Claimed OU circuit of Rosemary Ainslie

[MileHigh]

Hey TK:

And Aaron, simply turning up the power until your "control" resistor is at the same temperature (monitored how now? and how calibrated?) is not a good way to do it. The more valid comparison would be to do as I showed, where I monitor the time and temperature profile of the load under the Ainslie condition and under DC at the same or similar average input power. Clearly, if the Ainslie system is making more power somehow, the load should heat up faster and get hotter, with the same input power. It's the rate of heating that is most important, at the same known input power.
Your method is actually backwards and is subject to large error. Obviously.

Let's look at this issue one more time.  We want this one to be clear plus if we demonstrate that we can disagree with each other we earn the respect of the opposing camp.  lol

Aaron is actually right when he checks the final temperature as the reference.  Of course you have to be sure that you really are at the final temperature and not jumping the gun.

If the Ainsley system is generating more heat like you say above, the load will heat up faster and get hotter.  Here is the biggie:  The time it takes to reach 63% of the final temperature will be exactly the same for the control experiment or for the Ainsley circuit test.  The final temperatures may be different but the overall timing profile will be the same.

What's determining the time it takes to reach 63% of the final temperature, a.k.a. the thermal time constant?  It's good ol' "RC", but this time we are not talking about a resistor and a capacitor, we are talking about the thermal resistance between the coil-resistor and the outside world, and the thermal capacity of the body of the coil-resistor that's being heated up.

The thermal resistance to the outside world in the control test or the Ainsley circuit test will be the same.  It is related to how air currents convect heat away from the coil-resistor and how much thermal IR radiation it generates.  The thermal capacity for both tests will also be the same, it is related to the physical mass and materials used to make the coil-resistor itself.

So for the control test and the Ainsley circuit test it will take exactly the same amount of time to reach 63% of the final temperature.  The final temperature is of course when the heat injected into the system by the electrical power source is equal to the amount of heat removed by radiation and thermal convection.  It is an equilibrium point.  (The thermal circuit is a current source feeding a cap in parallel with a resistor)

So a general rule of thumb is to wait at least five time constants to consider the temperature to have stabilized.  It the same old exponential curve, where here the plot is delta-T vs. time.

The real data is in the final temperature of the coil-resistor.  The time profile for getting there is of no real importance.

MileHigh

[TinselKoala]

Actually, early on, Rosemary said that most any transistor would do. Several times. She also said that FG drive would be fine. But as we now know, those things are, well, just not true, are they. And how and why do we now know those facts, I wonder?

It's Aaron who said several times that the transistor has to have the repetitive avalanche rating and the high Vdss -- like the IRFPG50 -- which is why he said my 2sk1548 won't work, since right there on the data sheet it says "avalanche proof". But of course searching through Aaron's posts is futile, since he adopts the tactics of Orwell's Ministry of Truth -- "down the memory hole" with so much that has been posted on threads that he controls.

Now, if anyone can point me to a similarly high Vdss and repetitive avalanche rated mosfet that is NOT the IRFPG50, I would be glad to test it.

And in one of the very earliest posts in this thread I encourage critics to compare the data sheets for the two mosfets. Sorry, I failed to point out in that post that the difference in gate capacitance and turnon-turnoff times would likely mean that the 2sk would make better spikes used in this manner. But of course that elementary fact would have been clear to anyone who actually bothered to read the data sheets as I suggested, way back before the troll started pissing on straw men in the thread.

[TinselKoala]

Hey TK:

Let's look at this issue one more time.  We want this one to be clear plus if we demonstrate that we can disagree with each other we earn the respect of the opposing camp.  lol

Aaron is actually right when he checks the final temperature as the reference.  Of course you have to be sure that you really are at the final temperature and not jumping the gun.

If the Ainsley system is generating more heat like you say above, the load will heat up faster and get hotter.  Here is the biggie:  The time it takes to reach 63% of the final temperature will be exactly the same for the control experiment or for the Ainsley circuit test.  The final temperatures may be different but the overall timing profile will be the same.

What's determining the time it takes to reach 63% of the final temperature, a.k.a. the thermal time constant?  It's good ol' "RC", but this time we are not talking about a resistor and a capacitor, we are talking about the thermal resistance between the coil-resistor and the outside world, and the thermal capacity of the body of the coil-resistor that's being heated up.

The thermal resistance to the outside world in the control test or the Ainsley circuit test will be the same.  It is related to how air currents convect heat away from the coil-resistor and how much thermal IR radiation it generates.  The thermal capacity for both tests will also be the same, it is related to the physical mass and materials used the coil-resistor itself.

So for the control test and the Ainsley circuit test it will take exactly the same amount of time to reach 63% of the final temperature.  The final temperature is of course when the heat injected into the system by the electrical power source is equal to the amount of heat removed by radiation and thermal convection.  It is an equilibrium point.  (The thermal circuit is a current source feeding a cap in parallel with a resistor)

So a general rule of thumb is to wait at least five time constants to consider the temperature to have stabilized.  It the same old exponential curve, where here the plot is delta-T vs. time.

The real data is in the final temperature of the coil-resistor.  The time profile for getting there is of no real importance.

MileHigh

Um hm, your analysis is in depth and is correct, I have no problem with it. But I still maintain that the rate gives important information as well, even if, as you show, it is really the same information. Take a look at this graph, again. You, and hopefully everyone else, will see that the equilibrium temperatures are approached  asymptotically, and just as you say one must wait a while for the temperature actually to stabilize in its dynamic equilibrium with the surroundings. The "leaky" calorimeter idea is clearly crucial here. You can't be totally open to drafts and convection and you can't be totally thermally sealed either, to do it on the desktop in a reasonable time.
The rate information is important, well, because it makes nice visual impact, for one thing. See?
But I hope we are agreed that the way NOT to do it is to sit there, turning up the power supply a little at a time, until you think your resistor is at some particular temperature, and then reading the power at that point. That's just backwards. Even if one discards the rate info (a mistake, I maintain) one still needs to set the power at the beginning and keep it there, at whatever chosen value, and let the load temp equilibrate without further fiddling.

[Yucca]

Current methods of calorimetry used on this circuit are not ideal due to heavy thermal coupling to the atmosphere. The maths to straighten it all out for absolute values are difficult and its much simpler to do better calorimetry.

A hot drink glass vacuum flask filled with distilled water, deep styrofoam plug in the top with holes for load leads and thermocouple. Dip any exposed leads of the load in varnish. Run X joules in with straight DC. Run X joules in with claimed OU circuit. Same water volume and start temp in each case. For best accuracy run times should be equalised by varying (via voltage) the DC input power to match the pulsed RMS power. For even better accuracy two identical calorimeters should be built and ran close to each other.

edit:
TKs aymptotic equilibrium curves of the leaky calorimeter runs above do say underunity if they were conducted at approx the same ambient temps, pressures and relative humidity. But without real calorimetry obtaining absolute heat out figures is difficult and noisy.

[TinselKoala]

While we are on this topic, I would like to point out a subtle fact that maybe Aaron is missing: The circuit, operated in this oscillatory mode, produces waaaaay tooooo mucccchhh heeeaaaattt.
How do you, Aaron, or anybody else, account for the fact that in Ainslie's reported experiment, her load equilibrated at about 50 degrees over ambient in "about an hour"...
Whereas anybody who is experiencing these oscillatory experiences is experiencing a load that heats to, well, let's just say it gets HOT, and it gets HOT FAST.

The implication is pretty clear to me.  I'll even say it out loud. This mode of operation, call it the Aaron mode, is NOT the mode of operation of the Ainslie circuit reported in the article.

Plus, one may compare the results in the graph above with Ainslie's actual reported heat profile. As anyone can see, my Ainslie load, driven at a known and stable 4.5 percent ON with no "oscillations", got to about 50 degrees above ambient in about an hour. Just as Ainslie reports in her experiment.

But the circuit operating in "Aaron mode" makes the load heat to well over 100 degrees over ambient in just a few minutes. Just like using Yucca's mosfet-recycling circuit. And for the same reason. And with the same effect.

This is why I say the oscillations are a red herring, and when we are discussing them, particularly when they are generated in the manner being used, we are no longer discussing the Ainslie circuit or her experiment.