another small breakthrough on our NERD technology.

[Rosemary Ainslie]

Gravock - I'm amazed that you're in agreement with any part of what's written.  I'm knee deep in my justification of the disconnected state of the Q2 MOSFETS.  I'll need to finish that and will get back here.  I think I'll need to post over more of the paper before we get to nitty gritties.  BUT I'M DELIGHTED THAT YOU'RE PREPARED TO DISCUSS THIS.  I had no idea that anyone one at all - was seriously interested.  But I've got to finish off that ruddy argument for MileHigh et al - or I'll forever be plagued with that absurd argument that that the battery can deliver any energy at all during the off period of the duty cycle. 

Golly.  Am delighted with this.  Hopefully you'll manage the equations for us.  Someone must.  They're BADLY NEEDED.

Regards,
Rosie 

And right now I need to break off and do some shopping.  I've just seen the time.  Be back as soon s I've bought the food to feed the dogs and cooked  some lunch.  HLTT (However long that takes) and BRBST (be right back soonest thereafter) This is seriously fun for me.
Edited the edit.  Grammar.

[AbbaRue]

@All:

Interesting enough this article about negative resistance circuits (Negative Resistance = Power Source)
shows circuits very simular to the Rosemary circuit!
Main difference is he used the drain instead of the source. (Theoretically that shouldn't matter)
A very good article to download and study on this subject:

http://rfic.eecs.berkeley.edu/142/pdf/lect23.pdf

From page 12 on there are a number of circuits with similar connections to Rosemary's circuit. 
Some of these could be tested to further this study. 
Based on this article and it's implications I think this subject deserves further investigation and less mockery!
After all this is a Berkeley University of California, article.
Unless someone on this forum thinks they are above this Berkeley Professor.
Rosemary may have something here!

Also wanted to mention Naudin's studies into Negative Resistance Using only 2 leads of a 2N2222A transistor.

http://jlnlabs.online.fr/cnr/negosc.htm

The MOSFET "Q2" who's source is not directly connected to the battery reminds me of it. 
Further study suggests that Q2 is functioning as a MOSFET Varactor. (page 25 of Berkeley article)

I have tried replicating this circuit using following MOSFET's
BUZ11 50v 33A
IRF630 200V 9.3A
IRF820 500V 2.5A (best results)

For the heating element I used a 300W 130V halogen bulb measured at 5 ohms and 250 micro Heneries.

I used 555 timer circuit for the Oscillator input:
With variable 10k resistors across 6-7 7-8 leads, to vary the wave form output.

For the main voltage input I used two 12 V. lead acid rechargeable batteries for about 25V.

I got high voltage spikes returning to the battery when I connected an inductor in series with the halogen.
I connected a 1000v diode across the halogen so spikes would return to battery. 

I didn't get battery charging, but I found that the battery voltage went down incredibly slow for the amount of drain.
When same halogen bulb was connected directly across the batteries voltage dropped very fast.

PS: I think I will experiment with some of these Berkeley circuits.

[Rosemary Ainslie]

Apologies for quoting myself but just to show the relevance of this argument that is now spanning a 3rd POST 

Gravock ... I'm knee deep in my justification of the disconnected state of the Q2 MOSFETS.  I'll need to finish that and will get back here... But I've got to finish off that ruddy argument for MileHigh et al - or I'll forever be plagued with that absurd argument that that the battery can deliver any energy at all during the off period of the duty cycle. 

Regards,
Rosie

NOW.  All things being EQUAL.  When we CHANGE THAT SIGNAL at the gate of Q1 -  to an applied NEGATIVE SIGNAL... continued /...from the function generator, then the circuit is again OPEN.  That NEGATIVE voltage signal applied by the signal terminal - now induces a small NEGATIVE flow of current - to the GATE OF Q1.  Which, in turn, NOW REPELS the applied current flow from the battery supply source.  Those currents are in anti phase.  The current from the battery supply CAN NO LONGER CROSS THE BRIDGE.  THE CIRCUIT IS OPEN.  NO CURRENT CAN FLOW from the BATTERY SUPPLY.   

HOWEVER.  As MileHigh keeps telling you and as POYNTY keeps pointing to - as they both seem to be staking their LIVES on this their argument, - when the positive signal at the Gate of Q1 changes to a negative, then SIMULTANEOUSLY the applied signal at the SOURCE of Q1 is NOW POSTIVE.  NOW.  LOOK AT THE CIRCUIT SCHEMATIC.  You will see that the NEGATIVE TERMINAL of the signal generator is connected to the SOURCE leg of Q1 - and then onto the SOURCE of the battery terminal.  And this is LINKED to the Gate of Q2. THEREFORE, and in TRUTH, the signal that is NOW APPLIED TO THE GATE OF Q2 is POSITIVE.  THEREFORE ALSO - they are arguing - THE CURRENT FLOW FROM THE BATTERY SUPPLY SOURCE IS NOW ENABLED VIA THE GATE OF Q2 IN THE SAME WAY AS IT WAS PREVIOUSLY ENABLED AT THE THE GATE OF Q1.

I'm interrupting this argument to repost that schematic.  That we can all be on the same page.
R
GUYS this is going into LOOP MODE again. Something's seriously wrong with this software of yours Harti.  I'm hoping this AMENDMENT WILL TAKE.  I NEED TO TAKE OUT THAT WAVEFORM.  AND TRY AND POST THE CORRECT SCHEMATIC.  If I don't get back here immediately - it's because I cant.

[gyulasun]

 Hi AbbaRue,

Sorry for ’chiming in’  I could not resist to share some of my views wrt your post on negative resistance circuits.       

The lecture paper you linked shows the so called cross coupled oscillators with  devices like MOSFETs and BJTs (bipolar junction transistor) devices.  Such circuits exhibit negative resistance between the MOSFETs drains or between the BJTs collectors as a result of positive feedback.  Positive feedback results when one device’s  control electrode (i.e. the gate of the MOSFET or base of the BJT) is connected to the other device’s  drain or collector electrode and vice versa, as shown in Page 12 in your link. 

You wrote:  Main difference is he used the drain  instead of the source. (Theoretically that shouldn't matter.)

  Respectfully I disagree,  it Does Matter because without positive feedback these circuits do not manifest the negative resistance between the drains or between the collectors.  Here is a link
http://www.tubecad.com/2009/07/blog0166.htm
where the cross connection is shown between the gates and the cathodes  (of electric valves)  and here negative feedback is involved with such connections!  (I hope nobody is confused by seeing circuits with electric valves,  the cathode where electron emission happens can be viewed as the source electrode of a MOSFET or the emitter electrode of a BJT.)   These circuits are used for improving the linearity and distortion performance of (audio or other) amplifiers and this means no negative resistance creation possibility for such type of cross coupling because this would bring about oscillations at certain frequencies:  this would be a rather unwanted ’feature’  in linear amplifiers should such gate-cathode (or gate-source or base-emitter) cross coupled circuits create negative resistance.
So in Rosemary's circuit the cross coupling is shown between gates and sources of the MOSFETs and NOT between the gates and drains, hence there cannot be any positive feedback like shown in Page 12 of the paper, hence no oscillation due to negative resistance from the MOSFETs. (more on the oscillations in Rosemary circuit later below)

I have to notice also that the Naudin circuit you linked to uses a BJT in the ’inverse mode’  i.e. an NPN transistor’s emitter is biased to be positive (normally it should receive negative polarity) from a battery and its collector receives the negative battery polarity (normally it should be positive) so a reversed breakdown can happen in its C-E junction and this exhibits the negative resistance  (Cf  this with the Esaki tunnel diode or the Gunn diode etc which also have a negative resistance region  in their  voltage-current curve but by proper semiconductor doping.)   

So my note here refers to your text:  The MOSFET "Q2" who's source is not directly connected to the battery reminds me of it.  (i.e you meant Naudin’s 2N2222 circuit with its negative resistance)

So in Rosemary’s  circuit the source electrode of Q2 is indeed not directly connected to the battery but via the function generator.  I hope you agree with me that such function generators normally have 50 Ohm inner resistance so the source electrode is practically connected to the battery negative point via this 50 Ohm.  (And when a DC voltage shift is used at the output of the FG, the 50 Ohm still should be there by the FG inner circuit design, otherwise the specification for the output resistance is not fulfilled in that DC shift mode, what I doubt  so it is still 50 Ohm.  And the 50 Ohm output resistance can simply be checked by an Ohm meter part of a DMM, when the FG’s output level is turned down in the  few milliVolt range, not to fool the Ohm meter and also the DC shift feature is adjusted to zero should it be turned on.)  Can you agree with the connection of Q2’s source to the battery negative via the 50 Ohm inner resistance of the FG?

On you following single sentence:  Further study suggests that Q2 is functioning as a MOSFET Varactor. (page 25 of Berkeley article)   

I would like to notice that the varactor effect is shown between the gate and both the source and the drain electrodes (the latter two are connected together in Page 25 for both the N and P channel MOSFETs) and bias voltage for controlling the capacitance is connected to the body electrode of the MOSFET (a fourth electrode beside G, S and D) and in most power MOSFETs this fourth electrode is normally tied inside of the device to the source electrode. 

So I do not think Q2 can work as a varactor like it is shown in Page 25 BUT I can say that any MOSFET has a voltage dependent inter-electrode capacitance dependence and data sheets show this mainly for a 25V drain-source voltage and at zero gate-source control voltage and also they give a curve for all the three changing capacitances in the function of the full operational voltage range for a particular type.  And this changing inter-electrode capacitance dependence in the function of drain-source and gate-source voltages (in both AC and DC voltages' respect)  is of course true for Q1 in Rosemary's circuit as is true for any MOSFET used in any circuit.

Sorry again for mentioning these three topics in bold from your above post and share my opinion on them, I did not mean to steer up any quarrel or ’mockery’,  and I hope you or others here can agree with all the 3 notices of mine.  Of course you do not „have to” agree with me,  lol.

IT is very good you built the circuit and I would ask on your batteries’s  AmperHour  capacity, and what is the 555 pulse frequency you found perhaps the best, and how bright the halogen lamp was? i.e. any idea on its output power. 

Now briefly on Rosemary circuit oscillations:  what I think is that their battery bank is banged by heavy current pulses via the MOSFET switches and via the heater element with a frequency which corresponds to the battery bank’s resonant frequency (yes, batteries have a resonant frequency, see this link where 2 to 6MHz range is indicated for lead acid types: http://www.reuk.co.uk/Battery-Desulfation.htm ).  Because of the high current pulses, I suspect the batteries develop heat and it would be good to learn from the team what battery temperatures were involved during the hours long tests,  in their paper they mentioned the cooking  of two batteries if I recall correctly and from this I assume the high battery temperatures (unless something else happened to those two batteries.)

Also I suspect that this heavy pulse ’bombardment’ of the batteries involves creating chemical reactions inside, this may be in connection with the team’s  claim of no need for recharging during the some month long test period, of course  I hypotese here with the chemical change inside, suggesting it as beneficial.  It would be good to learn in what condition these very same batteries are now:  can they still be used as normal batteries with similar AmperHour capacity performance now that there is no test period? Have these batteries been tested since the long circuit operation weeks and months are over, it would be a must I think .
Maybe I am mistaken with this battery resonant frequency excitation;  though maybe you could increase your test frequencies in the MHz range too and check the battery temperature too (do avoid boiling though, it can be dangerous). The above link on desulfation mentions using small current pulses at the resonant battery frequency.
Your using a inductor in series with the halogen lamp means that the recovered bemf is much higher than without it, so more part of the input current is stored in that inductance wrt to  the lamp’s own 250uH inductance,  hence the high voltage spikes manifest and more part of the input energy can get back for feeding the circuit.  A notice here also that using a higher value inductor in the team’s  circuit means to alter,  to change the original or the shown circuit…

On your sentence: "When same halogen bulb was connected directly across the batteries voltage dropped very fast."   
Please notice that a direct connection is not at all equivalent to a switched connection (the latter happens via your MOSFET of course),  perhaps the best comparison in performance would involve a known quantity of water or cooking oil heating  (this latter can isolate better electrically) for a certain time, this has already been discussed but perhaps not applied correctly, I do not know. 

Thanks, 

Gyula

[TinselKoala]

The two "V"s in that circuit represent the two voltage measuring channels of an oscilloscope, do they not? The negative (ground) leads are shown connected at the same point, between the battery and the shunt resistor, as is proper. However.... unless special pains are taken, or a fully isolated scope like the Fluke ScopeMeter 199 is used, all the instrument "grounds" will be common through the instrument and line wiring itself. Thus, the Function Generator ground will also be electrically connected to the scope probe grounds back through the instrument wiring, effectively shorting out the shunt resistor, and if a voltage offset is used in the FG output things get even more problematic.

I don't know if the actual test setup that is claimed to have been used in the "experiment" had this problem... but if the circuit shown above was used, with ordinary instruments plugged into the wall, then it likely did have.

I illustrated some of the problems with this sort of cross-talk and "groundloop" in a video, as usual, long ago. The pertinent part concerning isolation starts at about 6:30; the first part is just about the Fluke and its ability to resolve short or long duty cycles.
http://www.youtube.com/watch?v=JU1YGaEBKwM

Note that the circuit still "runs", with larger oscillations and spikes, when the battery is "disconnected" from the circuit by removing the negative power connection. What's not obvious.... unless you are careful.... is that the battery is actually STILL CONNECTED to the circuit through the earth ground and the scope and FG signal grounds.