I have been prototyping custom dive lights and slowly learning about LEDs and drivers in between flooding and frying components. The current light is a cree XM-L2 driven off a single protected 3.7 Li-Ion cell. Reliability is huge because the final lights are fully sealed and potted, and the build process is quite labor intensive so any type of component fails the light is bricked and time/money is lost.
The last one used a Nanjg 105C 1.15A driver (from Mountain Electronics). Worked pretty good but started to flicker after some hours of use at depth even though battery showed 3.25V charge remaining. To be fair, I think saltwater may have reached some components. I have a Nanjg AK47A on order.
Also because of the design requirements, the LED is about 6” from the driver. Is that length of wire problematic for the PWM or likely to introduce noise? I know that most flashlight pills are designed so the driver is contacting the LED but don’t know if this is convenience or electrically necessary.
Does anyone have any recommendations for other drivers to try? Again, I interested primarily in reliability, followed by efficiency and cost. Ultimately I want to go with either a single mode driver, or two mode (50%/100%/with memory) driver. Does flashing custom firmware on the chips effect the reliability? I don’t want anything experimental as I will ultimately be making a small run (70 units) for sale, and would like to offer some kind of warranty.
Thanks! I will post pics as development moves along.
I would get with Randy from PFlexPro he builds some very reliable lights and has the analytical mindset to help you through any issues. He pots all his drivers and would be able to tell you about reliability of flashed firmware.
At 3.25v your battery is flat and the driver is doing as it should tripping the low voltage protection. Consider a multicell flashlight for increased runtime. See HKJs graph here.
You can have the driver firmware customized to do just about anything imaginable, like lower the low voltage settings. 3.25v is dead empty though, under load the voltage was around 3 volts which is where the 'standard' protection starts to cut in. You can probably safely run it down lower than that but so close to empty there is no current left for the cell to give. You need more than one cell if you are running it down that low in one use... and being LiIon and not subject to memory effect, it should be recharged after every use (or a fresh cell swapped in).
Is this the only light in use, or is it a must-not-fail emergency backup? It's a Bad Idea to expect a single light to cover both duties. If it's a backup, why was the cell run down that low in the first place? If it's the primary light, why does it have only one cell?
Welcome. Need more or bigger cells, or a canister set up. Small amounts of resistance(switch, spring, thin wire, etc.) cause voltage drops that show up more when the cell is depleted. Eliminate as much of this parasitic resistance as possible in your design. The cells are more at risk in this application than a fully potted driver.
@Dru_dragon and comfychair - Okay, makes sense. Kinda bummed, I hacksawed the light apart so I could get at the LED leads and test voltages, etc. The entire system is cast inside solid urethane resin (including cells; the only the way in is the charging port, which is disconnected whenever the LED is switched on). That’s the bad news… good news is that then it might have not leaked after all which is more of a concern to me. I understand about multiple cells but the light is actually integrated into a speargun muzzle so for design reasons it has to be very compact, needs to have a very high mechanical strength, and has weird design parameters so that the loading bands fit properly. It’s pretty specialized, for spearfishing in reef/cave structure and wrecks. I promise to post up pics later but the project is kinda on the DL until I am ready to go to market, building second round of molds hopefully in the next week. I ran it down as far as it would go since I try to test my prototypes until they break. Usable battery life was a good 2-3 hours on high power. I used it on multiple dives so I don’t have a firm number, it was the first ocean ready prototype.
@Rufusduck - since my focus is freediving/spearfishing I haven’t been developing cannister lights—although I do have some video lights I am working on that run off external batteries—because they are rather unwieldy. Focus is always on making everything smaller and more durable which is always the main thing for freediving equipment, but cave divers would probably also appreciate the direction. The video stuff I am building is mainly for my own purposes, it’s a pretty crowded market. I may use some larger gauge wires in the next iteration, I think these were 24 gauge.
@Remang - I took a look at that site but will give it a closer look, thanks.
All the lights I have made so far have been fully potted, including the batteries… I have been 3D printing the housings, putting in components and then filling with thermal epoxy. The lights in question are cast in silicone molds since they are not so much one off’s.
As a diver myself I appreciate the clarification. How did you find visibility off axis? Would it help to have an alternate light source to illuminate more than just where the spear gun is pointed? Is it possible to double up the cells in line but still wired in parallel since linear drivers are cheaper and less complex than buck drivers? A one hour surface interval is not likely to be enough time to recharge a single cell but two would last through a day’s bottom time it still recharge overnight.
Viz off axis is pretty good. You could hold a secondary light in your left hand but then it is more stuff to hold, I prefer to have my light on the gun, I mean you can’t shoot the fish if the gun isn’t pointed at it, although at night I have sometimes used a gun light and small light in the left hand. The typical solution is usually a janky bracket on the tip of a speargun to hold a little flashlight but then the balance/tracking of the gun in compromised and the beam is off to the side of the spear. I had to custom make the reflectors, it is a fairly narrow beam and no hotspot since the reflector is also potted in clear resin. Since it is designed for hole hunting the requirements are a little different than for a regular dive light, you could use it at night but it’s really to help you see into cracks and in murky/deep water during the day. Here in southern California if you want to look in a hole past 40’ it helps to have a light. For hunting fish, you actually don’t want a ton of light—if it penetrates significantly further than the effective range of the speargun (8’ - 20’ depending on design) it’s not good because then you just spook fish, and a warm light is preferred, tends to be less spooky to sea creatures. At 50% power the light was as bright as the little 4xaaa light I used to have mounted on that gun, I think at 50% power it would get 5+ hours. I am using warm 80+ CRI LEDS which is really nice, and atypical of dive lights unless it is a specialty video light, I’ve used a lot of dive lights and I really like the way things look under this one. The other thing is that very small dive lights tend to be poorly made, and no matter what they are rated they flood. I regularly hunt and dive deeper than many guys will even scuba and feel like freediving is much harder on gear because of the constant rapid ascents/descents. As far as multiple cells, I am probably going to work with a friend on making some speargun barrels more for East coast type wreck diving where a longer gun is used, and the cells would be in the rear of the barrel and it would be easy to have two or even three. Too many cells up front would be bad for the balance/tracking of the speargun. For these I want to sell the muzzles as an easy swap kit so they need to be very simple and self contained.
We dive every year out around the Channel Islands late May early June and for the most part wouldn’t see much at depth though there are rare days when you can still see the disc of the sun at 80’. If you’re altering the handle you could do a reed switch as a tape switch alternative and locating multiple cells there strikes me as a good idea.
Not a diver so this may not be feasible, but a physically separated cell pack run with heavy wires going to the light module would offer less alteration of handling characteristics. Should solve the multi-cell problem though it would double the things to attach.
Yeah, you guys a basically talking about a can light in a speargun; I have built one those, it is a bit complicated because of the separation of all the components, lots of waterproofing and wire runs… I will probably work on a better system for those later, that is the barrel light I planning to do. Currently I use a mechanical switch design, it is itty bitty, really robust, vibration proof (can live forever in a mesh dive bag that’s knocking around in the bottom of the boat without switching itself on) and reasonably cheap for me to put together. I did experiment with reed switches but in my experience they are kind of touchy and tend to become magnetized, also messed with hall effect stuff but they have parasitic drain and it is larger than my mechanical switch, more crap. The hall effect + optocoupler had some potential as a bullet-proof underwater dimmer circuit, I got something kinda working on the bench but didn’t go further with it. That would be more of a video application. Waterproof pots are really bulky and expensive. For the freedive/spear stuff durability+miniaturization are the main things.
Thanks, yes, lots of headaches to make them but once made they are super solid.
The light is an 80+ CRI 4000k emitter. They are a little harder to source than typical dive LED’s. I really like the way they look though; I have used a lot of dive lights and have come to conclusion that if I’m going to go to the trouble of getting down there I want a quality light. On the video for many shots we are deep enough to lose all red from ambient light, so the warmer LED really looks and feels nice since it puts those reds back in, making the footage pop.
When you want the ultimate in reliability, you need to go as DD as possible. For me, that’s always been 1-mode, NO driver chip, just a few parallelled 7135s for the desired current.
DD, or DD-through-resistors (sub-ohm), would be the most reliable, but with drastic differences in current from a freshly-charged 4.2V cell, to 3.7V at 90%, then linearly down to 3.0V of whatever’s the Vf of your LED.
I prefer the 7135-only approach, as it limits current to about the same exact level through most of the battery’s charge. And if one 7135 gives it up, the others are still parallelled and will continue driving the LED, albeit at less current (eg, 1400mA —> 1050mA if 1 of 4 7135s goes).
A controller chip can go mental and start in weird modes, or just refuse to work. I keep mentioning this, but my ’98 went mental and now starts in low instead of high, so I have to flip through strobe and sos to get to high. Oddest thing I’ve seen.
So, skip the controller and just keep the parallelled 7135 regulators, is my vote.
