I have been experimenting with the data collected by the Mizuno calorimeter and have derived an interesting system to compensate for the short comings of that device in an effort to uncover an accurate measurement of the true energy produced by that pulse drive waveform. The final results are remarkable and I believe accurately reveal the desired information.
I initially began the project by developing a computer simulation of the expected waveforms displaying the temperature of the coolant water as a function of time when subjected to input energy pulses. It takes two important time constants in order to duplicate the measured waveforms to within a reasonable degree of accuracy. I also discovered fairly soon in the project that a compensation signal was required in order to take into account any leakage powers that enter into the coolant as a result of pump power. Without this addition a strong bias is seen to the signal that entirely obscures any hidden real signals. The ambient temperature finds its way into the coolant water through the thermal resistances separating them and I had this included as a portion of my original model. This worked quite well and I had an excellent simulation of the expected time domain response of the coolant that matched the real life data fairly well. The correlation was adequate, but one issue remained which tended to complicate measurement of the true signal energy due to an imperfection of the calorimeter. The thermal time constant of the existing calorimeter should ideally have been infinite. Had this been the case no energy would leak away from the thermal capacitance of the device and thus any signal present would remain constant in storage forever. Unfortunately, leakage is present and the signals drain off over time until none remains after a number of significant time constants. In this case that important time constant is approximately 6 hours. It occurred to me that I could take this signal energy leakage into account by adding a technique that balances out its generation into my model. Since with this modification there is zero net energy deposited into the thermal capacitance, none will leak out over an extended period of time. This idea appears to work perfectly! So, my final model takes the ambient noise and adds to it a calibrated input pump power signal plus the time domain response due to an ideal signal of 20 watts amplitude along with a calibrated extra power pulse that represents the hidden signal. The suspected signal power can be adjusted until an ideal balance is seen at the calculated water coolant transient response. Of course the ideal response for my system is for the water to remain at the same temperature over the entire period of the test. The actual display that is graphed is determined by taking the real coolant temperature data as given in Jed's report and subtracting it from my calculated waveform point by point. The coolant values are never adjusted directly in any manner before this subtraction takes place so they are not altered. I then take the difference determined above and filter it through a three pole Butterworth digital IIR filter in order to clean up the graph for ease of display. I have generated a JPEG file that demonstrates the ultimate output of my balance system that I think most of the members of vortex would enjoy viewing. In the past it has been impossible to attach such drawings due to constraints of the mailing list. I will attach that drawing to a second post within a few moments. If it does not show up, anyone wishing a copy can email me directly and I will send it to them individually. Also, in the interest of open science, I will be happy to send a copy of my file to anyone who requests it. I used LibreOffice Calc for the platform. It will be possible to convert the file into one that can be used by excel if required. Dave
