Paradigmnoia
Nice! I am amazed at the difference that sealing leaks makes. I suppose that this is something that becomes particularly important when you have forced-air cooling.
Now that you have lots of experience with this type of calorimetry, have you come across anything in your non-LENR system that would account for the difference between the active and inactive runs in the papers by Rothwell and Mizuno?
I haven’t even sealed the panels yet. The small leak at the fan (just upstream of the outlet thermocouples) was probably a bigger issue for measurement than the possible leaks at panel joins which are overlapped and a convoluted way for air to sneak in (and be heated by the inside air anyways).
Eliminating the excess mass of bricks and the heavy acrylic box has shortened the warm up time by 3 hours, and the same for cool down. It was pretty much at steady state when I remembered to put on the fan cover. I’ll test that again today, with the fan cover on from the start. It is just a box made of foam, not really sealed but enough to keep the fan heat loss to a minimum. The exposed fan heat loss is about twice what the fan consumes in power.
There are several potential pathways to erroneous results in this method, but a method that results in getting positive errors in only the “activated” experiments is something that I cannot explain in a consistent, believable way. Most errors in the measurements or method will cause a constant offset in all tests. The air inlet area geometry and the outlet fan connection to the box are the most sensitive areas for transient errors. Since the Delta Temperature is directly proportional to the measured outlet power, getting a strong temperature : power response ratio is important to weeding out artefacts.
The main thing is now a round trip, heat-cool cycle is shortened from 6-8 hours to 2-3 hours, meaning various things can be tested about 3 times faster. Plus I still have a separate, original Mizuno configuration to compare to.