Introduction: Mega Torch - the Ultimate Guide

Mega Torch Build!

Welcome! This instructable has been 'written' mainly as a set of detailed videos (10 in total including the 'fake' advert above). As such, you will find the text is sparse but, this is more than made up for during the in depth descriptions in the vlogs.

I hope you enjoy this build and hope you can find the time to subscribe to my YouTube channel.

link to my YouTube channel

In this series, each video is between 5 and 15 minutes long and so its time to get a nice cuppa, sit down and lets get started!

Happy making!

Step 1: Safety

Picture of Safety

This is a low voltage project and so you'd be hard pressed to get electrocuted. On the other hand working with any batteries can present a danger as they can provide a lot of current in a short period of time. So if you accidentally short out some part of the circuit there is a likelihood that parts will get hot/melt very quickly so....

Please be careful!

In addition, when the project is finished please ensure you charge the lithium battery with one specifically designed to do the job. I recommend a GENUINE imax B6. These can be bought online but beware most on ebay/aliexpress are fakes. Buy from a reputable source if you are unsure. Here is the manufacturers link:

Step 2: Parts List

Picture of Parts List

The parts used in the project are varied and it might take a while to collect them together from the various sources but hopefully they are all easily obtainable in your local hardware shop or via eBay. I've submitted a pdf and an excel spreadsheet with direct links for all the purchased parts.

A few of the main parts are salvaged from old PC's and laptop batteries. There are a couple of videos in the steps which go into all the necessary details on how to salvage used 18650's from old laptop batteries. Not all will be good enough (capacity wise), so you may need to source more than the 18 pieces you need before you have ones which will be acceptable to build in a pack.

Step 3: 100W LED's - the Good the Bad and the Ugly!

This is the first video and it looks at a small selection of 100W available on the web. Some are OK and some are not! 100W LED's from a tier 1 manufacturer would be too expensive for a home build (typically $80 @ 2017 prices) and so, at best, the ones I tested here are seconds. What does this mean in practice? Well the LED's may have broken/shorted LED's in the matrix or possibly there are other manufacturing defects which will shorten the life. It's a bit of a lottery really you have been warned! However, perhaps I was lucky as 3 out of the 4 I purchased were 'good' so I'm not complaining!

This seems a good source: (well the 2 I got from them worked ok)!

Step 4: Recovering Laptop Batteries

In this part I take a quick look at most of the parts that will be used in the build then get stuck in pulling the old laptop batteries apart and testing the 18650 cells. This can also be a bit of a lottery but if you have obtained these from a free source then what can you loose?!

If you are in the market for new cells then I'd suggest these 2 websites (I have no affiliation to either). I think the LG Chem MJ1 3500mAh cell is the best price/performance/capacity at the moment (Feb 2017)

NB: I do not recommend ebay as a good source for batteries (period). waaaaaay too many fakes!

Step 5: Building the Battery Pack

In this 3rd video the 18650 cells are reassembled into a 6s3p battery pack. I complete a series of calculations to ensure the battery pack will be up to the task of delivering the required voltage and current and also run time duration.

Step 6: Powering the LED

In this video I set-up the DC-DC converter to power the LED and do some thermal testing to ensure the LED remains cool enough to ensure a long life.

EDITED: I can not recommend using Constant Voltage CV as a method to power the LED - they really should be driven by a Constant Current (CC) power supply. I've had a lot of warning messages posted to this instructable with dire warnings of disaster hence this edit.

As an alternative you can use this power supply (link below). If you want it should do the job and you can still implement the potentiometer to vary the current to alter the LED brightness as shown in a future step.

As for me I'm sticking with what I have at the moment I may do a follow up video showing the set-up for the above CC source.

Step 7: A Method to Vary the LED Brightness

In this step I show how to vary the brightness of the LED by using a resistor divider and a 10k pot. You will need 2 1/4watt resistors, one at 10k Ohm and the other at 12k Ohm. A bit of soldering required here but I'd class this as 'easy'.

Step 8: Finishing Off the Battery Pack

This is just a short video showing how I wrapped the battery pack with blue heat shrink tubing.

Step 9: Cooling Fan Control

The battery pack is 24v and the fan is 12v so I just used a DC-DC buck converter to reduce the voltage. This video shows me setting the voltage up and wiring the fan in.

Step 10: Low Voltage Tester

18650 batteries need careful looking after. They don't like to be under/over charged. The addition of a low voltage test module will keep a check on each of the sets of parallel cells and will set off an alarm if the voltage goes below a set level. This video shows me checking these out and finding some surprising and disappointing performance of the modules I bought. Buyer beware!

Step 11: Final Assembly and Testing

Finally, all the parts can come together and we can test the performance of the Mega Torch.

Step 12: Comments & Suggestions?

Picture of Comments & Suggestions?

Please feel free to comment and suggest any improvements.

Have you made an even bigger torch? I have seen some insane water cooled ones on YouTube (see link below) but they are not really as portable as my design is!


rafununu (author)2017-02-08

Many people working with Leds think in terms of voltage, which is the biggest error that can be made. Leds work with current. If you provide a constant current to a Led the voltage will be automatically adjusted by the driver to fit the Led's needs. If you give it a constant voltage, this will imply more heat and an increasing current up to the point where the Led will be destroyed. You can always decrease this voltage but the Led will stop emitting light very soon. A constant current driver is not cheaper as a converter but it's much more safer.

sjowett (author)rafununu2017-02-08

Thanks for your comments. Yes - on the whole using a LED driver with a constant current control is a good thing. At the same time in this instance I am under driving the 100W LED. 'White' LED's of this type are actually blue LED's under a phosphor coating with a forward voltage of ~3.5v each. So with a 100W LED (with 10xLED's in series) this gives a full forward voltage of 35V. My battery voltage of 24V requires boosting and I limited the boost to well under 35V - (I also checked the current & power going to the LED to check this was true).

However - as an alternative how about this board?

I did have one of these but I let the magic smoke out!

GTO3x2 (author)sjowett2017-02-12

Does "forward voltage" refer to voltage forwarded to or from? If to, how is this any different than what has been normally under stood as voltage (supplied/ required etc...)?

sjowett (author)GTO3x22017-02-12

Forward Voltage (Vf) is a term used on an LED's data sheet and refers to the voltage dropped across the LED while it is illuminated at the specified current etc. Effectively in my case this is the voltage provided by the DC-DC converter. You need to avoid going over this voltage as you will burn out the LED - Very quickly!

GTO3x2 (author)sjowett2017-02-14

Thanks. Yes, I don't see why it, then, isn't called "voltage drop". I'm seeing the case where scientist people are "in too deep" and losing their reference. This is applying to the health/medicine field too; e.g. I don't call alcoholism a "disease". Thanks, -out.

sjowett made it! (author)GTO3x22017-02-14

I guess we are getting a bit off topic but diodes require the 'extra' refinement in terminology to discriminate between forward voltage and reverse voltage as they are both critical parameters whenever you choose one for your circuit, (see spec sheet below).

At least we (engineers) don't then give it a Latin name to obfuscate like the medico's!

GTO3x2 (author)sjowett2017-02-15

I find the understanding of LEDs as opposed to filament lighting as confusing as working between time and frequency domains. I'm going to take that working with LEDs has its abstractness due to the nature of semiconductors.

My initial reference is that all current flows because of voltage differential or "pressure". The LED element of the circuit has its operating ranges and functions. This is why I have reservations about accepting "current limiting" as the key design focus. -that's not saying it isn't critical.

This is all because of my tinkering with accent LED lighting in my car. I also couldn't understand why I saw flickering in LED taillights from a DC system. I now know this is because of the use of pulsing (Pulse Width Modulation) to achieve the desired operation, luminosity (and, possibly, its durability). Similar to the typical maximum rated operating voltage of 250V for incandescent bulbs and other appurtenances (switches) commonly specified but not paid much attention to. It may also be that the spectrum and bounds for LEDs are much smaller, and this is why it is necessary to heed.

Sorry for the tangent, but I see there is a lot of confusion about the fundamental issue, nomenclature and use, and I am not the only one getting tripped up. I am going to also take voltage "drop" equal to nominal supply voltage since the reference level is zero and the differential is numerically equal in each case. I'm not sure reverse voltage is ever much encountered, but is a necessary specification to disclose. -similar to the rating of a very weak check valve.


Incandescent bulbs are resistive - not completely linearly, but relatively so (resistance changes pretty dramatically with temperature - which is good, makes them self current correcting).

You observation about "check valves" is an interesting analogy.

Ignore reverse voltage for the moment, because most diodes (except Zeners) are intended to be used in forward bias. Mostly.

If you look at the voltage vs current plot of a resistor, it will be a straight line. More voltage, proportionately more current. V=I*R.

Most semiconductors behave in non-resistive, non-linear V vs C relationships.

If you imagine a "check valve" that "sticks", that's kinda like a diode.

Nothing flows backwards. But you need some forward pressure to overcome the "sticking" before any current will flow. Then, once the valve opens, it presents almost no resistance and as much water as possible can flow... For diodes, the "sticking point" is called Vf (forward voltage).

Instead of a straight V/I plot, diodes have what's called a "knee". Current will be a horizontal line at zero up to this Vf, then the plot will go nearly vertical... With quite a sharp "corner" or knee.

For a regular diode, Vf would be about ~0.7V. Red LED ~1.7V, white LED ~3.2V etc.

Again, below Vf almost no current flows, the LED will be extremely dim or off. Above Vf (even a little), as much current will flow as your power supply can handle!

The LED might be rated at 20mA. Your battery can put out a lot more than that!

You need at Vf, but you something to regulate the current to below maximum.

This can be done with a simple resistor, or a more complex current regulator.

If you connect a 20mA white LED to a 9V battery, you'll probably see a brief flash, and then think you got a defective LED... The 20mA spec says all it can HANDLE is 20mA. At nine Volts, a lot more than 20mA will flow (remember our V/I plot?).

In a nutshell, if you want to drive a white 20mA LED from a nine volt battery, subtract Vf from 9V, you get 5.8V left over you need to get rid of, and you want everything balanced at 20mA. If you put 290 ohm resistor in there, at 20mA, it will have our desired 5.8V across it, leaving 3.2V for the LED. But - here's the good part - If more than 20mA starts to flow, the resistor will steal more voltage (V=IR), leaving less for the LED, and vice-versa. Things will balance nicely at the knee at ~3.2V and 20mA.

sjowett made it! (author)RicksterInstructables2017-02-18

I've added a couple of images grabs from 2 LED specification
sheets. The one for the 100W LED is similar to the one I used in this

The Voltage Vs Current 'curve' is nearly linear under the
rated Vf. Your statement that an LED is somehow like a sticking check
valve is odd given the data contained here - perhaps you can show data
for the type of LED behaves in the way you described?

Where you are correct is that once you exceed the rated Vf of an LED you quickly run into issues with excessive (current and resulting heat), but this is irrelevant to this project as I am well under this rated Vf.(35V). You may argue that there is a failure mode of the constant voltage DC-DC converter I used leading to uncontrolled voltage but I counter that by saying any current control DC-DC converter will have similar failure mode issues. The worst that would happen is I blow up a $10 LED.

Also, it may be easy to control a 20mA 'single' LED with a resistor but this type of current control would be disastrous for this project......for example, with a current 3A and a supply voltage of 40V.......this would need a 1.8Ohm resistor which would be dissipating over 16W!

Sure, I'd be happy to backup my claims with facts.

The datasheet you show is deceptive in that they plotted X only between min and max Vf. They didn't plot from zero volts. In the small range they plot, the line, though more exponential, appears almost linear, but look how steep it is! Current is VERY sensitive to voltage!

I've included a datasheet from a single Cree LED (easier to read than serial strings). They also plotted only over rang of Vf. So I filled in the plot. My red line extends to one volt before running out of paper. Notice forward current is near ZERO and the knee I described is clearly visible.

You are playing too close to disaster. You will get burned.

If you have any further questions, feel free to ask.

Constant current is the way to doubt - see the reply given above for a recommendation.

As for me I'm destined to get burned so this will be my last instructable.

Please don't make this your last Instructables based on my comments!!!

I was trying to be constructive, and am sincerely sorry if I offended you.

You've done cool stuff. Share it. Let others help you improve your designs!

That's what Instructables is all about!

sjowett (author)sjowett2017-02-20

.... I've also edited the step regarding the power supply and hopefully extolling the virtues of CC driving and I give my sincere aploogies for having the temerity to suggest that you can satisfactorily drive an LED with a constant voltage source. You can NOT (now sitting on the naughty stair).

That's the second thing about LEDS.

"Under driving them", below Vf, they do little or nothing.

If you're making them light up, you are driving them at Vf.

Oh, and as (most) LEDs get hotter, Vf decreases. So if you try to tweak the voltage to get the desired current... As it heats up, Vf goes down, current goes up... Meltdown, slowly dim or last gasp bright flash... Dead.

You really MUST understand that LEDs are CURRENT devices and drive them with a constant current source. This needn't be complex - it could be a simple resistor - but it is necessary if you want to make anything other than a flash-bulb.

sjowett made it! (author)RicksterInstructables2017-02-18

"Under driving them", below Vf, they do little or nothing.

If you're making them light up, you are driving them at Vf.

Both these statements are not correct - the luminosity is nearly linear to the current (see attached datasheets below). This is demonstrated in the videos as the voltage rises so does the current (and so does the luminosity of the LED).

FYI one of the datasheets is for a 100W led similar to that used in this project and the other is fro a "plain vanilla" 5mm LED. In both cases the characteristic curves are similar.

DB.Cooper (author)sjowett2017-02-16

"I did have one of these but I let the magic smoke out!


nothing ever works after you let the smoke out - ( except diesel. ) MWAHAHA! ROFLMBO!

the_3d6 (author)sjowett2017-02-09

It would be best to connect less LEDs in series and more in parallel, personally I probably would made 2 banks of 50 parallel LEDs connected in series, so 2 cell LiPo would have enough voltage to provide max brightness. It seems reasonable to control current using some mosfet with rated current of 20-30A at high frequency (50KHz at least), with large enough capacitor to smooth current peaks. These LEDs can use up to 150mA current each, so 50 in parallel would require up to 7.5A. At 50KHz, it means that ~150uF will flow there during each cycle (7.5F / 50000), so 4-5 100uF tantalum caps in parallel would smooth this current well enough.
The only problem is to monitor current, simplest way is to use some nichrome wire with 0.05 - 0.1Ohm resistance (nichrome properties are well known so it's easy to cut length to get quite precise resistance), connect it in series with the battery and measure voltage on it, then transform this voltage into current using I = V/R, and use it to limit max PWM width on microcontroller.

But such design would burn LEDs if PWM generating microcontroller will freeze in open mosfet state, so although it is probably the simplest way, it is not the most reliable one (there are ways to fix this problem, but then it becomes more complicated).

sjowett (author)the_3d62017-02-10

The LED used is a 100w led made up of a matrix of 10x10 1w LED's. As such I had no choice of LED arrangement in this design. Similar LED's of 20W and 50W also have a 10 in series arrangement ie 2 x 10 and 5 x 10 respectively.

the_3d6 (author)sjowett2017-02-10

Oh, I see... I though this was soldered from 100 separate LEDs. Then you don't have much choice, yes

GTO3x2 (author)rafununu2017-02-12

Then why aren't all the LEDs I have on DC not burned up?

Tecwyn Twmffat (author)rafununu2017-02-08

Yes, where the 150w buck boost power supply is specified, select one with adjustable current as well as adjustable voltage?

sjowett (author)2017-02-20

.......I've also edited the step about the power supply extolling the
virtues of CC driving of LED's and the true wickedness of EVER
having the temerity to suggest only controlling the voltage is
satisfactory. It is NOT. (now sitting on the naughty stair in shame)

starphire (author)2017-02-13

This appears to be a well-thought-out design; you are right to take cooling seriously, as there are far too many DIY types claiming their LED torches are adequately cooled with obviously undersized heatsinks (or no fans) simply because "it's only warm to the touch" or "it's been working fine for a dozen hours already". Then others use their design as a guide and thus perpetuate the fallacy.

I am wondering what "Tier 1" manufacturer you've been able to find with $80 LED modules. Can you provide a link to one? By my definition of the term, there is no such company manufacturing any 100W LED arrays in this package style. This applies to the smaller 10W and 20W arrays in similar square packages, as well. The "good" modules of this type go to large scale lighting OEMs, mostly for commercial lighting applications. The substandard ones go into low quality light fixtures, or sold as modules on aliexpress, ebay, amazon and so forth.

The most prevalent defect in these no-brand square LED modules is actually not detectable with low-current testing looking for unusually dim or bright LEDs. It is that the whole module does not emit nearly the amount of light they were specified to produce. A failed luminosity test sends them to the same scrap bin as the "defective" modules you tested for. It's an excellent business for these online resellers since neither they nor their customers will have the necessary testing equipment to determine that a "100,000" lumen module actually only puts out half that much light at full rated power. The 3 out of 4 "good" modules you got are very likely not as bright as claimed, so that would be the reason they were also sold at low prices through aliexpress. In other words, it's not as much of a lottery as it might seem since for most of us it still seems ridiculously bright to our eyes.

sjowett (author)starphire2017-02-14

Thanks for your comments Starfire, I guess using low current and looking for dull/bright spots is a first check and one that shows up the real dud's. As you say most amateur customers such as myself don't really have the equipment to really test the luminosity of these cheap LED's. However, I did try to get a better idea by ramping up the voltage and comparing that the the observed forward current which followed the spec in the PDF I show in the video. I also did the basic Lux meter test and got just over 14000lux at 1m on axis - hardly scientific but a reletive measure no-the-less. My feelings are that the defects may be down to any number of issues such as bad bond wires, poor alignment, bad phosphor coatings and so on, some of which may manifest in lower luminosity or some other unrelated parameter such as CRI

As for tier 1.....well it might be a matter of opinion but for example, at this link below there are 2 LED's on sale for $163 (I think the LED chips are Epistar). Hardly exhaustive but again - I'm not an expert , just a bit of a hacker/maker

In the end, we have to just say this is a bit of fun and I hope I have not misrepresented the LED's I got from Aliexpress/ebay. Yes, they not top quality but they do work and you can make a really bright flashlight/torch for not much money.

starphire (author)sjowett2017-02-14

I think of YujiLED as a specialty supplier, but they certainly do hold their contract manufacturers to high standards that could be considered tier 1. They also provide detailed datasheets on par with top tier manufacturers.
I didn't know that they had 50 and 100W packages in their lineup now, so that is definitely interesting for specialty applications like video production.
As for other sources of these COB modules, it is highly variable in terms of what you get. After being burned several times with truly crappy modules direct from China, I've paid extra for 10W modules with higher CRI through what I believed to be more reputable sources, only to find the light output around 30-40% of advertised (based on my own crude testing). At least they didn't have truly bad chips in the array, but I would have done better with a selection of $5 LED light bulbs from local stores!
I am very tempted to bring my spectrophotometer and integrating sphere out of storage to do a larger survey of the marketplace, but I can't really afford the time or money to do a usefully large sampling since it changes so quickly.

Sylvaind9 (author)2017-02-12

How about a simplified version with 7s3p battery which would give you a maximum full charge V of 30.1V and a minimum of 23.1V. doing away with the boost converter and the lipo alarm as the LED would essentially turn itself off when the cell voltage fell too low. A very simple circuit is then sufficient, with an on/off switch and perhaps a resistance voltage divider to drive the minuscule current needed for the fan. OK no brightness adjustment on this one. It might event work with 8s3p.

You absolutely must run a constant current power supply to LED arrays. Otherwise they will LITERALLY take fill load current. In even my 3s Lipo batteries on my multirotor your looking at 180 amps! That would nuke that LED in an instant without a controller. Constant current (boost or buck) power supplies are not that expensive, be safe!

There seems to be some confusion here. A 3s LiPo battery will not even light the type of LED used in this project (100W led with a rated forward voltage of 35v). There is simply not enough voltage to get the semiconductor to conduct. If you watch my detailed video on step 2 you will see the LED's begin to illuminate at around 25V, and without controlling the current it is perfectly safe to ramp up the voltage to around 29V .....without any current control. At the same time if you want to run this LED at 35v then YES you are right you must control the current.

Sorry, I didn't mean using 3s on your array directly. I was referring to a smaller array or just leds in general. You should never use a battery like a lipo directly on an LED due to their current delivery capability. However. You CAN use a 3s lipo on a boost driver for a large led array. It will draw more current though then the led is being driven at.

sjowett (author)Sylvaind92017-02-12

I guess it might work but I really like having the having brightness control. With 8s (33.6v when fully charged) I think the current would spiral out of control. My video testing the LED's showed that at 29/30V I was approaching 100W. This is lower than I expected but I believe the test was valid. 7s might be ok though but I'd still prefer to see some controller in the circuit, at least voltage and as some others have commented, current too!

GTO3x2 (author)2017-02-12

I don't know why the enthusiast group refers to these as "torches". I don't see a fire.

tonysuffolk (author)GTO3x22017-02-12

In the UK we use the term "torch" for what Americans call a "flashlight". For Brits a "flashlight" is a device to illuminate ones private parts. Useful for men who like to expose themselves at night. ;-)

GTO3x2 (author)tonysuffolk2017-02-14

Ahh, thanks. Our "flashlights" used to commonly have a momentary button for flashing besides the on/off slider. I "assumed" that was for Morse code.

sjowett (author)GTO3x22017-02-12

Ha Ha - sorry GTO3x2! TonySuffolk has hit the nail on the head - I was born and brought up in the UK.... Celebrate the differences!

jonathangab18112003 (author)2017-02-13

i can't see the step 3 until step 11 because my youtube was blocked

plz give me picture

I'm sorry I can't help as this instructable is predominantly based around detailed videos loaded up to YouTube (rather than my previous instructables which are predominantly picture based).

HarryB80 (author)2017-02-12

Yep i have done a lot with high powered LEDs and for long life you need 2 things constant current and good cooling ( heat sink).
Nice job on the torch...

sjowett (author)HarryB802017-02-12

yep! - Heat is the enemy of long life for LED's - thanks for the positive feedback Harry.

The_Technocrat (author)2017-02-12

Hay nice build. !!!
Where did u get the lens u used as projector ??

sjowett (author)The_Technocrat2017-02-12

Step 2 gives a fully detailed parts lists with links!

atitajevs (author)2017-02-09

But what if we make a car cigarette lighter operatable one, boom

sjowett (author)atitajevs2017-02-10


About This Instructable




Bio: An engineer who likes to tinker with electronics and make stuff!
More by sjowett:"Hydra" a MONSTER Bluetooth Speaker!Meet 'Holman' the Ultimate Bluetooth SpeakerMiniature Weather Station - ESP8266
Add instructable to: