My DIY Joule Thief Hurricane Lamp Project: Lots of pics!

[Darvis]

Thanks!! I found the parts.... So, wow, I really think I'm going to give this a try... the parts are cheap!!! And your thread was great as well...

I may need to ask you some questions when I get the parts, though, as I am just learning this stuff and I'm not entirely sure I'll know what to solder to what. My perfection of the lamp has purely been cosmetic and I've made my basic circuit more compact, figured out a few things here and there, but have not yet deviated from the classic JT parts as of yet...

You have all been a great help with this, I need to get some newer pictures up of some recent builds, including my dual circuit, dual LED lamp with selector switch and both warm and neutral LEDS!! Woohooo!!

[Darvis]

Complete and utter success with my first ZXLD circuit... unreal!!! Man, it really reduces the footprint. I have to admit, I was not prepared for how small the components really were and probably made this first one bigger than it needed to be, but it works!!

Here it is next to a traditional JT circuit:



This may be a totally bad practice, but I was having trouble getting everything to stay in place to solder, even with the helping hands, to, er... help, so I simply used a drop of crazy glue to hold the parts in place and it worked great...


Now I just gotta get it into a host.

[tylernt]

Yeah I was a little surprised when I got my first ZXLD381 too, sneeze and that thing is gone! :-P

[Justin Case]

I'm interested in how low the Vin can be and still start up the ZXLD381 and how low the Vin can be to keep the ZXLD381 running. I've gotten startup and good output from a AA alkaline with less than 0.9V open circuit, and 0.549V under load using the basic JT circuit.

[Darvis]

So I was able to get the new circuit integrated into a host....



From right to left: A 1.25" compression host with 2 JT circuits each driving one LED with a single pole, double throw switch with center off. This allows the use of one neutral LED (brighter) and one warm LED (low mode) depending on the needs. If you look close, you'll see the second LED towards the front, in front of the rear (lit) LED

Next to that is what was my standard build using a 1" compression fitting and an on/off toggle... this was about as small as I was getting it...

Enter the ZXLD (third from the right) and now I'm in a 1/2" compression fitting and have eliminated the switch all together... Small, clean, awesome!!!!!! (well, awesome to me anyway)

For size reference next to a 6P and an 18650:



I have gotten my taditional JT down to .58-60v before in a straight run, but found it needed about .90 (as you noted) to actually start the circuit. Not yet sure about the ZXLD, but I aim to find out.


Oh, and here's a reading lamp that only a mother can love:

[Justin Case]

Mine will start at ~0.5V from a bench power supply. The 0.9V referenced before is the OCV for a "spent" AA. I can run such a AA cell so that it sags to ~0.55V under load. Turn off the JT and immediately turn on the JT again. The circuit clearly starts back up since the LED lights up with a nice, bright, steady beam. Within that brief period of time, the AA cell certainly has recovered its OCV to no more than perhaps ~0.6V.

[Darvis]

Makes sense. My guess is that I've duplicated your results without knowing it, if that makes sense. Because I've not actually measured anything other than the voltage of the battery going in and then coming out, I've not seen the the OCV recovery part of it, etc.

I did experiment with different sized ferrite beads and windings to see if it would improve output or runtime (in other words, could I get down to .30v?) and found no real difference. There seems to be no gain from using more winds, or larger diameter wire- a good wind is a good wind it would seem... A bad, or crossed winding however, makes for some truly weak (or no) output!!! I learned that lesson the hard way. It has to be done right, but over doing it gains nothing...

I can say that with my traditional JT circuit, the light does start to dim around .70v (this is based on me pulling NiMH batts for recharging) It's my indicator of when they need to be dropped into the old charger. I've run alkalines all the way to almost off and towards the very end, they're just barely glowing. But honestly, until that ~.65-.70 mark, there is really no perceptable change in brightness, so the circuit does provide some degree of regulation as well.

This has become quite the cool learning project!!!

[Justin Case]

Do you know anything about the ferrite you have, e.g., AL value? How many turns are you using? It does seem to me that the basic JT circuit is fairly insensitive to coil inductance, as long as you are "in the range". I wound a powdered iron coil (Micrometals T44-52) with calculated inductance of only 2.8 uH and that circuit was very squirrely. It would work "bare" -- the parts all soldered directly to one another. But if I mounted the parts neatly onto perf board, it wouldn't turn on. Just a little change in how tightly the coil was wound seemed to affect the JT. When I wound more turns on the powdered iron core to get about 6 uH, the JT became reliable. I also have some ferrite cores (Epcos B64290L0044X038 -- T38 material) with very high AL (5100) and only 4 turns give a calculated inductance of about 80 uH.

Depending on what ferrite formulation you have, you might try something extreme like just one turn to get the min inductance to see if your JT will still work.

A Si transistor is not going to work well (i.e., at all) with only 0.3V. You'd have to go to germanium.

[Darvis]

I do, this is straight off of the website I got them from and I'm getting 9 turns on mine using standard twisted pair phone wire (22ga?):

OD(A) = 0.5 in / 12.7 mm +/- 0.25 mm
ID(B) = 0.281 in / 7.15 mm +/- 0.20 mm
Ht(C) = 0.188 in / 4.9 mm +/- 0.25 mm AL=440 +/- 20 % uH=(AL*Turns2)/1000
Actual measured AL using 10 turns #28 wireTemperature Stability (ppm /°C) = 12500Color Code = shiny blackApplication Freq Range
Wideband Transformers 5 - 400 MHz
Power Transformers 0.5 - 30 MHz
RFI Suppression 5 - 500 MHz

Now, I admit, I have NO idea what any of this means, I literally bought them based on their size. The website is actually pretty cool, it's called "kits and parts" and this bead is part number FT50-43

I will try what you suggested above and report back on what I get. I have the circuit rigged on a breadboard, so I can play with it as needed.

[Justin Case]

FT50-43 means a ferrite core, 0.50" outside diameter, material formulation 43. #43 looks like it is spec'ed for EMI/RF suppression. For the JT application, you probably care most about the AL value, which relates the number of turns to the inductance. For 9 turns, the inductance is 440*9^2/1000 ~36uH (+/-20%). To explore the low end, I might try 2, 3, and 4 turns to see if the JT still works.

[notrefined]

I'm also curious about the minimum startup voltage for the ZXLD joule thief...eagerly awaiting results

[Cypher_Aod]

Hey guys, i'd quite like to make a couple of joule thieves, and i have one in particular that i'd like to ask about.
Basically, i have an ANCIENT Luxeon I Red-Orange emitter which i'd like to joule-thief-ify.

Hopefully this LED would be perfect for joulethiefing because it's low Vf of only 2.31V and a max rated current of 385mA so it's unlikely to get killed by fresh cells.

[Darvis]

I'm also curious about the minimum startup voltage for the ZXLD joule thief...eagerly awaiting results

I'm going to be building one tomorrow and will begin tests after that, will definitely post results...

[Darvis]

Hey guys, i'd quite like to make a couple of joule thieves, and i have one in particular that i'd like to ask about.
Basically, i have an ANCIENT Luxeon I Red-Orange emitter which i'd like to joule-thief-ify.

Hopefully this LED would be perfect for joulethiefing because it's low Vf of only 2.31V and a max rated current of 385mA so it's unlikely to get killed by fresh cells.

All I can say is build the circuit on a breadboard and give it a try.... based on those specs, it should light right up. My guess is that the JT circuit is boosting to about 3.4v max as I can get LEDs with a required Vf of 3.2 to light, but I have one that requires 3.6 and it won't go. I never measured it and probably should...

I have a bunch of cheapy red LEDs that came with these neat chrome LED holders that I bought and they light right up direct on a AA battery- I've also wired them into the JT circuit on my breadboard and they light up and don't poof at all. So they have a much lower Vf requirement and still run well at over a Vf of 3.0. Maybe they have that range, but my guess is the Luxeon is much more robust and can take it.

[Cypher_Aod]

All I can say is build the circuit on a breadboard and give it a try.... based on those specs, it should light right up. My guess is that the JT circuit is boosting to about 3.4v max as I can get LEDs with a required Vf of 3.2 to light, but I have one that requires 3.6 and it won't go. I never measured it and probably should...

I have a bunch of cheapy red LEDs that came with these neat chrome LED holders that I bought and they light right up direct on a AA battery- I've also wired them into the JT circuit on my breadboard and they light up and don't poof at all. So they have a much lower Vf requirement and still run well at over a Vf of 3.0. Maybe they have that range, but my guess is the Luxeon is much more robust and can take it.

Sweet! I'll build a traditional-style Joule-Thief for now for maximum battery-drainage

[Justin Case]

All I can say is build the circuit on a breadboard and give it a try.... based on those specs, it should light right up. My guess is that the JT circuit is boosting to about 3.4v max as I can get LEDs with a required Vf of 3.2 to light, but I have one that requires 3.6 and it won't go. I never measured it and probably should...

That's strange that the apparently higher Vf LED wouldn't light. Vf is not a single value. It depends on If. 5mm white LEDs should be able to light up at around 2.5V. A basic JT circuit ought to be able to boost a AA cell to at least that, even with a very dead cell. It'll be at moonglow level, but it should light up.

I'm going to test coil inductance in the opposite direction -- increased inductance. I have some Epcos B64290L0044X038 ferrite toroids. I figure if I wrap 15-20 turns, I could get 1-2 milliHenries.

[Justin Case]

I wound 22 turns of 30 ga solid wire on the Epcos ferrite toroid and the JT was able to run on a very depleted AA Energizer alkaline. By the time the cell measured 0.502V under load, the 5mm white LED was basically in moonglow mode. But it's been running steadily for hours, whereas my basic JT with Micrometals T44-52 powdered iron core can't deal with this "dead" AA cell (the LED lights, fades, starts blinking, and goes out, all within a few seconds).

The rest of the two JTs are the same: 1Kohm resistor, 2N4401 transistor, and 5mm high brightness white LED. The calculated inductances are 2.5mH for the Epcos core and 6.9uH for the Micrometals core.

[IcantC]

This is a great idea! Plan to make a few for emergencies.

Now a q for you guys, how can I mod this diy

http://www.evilmadscientist.com/article.php/nightlight

So my JT can run a white led and run off one or 2 aa?

[Darvis]

Seems to me the easiest thing to do would to make that photo sensor circuit the input to the JT (the leads to the JT would take the place of the red LED in the design) It would control the power in and all you'd need to run it would be AA. I might have to give that one a whirl.

Jason... I did not get chance to play with windings as of yet, but plan to in the next few days.

I did confirm that my traditional (non-ZXLD) JT would fire up at .74v at least, more tests to come.

[IcantC]

Darvis, if you can please try let me know! I don't have any parts or the aptitude to troubleshoot the circuit.

I plan to order some some stuff and try a basic JT, the darvis JT and then the light sensing one.

[notrefined]

Darvis, did I read that correctly? The ZXLD based circuit will start and run from an OCV of 0.74v? Or is that start and run from a bench supply at 0.74V?

[Darvis]

Darvis, did I read that correctly? The ZXLD based circuit will start and run from an OCV of 0.74v? Or is that start and run from a bench supply at 0.74V?

No, that was for the traditional JT circuit that I had on hand. Thanks for pointing that out! I edited the post to make it more clear.. I still have to build a new ZXLD and do some tests.

[potpot]

Pretty interesting project. I'll be definitely building one (or more) of these for use as a battery drainer/night light around the house.

[cnjl3]

I have made a couple of JT's with the PVC repair coupler which included both a switch and battery compartment.
I decided to eliminate the battery compartment by soldering magnets to the end of my wires.
I used both 28 and 26 gauge wire on my ferrite bead - but I couldn't tell if one is brighter.
I like your idea of a "twisty" on/off so I went ahead and ordered some glass magnet switches so
that I can twist on/off with a magnet.
I don't know if you can see but I used part of a BlueBell ice cream stick to hold my components--YUM!

[datiLED]

Here is my take on the ZXLD381 circuit. This is the version with a smoothing diode and capacitor. The circuit boards are 14mm, 10mm and 7mm in diameter. For reference, they are sitting on a dime. The circuit on the left is using a warm white XP-G LED, and is driven at 50mA. The other two circuits are driven at 18mA.




I also have a thread about building battery drainers using the Joule thief circuits. I chose to use the Supercharged Joule Thief circuit from a weblog that appears to have been removed. It had lots of good information and testing data on the Joule Thief. Here is the schematic for the circuit. I use a 2N4401 transistor, and a small toroid from Mouser.

[tylernt]

Here is my take on the ZXLD381 circuit. This is the version with a smoothing diode and capacitor.

Correct me if I'm wrong, but I don't think the XP-G needs the smoothing cap. The 5mms definitely do, but the XP-G should be able to handle the 380mA pulses no sweat?

[datiLED]

Correct me if I'm wrong, but I don't think the XP-G needs the smoothing cap. The 5mms definitely do, but the XP-G should be able to handle the 380mA pulses no sweat?

The XP-G could certainly handle the pulses. For the cost of two extra components, I chose to err on the side of caution and design the PCBs with the smoothing diode and capacitor. Regardless of LED used, I build all of the circuits the same way, and change the inductor to set the current to the LED.

[wquiles]

Most of these great small circuits are meant for a 1.5 volt cell (presumably discharged), but has anyone tried using discharged CR123 cells instead?

[Justin Case]

My JT seems to work with a 3V Li primary (I used a Varta CR2 cell that measures about 3.1V open circuit). The SPICE simulation looks ugly but I guess that the rapid pulses save the 5mm LED.

[tylernt]

The ZXLD381 datasheet says "the pulse control circuit ... requires the load voltage to be at least 0.8V greater than VCC. (The device will function with load voltages smaller than this but output current regulation will be impaired.)" I assume this means that you'll be sending a lot more current to the LED than expected. You could compensate by using a larger inductor, but some experimentation would be required... ideally, with a 'scope, so you can see the current peaks.

I'm sure a JT could be tweaked to run efficiently at a higher voltage, but I lack the electronics experience to know what to tweak. Maybe just use two LEDs in series?