Tweet
Originally posted by gadh
View Post
I'm not sure what you mean by 'binding', but if your capacitance values are the values between turns on the resistor and you have 48 turns then I would expect the total capacitance to be 1/ ((1/c1) + (1/c2) . . .) up to 48. Thus if all of the turns have the same capacitance per turn (ct) between their nearest neighbors, we can use 1/ (47*(1/ct)) to arrive at the total capacitance of the coil because they are all in series with each other. So your measurements are a little confusing if the 'binding' means turns. I would expect 0.31 pF for 3 turns if any and all single turns were 0.93 pF between a nearest neighbor and an overall of 0.01979 pF for the entire coil. We use 47 instead of 48 because the capacitance is in the gap between the turns and their are only 47 gaps.
As far as the inductance goes, we cannot believe anything we have read unless it has been properly validated. The inductance of a coil is very specific to the geometry and the published geometries are simply incorrect and even impossible to achieve as stated, this has been openly admitted to. IIRC, the inductance was based on a calculation, not a measurement. So it is very good that you have actually done the measurement. I'd have to check my calculations, but I think your measurement is quite close to the calculation I arrived at for Glen's resistor.
Let us consider the self resonance of your resistor (the ringing frequency). This is the frequency that we would expect the coil to 'ring' at if it were solely connected to a battery source, raised to full current and then both ends of the coil were disconnected abruptly and simultaneously. The energy stored in the field would rush back into the coil causing an EMF to form on one terminal, and then that would setup a secondary event of causing current to flow through the coil in the other direction which creates a new reversed polarity field that when the current stops flowing would collapse and cause EMF to form on the other terminal. This process would continue back and forth until all of the energy is fully dissipated and that is what we call a decayed ringing or a damped sinusoid (See Laplace Time Domain f(t) damped sine equation here: http://people.seas.harvard.edu/~jones/es154/Laplace/Table_pairs.html and here:http://highered.mcgraw-hill.com/site...z80377_apb.pdf)
To calculate the expected frequency of that waveform we can use the resonance formula f = 1/2π√LC where L is Henries and C is Farads and f is Hz. Using your values, we have 1 / (6.2832 * (√(0.0000235 * 0.00000000000001979))) = 233,379,821.5 Hz (or 233.4 MHz). That's what we would expect if all other interactions were removed from the inductor.
However, in the MOSFET Heater preferred mode of oscillation we do not see a primary damped wave or inductive ringing pattern nor do we see primary frequency anywhere near 240 MHz. Even though we have seen frequencies displayed as high as 900 MHz, these have been secondary lower amplitude effects that are not considered to be part of the primary retriggering frequencies that are typically between 140 - 250 KHz. So obviously, we have something else happening here that is only partially dependent on the inductance of the resistor. A careful observation of Glen's waveforms will reveal that the energy contained in the magnetic field does not reverberate in the decayed ringing waveform mentioned above but rather it is fully dissipated within the first half cycle so that even the internal body diode does not get a chance to activate on the negative half of the cycle because the energy is already spent before it reaches the negative transition. This means that the energy to dissipation ratio is well matched in Glen's precise configuration. If the inductor were too large, then the energy stored in the field would be too great and negative ringing would ensue. If the inductance were too little then not much energy would be dissipated and the heat would suffer. When the ratio is well matched, all of the energy goes to heat.
We have already discussed the rise time of the current curve and the disadvantage of pumping current through the resistor after it has reached 99% of its peak value. At that point the inductor is fully charged, meaning that the field cannot take on any more energy and there is no longer any inductive reaction because the flux becomes static at that point. So once the curve levels off (or just before) it is time to turn off the MOSFET. At that instant, the resistance of the inductive resistor along with the capacitance of the MOSFET and the energy stored in the field all play a part in what is left to become the BEMF spike. The area of that spike is representative of the energy in it. If the MOSFET turns off fast, then the area is distributed as a high amplitude narrow spike. If the MOSFET turns off slow then the area is distributed as a low amplitude wide spike. Either way, the energy is the same. We do not want that energy to be passed through the MOSFET which at this point will look like a 2800 pF capacitor to ground. Instead, we want that energy to be reflected back through the resistor and fully dissipated in the coils so that no energy is left to form a reverse BEMF spike on the other end of the coil. I have noticed that in some of Glen's videos, the frequencies did shift enough that he lost that good match and some of the energy was wasted in the MOSFET diode.
This is why the retriggering (as described in other posts in this thread) seems to improve the function of the MOSFET heater, because it forces an early switching of the current and moves the duty cycle closer to 50%. There may be some clashing or collision of the energy in the coil due to the aperiodic operation as well. If a new trigger event begins before the BEMF has fully dissipated, then we end up with a current collision where the BEMF is moving conventional current back toward the battery (+) and the MOSFET turns on asking for current to be moved the other direction to the battery (-). The occurs because the inductor is acting as a separate energy source . Such clashes would create standing waves in the inductor which are well known to dissipate energy in antenna leads and load coils. The SWR (standing wave ratio) is typically kept as close to 1:1 as possible in those systems to prevent power loss. In the MOSFET heater we prefer just the opposite, we want the energy to dissipate in the heater coil and it is best suited to have it fully dissipated in a single half cycle.
If it's possible to disturb the vacuum and get energy to fall out of it, then I think these very narrow high amplitude BEMF spikes coupled with aperiodic magnetic confusion (SWR collisions) in the heater coil is a good way to shake it up and see what falls out.
Any actions above 10 MHz are also candidates for dielectric heating.
Plotting the curve of the MOSFET Ciss capacitive reactance (2800 pF) reveals that it becomes impedance matched with the vacuum (300 Ohms) at 189,470 Hz. Plotting the curve of the Heater Coil inductive reactance reveals that it does not match the vacuum impedance until it reaches 1.948836 MHz so it is unlikely that we are getting a resonant action there unless there is a slight shift in the values and we are aligning a subharmonic. Also, we find on the MOSFET capacitive reactance curve that it does not match the 11.2 Ohm resistance of the Heater Coil until it reaches 5.075093 MHz.
What all of these mismatched impedances tell us is that the MOSFET heater circuit is not a resonant system. If it were a resonant system it would have a near zero resonant impedance and an 11.2 ohm resistive impedance to the battery and would drain it with haste with a repetitive 2.14A draw at nearly a 50% cycle. That alone would dissipate 51 W of power in the 11.2 ohm resistor 50% of the time giving an average 25.5 W just sitting there resonating at a single frequency. This is not what we see happening in any of the recorded data available to us from 1998 onward. So the evidence is reasonably clear that this is not a resonant system - the specified parts just simply are not set up for any resonance. Not included here in these numbers is the wire and battery inductance and capacitance. Some of the high speed oscillations present in Glen's picture's could not be reproduced in the spice simulators until we added wire capacitance between the CSR and battery. So I was able to conclude that some of the interactions were related to the proximity of the positive and negative wires laying close to each other and having capacitance between them. However, those transitions were very small in comparison to the rest of the signal amplitudes and can probably be dismissed but I mention them here because those things can directly impact resonance.
Now, if by some arrangement you were able to reduce the effective capacitance of the MOSFET to 272 pF then it would be nearly impedance matched to both the vacuum and the Inductor for resonance 1.948836 MHz. You may want to try putting a 301 pF capacitor across a 1 Ohm resistor and place that assembly between the Heater coil and the MOSFET. The series capacitance of the 301 pF and the MOSFET 2800 pF Ciss would then be an effective 272 pF. The actual impedance of the entire system would be somewhere around 12.2 Ohms, but from a reactance perspective the inductor, the capacitor and the vacuum would be reasonably tuned and phase compensated. This last suggestion is purely experimental and deviates from the prior work done in this area. But is is a step toward a resonant system if that is what you want to experiment with.
The NE555 does not have a frequency range as high as 2 MHz and those parts that do have a range that high would need to be buffered to drive the MOSFET. So trying to push things up into that range could be problematic. However, you could use your existing 555 at a lower frequency multiple and try and strike the resonant system say 1/4 of the time (every fourth beat) and see if it syncs up. So if you set your NE555 for 487,209 Hz things could sync up on the 4th harmonic.
The frequency is a function of the charging and discharging of the timing capacitor. The pots and the series resistors control the RC constants involved. Diodes have been added to allow steering a separate charge and discharge timing. The capacitor charges from the supply line through that pot, resistor and diode and it discharges through pin 7 and the other diode pot and resistor. If you post a link here to the schematic you are using, I will specify the values and part numbers for your desired frequency range.
.
Comment