PhysicsProf Steven E. Jones circuit shows 8x overunity ?

[JouleSeeker]

I agree Chris --
1.  "This circuit is a very unusual circuit." 
2.  "We need to keep the current down for this effect."
3.  "Sometimes it seems to be looking really good then other times it does not. "

I have had the "glitch" in one session on the Tek 3032 myself a while back, where the circuit was performing very well, n>1 per the Tek 3032.
  I made measurements of Pin and Pout repeatedly, back and forth, and kept getting n>1 for about 45 minutes, varying Rr and Ro, and getting variations in n but always n>1.   
Then all of a sudden, it changed for unknown reasons to n<1.   Sorry I did not mention this sooner -- I thought the "glitch" was for it to fall out of the super-efficiency condition temporarily, or perhaps I caused an inadvertent short in the system -- I could not find what made it glitch.   Later, it looked fine again.


During this same session at the university lab, I tested the build by my friend Les Kraut.  He tried to replicate the sj1 circuit exactly.   His build showed n ~ 8, after the "glitch" showed up in my initial circuit-testing.


Later -- I have felt one of the resistors being very hot to the touch during a run, and this MAY be a reason for the "glitch". I was not at the Tek 3032 this time.   I suggest feeling the resistors if your circuit "glitches" again.

Clearly the goal is to understand our observations and to keep the device in the super-efficiency condition.
Thanks again, Chris, and my apologies for not mentioning this sooner.  I do not think this "invalidates" the circuit, but it is something we must try to understand.
Steven

[xee2]

Would you suggest a diode on the negative rail? This may help any ripple? I am trying to measure only DC Voltage, but measure the AC Current as this does bounce around crazy at points.

Adding diode here will help some.

[xee2]

@ hyiq

Kooler and I have been able to get Joule thief circuits to light an LED dimly on as little as 5 micro-watts of input power using a MPSA06. You may want to try that if you can get one. The MPSA06 seems to work better than a 2N2222 at low power.

[nul-points]

 
firstly, a quick thank you to Chris for re-sizing your circuit - still very readable - but no need to scroll across now - or reduce the page size & not be able to read the writing!


Steven

there have been some developments which you may find interesting

apologies that there is a mass of detail and calcs - in this case, i feel that 'the end justifies the means'!

i ought to mention in passing that i modified my circuit slightly to decouple the AC output from the DC biasing conditions of the oscillator by adding a tertiary winding on the transformer (see schematic below)

i don't believe that the following information is influenced by the output coupling method - current draw from the supply cell and LED illumination appear to remain at a similar level as before


i've been able to use an opto-coupler with my low-freq. 'inverted looped' sr1 circuit variant to get a handle on the o/p level issue

i used the opto LED in place of the discrete LED in the circuit**
(and confirmed that this didn't significantly alter the DC current draw of the circuit from the single AAA NiMH supply cell)

i used a DVM resistance range to bias and measure the opto transistor C-E 'impedance'

(transistors are viewed as having 'transconductance' between terminal current paths, but let's not get into 'terminology' just yet!)

i applied a time-constant to the DVM reading by connecting a capacitor in parallel with the opto o/p, to act as a DC 'smoothing' filter on the reading

Please see the details and calcs in the following post!...


[EDIT:  ** please note, i'm not claiming this is a precision method, or that this approach can't be refined to give more accurate quantified readings - i'm suggesting that this approach can be used as a simple comparative method to provide a ballpark value for DC Power equivalent of LED o/p in a suitable situation]

[nul-points]

PLEASE NOTE:
In the following calculations i'm using DC Power values - since all of these values are equally proportional with time, they're equivalent to using Joule values (ie. working with energy)


Steven

i realise that you already make use of this method yourself - the previous comment is only for the benefit of others! (get ready for howls of protest...) 


Measurements for the 'Inverted, Slowed, but NOT looped' sr1 circuit
===============================================
Vbatt: 1.243V
Iin: 107uA

DC Pin total: 133uW

let's just emphasize, here:
  with 1.243V across the oscillator supply, the oscillator will draw 133 uW of DC Power


Measurements for the 'Inverted, Looped & Slowed' sr1 circuit
===========================================
Vbatt: 1.243V
Iin: 65uA

DC Pin total: 81uW


after losses, the available o/p energy is stored in the buffer cap

some energy from the buffer cap gets transferred back to combine with the input energy to meet the total DC Power draw requirement of the circuit

since the DC Power drawn from the battery decreases to 81uW, when feedback is applied, then the DC Power contribution from the buffer capacitor:
  133 - 81 =  52uW

let's just emphasize, here:
  the contribution of DC Power In from the o/p feedback path to the oscillator circuit is 52uW


so now the question is: "what are the various energy losses in the system?"

- light
- heat
- e/m radiation


let's consider the energy conversion in the LED

firstly, find the equivalent DVM reading (of the opto o/p) to that caused by in-circuit LED replacement by the opto LED
(see schematic below)


Opto LED DC Power comparison (opto LED in series with variable resistor)
===============================================
Vbat: 2.77V
Iin: 84uA

DC Pin total: 233uW

Variable resistor set to 21.8 Kohms (to match opto o/p with in-circuit value)

Joule loss in Var Res:
0.000084 * 0.000084 * 21800 = 154uW

Therefore DC Power converted by LED: (233 - 154) = 79uW

ie. the LED is providing the equivalent of 79uW DC power


since the opto LED is 'diverting' 79uW (mostly as light, some heat),
the total DC Power supplied by the buffer cap:
  52 + 79 = 131uW (at least, ignoring losses)

let's just emphasize here:
  the total DC Power provided from the buffer capacitor is 131uW


OK, so the oscillator circuit DC Power requirement is:
  133uW

and when looped,

DC Power supplied by energy source A (NiMH cell):
  81uW

DC Power supplied by energy source B (Buffer cap):
  131uW

Total DC Power supplied to the oscillator + LED:
  81 + 131 = 212uW

let's just emphasize here:
  the total DC Power converted by the whole system is 212uW

  the total DC Power supplied by the NiMH cell is 81uW


so the efficiency value, 'n' is 212/81 = 2.62

...and that's ignoring losses (heat & e/m radiation)

[EDIT:  note that 79uW equivalent of DC Power is dissipated by the LED and is lost to the system, which would reduce the 'available' electrical 'n' to be 133 / 81 = 1.64]


now it's my turn to ask if YOU would mind kindly checking MY math, thanks Steven 


and finally (Phew!) let's quickly 'lead out' a possible objection...
(DA = a 'Devil's Advocate')

DA:
"Ah, Mr Nul-Points, that's all just fancy footwork - in reality when you looped the circuit back you probably just increased the total circuit impedance, so THAT is why the DC Power In from the battery decreased!"

NP:
"Not so fast, Mr DA - connecting a load in parallel with the oscillator would conventionally be expected to reduce the system impedance and tend to increase the DC Power draw from an external supply;
  what we're seeing here is REDUCED DC Power draw from external supply, when connecting a load to the sytem"


let me know what you think

thanks
np


http://docsfreelunch.blogspot.com