Joule Thief 101

[sm0ky2]

Firstly, as you have previously stated, the majority of your group's earlier explorations with the Joule Thief circuit were anecdotal observations.  You never seriously analyzed scope traces to figure out exactly what was happening with your Joule Thief replications.

Secondly, let's say you can split the battery voltage response into two ranges.  Let's say that the normal battery voltage range is 1.5 volts to 350 millivolts. In this range the Joule Thief will work just fine with a fixed base resistor and switch properly.  It's in this normal range where if you change the value of the base resistor then the transistor will keep switching normally and the Joule Thief LED will look pretty much the same.  Naturally, common sense is telling you that you can change the value of the base resistor within certain reasonable limits, and if you exceed those reasonable limits in either direction then the Joule Thief will cease to operate normally.

Then let's say that there is another voltage range, and that voltage range is between 350 and 200 millivolts.  In this range the Joule Thief does not act as a normal switching device and all bets are off.  Playing with the base resistor will do something including increasing the brightness of the LED.  But I will stress again that the Joule Thief is not acting like a standard switching device at this very low voltage range.

Needless to say, all of my discussion in my previous posting applies to a standard Joule Thief operating voltage range of somewhere between 1.5 volts and say 350 millivolts.

MileHigh

By these comments, it is obvious you have never played with a Joule Thief.

In your first "range of voltages", using a VR, you can find frequencies with brighter luminescence.
While these are not always directly related to any form of "resonance",
switching frequencies and internal diode capacitances can result in brighter "light".
You will hear Bill and TK refer to this as a "sweet spot".
greatly visible brighter-ness, with no large leap in current from the source.

Most LED's will function in this manner.
this can be misleading, but is nothing special or magical about the transistor function.
it is an exploitation of the operation of the diode.

If you think of the LED as having a range of "brightness" it fluctuates through.
Pulse the LED to turn it on, it will go to max brightness, then scale down to dimmer, and dimmer, until it cuts off.
This is a function of the diodes internal capacitance.
where it is on the "dimness" scale, when it switches back on again, will determine the average "light" you visibly see from the diode.
When the diode spends more time in the brighter end of this scale, the light you "see" will be brighter.
Since "most JTs" do not let the LED fully turn off, it turns back on during some dimmer state than its' full brightness.
In general, within some range, an increase in switching frequency will result in "brighter light"
But, in this case, we are increasing the scale of dimness. staying "brighter" longer, while actually decreasing freq.
a bit counter-intuitive, but this does happen.

fast acting diodes, that can exceed the switching rate of the transistor circuit, display less of this feature.
Because they can actually stop emitting light before the transistor sends another pulse.
At higher frequencies, the switching can become unperceivable to the human eye.
For classification purposes, these fast switching LED's should be placed in a separate category of JT circuits.

Let us take the question deeper,
rather than just how does base resistance affect LED brightness

let us ask:

How does base resistance affect switching times?

This will help you understand "how a joule thief works".

[sm0ky2]

https://www.youtube.com/watch?v=v9tsvbkOeW8


Brad

Taking note of the frequency at the bottom of your scope,
Perhaps, You can dig up the data sheet for your ferrite core, and calculate a lower octave of the SRF.
probably the 6th, will be right around the lower range of your VR.

I noticed that you do not have a great deal of accuracy, of control with that pot.
There is a thing I learned a while back about using parallel pots to achieve a greater sensitivity of control.
the initial resistance with two pots in parallel, is half of the resistance of one.
and with 3 pots, the Ohms are 1/3
so,. by having two pots set at 1k, ohms = 500
setting one at 1k, and one at 500, you will have ohms = 375
so by changing one VR, you are able to control the range of resistances by a greater detail.
If you had something like 10 VRs, in the 10k-Ohm range
you could get the total resistance down to around 1k, and tune it within a degree of a couple of ohms at a time.

a high-precision trim-pot can give the same accuracy, but possibly more expensive than 10 cheap ones in parallel.
without enough precision, you can zoom right passed a resonant node and not even notice it.

you may have crossed it in your video around the lower  range of your pot, and not even known it.
I would need to know more about your ferrite to say more on that, but either way, you are within range of hitting a node,
or can get there with not much modification.
If the pulse width of both on and off times do not balance out to equal lengths of time as you approach the desired freq.
(the pulses/spikes resolve into a sine wave)
then we can look for the disrupting impedances, and dig them out of the circuit.

there are a lot of connection points that seem like they would be problematic.
most of the circuits I tested were soldered, or twist-connected directly coil to transistor.
each piece of wire adds an impedance.
each bi-metal junction adds an impedance.
[ for example: a tiny thin piece of metal was usually all it took, in terms of "added impedance" to resolve the feedback phase of the base->ground loop.
literally, in some cases, a piece of aluminum cut from a can, slid between the - battery and the wire.]


most people assume because of the low numerical value of these impedances,
that they play no significant factor in the circuit.
In single pulse analysis, this may appear to be the case,
however, I assure you, when you pulse 30,000 times per second,
this tiny difference can add up quickly.
a tiny thing happening many, many times adds up to a lot of drain on your
already dying battery.
And will affect how long it can continue to operate under those conditions.

Notice, how the tailing number on your freq counter is slowly dropping.

What does this do to the duty cycle equation? (your scope may even have a button for this?)

When the freq stabilizes into a resonant node of the ferrite, reluctance (which is the primary circuit drain, not the LED)
approaches a minimal value, for circuit operation.
The frequency will maintain this range of values, close to the resonant node,
for long periods of time.
It will drift, much slower than in any other mode of operation.
Why? because the resistance to the changes in magnetic-flux that occurs in the coil, as induced in the ferrite
are almost non-existent.
the coil does what the coil does. and the ferrite responds to its' own ringing. which is an accumulation of the "now" signal, and the one that preceded it. - magnetically speaking
For some smaller ferrites, this 6th lower frequency node can hit down in the 14-16Khz range
In this case, you don't need a scope to know when you hit a resonant node, because the ferrite itself will make a noise.
I stress, that this is an undesirable condition, and when you reach this node, tune the frequency down just a touch.
you want a point, just under resonant frequency, within tolerable conditions for the circuit.

This is the ONLY manner in which the torroid ferrite inductor "should be used".
PERIOD.

This is basic electronics.

If your circuit does not operate the torriod in this manner, you have chosen the wrong torroid for that frequency application.
All you are doing is wasting more energy than necessary to induce the required magnetic field for your inductor.

This is the very definition of "efficiency".

It is "how the ferrite works".
however you want to switch it.
Joule thief, Triode, Radio-signal, mechanical piezo oscillator, whatever.

For those of you who do not understand this,
why are you even using a Torroid in the first place?
Your "joule thief" can light the LED with a coil wrapped around a ferrite rod.
or, if you want, you can use an air coil.
Has this thought even occurred to you?

Could it be there is much more to this device, than the Instructables Video has to say about it?
Why do you think Wikipedia defines it as "minimalized" ?

resonance has nothing to do with a Joule Thief

Go light your LED !


For the rest of us,. we can examine what really happens when the ferrite is switched near a resonant node.

eventually, the resistance will need to be adjusted to compensate for the drain on the battery.
Power is still being used, to magnetically charge the ferrite. Because we are not "at" the resonant node, but some non-zero value lower than it.
Also, the LED, if it is in the primary loop (not if inductively coupled), represents a significant drain on the battery.
In order to bring it back into the frequency range. (and sometimes this value has to actually go UP in resistance!!!)
the degree to which the resistance must be change, will be more of a factor of how long you let it run out of resonance,
then how long it was running "in resonance".
Because, the drain, out of resonance, is so much higher, it becomes the dominant factor of how much is "left" in the battery.
If left alone (no change in resistance), the frequency will drop further away from resonance, and the ferrite will require more energy to charge.
Thus, lowering the efficiency of operation.

a JT thrown together from the instructables video
will maybe run for a few hours/days, to a couple weeks. depending on the battery, and LED, and transistor used.
type of wire, particular ferrite. etc.

a well tuned JT, that is build to operate as a timed Armstrong oscillator, has no problem lighting an LED, powering a speaker, or micro-motor, for a months time, and in some cases much longer.


Aside from any resonance between the circuit, and ferrite
if your low voltage source, is a replenishable one, or from an environmental source,
it can operate indefinitely.
voltages of + or - 1v can be found almost everywhere.

EB+JT=PM? some people here think so. I myself never had an EB draining my yard power for any length of time to figure that part out...
Heavy loads - yes they will deplete the natural capacitance of your garden, and it will regenerate over time.
But a tiny LED? it may very well could run forever,. who knows...
is the BatCave is still glowing from the 4-ft fluorescent tubes, powered by the Fuji circuit, and some grounding rods?

What do we do with the solar-powered JT nightlight, that charges itself from the incandescent bulb in the room?
(wasted light harvesting?)

"what a joule thief is"

Is a basis for an entire new realm of electronic circuits, based on a combination of old-world technology, and new-age components.

this is what separates electronics from computers.
in electronics engineering, we are taught to destroy resonances.
in computer engineering, we must maintain them, or the whole thing falls out of timing.
to combine these two mind-sets, on the same platform, such as a circuitboard -
much thought must be put into maintaining internal resonances, and disrupting them on the main circuitry, to preserve tolerances.
all the while, without losing track of timing cycles that are data dependent.

a single mismatched impedance on a complex circuit board, that results in a resonance, can reverse polarities at amplitudes that cross the threshold voltages and currents of the components in the circuit.
That is why circuits are not designed to be "resonant".
If we did, none of them would work.
capacitors would blow, transistors would burn up, resistors would burn out, wires themselves would even become compromised.

If you've burnt out a transistor with a "dead" AA battery, while playing with your Joule thief - raise your hand!
I will not presume that all of these cases were the direct result of resonance.
But I, and a couple of others have already shown mathematically, how this can easily occur.
and Has occurred in lab-settings

If an entire circuit, from the JT, to the load, and back, were designed to operate at a node resonant with the SRF of the ferrite,
direct power transfer from battery to load is possible with minimal losses through the transformer.

You can take a look at any professional circuit (Chinese excluded), that uses ferrite torroids.
and there is some box you can place around that portion of the circuit, and isolate the frequency of the torroid signal.
you will almost always find them to be operating just below an SRF node. It is standard procedure.
I would go so far as to say, if it is NOT, the engineer did not know what he was doing when he designed that circuit.
(or in some rare cases, the specific frequency required has no economically obtainable matched-frequency ferrite)
Why anyone in their right mind would not want to operate a JT in this manner is beyond me.
especially after all that has been said on the subject.

To operate a JT far from resonance, is like driving your car with the brakes on.

[tinman]

Tinman,

Do you have the 'scope channel displaying the base/emitter voltage inverted? 

PW

Yes,CH1 was inverted.
We can see we have 800mV at the base. What i wanted to show was the switch on curve,but the resolution on the scope was set to high,and the top of the wave form(the switch off spike) would have been out of the screen if i had decreased the VPD limit-and i did not want it flowing into the LED trace. So i just left it at that.


Brad

[tinman]

[transistor 101]

1kOhm?  hmm....  2N2222? (yup,.. "standard"... uh-huh)

ok let's roll with that for a second shall we..

What is the current through the transistor?

Ie = Ic + Ib

So,. what happens when we increase base current, while collector current remains the same?
Current at the emitter increases!

Where does inductor current come from?
The emitter? hmm

For some(or one),this seems to be a little above there pay grade.


Brad

[tinman]

Yeah, after being on the bench for six years you are not only eating humble pie, you are manufacturing humble pie in mass quantities.  The clip is awful, an embarrassment.

An embarrassment for you MH.

I think right about now,you are running around trying to figure out a way of dismissing everything i just showed,without making yourself look like a bigger idiot.

Im done with your negativity and insults--and i just !once again! proved you wrong.
That was just a quick video,but there will be more to come that make things very clear to see.
But the video i just showed,shows exactly what i said it would show.
The reduction in base current increases pulse width,and increases LED light output-->all while the battery voltage is decreasing.
How are your !laws! and V/R limits going now MH?
How or where dose the extra magnetic field strength come from that gives rise to a higher potential
from the field collapse that drives the LED harder?,as !your! V/R limit is set in L1
How is my theory on increasing the current flowing through L2 being what gives rise to a greater magnetic field looking now?.



Brad