On 14/9/2010 I changed the transmission mode from FSKCW3 to FSKCW6 (by reprogramming the keyer microcontroller) and went back on the air on same frequency. The 3 dB improvement on SNR (because of mode change) didn't seem to be enough to make my signal detectable on US grabbers (at least during daytime where I am not sleeping and I am able to check the websites). So next day (15/9/2010), after a lot of trials and errors, I succeed to remain within the 100Hz QRSS window and stay stable on 10140.075 after the MEPT's warm-up period. After my comment on QRSS Knights mailing list for the reason of my frequency change, I received the same day, an email from Bill W4HBK in Florida. Bill wrote me that he was receiving my signal "often" and it was "one of the strongest from Europe" BUT he forgot to send me the report through the mailing list. He commented that my signal was appearing every day around 3:30 UTC (during my sleep) when the local US signals fading out. He also sent some screenshots of my FSKCW6 signal and was asking more info about my antenna. Now, it was clear, that my FSKCW3 experiment was more successful than I thought. My signal was detected from USA grabbers but I was sleeping "too much"... to realize it!
The next days, I continued to transmit on a 24-hour basis because I wanted to check if I will receive better results with the FSKCW6 mode. From the first day, the MEPT was working without any special box for ground or thermal shielding. Even that this didn't seem to be a problem, after the frequency change (I did) to 10140.75, I realized (by checking the online grabbers) that every time I was sitting on my desk, my "body capacitance" was affecting the MEPT's unshielded crystal oscillator and the frequency was detuned several Hz according to the distance between my body and the oscillator. Also the last days' weather change had as result several degrees change in the room's temperature and of course changes on transmission frequency. These frequency changes made the situation worse than before my QSY from .030 to .075. Now my signal, most of the day, was somewhere around .060 and it was always mixed with European MEPTs!
The idea of continuing to operate without ground and thermal shielding was not good at all. I should do something as soon as possible otherwise it would be better to stop the experiments. Even that the usage of a crystal oven is the optimum solution in applications where thermal stability is very important, in my experiments this would require a lot of time for oven's design and construction and would give more complexity to the project (something I didn't want). So I decided to do something (on the "quick and dirty" basis) that will improve the frequency stability without being complex.
Above you see the improvements... is not something special but is effective and very far from "rocket science". I just installed the MEPT inside a metal box that is grounded and makes MEPT unaffected from "body capacitance" and nearby metal objects. The metal box is installed inside a Styrofoam box that isolates thermally the metal box from the room environment. When we say thermal "isolation", we mean thermal energy (heat) transfer through the Styrofoam material in a very slow rate. Do you wonder how this affects the thermal stability of the MEPT? Inside the MEPT's circuit there are 3 transistors & 2 voltage regulators that produce heat during transmission. Under normal conditions and when the circuit works in "open air", the produced heat is dissipated in room's environment very fast and the transistors' surface temperature is a little bit higher than room's temperature. When the MEPT works inside the Styrofoam box, the produced heat is trapped in the air (inside the metal box) and the Styrofoam material does not allow the heat to be dissipated fast in the room. The result is to have a temperature inside the metal box many degrees higher than the room and a very slow rate heat transfer from inside to outside (of the Styrofoam box). In practice the metal - Styrofoam box works like an oven and makes the internal temperature unaffected from the rapid and small range temperature changes inside the room.
The modifications took place on 16/9/2010. Since then the Styrofoam box is warm and the MEPT is almost "locked" on 10140.080. No more frequency drifts because of proximity or room temperature changes. Of course, if I turn on the air conditioner or move the MEPT outdoors, the Styrofoam box is not enough to keep stable the MEPT's temperature (because of the great external temperature changes). In this case, a real crystal oven is the only solution and it will be one of my future projects. BUT for indoor usage, Styrofoam can give good results without increase complexity!
 SV8GXC, My 30m QRPp MEPT, http://sv8gxc.blogspot.com/2010/09/my-30m-qrpp-mept.html