A 1.5V High Power LED Flasher

bwalzer

Newly Enlightened
Joined
Oct 11, 2010
Messages
7
Note: This somehow ended up in the wrong forum the last time I tried posting. Second try at getting it into Bicycle...

I have found out something that I have not seen elsewhere so I thought I would pass the idea along. Using super/ultra capacitors you can make a high current flasher running off a single cell in a very simple way. Here is a description of how I used this idea to make a dual LED flasher that I am presently using on my bike to flash 1W LEDs:

http://59.ca/weblog/ledlight/dflasher.html

Here is a bit bloggier post that describes the thinking behind the circuit:

http://59.ca/weblog/ledlight/flasher.html

Comments welcome...

Bruce
progress.gif
 
I didn't realize anyone had built a 555 that would run down to 0.9vdc. Cool!

While I'm in favor of building any circuit that provides entertainment, it seems a bit inefficient to use all those parts when an extra battery would get rid of the need (mostly). Two AA's would easily flash red leds.

Charge pumps are handy, but a high frequency boost converter would be more cost and space effective.

regards,
Steve K.
 
@Steve K

I think it might be fun to try to defend the design...

A single alky C cell provides around 1.5 times more energy than 2 AAs at roughly the same cost. You only get to use a bit more than half the energy available in the AAs if one is using resistive current limiting (1/4 brightness comes out to be about 1.3V per cell). As a result we end up with as much as 2 1/2 times more usable energy per currency unit in the 2 AA vs single C case.

The efficiency thing is killer here (<55% for the capacitor multiplier) but in practice it is not as bad as it sounds. A capacitor based voltage multiplier gets more efficient as the battery voltage drops. Boost converters tend to not work all that well at lower voltages. As an example the Linear Tech LTC 3490 is probably about the same efficiency at 1.2V as the tripler scheme I used. An idea (I have not tried this) would be to charge a single supercap with a something like a 20mA capable boost converter and then connect the supercap to the LED to do the flash. I think the efficiency would be reasonably high and the current regulation could be done on the 20 mA side which is a lot easier than trying to do it on the .4A side. If you did have good current regulation you could use a small supercap.

I guess my hope for the cost thing depends on the fact that the cost of small supercaps is currently falling at a greater rate than multi amp capable inductors.

Thanks for the comments...
 
Zetex, the maker of this low voltage 555, also make a neat ZXSC310. It's a little boost converter than can run down to 0.8v. I've used it in standlight circuits on my bikes, where a AA is charged by the voltage across the light, and then runs four 5mm leds when the dynamo is stopped. Very handy little circuit.

You might play with the idea of making a little boost circuit with the 555. The 555 could be set to a fixed duty cycle, designed to work with the max battery voltage. No feedback. Just energize the coil by pulling the far side to ground, and then remove the ground, allowing the coil to dump the energy into the leds. The efficiency is a function of the switching frequency, the inductor losses, losses in the switching transistor, etc. In principle, if you are willing to use big components, you can really improve the efficiency.

As in most designs, there are a lot of variables that you can play with, and different things can be optimized (cost, size, efficiency, run time, etc.). The important thing is to decide what should be optimized, and then know how to adjust the design to achieve that goal.

It also helps to know about all of the amazing products on the market, like this new 555 or the little Zetex boost converter! I wonder why so few manufacturers make stuff that runs down to 0.8v?

regards,
Steve K.
 
Very interesting concept. Been awhile since I sold parts in a Radio Shack.

bwalzer: I took the liberty to exerpt one of your blogs and cite it here, hopefully sufficiently well to be OK with you.

Attribution to bwalzer From second blog page link in first post:

"This flasher suffers from a problem I refer to as the infinite battery life syndrome. You often see bicycle flashers in the wild producing barely perceptible light output. I think that is caused by a combination of human perception and the characteristics of LEDs. As the battery voltage gets closer and closer to the LED forward voltage the current decreases and as a result the battery life increases. Eventually the current gets down into the microamp range but the LEDs are still showing light. Since people are really bad at judging brightness there never comes a time where it is obvious that the battery needs to be changed. Since this is safely critical equipment I do not think this is good design."

This was re-re-rediscovered by a couple of us messing around comparing the Planet Bike blinky to other lights. We put inexpensive light meters to work as discovered just how bad close up perception of brightness is. Wondered why the dang things didn't show worth a hoot in video! Worked loads better with fresh cells!

Continuing the quote above:
"Presently the prototype is happily producing 63 mA (0.07V across R5 while charging) pulses at a battery voltage of 0.83V after running steady for a week and a half and now seems fairly stable. I guess one could add a voltage detector and do something on low battery voltage. I have no idea what this something should be."

If you gave up some voltage for NiMH, then with plots of time, output and voltage, you could get a good estimate of reasonable output runtimes. NiMH should not be discharged below 1 V. For a consumer item a low voltage warning and cutoff would be needed. For a DIY, simply knowing you have 20 hours of maximum runtime (or whatrever) and 2 hours per commute (or whatever), means 10 commutes and the cell is toast, so recharge the cell or cells every week and you are golden. Cell life is better without the deep discharge and you are running more time in the top part of the charge (more voltage). At some point the cells are several years old and owe you nothing so you can replace them on principle or redo the runtime test to see whether you need to repace them yet. Not a consumer friendly aspect but easy for a DIYer. A slowing of the pulse at some preset voltage could work. A two or three second interval piulse is noticeably slower.

Continuing the quote above:
"I know that 1 second period flashers are considered with some disdain in some quarters. The issue is that it makes it difficult for observers to infer the speed and direction of the bicycle. I can rebut this from experience as I have used such a flasher for some years. On a totally dark street there might be an issue but in an urban environment if you can draw attention to yourself there is almost always enough light to follow the path of the bicycle/rider afterwards. This is especially true if the bicycle is liberally plastered with reflective tape as mine is. Since attracting attention is important it seems reasonable to throw out as much light energy as possible during the pulse. The best way to do this is to pulse as infrequently as is practical."

I ride a LOT of unlit roads and streets. I not only need to attract attention, but help them track me day or night (loads of reflective material and an ANSI vest). Problem is they don't expect a bike anywhere at anytime and some take a long time to clue in. So for me this is a deal-breaker except as an additional marker light/backup system. I suppose one could devise a pair of these exactly out of phase to get two flashes per second. Twin NiMH recharge as a pair nicely, so that's a benefit. Twice the cost and complexity but twice a second is a fairly annoying and frenetic pace, and if they can't follow that, then their night vision needs checking, or their pulse does! :thumbsup:

Ride safe. Ride on.

Brian
 
@Steve K

Whoa, the ZXSC310 would make for an ideal supercap charger. You could charge a series string of them and flash multiple LEDs in series and/or green/blue/white LEDs. ... all nicely current regulated. Thanks! Well mostly thanks, this is distracting me from other geeky pursuits that should be done before winter. I will have to try to settle for the 50 hour battery life at least for this year. There is not enough time this season left to run down even the one battery.

I spent some time time thinking about doing some sort of inductor based thing on my own. It is hard to make it efficient. You need voltage to induce the current in the inductor. If you only have a volt to work with you need to use a small inductance value. To get any significant power transfer you have to switch fast. That gets into a whole lot of exotic losses. It ends up being beyond my design skills...

Everyone wants to do a CMOS process these days I suspect. For low voltages bipolar transistors still rule. I think this means that integrated bipolar is some sort of Zetex strength.
 
@BrianMc

"NiMH should not be discharged below 1 V."

Oh... right... So we have to slow down at something like 1.1V and cutoff at 1V? The change of rate would be reasonably easy. Dunno about the cutoff, I suppose you could just use a voltage detector... but that seems inelegant...

Two pulses a second are easily obtainable with this scheme. You just need to make some component changes. I use a one second rate by choice.

It is interesting you mention the alternate flash idea. I actually think having both LEDs go off at once is a feature. I think it is less visually confusing in the dark in the case where someone can see both lights at once, which probably happens a lot of the time.

Your comment for some reason made me think of how the aviation people deal with lighting at night. They generally tend to have a one second bright flasher to attract attention and then much dimmer continuous lights to show the orientation of the aircraft. There might be an idea in there somewhere for people like me who like to blink slow and bright...

Thanks for the comments...
 
I spent some time time thinking about doing some sort of inductor based thing on my own. It is hard to make it efficient. You need voltage to induce the current in the inductor. If you only have a volt to work with you need to use a small inductance value. To get any significant power transfer you have to switch fast. That gets into a whole lot of exotic losses. It ends up being beyond my design skills...

but the ZXSC310 is doing exactly that with an inductor! You do need to be able to switch fast enough to not end up dissipating the stored energy in the transistor, but I don't think this will be that hard if you are driving it with the Zetex 555.

The only advantage to making your own little boost circuit with a Zetex 555 is that you can really tweak it for efficiency or whatever. The little SOT-23 ZXSC310 is a pretty slick device that is effective in most applications.

regards,
Steve K.
 
@Steve K

OK, it seems that getting any amount of power efficiently with a boost converter is not really possible when going from 1V to 5V. So my dream of using something like the ZXSC310 charge strings of capacitors is pretty much dead at this point (perhaps a really tiny transformer based flyback converter?). One LED at a time still looks possible. I guess a 555 osc driving separate boost converters would work. Wouldn't need much of an inductor.

Another idea that came out of this is to simply stop tripling and just double when the input voltage is too high. Apparently the people who make charge pumps for white LED backlights do stuff like that...

@BrianMc

I dredged some NiMH manufacturer websites and it appears that running a single NiMH to something like .8V might be acceptable. The big problem comes with reverse charging when using cells in strings. Yet another advantage of the single cell solution... :)

Bruce
 
You are right a DIYer can just pop the cell take its voltage and know if there is enough reserve power, or not. Not user friendly to the average consumer, though.

1 V. Yeah, you get certain rules of thumb and special circumstances may mean they are a bit more flexible.

FYI. The dull on and flash is exactly the appearance of the flashing mode from the Taskled Drivers in my headlights. It uses a triple and pause to good effect. It is toward the end of the video in this post. I agree that would be a very good solution. I don't think it is a by product of the low light/slow frame rate of the cheap camera, but even if it is, it sure looks good!
 
@BrianMc

Actually I totally agree that some sort of low battery warning is important. Especially with NiMH where the power goes away shortly after the voltage drops. I just meant that nothing special has to be done to protect the cell...

Video? There is a video? Where should I look?

Thanks,
Bruce
 
@Steve K
OK, it seems that getting any amount of power efficiently with a boost converter is not really possible when going from 1V to 5V. So my dream of using something like the ZXSC310 charge strings of capacitors is pretty much dead at this point (perhaps a really tiny transformer based flyback converter?). One LED at a time still looks possible. I guess a 555 osc driving separate boost converters would work. Wouldn't need much of an inductor.

The datasheet doesn't get into the details of what controls the efficiency. I do know that I use a ZXSC310 in a standlight application that uses a AA nicad for the power source, and drives 4 amber leds in series. It'll run 8 hours or so, but I don't really know how efficient it is.

It may be that the inductor you use is saturating, or maybe the ZXSC310 is running into duty cycle limits? Beats me. All I know (I think....) is that there's not an intrinsic limit to the efficiency, other than the usual 100%. You might look around for a tutorial on switching power supply design and see if you can find the basic equations that govern the operation. I've got one, but it's not in reach right now.

regards,
Steve K.
 
The word 'post' was a hotlink. Oh Well I'll paste it here too then!

<embed src="http://img820.imageshack.us/flvplayer.swf?f=Mbeam" width="640" height="380" allowFullScreen="true" wmode="transparent" type="application/x-shockwave-flash"/>
Uploaded with <a target='_blank' href='http://imageshack.us'>ImageShack.us</a>

You'll have to wait for the pass at 0.5 A flashing. Even though the video dropped to 6 frames a second (camera doesn't like the low light much), I think the glow between flashes is real. A nice look to try to get.
 
Last edited:
Top