You are using the governor? If so I can't see how it is speeding up. I could imagine it slowing down if your battery was running down and going below the voltage needed to maintain a constant rpm. But not speeding up.

My setups have been pretty constant--although inverted flight is 0.1 second faster than upright. I attribute this to the apparent fact that my plane trim is more efficient inverted than upright. Takes less power too (just a little less).

The ESC knows exactly the rpm (for a brushless motor) since it has to do the commutation itself. The only issue is how it corrects for different loads.
A long term speedup makes no sense--if you are using governor mode. At least not to me!

Again the governor should account for this--but maybe Walt isn't using a governor mode.

The other off the wall possibility is that the throttle/timer control is setup to increase the throttle over the flight. Normally (or almost always) that feature isn't used with the governor mode. I do use it in my 1/2A when I don't use a governor to compensate for the dropping battery voltage. Of course it isn't perfect compensation, so it might overcompensate (or under) depending on how you setup the throttle rise.

Just a guess!

Yes I am using Heli High gov. mode and it seems to work the entire flight. In reading
all of your comments, the one that might be responsible is the heat or warm up factor.
The ESCs warm up so do the motors and batteries. I was told last year at the NATs
by E-flite engineers that the motors and batteries really don't reach their best efficiency until a much higher temp than I like to run. Currently I am using 68-70% of
the battery rating and the temp ranges from 116 degrees to 124 degrees F.My ESCs are mounted in the nacelles and have fair cooling. I also noted the same speed up on my single motor NAKKE with an AXI2828/10 with same batterys EVO lite 14.8v 4270MAH.

Walt   
 

Alan
I am using a ztron it is an older version with a pot. It does not have a voltage comp.
feature. It puts out a constant 1.285 ms. However I do use a JMP on my Gemini
twin which I have converted to Heli high Gov. mode. I have noted the same small lap speed increase during my flights. I have not measured lap times inverted,but I will
on my next flights.

Walt     

Well there could be one more possibility.

I have two "identical" CC Phoenix 35 ESC's. I set the newer one up to use on my motor test stand to make some measurements. Using the same JMP-2 timer that faithfully gives 8100rpm on all 14 magnet motors with the , I noticed that the newer one gave a slightly higher rpm (8200 from memory). I attributed that to the fact that the ESC has to convert the pulse width from the timer into a rpm setting, and perhaps the ESC clocks were slightly different--or whatever circuitry is used in the ESC was slightly different. Maybe as the ESC warms up during the flight, this circuitry slightly drifts making the ESC think the throttle setting is slightly increasing. It is also possible the timer/throttle could also slightly drift if its temperature changes. My timers are on the outside of the plane, so they take on whatever that temperature is, and as I note, that has varied all over the place with at most only slight (and unnoticeable to me) rpm variations. Maybe I'm not looking hard enough.

Or something like that!

As far as I know, in the Castle Creation Phoenix line, the ESC uses the pulse length from the timer/throttle to set an appropriate count rate---the ESC counts the magnets as they rotate by the armature. That is why the kV of the motor doesn't matter. As long as they have the same number of magnets, any motor will give the same rpm if hooked to the same controller--within the limits of the voltage available of course. A motor with a different # of magnets will just scale its rpm. Common magnet counts are 10 (the Brodak stock motor and high kV type motors), 12 ("CD ROM" type motors), and 14 (most AXI and Scorpion "standard" motors). Of course I am talking about outrunner brushless motors. Inrunners have a lot fewer magnet poles (2 in one I have).

"Clearly" there is no issue in the counting (because it has to have that circuitry to know when to do the electronic commutation for the brushless motor), but in order to get a rate (=counts/unit time) it does need a clock. Also that clock is used to convert the timer/controller pulse width into a digital value (or equivalent) to make the pulse rate comparison. So my guess is that this clock might be sensitive to temperature changes.