I read your nice article about the supercharged JT and wanted to see your circuit diagram, but it is too small to
read. Please can you send a better copy
You have to click on it more than once to get it to enlarge.
Thanks,I can see it well now!
What is the purpose of the 0 ohm resistor? Wouldn’t a direct connection be the same?
Yes, for measuring the lux, I shorted the 1 ohm resistor so that the LED output would be the actual light output. For measuring the LED current, I removed the jumper from the 1 ohm resistor. For each millivolt I measured across the 1 ohm with the DMM, it was the same as 1 milliamp LED current. After the measurement, the short can be used for full light output or the 1 ohm resistor can be removed and the 0 ohm jumper put in its place.
I think you’re placing R3 in the wrong place, the way you have, you are just measuring the current through the LED, but the current through the transistor is not measured (when it’s on), you should put R3 next to the battery to get all the current in the circuit. Don’t know if was clear.
Yes, maybe I should explain that. The R3 is zero ohms (a short) so it doesn’t affect the luxmeter measurements. The current was measured by an ammeter in series with the power supply. I compensated for the voltage drop this ammeter caused by increasing the supply voltage, so the voltage at the circuit was still 1.5V. So I didn’t need the R3 at all.
Before I got the luxmeter, the only way I had to measure the power to the LED was to put a 1 ohm resistor in series with the LED. Then for every millivolt I measured across the 1 ohm resistor, it equaled 1 milliamp of current. This is the same way many other JTers do it.
Yes, I did understand the 0 ohm resistor.
But when you use the 1 ohm resistor just in series with the LED, you are just measuring the LED current, so yo just calculate the LED efficiency, just see the LED manufacturer current(A) vs lux/lumens graph and I think you get similar figures to your calculations. But I could be wrong.
The manufacturer’s graphs are typical values, found by averaging many LED measurements. If you measure the LED you are using, the measurements may fall on this line, or they may differ by a significant amount. The only way to find out is to do the measurements yourself.
The way to find efficiency is to divide the power output (LED current times its voltage, which I assume to be 3.3V) by the power input which is 1.5V times the current, typically 70 to 80 milliamps for a conventional Joule Thief. Putting the resistor in series with the LED gives the LED current, but it isn’t exactly the same as if the LED had DC through it. So the LED current is relative, it will give a good relative current measurement, but not exact.
Hello, i was trying your joule thief circuit and i kinda got stuck at the part where you’re supposed to coil the wirings. could you go in depth into that, and perhaps provide a picture or two? thanks so much!
– electrical student
A lot of questions and answers about Joule Thief coils http://rustybolt.info/wordpress/?p=5773
Coil Terms explained http://rustybolt.info/wordpress/?p=6479
Joule Thief Inductance – Too much? Too Little? http://rustybolt.info/wordpress/?p=3497
Pictorial diagram of how to make a Joule Thief http://rustybolt.info/wordpress/?p=2583
Haha thanks so much for the info! But i was really looking for a picture or a video on how to coil that ferrite core.
Have only coiled normal joule thieves and their coils are relatively simple compared to this one. could u leave a video or perhaps share a picture of the final coil for reference?
thank you so much! ^^
Hello, thank you so much for your prompt reply! However link no.2 that you provided is actually dead (error 404).
alternatively, could you upload a picture or a video instruction on how to wind the wires around the ferrite toroid?
Was actually pretty confused as to how to coil 3 windings for the primary, and 1 for the secondary too. Any help on that would be very appreciated haha
thank you so much!
Oops, sorry, I had that URL marked private and forgot to make it public. It should work okay now.
For the average JTer, there is no signficant advantage to winding four windings instead of two. Two windings will work just fine.
You can search my blog for the word quadrafilar or quadrifilar (I may have misspelled it) and you should come up with more than one explanation. You may be able to find several videos on Youtube on how to wind a Joule Thief coil. I have posted only a few short videos and I don’t intend to do a video on that.
Hi Admin, are you able to do a Pictorial for this joule thief circuit. I try to replicate the circuit but not working at all. I have some experience in making conv JT to work but not this one. How many turn is requird? clockwise or anti clockwise wound affects the circuit? Anyway thanks and sorry to trouble you.
Yes, there are other blogs with pictures of the actual circuit.
http://rustybolt.info/wordpress/?p=2897 (This is a pictorial of the supercharged Joule thief.)
(for some reason the picture was blank but when I clicked on it, the picture showed. It was the same as the link “p=2583” above.)
http://rustybolt.info/wordpress/?p=128 (this has a good pictorial of a JT.)
http://rustybolt.info/wordpress/?p=5773 (many Q’s and answers about the Joule Thief, especially the coil)
http://rustybolt.info/wordpress/?p=4396 (picture of several different coils, which should work for both conventional and Supercharged JTs.)
The coil has two windings. With the toroid cores I’ve blogged (Fair-Rite 2673002402 or 2643002402) 8 to 25 turns is enough. Heavier wire is better. Both of them are usually wound at the same time, which is called bifilar.
Feel free to ask more questions if you need to.
Hi Admin, thanks again for your reply. I tried to buy the toroid core but both are discontinued product. Anyway I got a similar replacement. Base on your schematic, it is using quadrifilar which I tried and no way I could wound four 18 inch length of 30 AWG magnet wire. I have about 9 inch length left over on each of the wire. Also I do not understand the link up point from the schematic for the symbol “T1”. My understanding that a wound toroid only have 3 contact points but the schematic have 4 contact points which I tried and not working. That is why I hope you can do a pictorial like http://rustybolt.info/wordpress/?p=2583.
that should be very cool.
but I am wondering that is it really the help of such feedback design, perhaps core material and the transformer ratio will do the same.
I had have a power transistor 13002, 10mm toroid, and a superbright white LED, the Zi-C cell can run down to 0,45V (under load) and still on going, no load the cell was about 0.5v, 100ohm was the base resistor, however the light output is dimmed so obvious.
I have made Supercharged JTs with air core coils, and they work but the frequency is high, up to a megahertz. I have used many different cores for the Supercharged JT and they are all more efficient than the conventional JT. I have not tried using low permeability toroid cores for a Supercharged JT. The Fair Rite 2643002402 (medium to high mu) and 2673002402 (high mu) toroid cores are available from Mouser for 12 cents apiece or less in small quantities and make good Joule Thief cores.
how can i drive JT with a peltier TEG module..?
Here is one solution. http://rustybolt.info/wordpress/?p=6677
I made this circuit with 71 to 24 turns and different parts. (That’s what I calculated using the part number and the 18inch information.)
The circuit makes a highly audible whine at 13.3khz. I can change this by attaching a magnet to the coil. The mA do only slightly vary and the circuit now oscillates with 20.06khz. This is inaudible.
Did I do something wrong? Too many turns/windings?
I forgot: the voltage I read at the led is slightly less then the battery has. (1.206v @led and 1.211v @batt)
The circuit uses 38mA without the magnet and 40mA with. The LED Without the circuit could be lit, but way dimmer, and uses only 14mA.
I thought this circuit is voltage boosting?
It would be great if you could tell us the actual turns you used instead of some highly specific how-to. I don’t want to buy the ferrite core, I want to use what I have.
Well, if you use a DMM to measure the voltage you will get the battery voltage, minus a slight V drop across the coil windings. But if you use a ‘scope, you’ll see the high voltage spikes that cause the led to light.
I would stick to the 1 to 1 coil turns ratio. You don’t say anything about the toroid core. It should be ferrite, with medium to high permeability. You can get 5 of the green cores on Amazon or eBay for a few dollars, with free shipping. You might be able to get a small core from the base of a dead CFL light, but the tube is dangerous and easy to break.
Thanks for reply! I already thought of the possibility of a wrong/impossible measurement.
It is now working. I have no scope, but my DMM can measure hz. I did some alterations with very interesting results!
First I changed the resistor to a couple of different pots. I then found a 90kOhm pot to work best.
Actually, I left the core alone, doing it’s thing. (it’s a black ferrite core with unknown properties from the wires of a e-motor.)
Poking around in the pots while measuring the total amperage of the circuit, I found that, with high enough resistance, I could get the circuit to use as low as 0.5mA while still working!! And the LED was still lit quite bright!! O_O (This was with the pot turned to 91kOhm.) Later I made the choice to poke the pot until the circuit used about 10mAh, because starting at this value upwards the LED is getting useful. (Useful for a torch that is! As an indicator light it could be very much ok with 0.5mA.)
The circuit now doesn’t need the magnet to be inaudible. It is oscillating at about 10khz (10mA) or 6khz (0.5mA)
I am pretty much stunned. I could leave this circuit running with a new battery for over 90 days and the LED would still be lighting up my room faintl
Oh yeah, it’s a 1.5V alkaline battery now.
You said torch so i assume you’re in the UK? I have corresponded with one of your fellow Britons who lives in Chichester. He has done exactly what you said. He uses a very high gain transistor and a high value resistor, 56k or more, and puts a CdS photocell from the base to emitter, which shuts off the LED during the daytime. This makes the AA cell last for weeks. I’m not sure what the cell current is, but it’s much lower than the standard 1k resistor. You can search my blogs for his first name, Paul, to find out more.
Oh, made a mistake. It is oscillating at 22.8khz while using 10mA and 5.56khz at ~0.6mA.
Hey, the winding ratios of my SJTs have always been 1 to 1 just to keep things standardized and simple. Your different winding ratio might be a factor, and the core type and permeability may be another factor. My SJTs were always 2 to 3 times higher frequency than the same components in the JT circuit.
Seems like there’s a limit to the number of replys…
I’m flattered that you think I’m from Britain or any natively English speaking country. I’m actually from Germany and teached it myself by just using it. 😀
(I was using “torch” just because it sounds mightier.)
Anyway, I am not using a high gain transistor. I’m using a C33725, which could be this one: http://www.datasheetlib.com/datasheet/35980/bc33725_fairchild-semiconductor.html?page=2#datasheet
I just ordered extremely high, if not the highest, efficient 5mm LEDs and a couple of other electronics, including very high gain transistors. (30000 beta) The LEDs are from Nichia and are using only 20mA at 33kmcd/8lm. (they’re 15°) They will top whatever I build into my circuit and I’ll know the efficiency a bit more.
It’s already just using 0.6mA. I wonder how good it could get!
Oops! If your transistor had that high current gain, it’s most likely a Darlington transistor. Darlington transistors are a very poor choice for a Joule Thief, you can find my explanation be searching for Darlington in my blog, or go to the wikipedia wiki for it. Read about it and you will understand why it’s not good. If you want a high performance transistor, use the Fairchild KSD5041.
There are three current gain choices for the BC337, which is an excellent choice for a conventional JT. They are BC337-16, BC337-25, AND BC337-40. The -16 has the lowest gain range and draws less current. The -25 has the middle gain range and is the best choice for a JT. The -40 has a bit too high gain, and usually will not start with a new, fresh battery. If you change the 1000 ohm resistor to 3300 ohms, then it will start reliably. Honestly, I don’t know how the -40 will work with the SJT, because I have always used the -25 since it works so good.
I was stationed in Germany for more than a year, in the Army, in Wuerzburg. It was a nice place, but the army was not nice.
No, I was not using them. I tried but unfortunately I bought PNP versions.
Now I use BC337-25 which is slightly better than the BC547B. I also tried an SMD transistor BFR93AW but it is too dark, even though it looks like the circuit is more efficient. (I could light the led with only 20yA! Microamps!) Why these differences? Is gain the only important value?
BTW, I also use two separate COILS this time. (5278 97339) I did not wound one. Maybe tomorrow.
Anyway, now I use the BC337-25 and I have bright flashlight with only 5.3mA. Even at 1mA it’s still a useful torch in total darkness.
What do you mean when you say “high performance”? Is the KSD5041 the best possible transistor for a JT?
Okay, check out my blog about these high performance transistors. Beta or current gain is only one part of the higher performance high current switching transistors.
Here is my blog about using Darlington transistors.
I’ve also discovered that the type of diode makes a big difference. I’m now using a T4148 or 4148T diode, which could be the 1n4148T diode. Can’t find definitive answers.
And a 1.2V Akku again.
For the best diodes, the 1N4148 is just a regular PN junction diode. If you use a Schottky barrier diode, it is kind of like a half of a PN junction, so it has half as much voltage drop as a PN junction. Instead of 0.8 volts at 1 amp in a 1N4004, a 1N5817 may have 0.4 volts drop, and wastes half as much power.
I use the 1N5817 because I pulled them out of old PC boards. They’re more expensive than regular rectifier diodes, but they are better for this circuit. There are other Schottky diodes, I think the BAT46 is one. But it’s much lower current, maybe 1/4 amp.
Some newer power supplies are not using rectifiers because they may waste several watts as heat. Instead they use power MOSFETs, and turn them on and off depending on the voltage. These have a resistance of a small fraction of an ohm, so they can save several watts.
What do you think about Eddie currents building up in the core? Wouldn’t the circuit be more efficient if you use a laminated core?
Reply to 2016/09/03 at 20:37
That’s really helpful! I found a diode in my salvaged collection that has a voltage drop of just 0.32mV. For some reason I found that the bigger they are, the less drop they have.
If it’s a high current diode, like a rectifier, then your meter is probably measuring the leakage current, especially if it’s a Schottky barrier diode. Some rectifier diodes have a nonstandardized number, like SBD400, which usually is a Schottky barrier diode rated at 400 volts reverse voltage. The best thing to do is put some forward current through it and then measure the voltage drop. Put a 10k resistor in series then both across a variable power supply. Turn up the voltage until there is 10V across the resistor, which gives you 1 mA through both it and the diode. Then measure the voltage across the diode.
My meter is measuring the forward or reverse voltage drop. Why do you suspect current?
It is this one: http://mhtml.alldatasheet.com/html-pdf2/441872/VISHAY/SBYV27-200/219/1/SBYV27-200.html?lang=en
It’s not a schottky diode. The sheet states it has 0.88Vf at 3A. I suppose, since I’m measuring it with way lower current and it’s not an instantaneous but prolonged current, and including the charts in the sheet (saying about 0.55Vf), I may have a correct measurement.
But I still have to learn so much, I know nothing. (I do know that it can’t get any better than Vishay, though) 😀
Well, the voltage you gave was too low. You should measure at least a half volt for a PN junction rectifier, or about a quarter volt for Schottky. There is a possibility that the rectifier is damaged and is behaving like a low resistance. It is best to use enough current to make sure it’s not caused by the meter. I may not be understanding how you measured it because you didn’t give information such as how you measured it, which meter range you used, or possibly something else. It’s odd that they named the rectifier with the letters S and B, since that typically means Schottky Barrier, but this is not, it’s just a fast recovery rectifier.
Which meter range? It’s a DMM with diode test. It shoots through 2.4V (if I remember correctly. I measured it with another meter) and it shows me the forward/backward voltage drop. It’s a Extech instruments EX330. It’s quite good for it’s price.
There are other diodes that measure similar. I have two of the Vishay ones and some random ones. All other diodes are above 0.5V.
Well, maybe they are, as actually stated in the datasheet, special diodes with very low drop.
Your email address will not be published. Required fields are marked *