.

D

 

[bradsab]

Not sure I follow your diagram. Wouldn't a dpdt switch work for switching from the single spot LED to the twin floods? If I can wire it per my attached drawing, I could probably find a toggle rated at more amps that will fit. If I have to use a 3 pole, mini is the only thing that will come close.

 

[bradsab]

Well, DUH! Your drawing has the switch solution (or half of it). There is room for a 4PDT mini toggle, so I'll parallel two poles per LED giving me 10 amps. Question: when the high/low toggle is switched to off, is there any current draw from the batteries?

 

[bradsab]

The fins would help a little in air, but are probably unnecessary in water.

The fins have two purposes: 1. to save a little weight 2. I may make two of these and use one for a bike light. For a bike light, I'd use tighter reflectors instead of the wide floods. Not that I ride a bike at night... or anytime for that matter. I'm just having fun with this.

Before going further, I want to say that LEDs are really current devices, not voltage devices. You should aim for a particular current, not a particular voltage. At 9A, you should get the specified lumens, no matter what the voltage is (and it will vary a little from one LED to the next).

So, your driver will hold the current at a constant amperage (say 7A) until the batteries fall below that? What happens then? From what I'm reading, it sounds like true CC drivers abruptly shut down, while linear drivers continue to burn at lower outputs until the batteries are depleated.

The nomenclature of 'true constant current' is not as definitive as the originator intended. Both switchmode and linear drivers can be constant current. As far as I'm concerned there are four kinds of drivers, in order of decreasing desirability: Switch mode constant current, linear constant current, resistive, and other <expletive deleted>. Direct drive is actually a subclass of resistive, as the currents are determined primarily by resistance of wires, cells, and LEDs.

The difference between switching and linear constant current drivers is what happens to the efficiency as the input/output voltage ratio changes. With a switch mode driver, the efficiency can remain relatively constant over a large range. With a linear driver, the efficiency is fixed at Vout/Vin. Both should hold the current constant, when operated within specified limits.

The individual building the 'true constant current' driver is (at least as far as we can tell) building a really high quality switch mode driver that's adaptable to various input and output voltages with high efficiency. If that's what you need and you can wait for it and afford it, it's probably the best thing going.

If I have a 4.8V pack, and we drive at 7-8A, that's 75-79% efficient, right? As my 4.8V drops, won't I become even more efficient? Ni-MHs are supposed to remain flat, but hooked up the the SST-90s on the bench, the voltage drops to around 4.3V... 87% efficient? I doubt a switching cc driver would be much more than 90% would it? Are there any other trade offs other than the efficiency?

 

[Codiak]

I'll finish my prototype video light today which is fixed focus with a single SST-50.
(Final build will be quite different.)

However,
The SST-50 and SST-90 have a 120 degree absolutely smooth beam without reflectors or lens. I can't even discribe how the 50 floods a room with light though I will be posting shots on CPF soon. According to Packhorse the 90 will be twice as bright. Doubled again with a second lamp and wow I'll blind everythin on the reef.

The point is: For Video with 3 SST-90s you won't NEED any optics other than a protective window.

 

[bradsab]

I'll finish my prototype video light today which is fixed focus with a single SST-50.
(Final build will be quite different.)

However,
The SST-50 and SST-90 have a 120 degree absolutely smooth beam without reflectors or lens. I can't even discribe how the 50 floods a room with light though I will be posting shots on CPF soon. According to Packhorse the 90 will be twice as bright. Doubled again with a second lamp and wow I'll blind everythin on the reef.

The point is: For Video with 3 SST-90s you won't NEED any optics other than a protective window.

I'm using Ledil Lily wide reflectors for the floods. About 60 degrees with a fair amount of spill outside of that. 120 degrees would be well beyond the camera's FOV. Please post a link to your CPF thread.

 

[DIWdiver]

The fins have two purposes: 1. to save a little weight 2. I may make two of these and use one for a bike light. For a bike light, I'd use tighter reflectors instead of the wide floods. Not that I ride a bike at night... or anytime for that matter. I'm just having fun with this.



So, your driver will hold the current at a constant amperage (say 7A) until the batteries fall below that? What happens then? From what I'm reading, it sounds like true CC drivers abruptly shut down, while linear drivers continue to burn at lower outputs until the batteries are depleated.



If I have a 4.8V pack, and we drive at 7-8A, that's 75-79% efficient, right? As my 4.8V drops, won't I become even more efficient? Ni-MHs are supposed to remain flat, but hooked up the the SST-90s on the bench, the voltage drops to around 4.3V... 87% efficient? I doubt a switching cc driver would be much more than 90% would it? Are there any other trade offs other than the efficiency?

Yes, the NiMH cells will drop from about 1.2V to about 1.0V per cell during discharge. You can run them even lower, but it gets to be hard on the batteries. Battery folks generally consider 1.0V to be the end of life. In an emergency though, you can run them down to 0.8, even 0.5V. But the voltage starts falling faster and faster!

Yes, at 4.8V input and 3.6V output, you have 75% efficiency. At end of life (around 4.0V) you have a wopping 90% efficiency!

Yes most linear drivers will hold the current constant until the batteries reach some particular voltage, then will start to drop. The driver will always need to have some voltage across it, and mine is designed to keep up to 10A constant current down to 4.0V input, 3.6V output. It would continue to provide output as the voltage drops, down to at least 3.0V. After that I wouldn't be comfortable promising what would happen, but it would probably work down to 2.0-2.5V. At lower power settings, the minimum voltage is a little lower.

Any driver can be designed to act this way, or to cut off rapidly or even instantly. It's all in how it's designed. You're right to think that a switching driver is more likely to be designed to cut off rapidly or instantly, but you have to check the specs on the particular driver to be sure.

Switching drivers can achieve 90-96% efficiency over wide ranges (I've done it), but you have to make a lot of compromises to fit them in the tiny packages that light builders want. I think those compromises are going to make it hard to beat 90%. It would take careful design and probably expensive components to even make that.

But who knows? Maybe someone will soon make me eat my words!

D