Mathieu, in France, has finally got to the bottom of a problem that has been holding back the re-running of the dual cell Celani problem for quite some frustrating time.
In the Celani Dual Cell with Protocol V1.3 experiments, with no active wire contributions, this differential experiment is supposed to show identical temperatures on both cells with the same power input right? But there was something not quite right, they did not, there was a temperature differential with the same input power even before ICCF18 and so the approach for the first experiments was to use calibrations to compensate any lack of cell correlation.
Following ICCF18, Mathieu wanted design out this apparent flaw. In order to understand the system better he opted for new, more accurate thermocouples with uniform attachment because he was not satisfied with the previous temperature measurement of the surface of the tubes. The component chosen is listed on omega as SA1. They are able to sustain 175°C before unsticking from the tube. The documentation can be downloaded here. Compared to all the other means of measurement, the mode of attachment is designed to be identical in shape and have the same reflection and so multiple sensors would respond in the same way.
Installing the new sensors proved that Mathieu’s concern was well founded. The far greater precision of them, showed precisely the magnitude of the discrepancy (up to 10ºC). Additionally, these sensors are not reflective so they acquired the same temperature as could be independently confirmed by an IR thermometer. So, we knew they were reading the right temperature, so Mathieu started to investigate why it was that with the same input power into each cell, they were giving different measurements. The first step was to characterise the discrepancy. He did this by measuring the difference in reading for 10, 15 and 20 W of power input in a fixed 2 mbar vacuum. Temperature was seen to diverge as power input increased.
Then he swapped the integrated power supplies to see if it could be something to do with them being different in some way, it didn’t. Then he used different power supplies (Our Thurlby Thandars ) but it didn’t change the result. Then he decided to wrap the cells with aluminum foil and some kapton tape, but again, the temperature difference remained the same. Such a frustrating situation! Mathieu decided to wait for Jean-Paul Biberian’s visit because he was so stumped and wanted to take external advice.
They repeated the tests together with an identical nichrome wire to that of the control cell, sacrificing a brand new, unactivated, non-loaded Celani wire, just in case that was the reason. The wire was given to Jean-Paul for him to test resistance drop, so not totally wasted! They tested over and over again until it became clear that there was only one possible explanation for the observation. Could it be that tube thickness was the cause? In the extrusion process of the borosilicate tube, it happens that thickness and concentricity between the external surface and the internal one is not always certain. Bob suggested Mathieu used a spare tube to test his hypothesis by changing a single cells glass. Finally, the amplitude value of the differential temperature finally changed... This is quite a issue and must raise many questions we will try to discuss in the comments.
Two new tubes ordered
Mathieu devised a solution to the problem. It was decided to change both tubes and take further precautions in their manufacture. First, these two tubes must come from the same original extrusion. This will keep the variation in thickness to a minimum and keep the radial variation consistent. Then, the glass-blower was instructed to mark the tube to allow proper rotational alignment, Mathieu provided the example image below to the manufacturer.
This rigorous approach should keep all inconsistencies of the glass manufacture equivalent as much as possible between the two cells. This in turn should resolve the observed base temperature differentials.
Nearly ready to run
Ok, so we are waiting for the new glass now, but it is important to note that Mathieu has done a fine piece of research here in order to iron out an annoying inconsistency that had to be calibrated out in the first runs of these experiments. If all the investigation above pays off and the new glass delivers on design and expectation, we should have a re-run of this important experiment with greater integrity and less reliance on calibration curves. This will form an important pre-cursor to the mass flow calorimetry that he is similarly being very cautious to get right. Also after calibration, loading and a bit of active running, we'll be able to look for Gammas again - at last!
One other thing, we are going to do what many have asked for and foil wrap before calibration and active run of this experiment, we know you have wanted it and we do listen!
UPDATE #1 - New glass greatly reduces sytematic differential
Mathieu has installed the new glass and we are happy to report that the temperature differential is far less, reducing from 8.7ºC to less than 2ºC and may even not be divergent but an offset, though this will be properly assessed when he does the calibration curves.