Joule Thief TRS

[ion]

Can someone state clearly in a simple Technical Requirement Specification exactly what function or functions they would like the Joule Thief or any such circuit to perform. The engineering target seems to shift all over the place. Since different functions require different circuit topologies, we can properly design the optimum circuit once the goal is known.

Is the requirement:

a) to drain the remaining energy from a single nearly dead AA cell and transfer that energy to a supercap?

b) to drain the remaining energy from a single nearly dead AA cell and light a bank of series connected LED's What type of LED's and how many Lumens are required.

c) some other requirement or combination of the above.

If you  don't need to design for a single nearly dead AA cell, and can put several in series or parallel, the topology will change again. Can the batteries be fresh or is the idea to scavenge remaining energy from nearly dead cells?

I would welcome any input on this, then I can offer some optimized approaches once the problem is defined.

[Fraser]

Hi ION,

I would say the main goal is maximum transformer efficiency of the system. Whether or not somehow that efficiency could be made to exceed 100%, well I don't know, perhaps with some config it may be possible to coax in ambient energies through some magnetorestrictive property of the core or something, I've not seen evideince for this at the moment though.

[ion]

Hi Fraser

The most efficient magnetic structure that I know of is the pot core. No one is even playing with these. Properly designed they completely contain the magnetic and electrostatic field, minimize the DC resistance by having the windings efficiently on a spool. They are difficult to manufacture, and break easily, generally more expensive than other types, the window area is not optimum.

Visually a pot core is the E-E core rotated rotated and filled through it's axis of symmetry. It's magnetic structure completely surrounds the spool. You see some attempts at this with newer ferrite E-E core configurations with extensions on the outer edges of the core structure.

Once again, there is no one shoe that fits all. A transformer must be designed with the intended load in mind, then high efficiency can be approached. Thats why I think the JT has to come in several flavors depending on the intended load.

As you know engineering is a balancing act, and all of the parameters associated with transformer design must be fully understood before you can do the balancing. Perhaps the attached document will help in this.

Better to have a theory as a starting point to work from. Transformer and inductive component theory has been well researched over the ages with no one ever reporting  anything close to 100%. I personally have built a test fixture to pulse inductors and transformers and try to capture BEMF to bootstrap the power source with negative results.

The closest thing I remember reading about was a transformer core designed with thermoelements. This was theorized by Harold Aspden to run cold and extract heat energy from the ambient with the BEMF pulse. Never saw any construction details on the device.

It might be possible to exceed 100% but I would at least like to hear the theory of operation. The JT crowd has no theory and as far as I have determined, have not come close to 100% regardless of the claims.

I remain open minded and hope someone really has success.

Regards...ION

[Fraser]

Agree with all of that.

The "JT crowd" have not been scientific in their endeavours to date. Nothing I have ever read mentions input/output power. More can be learnt about micro social interaction than blocking oscillators when reading their posts, tiresome to say the least.

My mind does entertain the possibility of exotic behaviour in very fast rise times pushing the hysteresis curve of core materials into some peculiar asymetric gain mode, perhaps rectifying thermal noise as you mention?

The first thing needed to hunt for this is a proper way to measure input/output power. As I´m sure you´ll agree we should disregard how many christmas tree LEDs we can dimly light and pick up on a camera. We should gather real data!

My proposal is to sample across non inductive shunts over input and output using a DSO. Then adjust coil, core and pulse parameters and see just how high efficiency can be pushed. perhaps we need to venture into HV pulses of very fast rise and short duty?

[Grumpy]

Can someone state clearly in a simple Technical Requirement Specification exactly what function or functions they would like the Joule Thief or any such circuit to perform. The engineering target seems to shift all over the place. Since different functions require different circuit topologies, we can properly design the optimum circuit once the goal is known.

A joule thief with the following specifications would be a nice addition:

12vdc to 10kvdc conversion - output adjustable with a pot

output current of 5 to 10 ma

an accidental dead short of the output must not cause damage

(sorry - posted this in the wrong thread earlier)

[Fraser]

50w to 100w output at 10kv, what the hell do you need that for? I´m thinking abnormal glow discharge exciter, am I right?

[ion]

Grumpy

If you are going to be throwing those kind of accelerating potentials and power levels  into a vacuum tube, I hope you have plenty of lead shielding to protect yourself or you can look forward to some mutant offspring.

Pulse or continuous duty ? Regulation? Size? Weight? Efficiency?

To be real, I thought the original JT intent was supposed to be a cheap and easy to build device that will scavenge most of the power in a single cell for use in flashlights or "torches" as they say across the water.

Then the self runner idea started to proliferate, then more LED's, then charging double layer caps.

This is a far cry from a 10KV DC 100 Watt experimenters power supply. TV flyback transformers won't deliver these power levels.

But it can be done. If what you need is a lab type bench supply, for absolute robustness get hold of an old diathermy transformer or whatever you can find that will meet the kVA requirement. Make some diode strings with equalizing R's and C's and you are good to go. Usually a center tap full wave is the often used approach, though if you can make a full wave bridge, you'll more efficiently use all the copper in the transformer. Put a fast circuit breaker and variac for adjustability on the primary along with safety switches and fuses. Not 12 VDC, but a good general purpose bench supply.

For truly scientific work you will need some form of regulation or the load impedance will cause the voltage to vary. There are many ways to do this.

If you want to go switchmode, a flyback approach probably will be marginal under 100Watts. Best to use a push-pull drive and voltage multipliers operating at reasonable frequency say 20kHz-50kHz.

An old 100 watt tube amp can drive a pair of opposite polarity multipliers right off the plates of the tubes.

We are far away from a unified spec on the JT, but thanks for your input.

[MileHigh]

If you are talking about a classic JT setup then there is no specific voltage output specification for it.

The thing to realize is that a JT outputs a pulse of current, and the voltage that you see it producing is a result of the current.  It may seem counter-intuitive, but if you can get used to the idea that it's a pulse of current and not a pulse of voltage then things fall into place.

Suppose the JT coil has one amp flowing through it before the transistor switches off.  What voltage will it generate?

The answer is that it depends on the load on the JT outpput:

Load(ohms) - Volatage(volts)

0....................0
1....................1
156................156
3000..............3000
25k.................25K

Just think of the current being like a charging bull.  The moment the switch opens, you can't stop the charging bull of current and it will force its way through whatever resistance is in its path.

The thing to keep in mind is that the higher the value of the resistive load, the shorter the duration of the pulse.

MileHigh

[Grumpy]

50w to 100w output at 10kv, what the hell do you need that for? I´m thinking abnormal glow discharge exciter, am I right?

more like a portable "momentum conversion device"

Right now I am working with avalanche devices and don't need or want very much current - they won't handle it at high pulse rates.

I have a 1.5 kvdc converter (old Spellman version) that runs on 28vdc (it will run on just about anything including 12vdc, but output is reduced accordingly).  It has a 5k pot to adjust the output and is rated at a 1ma output.  Of course that 5ma is based on some unknown input.  I currentl;y run it off a bench power supply and it works pretty good but is not portable and requires 28v not 12v.

I have another power supply that is 10kv, unregulated.  It consists of a Franceformer (ignition transformer), diode bridge, and a homemade 15kv cap.  Current out is 24ma, which is acceptable, but again this is not portable.

(I wanted to avoid the use of an inverter, but that is an easy way.)

If you are going to be throwing those kind of accelerating potentials and power levels  into a vacuum tube, I hope you have plenty of lead shielding to protect yourself or you can look forward to some mutant offspring.

Pulse or continuous duty ? Regulation? Size? Weight? Efficiency?

...

For truly scientific work you will need some form of regulation or the load impedance will cause the voltage to vary. There are many ways to do this.

If you want to go switchmode, a flyback approach probably will be marginal under 100Watts. Best to use a push-pull drive and voltage multipliers operating at reasonable frequency say 20kHz-50kHz.

An old 100 watt tube amp can drive a pair of opposite polarity multipliers right off the plates of the tubes.

We are far away from a unified spec on the JT, but thanks for your input.

[/b]

Yes, ION, gamma rays and X-rays really suck.  Did you know they have detected X-ray emissions when pulling Scotch tape off a roll?

Regulated , continuous (the load is triggered in the high kHz range - considering several channels, each triggered at 100kHz to 200kHz)

The current can probably be much less, I just haven't tried it yet.   Testatika uses an electrostatic generator for comparison.

Push-pull drive and voltage multipliers sounds good - we can move this to another thread.

I just thought I'd reach for the sky since you asked - LOL!!!

[acmefixer]

If you are talking about a classic JT setup then there is no specific voltage output specification for it.

The thing to realize is that a JT outputs a pulse of current, and the voltage that you see it producing is a result of the current.  It may seem counter-intuitive, but if you can get used to the idea that it's a pulse of current and not a pulse of voltage then things fall into place.

Suppose the JT coil has one amp flowing through it before the transistor switches off.  What voltage will it generate?

The answer is that it depends on the load on the JT outpput:

Load(ohms) - Volatage(volts)

0....................0
1....................1
156................156
3000..............3000
25k.................25K

Just think of the current being like a charging bull.  The moment the switch opens, you can't stop the charging bull of current and it will force its way through whatever resistance is in its path.

The thing to keep in mind is that the higher the value of the resistive load, the shorter the duration of the pulse.

MileHigh

I think the name says what it outputs:  X number of Joules, they're any combination of voltage and current, that is really what is the output.

Perhaps a Joule Thief should be defined as any power between 0 and, say, 999 Joules.  If you go over 999 Joules, then it's a KiloJoule Thief, and if you go over 999 kiloJoules, then it's a MegaJoule Thief, etc.  IOW, there should be a dividing point when it comes to size and power.