Author Topic: Joule Thief 101 (Read 944441 times)

MileHigh · « **Reply #1095 on:** March 23, 2016, 05:53:40 PM »

Brad:

I will go out on a limb now and honestly I didn't want to go here because I expect you to demonstrate your own innate capability and do a test that makes sense. But since I got thrown off I will state what I have in mind and leave it at that.

When you decrease the base resistance the operating frequency lowers and thus the energizing time for the main L1 coil increases - obviously that can lead to a brighter LED. I was expecting some effect on the frequency but not as much as is shown in your clip, with the caveat that I still don't have confidence in your setup because it is running too fast and the discharge time through the LED is way too short. Nonetheless, a lower operating frequency means more time for the L1 coil to energize. The conundrum is that the operating frequency is supposed to be governed by the L/R-type time constant to energize L1 and the L/R-type time constant for L1 to discharge through the LED. The value of the base resistance doesn't really come into play for these two time constants, hence my expectation was that varying the base resistance would only have a marginal effect on the operating frequency. I am not sure why lowering the base resistance lowers the operating frequency, but like I said before, if your Joule Thief was running at a more normal frequency it would be interesting to see what happens then.

Let's examine your premise for your experiment. You said changing the value of the base resistor would compensate for a lowering battery voltage over time. It's hard to tell from a clip, but the LED does not appear to be getting appreciably brighter as you change the base resistance. Always the caveat being that I don't think your Joule Thief is operating at a normal frequency. The light meter shows increasing intensity, but only at a level of 40 parts per thousand. The eye is not very sensitive when it comes to detecting changes in brightness.

How will the LED intensity vary if you use a power supply and change the supply voltage from 1.5 volts to 0.75 volts? Will it be noticeable? How will the LED intensity vary if you run the Joule Thief at 0.75 volts and lower the base resistance? Will that compensate for the presumed loss in brightness or not? Again, the Joule Thief has to be running at a normal frequency.

That is the rough form of the testing that I think you should be doing. To me it looks like varying the base resistance will only have a very marginal affect on the brightness of the LED because it slows down the operating frequency. It is somehow affecting the positive-feedback triggering mechanism for the shutting off of the transistor. I would not be surprised if lowering the base resistance only barely makes the LED perceptibly brighter. In contrast, I believe that varying the supply voltage to the Joule Thief will have a very noticeable affect on the brightness of the LED. And that is the crux of the issue.

MileHigh

MileHigh · « **Reply #1096 on:** March 23, 2016, 06:46:58 PM »

Brad:

I know that you are going to jump around and gloat, but here is your posting #1076 verbatim where you give your "explanation" for how lowering the base resistance will increase the brightness of the LED:

Quote

MH
You need to have a good look at the basic JT circuit--the one you like,or call the JT circuit.
You have two coil's that provide current to build the magnetic field--not one.
As you decrease the resistance to the base,you increase the current flowing to the base,and this current is additive to the collector current due to the way the two coils are linked together,and thus the magnetic field strength can be maintained by reducing the base resistance value, so as to provide the same amount of power flowing through the LED from the kickback as the battery voltage drop's.

I will draw up a quick schematic of my test setup,and post it here. When you see the test setup,you will know that the measurements taken,and the statements i provided are correct.

Brad

So your "explanation" isn't even remotely close to what you are observing at all. You are observing an increasing period in the energizing cycle for the L1 main coil, and that is what is making the LED only slightly brighter. From your clip the brightness increase appears to be barely perceptible.

Quote

As you decrease the resistance to the base,you increase the current flowing to the base,and this current is additive to the collector current due to the way the two coils are linked together,and thus the magnetic field strength can be maintained

That statement looks pretty damn wonky to me. The coils are opposite wound on the same toroid, right? Looking at the schematic, let's say when current flows from top to bottom through L1 during the transistor ON cycle we are building up magnetic energy in the toroidal core. During this time current is also flowing from top to bottom through L2 to keep the transistor switched ON. L2 is opposite-wound to L1 and that means when current is flowing from top to bottom through L2 that it is producing magnetic flux that is opposite to the magnetic flux produced by L1. So when current flows top to bottom through L2 it reduces the magnetic energy stored in the toroidal core, not increases it. That would help to make the LED dimmer, not brighter.

So you can't claim any "victory" because your light meter can detect increasing brightness in the LED. You just stumbled upon this effect, and your theory for why the LED was supposed to get brighter makes no sense at all.

MileHigh

Pirate88179 · « **Reply #1097 on:** March 23, 2016, 07:34:03 PM »

MH:

Quote: "When you decrease the base resistance the operating frequency lowers and thus the energizing time for the main L1 coil increases - obviously that can lead to a brighter LED."

It has been my experience, as I have stated, that decreasing the resistance to the base increases the operating frequency as in the example of the low voltage battery causing the led flashes to be seen by the human eye...decrease the base resistance and now the led flashes on/off faster than you can see and appears "on". I have seen this many times first hand on my own units.

Maybe I am not understanding you here?

Thanks,

Bill

TinselKoala · « **Reply #1098 on:** March 23, 2016, 07:48:35 PM »

Here's something else to consider. In the standard JT, the toroidal windings actually form one continuous winding in the same direction, with a center tap. You can save yourselves a lot of trouble by simply taking one strand of wire, winding the toroid or ferrite rod etc. with one continuous single-layer, connecting the ends to Collector and Base resistor/capacitor (if used). Then use sandpaper to remove the insulation of the magnet wire in a "stripe" along the outside of the toroid or along the length of the rod, and locate the tap (which goes to positive rail) in the best place by experimenting with connecting to the uninsulated stripe. Once you have found the correct ratio location, solder the "center" tap to the bare stripe at that spot. This is similar to the way that a traditional inductively-tuned crystal set has taps located along the coil for tuning to particular stations.

MileHigh · « **Reply #1099 on:** March 23, 2016, 07:58:40 PM »

Bill:

It's a great issue you raised and I don't have an answer, but only some follow-up questions to ponder:

When your Joule Thieves were running off of a very low battery, they might have been flashing the LED at 25 Hz, and then you lowered the base resistance and the flashing frequency increased. What is the technical explanation for that?

How come Brad's Joule Thief is running at 20 kHz when one of yours might have run at 50 Hz?

Precisely why does the frequency in Brad's Joule Thief decrease when he decreases the base resistance? What is the technical explanation?

Why is there apparently so little energy in the LED discharge from Brad's toroidal coil? The discharge takes five microseconds, it's doesn't seem to make any sense.

Knowing Brad, who knows what the winding setup is like on his toroidal core. It's possible that the effective number of turns for the L1 coil are very low, and the core is only storing a fraction of the magnetic energy that it can actually store.

What is implicit in the questions I am raising is that I am trying to push the boundaries past doing a setup and passively observing how it performs, and switching over to actively trying to understand how something works. It's a huge shift from just observing something and pretending that you are experimenting and truly experimenting to test a hypothesis and actually understand the underlying processes that explain how something works.

Now, if I was in Brad's shoes, the first thing I would notice is that my decent-sized toroidal core is dumping a 5-microsecond pee-fart of energy into the LED when it discharges. Something seems amiss there. What goes hand-in-hand with that is the very high frequency that the setup is running at. I would undertake to check the core material, check the winding configuration, clean up my overall wiring, and put my Joule Thief "in the shop" so that when it came out of the shop it was running at a frequency and generating waveforms to my satisfaction. I wouldn't do anything until I was satisfied that my Joule Thief was operating in what I considered to be a normal fashion. But I am not in Brad's shoes and I hold out no hope that he will pay any attention to these matters.

MileHigh

MileHigh · « **Reply #1100 on:** March 23, 2016, 08:07:56 PM »

Quote from: TinselKoala on March 23, 2016, 07:48:35 PM

Here's something else to consider. In the standard JT, the toroidal windings actually form one continuous winding in the same direction, with a center tap. You can save yourselves a lot of trouble by simply taking one strand of wire, winding the toroid or ferrite rod etc. with one continuous single-layer, connecting the ends to Collector and Base resistor/capacitor (if used). Then use sandpaper to remove the insulation of the magnet wire in a "stripe" along the outside of the toroid or along the length of the rod, and locate the tap (which goes to positive rail) in the best place by experimenting with connecting to the uninsulated stripe. Once you have found the correct ratio location, solder the "center" tap to the bare stripe at that spot. This is similar to the way that a traditional inductively-tuned crystal set has taps located along the coil for tuning to particular stations.

That sounds really cool. I can easily envision a "spikey Mohawk" toroid with the single winding and a dozen taps with soldered wires coming off of the single coil. So then not only can you experiment with the best position for the center tap, but you can experiment to your heart's content with the number of turns in the L1 main coil and the L2 feedback coil.

sm0ky2 · « **Reply #1101 on:** March 23, 2016, 09:19:16 PM »

you generally don't want to saturate the ferrite
the ferrite may reach saturation at peak amplitudes, through regenerative feedback
but driven from the battery, to saturate the ferrite, uses much more energy than necessary to power the load.
it is better to operate under full saturation, or at a point approaching saturation.
since we have no way of limiting ourselves to the exact point of saturation,
we almost always go well beyond it, wasting energy into space.
for the small application of tracing scope images, or lighting an LED,
we don't need a lot of flux, just slightly more than the coil alone provides.
(and even that is not always necessary - see air-coil JT)

Magluvin · « **Reply #1102 on:** March 23, 2016, 10:11:40 PM »

Quote from: Pirate88179 on March 23, 2016, 07:34:03 PM

MH:

Quote: "When you decrease the base resistance the operating frequency lowers and thus the energizing time for the main L1 coil increases - obviously that can lead to a brighter LED."

It has been my experience, as I have stated, that decreasing the resistance to the base increases the operating frequency as in the example of the low voltage battery causing the led flashes to be seen by the human eye...decrease the base resistance and now the led flashes on/off faster than you can see and appears "on". I have seen this many times first hand on my own units.

Maybe I am not understanding you here?

Thanks,

Bill

"
It has been my experience, as I have stated, that decreasing the resistance to the base increases the operating frequency...."

That has not been my experience at all. Decreasing the resistance gives me lower freq, longer(stronger) on and off times. It is when I 'increase' the resistor that I get shorter(weaker) on and off times thus higher freq of operation.

Mags

Magluvin · « **Reply #1103 on:** March 23, 2016, 11:10:03 PM »

Quote from: sm0ky2 on March 23, 2016, 09:19:16 PM

you generally don't want to saturate the ferrite
the ferrite may reach saturation at peak amplitudes, through regenerative feedback
but driven from the battery, to saturate the ferrite, uses much more energy than necessary to power the load.
it is better to operate under full saturation, or at a point approaching saturation.
since we have no way of limiting ourselves to the exact point of saturation,
we almost always go well beyond it, wasting energy into space.
for the small application of tracing scope images, or lighting an LED,
we don't need a lot of flux, just slightly more than the coil alone provides.
(and even that is not always necessary - see air-coil JT)

I may have to order a core with low srf, and will know what the srf is by doing so. I have many toroid cores but I dont know what I have spec wise. Next purchase is a func/sig gen.. Havnt found resonance other than the battery with the caps at 1.4 or so mhz, depending on which of the disks I put in, of the same lot and value. Scanned the no signal zone beyond clean sine wave range. Was only able to get the battery to show up at near 1.45mhz and the led lit some. And Im pretty sure it is the battery ringing, as it changes with the added caps and across the battery is the only component oscillating according to the scope, individually checking components with probe and gnd lead of scope.

Mags

Magluvin · « **Reply #1104 on:** March 23, 2016, 11:49:59 PM »

Hmm. This is the second time Ive heard of the led going into blink mode. Smoky said it once earlier. Is that when the battery is super low? Because it seems to me that would indicate the battery is at a level that it needs some refresh time before being able to fire the transistor again. Im getting high enough freq levels at very low battery levels that you cannot see the blinking with the eye. No way. I havnt taken a battery ALL the way down yet to see if I can get it to blink. Will run one down hard to see, soon.

But as far as I can see so far, I dont think it is circuit operation that is giving visible blink, unless the windings were more than we have been playing with here, meaning way more inductance, low freq. Other than that, Id bet on battery weakness at the time of slow visible blink.

Mags

tinman · « **Reply #1105 on:** March 24, 2016, 12:08:18 AM »

Quote from: Magluvin on March 23, 2016, 11:49:59 PM

Im getting high enough freq levels at very low battery levels that you cannot see the blinking with the eye. No way. I havnt taken a battery ALL the way down yet to see if I can get it to blink. Will run one down hard to see, soon.
But as far as I can see so far, I dont think it is circuit operation that is giving visible blink, unless the windings were more than we have been playing with here, meaning way more inductance, low freq. Other than that, Id bet on battery weakness at the time of slow visible blink.

Mags

Quote

Hmm. This is the second time Ive heard of the led going into blink mode. Smoky said it once earlier. Is that when the battery is super low? Because it seems to me that would indicate the battery is at a level that it needs some refresh time before being able to fire the transistor again.

That is correct Mag's--the battery needs time to recover to a voltage where it can start to switch on the transistor again. If the battery voltage has not recovered enough by the time the next cycle is ready to start,then the transistor will not start to conduct,and you skip a cycle or two until there is enough voltage recovery on the battery. This you can test using a cap that is fedd by a battery or PSU with a resistor between the two.

You can also decrease the required supply voltage,and still have the JT running,by decreasing the base resistance. This i will cover in my next video.

Brad

tinman · « **Reply #1106 on:** March 24, 2016, 12:13:23 AM »

Here is the next video on bench testing the JT circuit.

This comment posted on my video from a viewer.

Quote

the joule thief schematics you are using is more like a boost converter, the classic Joule Thief has the LED connected in parallel to the coil in reverse to use just the inductive spikes, of course the voltage of that battery is not enough for the LED voltage conductivity, but it has a different approach to the circuit.

Just go's to show that there is not just one common JT circuit.

Anyway,we look at the base current in this video,and we can see the effects it has on the operation of the circuit.

https://www.youtube.com/watch?v=72BqF8bkk-k

Brad

Magluvin · « **Reply #1107 on:** March 24, 2016, 12:27:29 AM »

Quote from: tinman on March 24, 2016, 12:13:23 AM

Here is the next video on bench testing the JT circuit.

This comment posted on my video from a viewer.

Just go's to show that there is not just one common JT circuit.

Anyway,we look at the base current in this video,and we can see the effects it has on the operation of the circuit.

https://www.youtube.com/watch?v=72BqF8bkk-k

Brad

lol. Like I said way earlier before I had built my first Jt, the led across the winding seems a better choice as to not include the battery current in the led drive loop, as that would deplete the battery at all times while the circuit is running, vs having a real off time while the coil dumps to the led.

Mags

tinman · « **Reply #1108 on:** March 24, 2016, 12:33:46 AM »

Quote from: MileHigh on March 23, 2016, 05:06:00 PM

MileHigh

Quote

I am going to have to take my statement back and qualify it. On closer inspection of your scope shot I can see that indeed the base waveform is showing that the transistor is ON and you are correct that the level is 800 mllivolts.

Well this is good

Quote

What are the root causes of the misunderstanding?

Oh here we go

Quote

The first is that I wasn't observant enough to see the fact that in your close up shot that the trace is just hugging a point above the zero volt line. I thought it was zero volts, not hugging just above zero volts.

I thought

Quote

A not too distant second factor is that your presentation skills are generally very poor and it's easy to get thrown off because of that.

Ah-there we are,the big old switcharoony--now it's my fault because of your lack of observing skills.
Point 1--in the video,i went through all the scope setting value's.
Point 2-the screen shot you posted yourself has all these values clearly presented on that screen shot. The VPD on CH1 clearly show's a value of 5volt's,meaning that 800mV would be just above the zero volt line. The Vmax on CH1(800mV) is also clearly visible in the calculated value's on the right hand side of the scope shot.
Point 3- i told you all this in a previous post--but still you were lost.

Quote

A third factor is in your original clip it looks like a zero-volt baseline with a big positive spike - sloppy presentation and poor communication skills come back to haunt you.

It is your lack of being able to interpret and observe what the scope is showing you,and what the scope settings are,that is being the cause of all your drama's.
No one else here had trouble understanding what was going on in the video.

The circuit will run just fine on low voltages when the base resistance is set to the !correct! value--another reason that a fixed base resistance is no good.

Quote

In Magluvin's capture it takes about 30 microseconds for the LED to discharge and it runs at 4.2 kHz with a total period of about 238 microseconds.

Is my setup exactly like Mag's ?

Quote

There is no reason that your setup shouldn't have comparable timing. You are supposed to get a nice clean "crisp" set of waveforms like in Magluvin's capture.

Are you serious

Do you think that two different LED's will work in the very same way?
Do you think that two different core's will operate in the same way.
What about different transistor's?
What about different numbers of turns for each coil,or the wire size used?.

How you can say that the two different circuit's will/should operate very close to the same is beyond me.
Like i said MH,you really are not well versed in JT circuit's.
The fact that you do not know why the transistor stays on longer,and the frequency lower's when the base resistance is decreased,truly shows you have so much left to learn.

Facts are facts MH,and your fairy tails have been proven to be just that--fairy tails.

Brad

tinman · « **Reply #1109 on:** March 24, 2016, 12:39:35 AM »

Quote from: Magluvin on March 24, 2016, 12:27:29 AM

lol. Like I said way earlier before I had built my first Jt, the led across the winding seems a better choice as to not include the battery current in the led drive loop, as that would deplete the battery at all times while the circuit is running, vs having a real off time while the coil dumps to the led.

Mags

Exactly
And i too found having the LED across the L1 coil to be much more efficient.
But having the LED across the base/emitter put's energy back into the battery

Brad

News:

User Menu