Making a work lamp

Hey guys, I work in aircraft production and have to spend quite a lot of time inside the wing tanks, the torches we're provided with are bayco nightsticks (http://www.emergencyequipmentnews.net/2008/06/14/bayco-nightstick-new-models/ the second one on this page)

And they're near useless.

The bayco claims to be 400 lumens in flood mode (the four leds on the side of the tube) ideally i need something at least twice as bright.

The space working inside is kinda the same size as the airconditioning vents you always see people crawling through in movies for some idea of the situation the torch will be used in. So i dont really need a long throw just a good clear wide flood for inside the tank.

The tube with the leds on the side works very well as a design as it alows the torch to be laid between the stringers and aimed at the place youre working in so going to keep that kind of shape, maybe a pipe the leds out one side with caps either end with octagonal plastic/rubber bumpers so it can be angled at a good range of angles and not roll.

The other big requirement is battery life, i need it to last a solid 12 hours. (The bayco lasts about 3 before needing a new battery, which are all screwed from abuse).

Also due to the werid angles you often find yourself to relative to the torch I'd prefer not to use PWM to control the brightness because the flicker can drive you mad when its at the extremes of peripheral vision. (We have some cfl lead lamps and thier flicker is infuriating)

I'm reasonably competent at making things both mechanical and electrical but looking at LEDs there just such a wide range that i have no idea where to begin.


I was hoping you guys could help me work out what type and number of LEDs would be best (more dimmer leds better than fewer bright ones?) Along with the kind of driver and batteries that could give the kind of life i need.

Optics are something i do have zero knowledge of so advice there would be great, but im thinking as its a short range flood they wont be that important.

Any advice you could give would be a great help

I have been thinking of making a work lamp too. I wear a hard hat all day and I want something that I can plug an external battery pack into for super long life. Can't stand my regular light dimming with in the first few hours of work.

Make sure to post pictures of your project. I would love to see what you come up with.

Dale P.

It sounds like you're in kind of the same boat as me--I do a lot of hobby work on race cars and need a GOOD flood light. What's on the market just doesn't cut it.

So, let's brainstorm a little. I know from experience that a triple XPG mule run by a 18650 makes a great hand held flood. Thinking out loud (and being somewhat ignorant) would a light that contained maybe two to three XPG triples in a traditional flood light "stick" work? You'd obviously need more than a single 18650 to power them...maybe dual 26650s? I mean one XPG triple makes more than 400 lumens--it's just so floody the human eye can't pick up the real lumen count.

Thoughts?

I also have an issue because I weld often and my eyes must be getting old because even with my auto darkening helmet I have trouble seeing the weld path. So, something like we're talking about would be a tremendous help welding AND in repair.

800 lumens for 12 hours is a lot of battery power as 1000 lumen lights run about an hour or two off 18650 batteries you would need 6 to 8 of them or perhaps an easily swapped external battery pack system. Depending on the angle of flood you may be able to get by with a single XPL or XML-2 emitter but usinge several emitters would improve the runtime as driving them at lower current is more efficient and at low currents the XPG-2 or 3 or XPE-2 series would probably be about as efficient. It is quite possible that the Bayco is overrated on output and that a 500-600 lumen light would suffice well enough and it looks like with that output you can get close to 12 hours with 6x18650 batteries. Something like the Fenix HP-30 Headlamp as it uses an external battery pack I think is unpluggable you could design a larger battery pack for it.

Because of the eye's logarithmic sensitivity to light, if you think you want twice as much light, you probably really want four times as much.

I've built a number of drivers over the years. The first few had many steps with factor of two power per step. After using one of these for a while, I switched to 100-30-10% settings. My most recent has only two levels - 100% and 25%.

Most flashlights are overrated, and the less reliable the manufacturer, the more they are overrated. The first thing they do is look at LED lumens (what the LEDs actually put out). What actually makes it out of the light is generally 10-30% less than that because of losses in lenses/reflectors. It is my belief that only manufacturers that cater to technically sophisticated customers actually rate lights for lumens that actually make it 'out the front', and they almost always claim 'OTF lumens' and not just 'lumens'. Thus, I think we should consider Bayco's claim to be in LED lumens.

400 lm from modern LEDs should take around 3W to generate. To run for nine hours would be around 27 W-H, which can be had from 3 good 18650 cells.

Another thing that dicey manufacturers do is to fudge the runtime. They run the light on full power, but don't end the test when light output starts to drop. They either end the test when the light output has reached some arbitrarily low value, or they don't even run a test at all. Some battery/driver combinations will run for many hours or even days with light outputs that are tiny and useless.

Based on the size of the light, I don't think Bayco's claim of 9 hour run time is necessarily stretching the truth this way. That light might well be able to house 27 W-H worth of NiMH cells. It's at least close. If the battery is in good condition. A badly abused battery will not come close to meeting this claim.

So back to designing a new light. I think you are going to want 1000-1600 LED lumens. For flood lighting you will want a shallow to medium depth reflector depending on how wide a flood you want. Narrower flood means deeper reflector. For 1000 lm and 12 hour runtime, you need about 90 W-H of battery. For 1600 lm you need about 144 W-H. At the moment the best choices for packs of this size are:

1. NiMh AA cells. Around 3 A-H at 1.2V gives 3.6 W-H per cell. You'd need 25-40 of these cells. They are economical and very safe, but heavy and don't hold a charge a long time.

2. NiMH AA LSD (i.e. Eneloop) cells. LSD stands for low self discharge. These cells can hold a charge for a year or more, which normal NiMH cells can't do. Around 2 A-H at 1.25V gives 2.5 W-h per cell. You'd need 36-57 of these cells. They are very safe and hold a charge a long time, but large and heavy.

3. NiMH D cells. 10 A-H at 1.2V gives 12 W-H per cell. You'd need 8-12 of these cells. They are very safe, but large and heavy.

4. LiIon 18650 cells. 3 A-H at 3.7V gives 11 W-H per cell. You'd need 9-13 of these cells (though 12 or 15 would be a lot more practical than 13). They are much lighter than NiMH cells, but you have to be very careful to buy good quality cells. There are many crap cells available that don't come close to their claims. LiIon cells will hold a charge a long time, but are more easily damaged by abuse. They also pose a much higher risk of fire.

5. LiFePo4 26650 cells. 3.3 A-H at 3.2V gives 10.5 W-H per cell. You'd need 9-14 (though 15 would be more practical than 14). These are larger than LiIion or NiMH cells of comparable capacity, but they offer better safety and abuse tolerance than LiIon and less weight than NiMH.

These numbers may be a bit disencouraging. There's also the possibility that, because of the age of the battery packs, the light just don't deliver the stated 400lm anymore, even with a freshly charged battery, because they don't deliver the necessary voltage/current anymore. (Or even that the leds don't deliver their initial output anymore if they haven't had enough cooling). So possibly, we don't need to talk about 1600lm.
Also, could you possibly be happy with a (slightly) decreased beam angle?
400lm doesn't sound that bad to me, if you're in a narrow place. Or do you have to look at both, things that are near, and others farther away?

Hey sorry been away for a bit so not had time to look at this.

Yeah the batteries for the baycos are ****ed and so abused they're worthless.

As im not holding the lamp or wearing it weight isnt so much an issue. But tube diameter could need to be below 2 inches. But i think that gives a lot of room to pack in cells.

I was thinking instead of clustering the leds in one place to spread them out to improve the coverage, throw isnt important as due to the confined spaces the lamp is never more than 1ft from the work.

Sort of

|--V-----V-----V--|

With the V being the leds on the tube

Righto aome fag packet maths.

Cree xp-g3 (used because i think its the most efficent led atm)

190 lumens at 350mA

So 4 should be 760 lumens so a nice bump and i can spread them around a bir on the stick.

4x350mA = 1,400mA
X10 hours =14,000mAH required.


I was thinking of buying an anker 20100 power bank to harvest the cells from (they cost 20 quid and contain 6 panasonic 3,400mAH cells).

So 6x 3400 = 20,400mAH in the cells

So that gives me aprox 30% headroom for efficency losses, but given the size of the torch and the cells i could probbaly easily double that to 12 cells

Am i right here or wildly off base?

You may consider the XP-L is it is a larger die would be floodier and should be as efficient if not more so.

Ok xool ill look at that now.

Was thinking 12x3400ma 18650 wired up as 6 sets of 2 in series in parallel (6 pairs of 2).

http://www.ledsupply.com/led-drivers/boostpuck-dc-boost-led-driver

Paired to that driver. This was for 4 xp g3 leds though.

Drivers are something im totally lost on ill admit

What kind of amperage would you run the xpL at?

Looks like at 350ma they produce 200 lumens so more efficent!

But cree has the typical drive current at 1050ma which may be waay to much to keep going for 10 hours (plus heat may be an issue as i csnt have ANY external metal

Would you keep it at 4 x 350 leds, or would you use less with higher drive current?

Tefal said:

What kind of amperage would you run the xpL at?

Looks like at 350ma they produce 200 lumens so more efficent!

But cree has the typical drive current at 1050ma which may be waay to much to keep going for 10 hours (plus heat may be an issue as i csnt have ANY external metal

Would you keep it at 4 x 350 leds, or would you use less with higher drive current?

Click to expand...

I would say that you could get 3 or 4 LEDs and hook them up and experiment with the output and you may find that several output levels would be better a higher output for short times and lower output for conserving battery life concentrating on a medium mode for your most used output.

Tefal said:

Righto aome fag packet maths.

Click to expand...

What?:thinking:

Yes, the math looks okay. The numbers are about right even if the formula isn't exactly. Doing it properly the numbers are even a bit better. That driver should be good, too.

I would run the cells in 3S4P and 5 LEDs instead of 2S6P and 4 LEDs. That will get you a lot better efficiency from the driver (there's a chart in the datasheet of efficiency vs input voltage) and slightly better from the LEDs. I don't think that driver will work with 3S cells and 4 LEDs, especially if they are underdriven.

Any LED gets less efficient the harder you drive it. So in general, at any given drive current a bigger LED will give you more light, and more LEDs with the same total power input will give you more light. Of course this rule only holds if you are comparing LEDs of similar technology, or at least of similar efficacy ratings.

The reason the big guys don't do this is price. But if you are building your own light, and are willing to pay for efficiency, it's the way to go.

Heat is a concern, but not a huge problem. You are putting about 350mA*3V*4LEDs=4.2W into the LEDs (or 280mA and 5 LEDs for the same power). Something like 60% of that is turned to heat, so about 2.5W. A quick calculation says that a polycarbonate tube 3cm diameter, 2mm thick, and 25 cm long would see a temperature drop of less that 1.2 degrees C at 2.5W. Other plastics should be in the same ballpark. As long as you distribute the heat well within the tube you should be fine.

If anyone wants to check my calculation, I looked up the thermal conductivity as 0.19 W/m-K.

DIWdiver said:

What?:thinking:

Yes, the math looks okay. The numbers are about right even if the formula isn't exactly. Doing it properly the numbers are even a bit better. That driver should be good, too.

I would run the cells in 3S4P and 5 LEDs instead of 2S6P and 4 LEDs. That will get you a lot better efficiency from the driver (there's a chart in the datasheet of efficiency vs input voltage) and slightly better from the LEDs. I don't think that driver will work with 3S cells and 4 LEDs, especially if they are underdriven.

Any LED gets less efficient the harder you drive it. So in general, at any given drive current a bigger LED will give you more light, and more LEDs with the same total power input will give you more light. Of course this rule only holds if you are comparing LEDs of similar technology, or at least of similar efficacy ratings.

The reason the big guys don't do this is price. But if you are building your own light, and are willing to pay for efficiency, it's the way to go.

Heat is a concern, but not a huge problem. You are putting about 350mA*3V*4LEDs=4.2W into the LEDs (or 280mA and 5 LEDs for the same power). Something like 60% of that is turned to heat, so about 2.5W. A quick calculation says that a polycarbonate tube 3cm diameter, 2mm thick, and 25 cm long would see a temperature drop of less that 1.2 degrees C at 2.5W. Other plastics should be in the same ballpark. As long as you distribute the heat well within the tube you should be fine.

If anyone wants to check my calculation, I looked up the thermal conductivity as 0.19 W/m-K.

Click to expand...

Haha, sorry my typing on the phone is terrible back of a fag packet maths = quick and dirty workings like youd scribble on abit of card from a ciggeret packet guessing thats not a thing in the USA

Thanks for the tip on the 3s4p and 5 leds for efficency.

But am i right in thinking that by using 3s4p instead of 2s6p id be cutting the capcity down from 20,400mah to only 13,600mah and thus killing my chance of getting it to 10-12 hours especialy with the extra led?

I figure i can put some heatsinks in and maybe drill a few holes for airflow/exchange over the heatsink and seal it off from the rest of the tube so nothing can get to the electronics.

In the US we use the 'back of an envelope'. And 'fag' is a sexual slur.

Because you are using a switching regulator, the current coming from the battery is different from the current going to the LEDs. The proper way to do the calculation is to compare energy, not current capacity.

The 5 LEDs in series will have a total voltage drop around 15V. At 280 mA they draw 15V * 0.28A = 4.2W. To run for 12 hours you need 4.2W * 12H = 50.4 W-H.

Each cell has about 3.7V * 3.4A-H = 12.58 W-H/cell. So you need 50.4W-H/12.58W-H/cell = 4.0 cells. To account for losses in the driver, you probably need 5. Any convenient number of cells higher than 4 should get you at least 12 hours. You have to arrange the cells and the LEDs in an arrangement that works for the driver. Placing the cells in 3S configuration works really well with the driver you chose, but you can't put 5 cells in a 3S arrangement. Six cells in 3S2P works great. It gives you ample overhead for efficiency of the driver and aging of the cells.

Now you look at the output requirements of the driver. The data sheet says you need the LED voltage to be at least 3V higher than the battery voltage. Right off charge, the battery will be about 4.1 V/cell, or 12.3V total. So the LED voltage should be at least 15.3V. Because the cell voltage will very rapidly drop to 4.0V, or 12.0V on the battery, I'm thinking 15.0V should be enough for the LEDs. At about 3V per LED, that's 5 LEDs.

If you use the latest LEDs, and seriously under-drive them (like 280 mA in an XP-L), you may see that they don't even drop 3.0V. Either use 6, or live with the consequences. I'm guessing that the consequences would be that the driver would just be less efficient until the battery voltage drops a bit and you get your 3V margin.

By the way, if you were using a LINEAR driver, like any of the cheap and plentiful AMC7135 based ones, the battery current is virtually the same as the LED current, so calculating amp-hours gives the correct result, and calculating energy does not unless you figure in the losses in the driver, which can be considerable.