The data for the last week was promising but far from incontrovertible. The excess energy was above the confidence limits, but still low enough that some unknown systematic change could account for it.
Because of hard drive problems, we ended up resetting the instrumentation and letting the cells cool down. When we were ready to warm it back up, we took the opportunity to do do a test requested by an enthusiastic regular commenter David Roberson.
We stepped up the power on the active wire (red channel) and then stepped the power up on the NiCr wire (blue channel) till it was at near full power. Then we cooled the cell, again, before stepping to full power on the red channel and then full power on the blue channel.
David has prepared a transient analysis routine based on differential equations and has been comparing the rise and fall of power steps to the equation model he has fit to it.
I have run the latest calibration power steps in my program and the system is well behaved in the time domain. I got an excellent quadratic fit where the R^2 value=.999912 using data derived from the end of each of the power levels.
There appears to be a step in power out of .4 watts when the power input was 68.9 watts that occurred near the end of that input level. The mica was running around 203 degrees at that level of input. This result needs to be further verified.
I have concentrated my runs on cell 1.0 thus far. The major power steps that we planned might reveal additional information.
The fact that the curve match was excellent over such a large range of input powers suggests that there is most likely only a small quantity of excess power being generated during this test sequence. Perhaps the cell needs more time to operate before it becomes active.
I just completed running my program on cell 1.1 data and it is acting endothermic according to the simulation. I was able to obtain an excellent calibration curve fit to the quadratic equation with the R^2 = .999984 which is better than I obtained for cell 1.0. The excess power step for cell 1.1 obtained when both wires are conducting maximum current is approximately -2.2 watts which I believe is within range of what I was seeing on earlier tests.
An interesting difference is that the excess power output of cell 1.1 with just the Celani wire driven measures approximately zero.
When driven at maximum input power cell 1.0 was generating zero excess power while cell 1.1 appears to be absorbing power. This situation should be further explored.
I am working upon an alternate explanation as to what is causing the cell calibration to drift with the assumption that it might be as a result of varying amounts of IR being emitted by the wires. A better understanding ot this issue will be forthcoming as I continue to review the system behavior.
Both cells follow the calculated time domain curves quite well and I do not detect any significant deviations. The quality of the curve fit as compared to the actual data is remarkable. If excess power were to be generated at a significant amount for any changing cell temperature, then the curve fit that is observed would not be this ideal.
I am attaching two Excel files for you to use as you prepare the document. The curve fit file which is number ...FC0101 (72).xlsx looks very impressive. This is where I took your downloaded data and used it to obtain the a,b, and c=0 parameters. The first sheet is a repeat of the download itself while the second one demonstrates how I obtain the coefficients.
The second file is a display of the time domain responses to the large steps that you ran. It is remarkable how closely the actual data follows the delayed calculated values. Either or both of the curves look very good but undressed.
Here are two example Excel files and a more detailed write up of his methods all zipped up together.
Now that we are back up to nearly the input power level we were before, it is definitely interesting that both cells are showing significantly lower output than before.
We approached where we were before by small steps. Cell 1.1 tracked somewhere near zero P_xs, but Cell 1.0 lingered behind in the negative 2 to 3 watt range. Think it is an exothermic phase? Is this the zero output level and our calibrations were high?
We'll let this cook for a while, I think. Then we will attempt a suggested experiment by long time active follower Ecco; a high temperature and pressure run to test for more loading.
In other news, we have preliminary calibration data and analysis of the stainless cell in the Air Flow Calorimeter. I will have to put that in the next post, though. And, before I can share the data live, I would like to rearrange the website to list the different experiments more clearly. Any suggestions?
UPDATE#1 - back to higher Pxs
So the past seven hours started with a large uptick in PXs that seamed to be precipitated by a sharp (but very small) change in T_Ambient.
This seams to have stabilised around 2.75W with spikes above 3W.
Might be good to let this one run for a while. Here is a composite long time graph (in Indian time) spanning back to the last power step ups made in post Roberson test.