Storing Cookies (See : http://ec.europa.eu/ipg/basics/legal/cookies/index_en.htm ) help us to bring you our services at overunity.com . If you use this website and our services you declare yourself okay with using cookies .More Infos here:
https://overunity.com/5553/privacy-policy/
If you do not agree with storing cookies, please LEAVE this website now. From the 25th of May 2018, every existing user has to accept the GDPR agreement at first login. If a user is unwilling to accept the GDPR, he should email us and request to erase his account. Many thanks for your understanding

User Menu

Custom Search

Author Topic: Testing the TK Tar Baby  (Read 1998293 times)

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #45 on: March 27, 2012, 03:17:14 PM »
TK,

The referenced RA paper is at the following:

http://www.overunityresearch.com/index.php?action=dlattach;topic=13.0;attach=6766

PW

Thank you, I have the paper in my files and I have read it several times. This "paper" is the one that was summarily rejected five times by the IEEE journals they submitted it to, isn't it? Am I going to have to read that word salad yet again? They even list an incorrect make and model for their FG in that paper.

And isn't the Instek FG that they used, one of these? Note the frequency range of the units.

http://www.tequipment.net/InstekGFG8216A.html

.99 pointed this out to Rosemary early on during the present incarnation of the zombie tarbaby.:

http://www.overunity.com/11675/another-small-breakthrough-on-our-nerd-technology/msg310454/#msg310454

http://www.overunity.com/11675/another-small-breakthrough-on-our-nerd-technology/msg310470/#msg310470 ( with comparison photos)

The FG used by the NERD RATS  in the video demostration is an Instek GFG8216A, with a minimum frequency of 0.3 Hz (or 0.1 Hz depending on where you look) according to its manual and data sheet. How then was it used to make the very slow frequencies that they have claimed? I'm afraid I'm going to have to insist on some real evidence that these long cycles were achieved, if  I am to make some special effort to operate in that regime. Meanwhile, the evidence that we do have indicates they used a 50 percent duty cycle at 10 Hz.

An internet search for "iso-tech 324 function generator" or "isotech 324 function generator" returns mostly hits for the INSTEK 8216A. I have not been able to find any discrete listing for a "iso-tech" or "isotech" 342 function generator.  I'd love to see the data sheet for that unit.

http://www.iso-techonline.com/products/iso-tech-oscilloscopes-function-generators.html#tab4
Note the lowest frequency settings of these Iso-tech function generators. No model "324" is listed. Maybe it's obsolete... but I can't find one listed on the surplus/used market either.

It also looks to me like the front panel of the INSTEK unit is telling us that the "shield" or outer conductor of the FG's output terminals is grounded, not floating.

picowatt

  • Hero Member
  • *****
  • Posts: 2039
Re: Testing the TK Tar Baby
« Reply #46 on: March 27, 2012, 04:04:33 PM »
TK,

As to the rejection, well, I'll leave that alone...  the point is, which waveform to use.

The one you describe in your posted scope shot is very similar, only the duty cycle/period is different.  Does the hi level of the gate drive signal on the scope shot you posted exceed the zero line?  If not, then Q2 is just being alternately turned off and biased on by the FG as its waveform toggles and Q1 is not being utilized in any active way, except for its intrinsic capacitances and body diode.

Your optocoplers are hard switching the gates, which is not the same as the bias being used on the RA common gate amplifier circuit to just barely turn on Q2 for somewhat linear operation.  Q2 is configured as an amplifier, not a switch.  There are other ways to bias Q2 into a similar region besides applying a negative voltage to its source via a resistor, but again, the flavor of the circuit would change. 

I suggest an FG setting at a convenient to scope period with an open circuit excursion between -15 and zero volts, unless more data is available regarding the FG max positive swing voltage.  I believe your Interstate can be set to place the complete waveform at the zero volts level and below (I believe that's what I saw in one of your videos).  In that setting, amplitude can be used to set the negative swing value instead of using the offset control (very cool FG).  Just keep in mind that you only want to bias Q2 on slightly, 100-250ma., and not fully turn it on.  Also, keep in mind that there will be a lot of dissiation in Q2 when biased that way...  proper heat needed.

Hard switching Q2 would definitely be more efficient regarding less loss in Q2 and more power at the load, but the "magic" is supposed to be in the oscillations.  Hard switching the MOSFETS and using a lower Rdson MOSFET would eliminate the need for a heat sink, but again, the flavor of the circuit would change.

PW



 

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #47 on: March 27, 2012, 04:36:44 PM »
TK,

As to the rejection, well, I'll leave that alone...  the point is, which waveform to use.

The one you describe in your posted scope shot is very similar, only the duty cycle/period is different.  Does the hi level of the gate drive signal on the scope shot you posted exceed the zero line?  If not, then Q2 is just being alternately turned off and biased on by the FG as its waveform toggles and Q1 is not being utilized in any active way, except for its intrinsic capacitances and body diode.

Your optocoplers are hard switching the gates, which is not the same as the bias being used on the RA common gate amplifier circuit to just barely turn on Q2 for somewhat linear operation.  Q2 is configured as an amplifier, not a switch.  There are other ways to bias Q2 into a similar region besides applying a negative voltage to its source via a resistor, but again, the flavor of the circuit would change. 

I suggest an FG setting at a convenient to scope period with an open circuit excursion between -15 and zero volts, unless more data is available regarding the FG max positive swing voltage.  I believe your Interstate can be set to place the complete waveform at the zero volts level and below (I believe that's what I saw in one of your videos).  In that setting, amplitude can be used to set the negative swing value instead of using the offset control (very cool FG).  Just keep in mind that you only want to bias Q2 on slightly, 100-250ma., and not fully turn it on.  Also, keep in mind that there will be a lot of dissiation in Q2 when biased that way...  proper heat needed.

Hard switching Q2 would definitely be more efficient regarding less loss in Q2 and more power at the load, but the "magic" is supposed to be in the oscillations.  Hard switching the MOSFETS and using a lower Rdson MOSFET would eliminate the need for a heat sink, but again, the flavor of the circuit would change.

PW

Now I am beginning to think you either haven't watched all the videos or I am not explaining things very well. Using my FG to produce a strict negative-going pulse train, from 0 Volts to minus whatever, adjusted by varying the _amplitude_ control not the offset, I am making oscillations just as you say. When I turn the amplitude up from "idle" (which is a minimal voltage swing from about 0 to -0.5 volts) to "operating" you can see the mosfets turn on, and when the FG is directly connected not using the optoisolators the oscillations commence immediately when the signal is high enough for any response. I can put the amplitude all the way up (max negative excursion) but since there is a low impedance to the FG's output in the circuit, the voltage doesn't go way up, but the oscillations persist throughout the range of the FG's setting as long as the amplitude is enough to "tickle" the Q2 gates. In other words, I am operating just as you suggest in your second-to-last paragraph, as far as I can tell.

The purpose of the optoisolator trial is to try to figure out some way to operate the system without the FG's current path, while including the oscillations, since those are the only "magic" part of the behaviour of the NERD circuit (except how to get a 2 minute wave period from a FG that has a minimum frequency of 0.3 Hz). I don't think that the FG's path is absolutely required for feedback oscillations to happen; humbugger's excellent sims from last year showed that all you really have to do is "poke" the thing with a single short pulse and it will oscillate until the cows come home or the oxtail soup is done.

If it turns out that the FG's current path MUST be in the system, acting as a power source,  for the _claimed_ Ainslie battery charging effect to occur....   what does that tell us?

18 months of continuous testing every working day for 18 months as she claimed (12 Watts dissipated in the load) could have only used less than 4 times her total battery capacity anyway...... even if she really did dissipate 12 Watts continuously for 5 hours per day for 300 days.

So I think that you are trying to hold me to some standard that the Ainslie team certainly hasn't met themselves. You want me to reproduce a claimed frequency and pulse duty cycle that is clearly not within the range of the equipment they say they used.

(Although it IS within the capability of my Interstate F43, which can go down to... let's see.... 0.004 Hz x 10 on the knobs for 0.04 Hz, then use the .01f setting, we get 0.0004 Hz. What is that in period, I wonder? 0.0004 cycles PER one second == ? seconds per cycle, so we invert and divide, to obtain a period of 2500 seconds... can that be right?)

Right now, based on the heating and other behaviours of the Tar Baby, I can say this: operating the TAR BABY in full oscillation mode where I reproduce the gate and drain traces that Ainslie has _actually_ shown, not just talked about... there is very little power in the load and it doesn't warm up much. However, during "tuning" one encounters states where the mosfet(s) are cleanly switching AND states where the mosfet(s) are fully 100 percent of the time ON, and the load is exposed to the full voltage of the battery pack through a low-resistance pathway... hence it draws several amps of real current and heats up quite well.

It would be very surprising to me to find that a "real" NERD RAT device would behave differently, even with the magic mosfets and the magic signal generator. 
ETA: Also, the "efficiency" of the circuit as a heater is in question. When the circuit is operating in "oscillating" mode, how much of the power from the battery will be dissipated in the mosfet(s) and how much in the load? When the circuit is switched cleanly or is in "constant on" mode, how much of the power from the battery in _that_ case is dissipated in the mosfet(s) and how much in the load? This is a separate issue from the battery recharging issue... for which there is no _real_ evidence EVEN IN THE NERD RAT DATA as .99 and humbugger have extensively shown in their analyses.

I hope you don't mind, .99, if I attach your report here... maybe PicoWatt hasn't seen it.

WilbyInebriated

  • Hero Member
  • *****
  • Posts: 3141
Re: Testing the TK Tar Baby
« Reply #48 on: March 27, 2012, 04:55:52 PM »
Your alternate FET's should work, but for duplicating as close as possible the waveforms and Fosc of the RA circuit, the PG50 would be a closer replication.  You should be able to swap them out with the 830's but the lower capacitances of those devices will flavor the results slightly.

careful... i tried telling tinsel-lokin the same thing years back... they all jumped on me and called me a troll. accused me of strawman. or maybe it was because i used the same tone with tinsel as he uses with others that got him so upset...

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #49 on: March 27, 2012, 05:08:33 PM »
careful... i tried telling tinsel-lokin the same thing years back... they all jumped on me and called me a troll. accused me of strawman. or maybe it was because i used the same tone with tinsel as he uses with others that got him so upset...

Ahh... I've been waiting for you to show up. I knew you would. This will be my last response to you and it consists of one question:

When will you show your Ainslie circuit replication charging up its batteries and boiling water?

Please don't return until you can answer my polite question above.

http://www.overunity.com/11675/another-small-breakthrough-on-our-nerd-technology/msg310476/#msg310476

For the readers who may be interested, Wilby is referring to my prior work on Ains-lie's previous claim, the Quantum circuit and the claim of COP>17, with yet a Different circuit than is being examined here. For a short while, while my IRFPG50 mosfets were on order, I posted results from a 2sk1548 mosfet, and I stated that I would be surprised to find a significant difference between the two. Later when I actually DID have a PG50 for testing, I mounted the two transistors side-by-side on a huge triple-pole double throw slide switch so that I could switch them back and forth _live in the powered circuit_ for comparison purposes. I discovered that the 2sk1548 worked BETTER than the IRFPG50 in producing the high-voltage inductive spikes that Ainslie then was claiming were responsible for her alleged battery charging. I posted a note saying that I WAS surprised to find this difference..... but since I found no "overunity" behaviour from either mosfet, I didn't think it was a "significant" difference in the context. And I was right then and I am still right, but Wilby can't find anything else objectionable in my work so he's been stuck on that same single issue for YEARS.
(In that work I was the first to: identify the inverted duty cycle problem with her timer; show that using her timer AS PUBLISHED one could reproduce the time-temperature profiles she posted; boil water with the heat in the load; siphon off the "spike" using a fast diode and charge an external capacitor to hundreds of volts, running NE-2s and so forth on a 24 volt battery input; charge EXTERNAL batteries with a "fluffy charge", and many other of the reported effects of that circuit. And my batteries still measured 24 volts !! Lol. All this is documented and published in many videos on my YT channel. And all of it using BOTH the 1548 and the pg50 for comparison purposes.)

WilbyInebriated

  • Hero Member
  • *****
  • Posts: 3141
Re: Testing the TK Tar Baby
« Reply #50 on: March 27, 2012, 05:15:59 PM »
of course i would. i told you i'll be there a long time ago didn't i? don't you see tinselkoala/alsetalokin... my righteous indignation at what you did to me, what you implied and what you lied about is directly proportional to your self righteous indignation at rosemary for what you accuse her of.

i wouldn't even consider it until you offer up that mea culpa you owe me from our last go 'round... you remember... the go 'round where i told you the same thing picowatt did about substituting whatever you have on hand and expecting it to be a "replication". and recall what happened when you did actually get around to SCIENTIFICALLY testing the difference between the two transistors. you found they didn't behave the same... and recall where you cursed and ran away and said you were "done".  you remember don't you? i do...

furthermore, i wasn't talking to you. i was talking to pico. so unless you have that mea culpa ready. don't speak to me.

for the readers that are interested: the record of what was said, which tinselkoala/alsetalokin doesn't want you to have the link to is here.  http://www.overunity.com/7620/claimed-ou-circuit-of-rosemary-ainslie/msg197621/#msg197621

Quote from: utilitarian on August 16, 2009, 09:59:44 PM
Quote
Similarly, can you demonstrate how NOT using the specified mosfet will result in a different conclusion to the experiment, i.e. that the circuit is overunity, rather than underunity as demonstrated?

that wasn't the conclusion being posited by tk. he had concluded (apparently before he started his hack of a 'replication') it was perfectly ok to substitute a mosfet based on data sheets. i called him on it. he didn't actually voice this conclusion until page 2.

Quote from: TinselKoala on June 17, 2009, 08:13:03 AM
Quote
OK, several points to address.
    First, yes, when I can find them I will use identical components to Ainslie's circuit. Her MOSFET is kind of pricey and will have to be ordered; the one I'm using is...well, you can look up the data. It's pretty close, good enough for prelim testing. I will replace the shunt with .25 ohm today.

he then made an asinine hypothesis (see below for how that turned out) about the irfpg50 performance, never once specifying 'over unity performance'. i called him on it. he then asked if i could show how it would perform any different on page 11.

Quote from: TinselKoala on July 01, 2009, 08:28:24 PM
Quote
The answer to that, of course, is that I say what I say on a discussion forum, and she says what she says in published articles, the EIT paper, and patent applications. Her claim is false, mine is a slight exaggeration. My claim can be corrected simply by switching out the mosfet. Hers cannot be corrected so easily--her claim depends on an erroneous data input into calculations and would require re-running the experiment.

    Would you care to make a little wager, Wilby?

    If you can show a significant difference between the performance of the IRFPG50 mosfet used by Ainslie, and the 2SK1548 mosfet that I used in my replication, using the published circuit and parameters of Ainslie, I will gladly make a public apology to you. On the other hand, if the performance is substantially the same, you get off my back.

    If you really think the mosfet makes a difference, you should take the bet.


    (EDIT I was going to offer to bet money at odds, but I realised that would be unethical--like taking candy from a baby--. Sorry.)

please take note of these words...
My claim can be corrected simply by switching out the mosfet.  Hers cannot be corrected so easily.
here he is referring to his claim of no difference in mosfet performance, let alone OU, and her claim of over unity which he was calling erroneous due to the duty cycle issue.
and these words...
significant difference between the performance of the IRFPG50 mosfet used by Ainslie, and the 2SK1548 mosfet that I used.
note no mention of over unity performance, just performance. why oh why do i have to hammer this in? oh yeah, the slow ones... and the liar.
they (tk and his merry band of sycophants) all jabbered for a while pretending how smart they are and then, asymatrix quantified it once again as being non relative to over unity performance by saying this on page 24.

Quote from: Asymatrix on July 08, 2009, 12:12:43 AM
Quote
Please tell the class why a slightly different FET will make a huge difference, let alone create OU.

tk did not amend this to being specifically relative to OU performance. while they (tk and his merry band of sycophants again) continued to jabber about how smart they are and how little i know, etc. i waited for him to get around to actually testing this experimentally. when he finally did, on page 42 i might add. he found this out.

Quote from: TinselKoala on July 13, 2009, 06:50:12 PM
Quote
The long turn off time of the IRFPG50 really messes with the signal at these excessively short (using the FG) or LONG (using the 555) duty cycles.
    The IRF unit does seem to heat up less than the 2SK, but that's just an early impression.
    I think if you are into spikes in your signal, the 2SK might be a better choice here too. It turns on and off better than the IRF unit (not surprising, is it, looking at the data sheets and considering the gate capacitances). And since it turns on and off with faster rise and fall times, it produces a higher inductive pulse from the coil. I think. Maybe.

    So there goes my hypothesis that the two transistors would perform pretty much the same. I was wrong about that. The 2SK1548, when properly cooled, outperforms the IRFPG50, as far as I can tell. And it's smaller. And quite a bit cheaper. And locally available.

take note of these words...
So there goes my hypothesis that the two transistors would perform pretty much the same. I was wrong about that.

he did however, try to claim he 'meant' over unity performance later. much later. page 108 actually.

Quote from: TinselKoala on August 11, 2009, 07:07:38 AM
Quote
"Significant difference" in this context clearly means OVERUNITY performance. And the two mosfets do not differ significantly in this respect.
i called him on that too.

picowatt

  • Hero Member
  • *****
  • Posts: 2039
Re: Testing the TK Tar Baby
« Reply #51 on: March 27, 2012, 05:47:25 PM »
TK,

I am not trying to hold you to "any" standard.  I merely thought replicating the circuit first and then modifying it would be a logical way to go.  I meant no disrespect at all.  And no, I have not watched all your videos, I'll try to find the time to catch up.

By applying the DC bias to the gate, the AC parameters change a bit.  Also, do you have a 50R or any resistor other than the CSR in the Q2 source leg to the batt negative?  Q2 will self-regulate its current based on the source resistor value used , the positive voltage applied to the gate (or negative voltage applied via the (50R) source resistor) and its threshold voltage (which varies with temp).

As I said, I doubt Q1 is being turned on at all.  One could possibly place a 50R in the Q2 source leg to the CSR and apply 12 volts to the Q2 gate via a 100K to 1meg resistor connected to the first 12 volt battery tap up from ground.  That will set the DC bias about right for Q2, but now the first battery is providing I bias as well as I load.  Also, the drain of Q2 would now have 12 volts less swing (headroom) available.  And again, AC conditions change a bit as the internal resistance and strays from the batt would now be in the DC and AC loop and I load will be inducing ripple in what would now be the bias source.  The large gate resistor value will, in concert with Ciss, decouple the I load ripple somewhat, but in the end, it sounds like it would be another oscillator to me.... again, just a different flavor.   

One could use a separate 12 volt battery and a 50R to apply the negative voltage to the Q2 source instead of the FG, but then the bias battery will eventually die.  Properly sized, it would allow the main batts to be tested for a proper rundown test to see if the batts do indeed "recharge" or "maintain charge" during oscillation.  Assuming a fair degree of feedback in the FG output amplifier, the AC impedance of the FG output stage is probably quite low, relative to the Rgen of 50R.  Placing a cap across the new bias battery would reduce the batt's AC impedance to closer simulate the FG.

As far as the FG recharging the batteries in the RA circuit, I don't see how that is possible.  Adding another battery in series with a 50R in its place will not charge the higher voltage batt bank either.  As to the current state of the RA batteries, based on her observed swings to zero volts, I suspect their internal impedance is quite high at this time.  Even without a load applied at all, they would have seen 18 months of self-discharge and subsequent loss of capacity (increased internal resistance).  The batt's open circuit voltage could still measure just fine.  It would be interesting to just hang a 10R across the batts and measure the drop in the battery voltage to determine their internal resistance.

Again, no disrepect intended... it's all good... 

More coffee and then back to work for me...


PW

fuzzytomcat

  • Hero Member
  • *****
  • Posts: 676
    • Open Source Research and Development
Re: Testing the TK Tar Baby
« Reply #52 on: March 27, 2012, 06:02:45 PM »
Thank you, I have the paper in my files and I have read it several times. This "paper" is the one that was summarily rejected five times by the IEEE journals they submitted it to, isn't it?

Hi TK,

Actually the paper that was rejected five times in various versions was this one .....

Open Source Evaluation of Power Transients Generated to Improve Performance Coefficient of Resistive Heating Systems
R.A Ainslie, H.W Gramm, G.A Lettenmaier, A.Palise, A. Gardiner, D Martin, S. Windisch

( 23455916-Open-Source-Evaluation-of-Power-Transients-Generated-to-Improve-Performance-Coefficient-of-Resistive-Heating-Systems.pdf )

For the record ....

FTC
 ;)

picowatt

  • Hero Member
  • *****
  • Posts: 2039
Re: Testing the TK Tar Baby
« Reply #53 on: March 27, 2012, 06:10:18 PM »
Wilby,

Wow, you guys really need to bury the hatchet... life is too short.

You do have to give TK credit for being one of the first, if not the first, to identify the inverted duty cycle issue in the COP=17 circuit.  And, oddly, how that inversion, if the calculations are also inverted, produced a similar COP result.

But yes, I would typically replicate first and then modify second.  I am asked to improve on designs all the time, and I always start with what the customer is currently using to make baseline measurements to improve upon.

PW




TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #54 on: March 27, 2012, 06:21:48 PM »
TK,

I am not trying to hold you to "any" standard.  I merely thought replicating the circuit first and then modifying it would be a logical way to go.  I meant no disrespect at all.  And no, I have not watched all your videos, I'll try to find the time to catch up.
No problem, I know you mean nothing more than to have an honest vigorous and constructive discussion. And you clearly have a better understanding of these matters than I do. I'm a psychologist, after all, not an EE.
Would you rather that I wait until I have all of the exact components in the lab before I begin work at all-- and where do we draw the line? I mean, if I had started with the PG50 and pegboard and clipleads all over the place, and obtained oscs immediately.... would we have learned as much as we know now? I don't think so.
My style, as you might have perceived, is not to attempt to PROVE anything. On the other hand, if I can't DISPROVE something....  see what I mean? I am applying the scientific method, and you can rest assured that I WILL reproduce the exact circuit down to the pegboard and clipleads, Instek and "aluminium heatsinks" and all, if that is what is _truly_ needed to examine the NERD RAT claims for accuracy and truth.
And there's really  no need to watch "all" of my videos... there are only about a dozen in the present Ainslie series and 40 or so in the previous 2009 work, out of nearly 200 videos of my scientific tinkering that I've uploaded.
Quote
By applying the DC bias to the gate, the AC parameters change a bit.  Also, do you have a 50R or any resistor other than the CSR in the Q2 source leg to the batt negative?  Q2 will self-regulate its current based on the source resistor value used , the positive voltage applied to the gate (or negative voltage applied via the (50R) source resistor) and its threshold voltage (which varies with temp).
I am using the exact circuit attached below, except that I am not yet monitoring across the CVR, I am using 0.1 ohms 5W "sand" resistor as the CVR, and I am using the 830a mosfet, as I still only have one PG50... which by the way oscillates fine in the "single mosfet" version of this circuit which is diagrammed-- but not shown--  in the Ainslie video demonstration.
Quote

As I said, I doubt Q1 is being turned on at all.  One could possibly place a 50R in the Q2 source leg to the CSR and apply 12 volts to the Q2 gate via a 100K to 1meg resistor connected to the first 12 volt battery tap up from ground.  That will set the DC bias about right for Q2, but now the first battery is providing I bias as well as I load.  Also, the drain of Q2 would now have 12 volts less swing (headroom) available.  And again, AC conditions change a bit as the internal resistance and strays from the batt would now be in the DC and AC loop and I load will be inducing ripple in what would now be the bias source.  The large gate resistor value will, in concert with Ciss, decouple the I load ripple somewhat, but in the end, it sounds like it would be another oscillator to me.... again, just a different flavor.   

I'm fairly sure I can get both Q1 and Q2 mosfets to turn on depending on the FG's settings. Usually alternately, occasionally simultaneously, mostly just one set during the negative gate drive mode. At least all the heatsinks can get warm ! I did manage to blow just two of the "gang of four" during an overheat event, and the single Q1 hs was quite hot at that time too.
Quote
One could use a separate 12 volt battery and a 50R to apply the negative voltage to the Q2 source instead of the FG, but then the bias battery will eventually die.  Properly sized, it would allow the main batts to be tested for a proper rundown test to see if the batts do indeed "recharge" or "maintain charge" during oscillation.  Assuming a fair degree of feedback in the FG output amplifier, the AC impedance of the FG output stage is probably quite low, relative to the Rgen of 50R.  Placing a cap across the new bias battery would reduce the batt's AC impedance to closer simulate the FG.
A real problem here is that test results should be unambiguous enough and with transparent enough methodology that the "main suspects" can't hand-wave their way out of the conclusions, as they have so many times in the past. I've even, in the very first setup of this circuit that I did using 2n7000 mosfets--nice oscs there btw, see the video-- I coupled the FG to the circuit using a series cap--- which passed the gate signal fine, switched the mosfets "normally"... but killed the oscs completely.
Quote


As far as the FG recharging the batteries in the RA circuit, I don't see how that is possible.  Adding another battery in series with a 50R in its place will not charge the higher voltage batt bank either.  As to the current state of the RA batteries, based on her observed swings to zero volts, I suspect their internal impedance is quite high at this time.  Even without a load applied at all, they would have seen 18 months of self-discharge and subsequent loss of capacity (increased internal resistance).  The batt's open circuit voltage could still measure just fine.  It would be interesting to just hang a 10R across the batts and measure the drop in the battery voltage to determine their internal resistance.
In one of the recent videos I show the FG charging a small battery a little bit, but definitely doing so. But I now realise that it doesn't have to... since there's no evidence of battery recharging in the NERD data when it's examined closely. I agree, a proper and simple load test on her batteries would be revealing but we both know we will never see that done -- on the NERD device at least. Also I'd like to know how two of them "caught fire" as she claims.
Quote
Again, no disrepect intended... it's all good...

More coffee and then back to work for me...


PW

And I never thought you were being disrespectful at all. You have constructive criticism and suggestions and you seem to know what you are talking about. I'm not afraid of looking stupid and I do take correction when the correction is correct. Trolls who make distorted comments about nothing at all pertinent, and who have never demonstrated possession of opposable thumbs much less technical prowess, on the other hand, I cannot abide at all.

ETA: oops, I forgot to attach the circuit. Sorry... here it is:
(I can also place in or out a series resistor of 0.3 ohms between the FG + and the gate, because a resistor was shown here in one of Ainslie's many diagrams of the demonstrated circuit.)
« Last Edit: March 27, 2012, 10:21:27 PM by TinselKoala »

WilbyInebriated

  • Hero Member
  • *****
  • Posts: 3141
Re: Testing the TK Tar Baby
« Reply #55 on: March 27, 2012, 06:28:19 PM »
Wilby,

Wow, you guys really need to bury the hatchet... life is too short.

You do have to give TK credit for being one of the first, if not the first, to identify the inverted duty cycle issue in the COP=17 circuit.  And, oddly, how that inversion, if the calculations are also inverted, produced a similar COP result.

But yes, I would typically replicate first and then modify second.  I am asked to improve on designs all the time, and I always start with what the customer is currently using to make baseline measurements to improve upon.

PW
:) LOL yeah, there's really no hatchet for me. my righteous indignation at tk is more of me holding up a mirror to him... but he can't see it. or maybe he can and that's why he got so pissed off a year or so back and left 'till recently. i guess i can't blame him... i'm sure he realizes by now that it was even funnier that i made him do all the work to prove himself wrong after he and his sycophants tried so hard to get me to do it. harvey told him a long time ago that his type was easy to manipulate... i think it sailed right over tinselkoala/alsetalokin's head. :)

anyways, you're right, life is too short. i'm going fishing... you have fun here, you'll see soon enough.

picowatt

  • Hero Member
  • *****
  • Posts: 2039
Re: Testing the TK Tar Baby
« Reply #56 on: March 27, 2012, 06:53:51 PM »
TK,

As for frying one of your quads, that is what happens when MOSFETS (or bipolars for that matter) are paralleled withour individual source (or emitter) degeneration resistors.  Likely the one that popped was the one that had the lowest turn on threshold and was doing most, if not all the work.  Source degerneration helps compensate for the variations in the threshold voltage of the paralleled devices.  It is very likely that in the RA circuit not all, or even only one, of the MOSFETS in the "quad array" is doing all the work, with the rest just providing capacitance for AC current.  If turned on hard enough, the rest will eventually carry some load, but the lowest threshold device will continue to carry the most current.

Seriously, off to work now...

Wish I was going fishing...

PW


TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #57 on: March 27, 2012, 08:10:25 PM »
TK,

As for frying one of your quads, that is what happens when MOSFETS (or bipolars for that matter) are paralleled withour individual source (or emitter) degeneration resistors.  Likely the one that popped was the one that had the lowest turn on threshold and was doing most, if not all the work.  Source degerneration helps compensate for the variations in the threshold voltage of the paralleled devices.  It is very likely that in the RA circuit not all, or even only one, of the MOSFETS in the "quad array" is doing all the work, with the rest just providing capacitance for AC current.  If turned on hard enough, the rest will eventually carry some load, but the lowest threshold device will continue to carry the most current.

Seriously, off to work now...

Wish I was going fishing...

PW

I know, I know something about parallelling mosfets. You are now pitching a bit under my level, but that's OK because I know we have people of all levels reading here. Did you see the paper below? I think Fuzzy first found this one. Odd that that Morris didn't notice the massive overunity performance of his five parallelled IRFPG50 mosfets. Of course ... he followed your suggestion as to the proper way to parallel mosfets, and... he didn't get four of them in backwards, like the NERDs did.
I even told them what to expect when their "gang of four" had to carry 10 or 12 amps, much less the 4200 Amps like was implied by the famous "25.6 megaJoules" test that started my entire objection and involvement.

You might also be interested in my TinselKoil. It uses an H-bridge of 4 power mosfets to switch rectified line current at 170 vdc (roughly) through a 4-turn primary winding of a 300 kHz solid state Tesla coil resonator. The mosfet gates  are driven by phase transformer toroids that are in turn driven by a current amplifier H-bridge of silicon transistors, which in turn is driven by a PWM driver chip (TL494 IIRC). All developed by me, empirically and in public view, and all impossible to accomplish without protective circuitry and proper layout of the power stage to avoid stray inductances and achieve clean switching. Videos on my YT channel.


By the way... my Tar Baby oscillates just fine.... with the Q1 mosfet ENTIRELY REMOVED from the circuit. There is little difference in the scope trace and none in behavior except that it doesn't heat the load much that way. I'm pulling and testing all the mosfets now to see if there are any that are open... a shorted one shows up right away in-circuit but the opens are harder to detect.

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Testing the TK Tar Baby
« Reply #58 on: March 27, 2012, 09:47:27 PM »
More on removal of Q1:

With Q1 in place, if one uses a bipolar drive signal instead of strictly negative going, one still sees oscillations on only one phase--- at least I do, but my Q1 is mounted with short leads -- but an interesting thing happens. Something that might not be noticed by someone who thinks oscilloscopes are for drawing pretty colored lines with.

The first shot below is using a bipolar gate drive signal of +/- 5 volts, about. This shot is with the lone mosfet Q1 _ENTIRELY REMOVED_ from the circuit by pulling it out of its socket and laying it down on the bench about six inches away. (Of course you can't see the invisible wires, silly.)

The second scope shot below is exactly the same except with the Q1 mosfet back in its socket. See any difference in the oscillations?

What about NOT in the oscillations?

Now, silly old fumblefingers me, barely tall enough to reach the controls.... I sometimes make mistakes in my knob settings or hookups or whatever. But I try to assume FIRST that I have made some error when I see something strange, so I go back and check my work and try to correct what I've done and interpret correctly what I see, by experimenting all around in the problem space, NOT simply "trying to replicate exactly" somebody else's claim of an a-priori impossibility.

Is it possible that other people, perhaps even more naive than I am, could make mistakes, and perhaps not notice them at all?


picowatt

  • Hero Member
  • *****
  • Posts: 2039
Re: Testing the TK Tar Baby
« Reply #59 on: March 27, 2012, 09:48:18 PM »
TK,

Didn't mean to throw you a low ball, but, as you say, others may find that useful.

Yes, I have seen most of your HV videos, including your H-bridge driven coil.  It really is a thing of beauty and quite the accomplishment.  Fire up all those HV projects, turn off the lights... talk about mood enhancement...

Many (too many) years ago I built a VDG with a 3.5' foot upper terminal complete with cardioid shaped bottom section, 5" or 6"  belt, dry nitrogen filled column and active corona spray.  When it would discharge into its 24" discharge globe, well, let's just say you did not want to be around it.  Always enjoyed a good HV display, DC or AC.

With Q1 out of circuit, you probably lose around 20% of the AC current path.  If you monitor AC current at the CSR, that is likely where you will see the difference with/without Q1 in the circuit, i.e., a bit more AC current with Q1 in circuit.  Even the 830's have what, 600-800 pf of Ciss (I know you posted the data sheet, it was something like that if I recall).  I would think it would show up on a scope trace at the drain as a slight change in the osc amplitude (with Q1 in circuit). 

Break over,

PW