The Deliverator – Wannabee

So open minded, my thoughts fell out…

Doing the Math on an Ultracapacitor Flashlight

Posted by Deliverator on March 12th, 2009

Well, I canceled my Ultracapacitor powered screwdriver order. The company from which I ordered,, clearly promised shipment within 24 business hours, but several weeks after placing my order I had received nothing and hadn’t received any sort of shipment notification or update on order status. I went back to their web page and found that they were now sold out and wouldn’t be receiving more from the factory for at least 4 weeks. Again, they didn’t notify me of this, I had to pro-actively visit their site to find out. I emailed them and inquired whether my order had shipped before they sold out and it had not, so I asked them to cancel my order for now and refund my money, which they did. Paypal only gives you 45 days to file a dispute, so it is not in the interest of the consumer to sit around and wait while a e-tailer claims delay after delay. I don’t think this site is any sort of scam, I just don’t think they were prepared for the level of interest which the first Ultracapacitor powered consumer product would bring. Update: The company refunded my money promptly and even offered me a discount should I reorder once the screwdrivers are back in stock.

Another consumer Ultracapacitor product which interests me greatly is the Light for Life Flashlight. I like LED flashlights in about the same way as Imelda Marcos likes shoes. The Light for Life Flashlight is a large foot long club style flashlight, similar in a lot of ways to the ubiquitous 3 D cell Mag-lite, popular with security personnel the world over for its use as a multi-purpose device, to not put it too finely. The Light for Life is a bit wider in the barrel than the Mag-lites and also is made of plastic. This combined with the lighter weight of Ultracapacitors may make this flashlight a little less popular with the club em over the head crowd. On the other hand, the light for life promises around an hour of use at ~100 lumens with a ~300 lumen high brightness mode is also available although using it in that mode seriously curtails runtime. The real selling point of the Light for Life is that it takes a mere 90 seconds to recharge in the included cradle, which could be a real boon for a security guard walking patrols or just someone who wants a flashlight always ready to go at a moment’s notice without needing to search for a set of charged or fresh batteries.

I was quite curious about the Lite for Life, as I tend to haul my insomniac self out for an hour long walk usually between 1 and 3 AM. I am a hefty guy and it is rare that I encounter anyone on my walks, but I have noticed a definite uptick in shady vagrant types in nearby Factoria during the day and I have encountered some of them on the hill at night. The economy has gone to absolute hell and poverty makes people do stupid things. I’ve considered purchasing a stun-gun or a flashlight with built in stun-electrodes for personal defense, but I think a flashlight with a bit more heft to it might be almost as good in my case. I would never consider carrying a pistol. I usually walk with a LED blinker dangling from the backside of my jacket to make me visible to drunks approaching me from behind and I carry a smallish 3W LED flashlight in one hand to make myself visible to any cars I encounter ahead of me.

The release of the Light for Life has been delayed several times and is now delayed till June. The high price and delays of the Lite for Life made me a little skeptical of the product. Dan Rutter of Dan’s Data did an excellent technical analysis and running of the numbers on his blog and ultimately convinced me that I should just build my own. I started looking into the practicalities of what parts to use and found several faults with Dan’s analysis.

Dan’s energy calculations assumed a very high capacitance Ultracapacitor of 1200F and 2.7v. He notes that these are somewhat bigger than a D cell battery. The MC line of Ultracapacitors from Maxwell fall into this capacitance range, but they are almost twice the diameter of a D cell, which hardly makes for easy hand holding, assuming the capacitor is in the barrel of a flashlight. Maxwell makes another line of Ultracapacitors, the BC or “Boost Cap” line, which closely match the length and diameter of D cell batteries. These have a mere 350F capacity by comparison and a recommended voltage of 2.5 volt. You can likely over-volt these by a little bit without frying them, but you would be risking explosion (not to mention your expensive ultracaps). Energy storage goes as the square of the voltage, so higher voltage is desirable. Lets say you put 4 of these D cell sized Ultracaps in parallel, yielding 1400 F and push the voltage margin a little to 2.7 volts.

E = 1/2 * C * V^2 = 5103 Joules of stored energy. A high efficiency, high power LED such as a some of the new Cree emitters deliver a nice 100 lumens per watt. A watt is a joule/second, so you would get about 85 minutes out of such an arrangement, right? WRONG! WRONG! WRONG!

While the ultracapacitor bank can definitely hold that much raw energy and one can in theory drain a capacitor dry, in practice it is much more complex. If one connected up a capacitor directly to a LED, the LED would happily fry itself. There is next to no internal resistance in the capacitor and the LED will gladly try to draw a huge amount of current through itself and overheat and fry in a scant second or two. A cheap LED flashlight will typically use a resistor to limit the current flowing through the LED. Even in battery powered flashlights this is sub-optimal, as battery voltage tends to drop somewhat as it is depleted and different battery formulations tend to have different fully charged voltages. Cheap resistor limited flashlight tend to dim noticeably as the battery discharges and is a poor solution when trying to drain every last drop out of a battery. Better quality flashlights use some type of switching mode power supply and constant current regulator circuit to keep the LED shining at as constant a brightness as possible over the life of a BATTERY, without frying the LED. The key word there is BATTERY, as virtually all the commercially available circuitry for doing this is geared towards BATTERY use.

The voltage of a capacitor drops linearly as it is discharged. That 5103 Joules number we got assumes that we can drain every last bit of charge from out capacitor bank. The problem is that transistors need a certain amount of voltage difference to operate. An LED happens to be one form of transistor. A high power Luxeon LED emitter needs ~2.8 volts of potential difference just to start producing light and most seem happiest at around 3.6 volts.

Thus, we need a way to boost the voltage from our capacitor bank to drive the LED. A type of regulator circuit which boosts a range of lower voltages to a target higher voltage is known as a boost regulator. Another type of regulator is a buck regulator, which takes a range of higher voltages and puts out a lower voltage. Combine the two and you get a regulator that can operate over a wide range above and below a target voltage. Unfortunately, when combined you loose a lot of efficiency, and so you usually want to try and use just one type. In our case, we are only working with voltage below our output voltage, so a boost regulator is what we need. We also want to limit the current to prevent the LED from overheating and frying itself. One popular commercially available LED driver that accomplishes both of these goals is the MicroPuck from LuxDrive. It can be configured as a number of different regulator types to suit various needs. Driver circuits are made up of transistors (and other assorted flotsam) and thus have a minimum voltage to operate. I searched around quite a bit and one of the nice things about the MicroPuck is that it stops working at the extremely low voltage of .8 volts. This still means that all that energy left under .8 volts in our capacitor bank is unavailable for use for purposes of driving the LED.

E = 1/2 * C * V^2 = .5 * 1400 * .8 * .8 = 448 Joules left behind.

So, our initial theoretical energy of 5103 Joules has been reduced by 448 to 4655. In addition to not being able to squeeze the capacitor bank dry, the driver circuit is somewhere between 70 and 80% efficient, depending on input voltage. This means we loose another chunk of our total available energy to the driver circuit. So, lets say we only have something like 3200 Joules that actually make it to the emitter. That is 3200 watt seconds. Divide by 60 and you get about 53 minutes of light.

To some, a flashlight which only lasts about an hour doesn’t make much sense, but to me, a flashlight which lasts the length of my walk and can be recharged and ready to go in a matter of minutes without needing to swap batteries and which will likely NEVER need replacement makes perfect sense.