Talk:Joule thief

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Electronics (Rated Stub-class, Low-importance)
WikiProject icon This article is part of WikiProject Electronics, an attempt to provide a standard approach to writing articles about electronics on Wikipedia. If you would like to participate, you can choose to edit the article attached to this page, or visit the project page, where you can join the project and see a list of open tasks. Leave messages at the project talk page
Stub-Class article Stub  This article has been rated as Stub-Class on the project's quality scale.
 Low  This article has been rated as Low-importance on the project's importance scale.
 

Origin of name[edit]

The name "Joule Thief" was originally coined by Clive Mitchell in his write up of the EPE article with modifications shortly after initial publication - see http://www.bigclive.com/joule.htm - I note that whilst the Wiki article references his site, it makes no mention of his coining of the name - the article as published in EPE was entitled "Micro-torch Circuit".Nickds1 (talk) 23:41, 29 June 2013 (UTC)

Superior Device Already Patented[edit]

This article is biased in favor of one particular topography for running at low voltages. There are better ways to perform this task. For example US Patent 4,734,658 describes a system that runs from as little as 0.1 volt using modern, silicon devices (a JFET). I have added this in as an extra paragraph. 108.65.44.84 (talk) 23:57, 20 March 2012 (UTC)Duncan

You're missing the point - this article is about the "Joule Thief", not anything else - there are "better" (subjective term, that) ways of doing most things... Nickds1 (talk) 14:04, 26 July 2013 (UTC)

JFETs Have Very Limited Current Capability[edit]

The JFETs commonly found (one example is the MPF102) have an Idss of 2 to 20 milliamps. This is much less than the 2N3904, which cannot drive a LED to full 20 mA brightness. This current limitation is probably the main reason why the JFET is seldom seen in a Joule Thief circuit. The other reason is most obviously because they are much less obtainable than the PN2222, 2N4401 or BC337. I have built JTs using the TN0702, which seems to do fairly well. I also have use the 2N7000 MOSFET, which may need more than 1.5V to start (but I solved that by putting a 1.5V button cell in series with the gate).

One Youtube video shows a Joule Thief with many more turns on the feedback winding - more than 5 times as many as the primary. It also has _no_ current limiting resistor. This JT is capable of starting well below 0.4V, and will run down to less than 0.25 volt. I was surprised at its performance when I made one. However it will draw excessive current and most likely overheat if it is connected to a fresh AA cell or any source greater than 3/4 volt. Acmefixer (talk) 15:22, 31 May 2012 (UTC) Watson (contact is my Yahoo.com email address acmefixer)

Please Clarify this Article[edit]

Joule Thief LED Voltage Booster.jpg
Joule thief.jpg

2011 Dec 29 I checked back to see what is happening, and I just noticed that this new picture has been put up in the text. This picture should be removed immediately. One can see that there is no toroid coil, instead the two chokes (inductors) that look like light blue resistors next to each other are used as the coil. The one connected to the collector is acting as the primary winding and is creating a magnetic field, and by positioning the other choke next to it, it is picking up this magnetic field, enough to act as the feedback winding to keep the circuit running. This is about as shoddy a setup as one could possibly make. As soon as one choke moves away from the other, the coupling will stop and the circuit will stop working. It is not a toroid, and it is unacceptable to use this kind of mess for a JT. Not to mention that each of the two chokes probably cost more than a dollar (US) each, it is much more expensive than a toroid and a few feet of wire. I am going to edit the caption and add this information. Watson, acmefixer (at) yahoo.com 208.127.8.215 (talk) 19:13, 29 December 2011 (UTC)

I checked back and someone deleted the warning from the picture I talked about above. I will delete the picture. 2012 Jan 09 Watson, acmefixer (at) yahoo.com 204.75.251.6 (talk) 18:37, 9 January 2012 (UTC)

I've re-restored it. The Joule Thief is a "quick hack" and the air-cored transformer form is a perfectly adequate way to build it - maybe it could use more gaffer tape to hold them in position. There is no need for a toroid. Chokes are incidentally very cheap, and don't need to be wound by hand. It's a question of what you have to hand. Andy Dingley (talk) 19:15, 9 January 2012 (UTC)

I took a pic of a generic Joule Thief, uploaded it and inserted it in place of the other picture. I agree there is no need for a toroid, but everyone else uses one, so I used that; the idea is to give a representation of a typical JT. I have never, ever seen a JT built with two chokes mutually coupled by their proximity to each other, until I saw this one. I'm not saying there is anything wrong with this, it is just nearly unheard of. The problem I see with two chokes is that the coupling has to be the correct phase and magnitude (determined by the position of the chokes), if not the circuit won't oscillate or will oscillate poorly. Another detrimental factor that the small chokes have is high DC resistance. A JT coil should have less than 1/4 ohm DC resistance for the primary winding. Most small chokes have several ohms resistance, which is very detrimental to the efficiency. BTW, there is no transformation, the primary winding does the conversion, the feedback winding is there only to keep the circuit oscillating, therefore technically it is not a transformer. 2011 Jan 09 acmefixer (at) yahoo.com Acmefixer (talk) 21:53, 9 January 2012 (UTC)

The two-choke photo is mine and the circuit works acceptably well. As Andy Dingley pointed out there is very little difference between the chokes and a 1:1 air-core transformer, the chokes are cheap and readily available and much more compact than the more common torroid. Since the schematic in the article does show the doubly-wound transformer, it is a good idea to show that in the picture. If you could retake your photo against a white background, it would be a much better addition to the article. -- Autopilot (talk) 03:36, 10 January 2012 (UTC)
So how can you make a joule thief with two chokes? Is there a blog or something of Autopilot's work? 107.3.187.44 (talk) 06:11, 11 February 2012 (UTC)

...And neither of these pictures match the schematic posted in the following section, I'm assuming that there is no one single joule thief schematic, but the article does not make this clear, neither does it call out the winding ratio on the toroid coil, is it 1:1? Finally, I've seen a few circuits that use a single transistor+transformer to create square wave "AC" power at high voltage, can someone clarify how the Joule Thief is different, it seems to be pretty much the same principal, and could therefore just be sub-section of an article on DC power conversion with transformers and rectifiers. BrainSlugs83 (talk) 09:33, 9 February 2012 (UTC)

"neither of these pictures match the schematic"
Then you need to revise your concept of "match", per simple network theory and Thevenin. The circuits may not be topologically equivalent, but they are electrically equivalent. Andy Dingley (talk) 10:57, 9 February 2012 (UTC)
Thanks for the response Andy, it would be great if the description could call it out for people who don't have a strong understanding of transformer operation. The main difference I'm seeing is the placement of the 1K ohm resistor in one connects directly between the transformer and the transistor and the in the other between the transformer and the power rail. It makes sense that this could be equivalent, but I can't conclude that with certainty. BrainSlugs83 (talk) 00:36, 10 February 2012 (UTC)
Hey guys, I've seen Joule Thiefs with and without toroids as well as chokes and even small transformers, basically anything with a dual (or more) winding coils that produce a collapse in the secondary, is technically a joule thief, with and without resistors, and often various types of coils, and ratios of windings between the primary and secondary and beyond... which brings me to the reason for my logging in to comment here, it also states in this article that it gets its name because it steals every little bit of power from single cell batteries (paraphrased), but this is not true, what it does is steal power from an "unknown source", which is identified only by a voltage spike upon the field collapse, that has not "known" referenced source, and then channels it back into the obvious source and this is how it sustains even all but completely depleted batteries, that even after a rest those considered dead by a joule thief, are found to be revived for various lengths of time, and does so a variable number of times, until there is absolutely no potential detected within a power source. I'm sorry I do not have the references for these statements handy, but I will retrace my research as time allows, and jot down all the citations, I hadn't intended on contributing at the time I had accumulated this understanding, and since its inception there have been a plethora of spin off joule thief circuit variations, some also being dubbed with various partial names such as joule ringer, for example, and then some variations that use capacitors as power sources exclusively, once "juiced" by a regular power source and it will run for various lengths of time before needing another "tap" of a battery or power source. My point was to bring this awareness to the table for now, and once I collect the sources for citations, I will gladly provide them to bring this article up to expectations and being much more comprehensive while still maintaining the full integrity of the topic. I have not yet read through all the protocols for properly editing wiki articles, and do not wish to make an attempt at a live article myself, until I am confident I have learned to guidelines of doing it RIGHT, so I am putting what I have learned about this on the table for others to run with, while I was researching for when this circuit was originally discovered, or at least its unique characteristics, and I avoided using the ambiguous term being currently entertained as that source of power creating the very high voltage spike, that being "Zero Point" which is just yet another term for radiant energy which is also another term for the physics ignored "Etheric Field" they do not acknowledge the Ether or Aether, but the originator of this type of circuit, Nikola Tesla, knew it quite well, and this was also what I was searching for to night, the relationship of the discoverer's own work/research to confirm that this was derived from Tesla, as well as confirm the earliest date floating out on the web, being around 1976...so as I culminate all my own research, I will collect all the sources that are verifiable and reference them here at a later date.
I sincerely hope my input is helpful and inspires someone to take the ball and run with it, as there are numerous, real life workable applications for this ever evolving circuit, and a few reliably sourced websites that have published a lot on their own experimentations with it, and the results, but I do not believe this is acceptable for wikipedia "citations" or sources, unless they also source what they published about it!
I sincerely apologize if my input is in any way inappropriate, if so then delete it...as I said, life has disallowed my getting around to learning all the ropes here and just wanted to expand the horizon of this topic to do it justice, as it did not appear the current content contributions are fully aware of much of this, and a lot more that I was reluctant to include for not being sure I can even come up with valid sources, beyond being witness to how it can be applied, and the completely wide open field of experimentation people have been doing with this since learning about it myself, not to mention using it in various real life applications throughout my home! One example you can reference is a video on you Tube called Jeanne's light, or this other one by lasersaber called joule ringer and joule ringer 2.0 and 3.0 for which he lights his workshed 24/7 with it, or powers almost a dozen CFL lamps using about 12 volts on a joule thief variation and swapping the transistor for a 2N3055 power transistor. There are even some using a flyback transformer for the secondary and tertiary winding coils, as I said, I will come back with the facts once I document them all for this purpose, I hadn't realized I would be doing this at the time I learned of it all...Sorry for the length, but I'm not sure how clear it would be if I were too brief! --The Smoking Gun (talk) 07:18, 9 June 2014 (UTC)

[edit]

I also added typical input and output currents and efficiency. Watson, acmefixer (at) yahoo.com 208.127.8.215 (talk) 19:42, 29 December 2011 (UTC)

Hi, It's very strange that nobody really talks about JT to say what's really going on. Even wikipedia does not say, doth not seem to comprehend. The voltage spike is the Back EMF biasing or amping caused by the back stroke diode effect through the transistor B-E and the resistor. Negative entropy, it's shorting, biasing, amping. The spike is also juicing or at least assisting the battery. The Joule Thief is no Joule Thief at all. Some Geek likely called it JT just to get past the naysayers, the reptilians ;) Sirmikey1 (talk) 10:00, 20 January 2010 (UTC)

I did not SEE this or I would have refrained from my poor attempt above as I set out to collect the references so that I may properly edit and add exactly what you have stated, and I believe you are using the best terms to cover the facts while skating around the "no talk" aspects of where this spike comes from that I find in most sources of info about this. I apologize if I jumped the gun and used no better terms than those in use, but I was not really prepared to DO this just now, but noticed what you did about the original article... that it was "missing" a lot of proper terms and lacked understanding ...which although I have, I could not verbalize at this moment! But I'm not sure why I did not see this in the article initially though! Sorry for my two cents if inappropriate however the terms being used here ARE EXACTLY what occurs in this circuit and are perfectly accurate descriptions for the lay person to understand the article fully!

--The Smoking Gun (talk) 07:37, 9 June 2014 (UTC) The above terms amping, juicing, etc. are non-technical jargon and should not be used in a technical article. 208.127.8.41 (talk) 08:44, 24 December 2010 (UTC)Watson

http://www.nifty-stuff.com/LED-boost-circuit-joule-thief.php seems to claim that the circuit is just for feeding the LED. Therefore, I guess, the end terminals have been drawn into a very wrong place: rather, the LED should be replaced by terminals and the current ones should be cut? I don't comprehend the circuit fully, so I don't dare to fix the article. Anyone? — Pt(T) 23:14, 9 March 2010 (UTC)
@Sirmikey1 -- What are you arguing, exactly? Are you saying that the article is incorrectly named? That the description is insufficient? That the device's name is nondescriptive? Please clarify. Tylerl (talk) 06:09, 17 April 2010 (UTC)
The circuit in the link to "Nifty Stuff" above is quite clearly not the same circuit as in the article, though very likely the same function. SpinningSpark 09:43, 3 May 2010 (UTC)
The article is correctly named; but needs a lot more explanation before it will be easy to understand. if i build the referenced circuit, what will it perform like? where are the actual outputs? what controls the switching speed? for the values given, what is the nominal switching speed? does it drive the LED only? does it give stable voltage output? so many questions are unanswered. -- 99.233.186.4 (talk) 01:47, 1 September 2010 (UTC)
Not to mention: Does it need rectification? And, what are the transformer wirings? Also a graph depicting the voltage inputs and outputs of an ideal circuit would be great, so that over the life of a battery we can see at what point will it drop to 20v, 10v, 5v, 0v etc... BrainSlugs83 (talk) 09:32, 9 February 2012 (UTC)
All of those things are very difficult to put in because they vary so widely due to the nonspecific criteria of what qualifies as a circuit to be considered a joule thief, as I have noticed so many variations that retains this name, although more recently some have added terms like "ringer" as in joule ringer, in order to differentiate it from a standard joule thief. As I stated above I will collect all my references and come back to list them much more coherently than my input tried to get across, and then we can decide on it when it is all in one place for all to agree upon, and then anyone who knows the proper method of finalizing the article may then do so with all the actual facts and proper terms and be able to include the variations, and have sources to reference it all as well. It will take a bit of time though! The Smoking Gun (talk) 07:37, 9 June 2014 (UTC)

Dubious[edit]

What do a Joule thief and a nickel-cadmium rechargable battery have in common?! There's no inductance nor an oscillator in the latter. Also, the claim needs a reliable reference. — Pt(T) 11:03, 8 April 2010 (UTC)

Though I don't know for certain, I assume the author meant that circuits using NiCd batteries employ this type of DC-DC converter to lengthen the useful life of the battery. I've never heard of it being the case, but that's the only sensible way I can read the statement. Tylerl (talk) 06:21, 17 April 2010 (UTC)
I've deleted the claim, cleaned up the article, added a description of its operation and amended the diagram. SpinningSpark 15:26, 8 May 2010 (UTC)

Operation Principle[edit]

I consider the description of the operation of that circuit to be misunderstandable or even wrong. The winding at the base of the transistor should be considered the secondary and the winding at the collector the primary winding in a model of a nearly unloaded transformer, as the base current is low compared to the collector current. From the transformer law, the voltage across the base-connected winding equals the voltage across the collector-connected winding (if you ignore ohmic losses).

On start-up with a 1.5V battery, some current goes through the resistor (which is around 1k typically) into the base of the transistor, turning it on and giving rise to a collector current. The collector current causes a voltage drop over the collector-connected winding, which will cause a voltage rise on the base-connected winding, turning the transistor on even harder, so that the transistor enters saturation with a collector-emitter-voltage of around 0.2V and 1.3V across the transformer. This causes the base current to be (1.5V(from the battery)+1.3V(via the transformer)-0.6V(transistor base-emitter voltage drop) = 2.2V)/1kOhm = 2.2mA. The collector current rise is limited by the inductance of the transformer. As typical small-signal transistor has a current gain of around 200, so the collector-emitter path keeps saturated until the collector current has risen to around 400mA (this is the on-state of the blocking oscillator) or the core saturates and ceases to produce the base voltage boost. At that point, the collector-emitter voltage starts to raise, and thus the voltage over the collector-connected winding decreases. This also decreases the extra voltage induced in the base-connected winding to decrease and thus the base current to drop. A dropping base current on the other hand makes the collector-emitter voltage rise even higher, which will further reduce the base current, and quickly turn off the transistor. Now the energy stored in the ferrite bead is discharged through the LED, the voltage across the collector-connected winding now is around 1.7V that gets added to the 1.5V from the battery to obtain the 3.2V forward voltage of the LED. The result is 1.7V also induced in the primary of the transformer, which pulls the base voltage down to -0.2V, keeping the transistor turned off (this is the off-state of the blocking oscillator). At the time the energy of the coil is discharged and the current drops down to zero, the oscillation restarts like on power-up. —Preceding unsigned comment added by 87.123.67.77 (talk) 16:56, 21 November 2010 (UTC)

Operation theory has been described correctly by "Vintage Dave" in the mean time. Thanks! —Preceding unsigned comment added by 87.123.119.98 (talk) 16:04, 23 December 2010 (UTC)

I suppose this will either be edited or fall on deaf ears as usual so listen up. a three volt LED should not normally light without at least a three volt power source. the articles explaination of back EMF flyback conversion in the inductors/transformer, whichever you decide it is at the time, does not adequately explain what is happening. what it does to is try, to explain the mechanism with general ohm's law. problem is, the 'transformer' is not a linear circuit component. and, don't bother to post some dismissive tripe as I can see directly above this, I won't be back to check, I just thought you should know. non-linear. ok bye now. —Preceding unsigned comment added by 90.207.180.68 (talk) 18:20, 9 February 2011 (UTC)

I made a few changes but gave up when I saw that there was too much to simply be edited. It needs to be rewritten.

I disagree with the statements that the internal resistance of the battery must have any influence on the circuit operation. There are at least two reasons. A joule thief will operate without any problem from a highly regulated power supply with essentially zero internal resistance. On top of this, the power supply is bypassed by a large value capacitor - 100 uF - at the power input to the circuit, so it has zero impedance for all practical purposes.

I disagree with the statement that the 'enormous' current flow causes the battery voltage to drop, not only for the above reasons, but also because the transistors listed do not have the capability to conduct enough current to cause the voltage drop. The circuit will work with a BC547, but it cannot conduct more than a hundred or so milliamps, much less than the 'enormous' current stated.

In the graph is shows V_BE to be a large negative spike approaching negative 20 volts. All silicon BJTs (bipolar junction transistors) have an emitter to base breakdown voltage of 5 to 6 volts, and any voltage higher than 7 to 9 volts will cause breakdown and permanently damage the transistor. This graph might apply to a vacuum tube AKA thermionic valve or a FET, but definitely not a BJT. By acmefixer@yahoo.com —Preceding unsigned comment added by 208.127.8.75 (talk) 15:01, 13 March 2011 (UTC)

Upon further inspection of the graph, I see that most of the other waveforms are impossible to obtain. For instance, it shows the voltage across the LED to spike at nearly 20 volts. The LED is a diode with a forward voltage of a few volts, 3.2 volts for the typical white LED. The current put out by the coil could never be enough to cause the LED voltage to rise above its forward conducting voltage by more than a volt or two. If it did, the current would be so high as to cause the bond wires to overheat and melt like a fuse. Another graph shows the battery voltage to drop to a half volt. As I have stated, this could not happen with a regulated power supply with an internal impedance of zero. The other graphs have discrepancies and this leads me to conclude that the graph was synthesized from some made-up numbers and is definitely not from a real Joule Thief circuit, and should be ignored and/or deleted. Above added 2011 Mar 13 by acmefixer@yahoo.com.

I changed the "switched mode power supply" to voltage booster, because the circuit is not a power supply, and is not regulated as is a typical switched mode power supply. It is simply a blocking oscillator and there should be a link to the wiki article on blocking oscillator, which has a good explanation of its operation. Above added 2011 Mar 13 by acmefixer@yahoo.com. —Preceding unsigned comment added by 208.127.8.75 (talk) 14:52, 13 March 2011 (UTC)

Also, the term "transformer" is used in the text. There is no transformation because the LED load is connected to the same winding as the transistor's collector, and the base-connected winding is not needed for the voltage conversion process. The base-connected winding serves only to invert the feedback signal to maintain oscillation. This base-connected winding can be removed and a second transistor added to invert the signal and maintain oscillation. The proper term for this two winding inductor is a Coil. Another pointless statement in the text is the part about the core saturating. Anyone can make a fully functional Joule Thief with a coil made of 5 meters or 16 feet of 24 AWG (.5mm) solid wire bifilar wound into a 'donut' about 25mm or 1 inch diameter. This has an air core, and an air core has nothing to saturate. This invalidates all the statements about saturation. Above added 2011 Mar 13 by acmefixer@yahoo.com. —Preceding unsigned comment added by 208.127.8.75 (talk) 15:17, 13 March 2011 (UTC)



The text gives no information on the influence of the transistor on the performance. The JT circuit puts a heavy demand on the transistor because of the low supply voltage. For full brightness, the transistor must have very low saturation voltage at several hundred milliamps peak current. Thus the transistor should be rated for low Vce(sat) at more than 200 milliamps. Also, the current gain at very low voltage must be adequate to keep the transistor in saturation (base current is limited by the 1k resistor). The 2N3904 and BC547 are substandard when it comes to these requirements. The 2N4401 and PN2222A are pushed to their maximum and can almost light the LED to a full brightness, typically 15 milliamps (full is 20 milliamps). The BC337-25 does better, and is a good choice for a Joule Thief. The type of transistor also determines the efficiency of the circuit (typically 40 to 60 percent). Contrary to what the text says, some of the input power is wasted in the transistor, both in losses across the emitter to collector, and in the current needed to drive the base. So using a better transistor helps reduce those losses. Added 2011 Jul 2 By acmefixer@yahoo.com



Indeed. Additionally, the designation of the primary and secondary windings are inconsistent. There are so many holes in the current explanation that it cannot be considered an improvement on the earlier one that attributed the working of the circuit (broadly correct, in my view, at least with supply voltages greater than 0.7v) to the non-linear magnetic characteristic of the ferrite bead.
I was intending to do a re-write of the section, but have realised my own understanding of the operation of the circuit is far from perfect... examining a real example of the circuit (rather than a poorly simulated one), reveals a number of behaviours that require further investigation.
For example: with supply voltages in the range 0.8V to 1.7V the transistor "on" time decreases as supply rises (as would be expected for a system limited by magnetic saturation of the inductor), but the "off" time increases... and I've not figured out why. Once the energy in the core has been dumped into the LED the transistor should be free to turn on again. If that energy is limited by the maximum flux the core can hold, shouldn't the "off" time should be independent of supply voltage?
Also, at supply voltages below 0.7V, the "on" time reduces as if the resistance theory was correct, though the supply voltage does not exhibit a corresponding dip... particularly if, as suggested above, a decoupling capacitor is employed. Presumably the feebler drive to the base means the transistor is not fully saturated and Vce rises enough to tip the circuit into the "off" state, but I haven't had time to confirm that.
Finally, there is the issue of the way it behaves at very low supply voltages. My versions continue to oscillate down to 0.35V (though the result is insufficient to light the LED), but won't self-start with a supply voltage below about 0.55mV. This suggests there is a further mechanism, possibly involving cross-capacitance between the two windings, that sustains oscillation as the supply is reduced. An experiment with separate, rather than bifilar, windings might establish the veracity of that idea. — Preceding unsigned comment added by Fazerider (talkcontribs) 01:05, 16 March 2011 (UTC)

Analog EE here -- the JT won't start until the supply voltage is high enough to forward bias the base emitter junction, but once it starts oscillating the peak positive voltage on the winding that drives the base is nominally 2 times the supply voltage (assuming Vce when the transistor is on is close to 0 volts and that the coil is wound with the same number of terms on the primary and secondary). So a .35 volt supply will still make 0.7 volts of base drive and (barely) keep the oscillations going. Cross capacitance between the winding has no bearing. The duty cycle the circuit as the supply varies depends on the gain of the transistor, the value of the base resistor, the Vbe and the saturation current of the core. Various elements dominate - with some supply voltages the transistor collector voltage may start to rise when the collector current rises higher than Hfe times the base current. The Hfe of the transistor is not a constant - as higher supply voltages and thus higher peak currents the gain drops, and even though the base drive is higher when the supply is higher, the transistor may drop out of saturation. If the core of the coil is small it may saturate before the transistor does - but perhaps only when the supply voltage is relatively high. At very low supply voltages the base drive will, as noted above, drop, and because of the ~.5 to .7 volt forward drop of the base emitter junction, and the logarithmic relationship between Vbe and Ib, base current will vary extremely non-linearly with the supply voltage (especially as you get below .8 volts). Which is a pretty long winded way of saying that no simple formula or general rule will describe the operation of the JT in detail - and the formula given on the article page is mostly BS. —Preceding unsigned comment added by 98.248.150.196 (talk) 02:03, 11 April 2011 (UTC)

This is another EE. I agree that the formula is BS. At the very least, I want to see how the formula is derived. On the other hand, I think the description of how the circuit works is very good, and it's as easy to understand as it can be. In response to the March 16th comment of "Why does the on time decrease when supply voltage goes up?"... Look up duty cycle, dude. April 18, 2011 — Preceding unsigned comment added by NewMeat1 (talkcontribs) 04:27, 19 April 2011 (UTC)

Thanks, Analog EE.
A couple of days after I wrote that I realised that regarding the base drive as an a.c coupled, inverted version of the collector voltage summed with the battery voltage made things clear. This is effectively clamped by the forward conduction voltage of the base-emitter junction (at least, while the "off" pulse is long enough to make the mean d.c level sufficiently high). Interestingly, a longer "on" state means that the reverse voltage "off" spikes the base is subjected to get more extreme with lower supply voltages. With some suggested loads requiring slightly higher voltage than a white LED, the transistor is likely to fail as battery voltage declines... the reverse Vbe limit being exceeded, rather than that of the Vce rating, being the cause of destruction.
Despite now understanding the low voltage operation I did experiment with separate windings to find out why bifilar is normally specified. The performance was indeed identical, but making the actual windings was considerably more fiddly!
The current explanation of operation in the article is a great improvement, though seems vague about which switch-off mechanism applies... it could make it clear they both do: with a sub 0.7v supply the base drive becomes insufficient to maintain the transistor in a saturated state as Ic rises, whereas with a battery voltage above this, magnetic saturation of the ferrite becomes the dominant effect.
The only remaining puzzle for me (notwithstanding Newmeat1's ill-aimed misfire) is the increased "off" time with an increasing supply voltage.
—Preceding unsigned comment added by Fazerider (talkcontribs) 21:00, 19 May 2011 (UTC)

Regarding "If the load on the circuit is very small the rate of rise and ultimate voltage reached is limited only by stray capacitances, and may rise to more than 100 times the supply voltage." and the rest of the following paragraph. It is stated that the voltage of the collector connected winding may rise high enough to exceed the Vceo of the transistor. Long before this happens, the voltage of the base connected winding, which is the same but opposite polarity, will exceed the maximum emitter to base voltage, and the emitter-base junction will break down or zener and dissipate the energy. This may cause the current gain of the transistor to be permanently reduced (damaged). (This assumes that the windings are the same number of turns.) Oct 13, 2011 by acmefixer@yahoo.com--208.127.8.215 (talk) 12:16, 13 October 2011 (UTC)

I see all that inaccurate simulation nonsense reappeared.
I took some oscilloscope screen shots a while back. Perhaps if I find the time I'll dig them out and add them, but for now I've removed the rubbish, tidied some of the remaining description to reflect better that the principal mode of operation depends on the ferrite saturating and fixed some inconsistencies where the windings were described inconsistently... settling on the convention that the primary is connected to the collector and the secondary to the base drive circuit. Oh, and also amended the failure mode description.
— Preceding unsigned comment added by Fazerider (talkcontribs) 16:08, 21 October 2011 (UTC)

In the paragraph starting with "At lower supply voltages a different mode of operation takes over: The gain of a transistor is not linear with VCE." The BJT (bipolar junction transistor) doesn't change any as the supply voltage decreases. As far as I know the models (ebers-moll, gummel-poon or whatever) apply to all supply voltages. The limitations that were not predominant at higher voltages become predominant at lower voltages. The demands of high current and low voltage mean the transistor's internal resistance of a few tenths of an ohm to a few ohms become a limiting factor. Later in the paragraph the term pinch-off is used. BJTs do not have a channel that can be pinched off, as can happen in a FET. Almost all Joule Thiefs use a BJT.

As I believe I've mentioned before, the saturation of the core does not have any effect in Joule Thief that uses an air core coil. It works essentially the same as the ferrite toroid core. I'm reluctant to believe the statement that core saturation has much effect, because the power levels of a Joule Thief are only a few tens of milliwatts and this is not enough power to saturate any but the tiniest of cores. Posted by acmefixer@yahoo.com 204.75.251.6 (talk) 21:51, 2 November 2011 (UTC)

I think "To summarize, once the current in the coils stops increasing for any reason," should be "To summarize, once the magnetic field in the coils stops increasing for any reason," to cover core saturation. David R. Ingham (talk) 03:16, 15 June 2014 (UTC)

@ David R. Ingham. Good point, I've amended as you suggest. @ others who cite behaviour with large ferrite cores or air cores as evidence that core saturation is irrelevant, the waters are getting muddied by referring to circuits that should properly be in the blocking oscillator section. Clive Mitchell's design was quite specific about the core and windings for the inductive element and those small anti-parasitic beads typically exhibit the onset of saturation at around 0.5 Amp.turns. It is unfortunate that the catchy title Mr Mitchell devised for his circuit is being applied to almost anything with a similar design without regard to the component characteristics. — Preceding unsigned comment added by Fazerider (talkcontribs) 23:18, 29 June 2014 (UTC)

Such a simple wording change makes it sound first of all as if the transistor is controlled directly by the magnetic field, as if it were some sort of magnetometer. Secondly there's no sourcing, nor even a credible attempt at giving the figures, to demonstrate this claimed core saturation. Andy Dingley (talk) 23:33, 29 June 2014 (UTC)