Best Cheap, Small Unregulated Boost?

[milkyspit]

I'm designing a light using a few Nichias driven by a single CR123. Because one of the design parameters for the light is price and the runtime will be long (another design parameter), it will be unregulated. At the 3V output of the CR123 I can get a useful amount of light for my purposes out of the Nichias, but it's nowhere near the Nichias' full potential AND there's little room for decreased battery performance due to cold weather, partial discharge, etc. If there's a fairly cheap way to boost the voltage just a bit -- like perhaps one more volt? -- that would take care of things nicely.

Being that I'm willing to forego regulation, is there a cheap, simple way to boost the voltage a bit, and how?

 

[yclo]

Run a search on the "satcure circuit".

 

[Steelwolf]

SatCure circuit may not work well with 3V input unless some of the component specs are adapted, the main 2 items would probably be the inductor coil and the resistor, since the flyback winding on the coil is the one that senses when the inductor is fully loaded and ready to discharge, and the resistor influences the point at which the transistor switches over for discharge.

There are some circuits out there that rely on the standard astable flip-flop circuit to switch on and off a standard inductor coil. I would suggest that type of circuit as more suitable since there is no need to specially wind the flyback inductor. All parts can be bought off the shelf and soldered straight in to place. None of that fiddling around with a tiny little ferrite bead and some thin enameled copper wires. You can probably reduce the parts count with a 555 timer chip too.

I have a schematic or two if you're interested.

 

[Lucien]

Hi milkyspit,

If you're comfortable with putting together a crude pcb and fiddling with some components, there are some ICs that you could use.

The ZXSC300, used in the ZLT circuit, is a simple unregulated boost circuit. Unfortunately the project website seems to be down, but you could try doing a search for "ZXSC300" or "ZLT" here on the forums.

The ZXSC400 (or 410 or 420) are similar to the 300, but use a sense resistor which allows for regulated behaviour. Right now I'm fiddling with this chip to drive a 1W LS. I've got it working, but wouldn't call it finished yet. Efficiency is lower than the 300 though.

There's also the LT1932 from Linear Technology, which seems to be specifically designed to drive 5mm LEDs. I haven't tried this one out but the circuit seems simple enough. They also have another similar IC, the LT1937, but I haven't really looked it over yet.

With all these cirucits, you'd need an inductor, but I think you might be able to get away with a store bought one instead of a hand wound one (as was used in the ZLT project). Right now I have a few J.W. Miller ones on order to attempt this. (Meaning I haven't yet)

On the subject of cost, I think the ZLT circuit was quoted at about $10. Not sure how much the others would add up to.

Good luck with your project, and do let us know how it goes /ubbthreads/images/graemlins/grin.gif

 

[milkyspit]

[ QUOTE ]
Steelwolf said:
I have a schematic or two if you're interested.


[/ QUOTE ]

Yes, I'd like to see the schematics. /ubbthreads/images/graemlins/cool.gif Where should I look for them?

 

[milkyspit]

[ QUOTE ]
Lucien said:
Hi milkyspit,

If you're comfortable with putting together a crude pcb and fiddling with some components, there are some ICs that you could use...

(snipped)

Good luck with your project, and do let us know how it goes /ubbthreads/images/graemlins/grin.gif

[/ QUOTE ]

Thanks, Lucien! /ubbthreads/images/graemlins/smile.gif My project involves building a minimal-cost, maximal-runtime Nichia candle -- in other words, not terribly bright, but enough to illuminate a room during a power outage or for the romantic in you /ubbthreads/images/graemlins/wink.gif -- driven by a single CR123A cell. A secondary goal of the project is to achieve something that could be used with CR123A's too depleted to run a Space Needle II, for example, and extract all the remaining juice out of them. In this sense, the "candle" would also serve as a battery recycling tool.

I actually have another thread over here in which I talk about some aspects of the project. No prototypes to show yet, unfortunately, though I've already had some success piecing together various combinations of parts for preliminary testing purposes.

What I'm most comfortable with so far is breadboarding circuits and/or wirewrapping onto a piece of perfboard. I'm not much of a soldering guy yet, though it will come eventually. /ubbthreads/images/graemlins/smile.gif I've wirewrapped circuits since I was roughly 11 years old, and have gotten pretty good at it. Does anyone still do that, or is it a lost art?

 

[JimmyR]

Wire wrapping isn't an art, it's therapy.:)

How's Arc get that DC-DC inverter in such a tiny space ?? Is it a secret or can I or anyone get the parts and build it ( 1.5v inverter) ?

 

[LED_ASAP]

If your main goal is a LED backup candle for power outage, you probably don't need to keep it in your pocket all the time. You can have the luxury to make it a little bigger (enough to put in two 123 cells). This way a simple resistor will do the trick nicely and you can feed it with your half-spent cells. You can even put the batteries in the base of the "candle" and add a piece of white PVC tubing over the LED to give a difused light for area lighting---more in the shape of a candle /ubbthreads/images/graemlins/smile.gif

 

[milkyspit]

[ QUOTE ]
LED mods As Small As Possible said:
If your main goal is a LED backup candle for power outage, you probably don't need to keep it in your pocket all the time. You can have the luxury to make it a little bigger (enough to put in two 123 cells). This way a simple resistor will do the trick nicely and you can feed it with your half-spent cells. You can even put the batteries in the base of the "candle" and add a piece of white PVC tubing over the LED to give a difused light for area lighting---more in the shape of a candle /ubbthreads/images/graemlins/smile.gif


[/ QUOTE ]

It's a tough call. I've thought about something similar to what you suggest, and while it's certainly cheap (one of the criteria) and would undoubtedly work, I struggle with the issue of LED brightness. Often people end up replacing the batteries in unregulated lights because the output drops too low, yet there is still a fair amount of life remaining in the cells. Theoretically at least, I was hoping to build something that while not necessarily equally bright over its entire runtime, would at least take the cells deep into their discharge curve before someone felt compelled to change batteries. The reuse of discarded batteries from higher-powered lights is intuitively very attractive to me. Then I struggle with cost issues, and so the cycle goes. /ubbthreads/images/graemlins/icon23.gif

Maybe a switch allowing the user to dial the light to higher brightness levels (by selecting successively smaller resistances) as the batteries get too low for the output at the previous level? This would be somewhat similar to someone in days of old raising the lever that lifts more of the candle out of the holder, until the entire candle has been consumed. The downside is that it would be pretty easy to dial too high an output early in the discharge curve, and fry the LEDs in the process. /ubbthreads/images/graemlins/frown.gif

Is there a semiconductor that could limit the voltage or current, so perhaps a safeguard could be built into the light preventing failure due to overdriving?

 

[paulr]

How about if you apply a bigger voltage boost (to 7 or 8 volts) and then use two nichias in series instead of parallel.

In practice I've found that a single LED (CMG Ultra or Arc AAA pointed at the ceiling) is enough to light a room enough to find your way around. Unless you're trying to build your own circuit because you like building circuits, maybe simplest is just get an Ultra. Countycomm and Batterystation both sell them for $15, and they'll run quite a long time on one AA.

An even simpler do-it-yourself idea: if you really want the "candlelight" effect, use yellow LEDs! Those have lower Vf than white LED's, so you can just run them direct from a CR123 using a limiting resistor. It will look more like candlelight, and no voltage boost needed.

 

[milkyspit]

[ QUOTE ]
paulr said:
How about if you apply a bigger voltage boost (to 7 or 8 volts) and then use two nichias in series instead of parallel.

In practice I've found that a single LED (CMG Ultra or Arc AAA pointed at the ceiling) is enough to light a room enough to find your way around. Unless you're trying to build your own circuit because you like building circuits, maybe simplest is just get an Ultra. Countycomm and Batterystation both sell them for $15, and they'll run quite a long time on one AA.

An even simpler do-it-yourself idea: if you really want the "candlelight" effect, use yellow LEDs! Those have lower Vf than white LED's, so you can just run them direct from a CR123 using a limiting resistor. It will look more like candlelight, and no voltage boost needed.

[/ QUOTE ]

Paulr, some good ideas here! /ubbthreads/images/graemlins/smile.gif I certainly already have enough Infinity Ultra-G's (7 at last count), but find in room illumination that I personally need a little more light. A TurtleLite II (which uses 2 white LED's) driven by lithium AA's gives me a nice level of illumination, and tends to run at roughly that level for a looooong time. /ubbthreads/images/graemlins/smile.gif Based on the warm fuzzy feelings from that experience, I'd like to achieve something similar in level of brightness.

More importantly, I want to stick with the 123 cells, at least for now. A surprising number of CPF'ers have embraced the project specifically because they discard lots of 123's from their other, higher-powered lights, and would love to get a little more use out of them. I know it would be great for my own EL Blaster VI and SNII discards.

The yellow LED's sound like an intriguing choice. I wasn't necessarily trying to duplicate the spectral characteristics of a real candle, but the lower Vf seems just right for the project. I had thought about red and rejected it as being too disturbing as a sole means of illumination for perhaps several hours, or even days; it seemed too much like being stuck in a nuclear submarine. /ubbthreads/images/graemlins/grin.gif I'll give yellow a try. Thanks! /ubbthreads/images/graemlins/thumbsup.gif

 

[paulr]

An Arc LS might work ok with an SNII discard. If you want a ready-made converter you might try a Micropuck.

 

[milkyspit]

Paul, thanks. I'll check out the specs for the Micropuck, although if it's intended for an LS it probably puts out too much current (I'm looking for really long runtimes on partially discharged 123's.). And the Arc LS itself completely violates the criterion of being inexpensive. (Wish I had one! /ubbthreads/images/graemlins/grin.gif ) I feel like I'm shooting down all your ideas, and you've been great to offer several! /ubbthreads/images/graemlins/thumbsup.gif Sorry about that. So far, I like your yellow LED idea the best. Now where can I get my hands on some? /ubbthreads/images/graemlins/help.gif

 

[paulr]

The Micropuck is like a Madmax, it's voltage regulated, not constant current. It specifically says you can run a cluster of regular LED's with it.

I think dat2zip's Sandwich Shoppe had yellow LED's available but I don't see any there now. Maybe you can email him and ask. There's some other threads with url's for various other LED vendors. They're not scarce.

 

[Doug Owen]

[ QUOTE ]
milkyspit said:

Is there a semiconductor that could limit the voltage or current, so perhaps a safeguard could be built into the light preventing failure due to overdriving?


[/ QUOTE ]

Funny you should bring this up.

How about Mr Al's Low Drop Out (LDO) linear *current* regulator? A LM334 (3 lead 'transistor case' IC), a small transistor, three resistors, a capacitor and your LED(s). As long as the battery voltage is at least .1 Volt above Vf, you get exactly the current you ask for. Parts cost a bit over a buck each for ten sets, under two if you decide to go fancy and put a high/low ('turbo') switch in so you can have a dual level (or 3 or 4....) level light. This changes the current regulation point, so it's not something you fiddle as the battery dies. The regulator (except the switch) can easily be built in a half inch cube, given reasonable soldering skills.

I've been fiddling with 3 NiMH cells and parallel 5 mm LEDs with several levels. Looks very promising so far, very good run times with high efficiency due to flat discharge voltage and close match between battery and Vf. Over 50 hours straight at 30 mA, over ten at 120, with 1800 mAH cells. Typically when it goes dark, one cell is very near zero, the other two close to 1.1 or 1.2 each. The end come fairly fast, less than an hour between 'is it getting dimmer?' and 'it's getting dimmer!'. Leaving it on until really dark a number of times hasn't killed the 3 cells in the test so far, in fact different ones 'die first' from time to time. A few hours in the charger after the LED turns green for the weakest cell seems to get it 'off the hot seat' next time. Anyway, different solutions for a different problem. You might consider the idea for driving red or yellow LEDs with a single cell, or whites with two. With a switch it's a bit like a storm lantern, you set the wick to get the light you want and fuel consumption follows.

Doug Owen

 

[Doug Owen]

[ QUOTE ]
paulr said:
The Micropuck is like a Madmax, it's voltage regulated, not constant current. It specifically says you can run a cluster of regular LED's with it.


[/ QUOTE ]

Unless yours is very different from mine, the Micorpuck is not a *regulated* anything (or anything regulated?). It's an open loop up converter, as the battery voltage goes down, the light level does as well. As I reported here some time back:

micorpuck performance

LED Current more than doubled from 1.5 to 3.0 Volts, not what most would consider regulated.

Doug Owen

 

[milkyspit]

[ QUOTE ]
Doug Owen said:
[ QUOTE ]
milkyspit said:

Is there a semiconductor that could limit the voltage or current, so perhaps a safeguard could be built into the light preventing failure due to overdriving?


[/ QUOTE ]

Funny you should bring this up.

How about Mr Al's Low Drop Out (LDO) linear *current* regulator? A LM334 (3 lead 'transistor case' IC), a small transistor, three resistors, a capacitor and your LED(s)...

Doug Owen


[/ QUOTE ]

Doug, you never disappoint. /ubbthreads/images/graemlins/thumbsup.gif I especially like the sound of "3 lead" -- sounds about my speed! /ubbthreads/images/graemlins/grin.gif I'm thinking I can wirewrap the circuit together in prototype version even without a circuit board, and glue the components to the host housing wherever there are bits of space. The biggest downside would be the non-portability of the circuit (it's glued in!), but if the parts are really that cheap I don't think this will be much of an issue. Of course, a little later I could design a fancy little PCB and solder everything in place.

This might be a good match to the 2xCR123A variant of my project. I wonder how long the cells will be able to run a couple Nichias?

I now have two variants to explore: one is the minimalist single CR123A with yellow LED, and the other is the LM334-based circuit.

Hopefully no significant heat generation from the LM334 and other components for the small loads I'll be driving?

 

[Steelwolf]

Wow, go away for a day and come back to so many replies.

Milkyspit: I'll put up the schematics on a website and post the address, just as soon as I can find them. I like the idea of using only 1 123 cell at a time, since this eliminates the potential of reverse charging, especially since you're trying to suck the cells dry. At those levels, reverse charging becomes a very likely problem.

Since you're using depleted cells, the SatCure might work well. I think I even have a revised version from another CPF member that regulates the output. Haven't tried it yet though.

I wonder about the micropuck. Can you run say one or two 5mm LEDs on it?

 

[Steelwolf]

http://sg.photos.yahoo.com/steelwolf_sg

I've put up some pictures there which have the typical flipflop circuit, as well as the regulated SatCure circuit. There were a couple of pdf files too, but Yahoo! doesn't seem to allow guests to view those files. And some of the pictures are a little small, so if anyone needs a bigger picture, drop me a PM with your email.

I'll like to emphasise that none of these files and pictures are actually my work. I just accumulated them over the years as people post them. Some can still be found on active websites, others have disappeared. Some I have no idea whom they originated from so I can't give proper credit. Most were works of CPFers, so if you do see your stuff here, I hope you take it as a compliment of your ingenuity. But if anyone is not happy and wants the pictures removed, just let me know and I'll take them down.

Milkyspit: Good luck with your project. BTW, I'm not sure how efficient each circuit is.

 

[Doug Owen]

[ QUOTE ]
milkyspit said:

I especially like the sound of "3 lead" -- sounds about my speed! /ubbthreads/images/graemlins/grin.gif I'm thinking I can wirewrap the circuit together in prototype version even without a circuit board, and glue the components to the host housing wherever there are bits of space. The biggest downside would be the non-portability of the circuit (it's glued in!), but if the parts are really that cheap I don't think this will be much of an issue. Of course, a little later I could design a fancy little PCB and solder everything in place.

This might be a good match to the 2xCR123A variant of my project. I wonder how long the cells will be able to run a couple Nichias?

Hopefully no significant heat generation from the LM334 and other components for the small loads I'll be driving?


[/ QUOTE ]

You could easily WW them, I've had good luck with open wiring the few parts together (soldering). A really small module can be made by epoxying the to 'transistors' together, face to face and hooking the other parts to the six pins at one end (1/8 watt resistors make it easier). Such can easily be built on the back of the nine volt battery clip (or top of a dead one if you cut it open to salvage the connector), complete with LED. Two LEDs if you're going to run nine volts. It's easy to make it portable, but as you say, why? I see no real need for a PCB. Multi level versions use the switch body as the base, at least the few I've built.

If you're big on WW, you might consider a 'cordwood module'. Glue the resistors and cap to the sides of the transistor and IC with leads all the same way, cut 'em short and WW to the ends. A shot of hot melt glue (or epoxy) will easily protect it. While I've put in my time with a WW gun, I'm basically a soldering type myself. Not even big on crimped connections, not sure about IDC........

You can easily figure run time, divide your remaining mAH capacity by the current you set. This will drive you at full blast (what you set) until cell voltage reaches under 1.5 per, very very close to dead. I'd expect a full 52 hours (1300 mAH divided by 25 mA) at full blast for one Nichia, twice that at half that current, three times......

At the kinds of levels you're talking (say 50 mA total at six Volts total 'worst case') heat should not be an issue. (total of less than a third of a watt total for all losses).

Doug Owen