I agree “some” voltage sag is to be expected and is perfectly normal during flight, however; not to this degree. generally on my aircraft i expect a sag to maybe 3.6V/cell maybe 3.4V/cell for micros, but it should then plateau to a steady rate whilst the bulk of the battery’s capacity is being used. The only time this doesn’t happen is if you are drawing more current than the battery is rated for, which in this case would have to be over 21A continuous.
To better explain the issue I have just done a timed steady hover in my living room and here are the results, I’m sure you will agree that something is not right here.
Battery starting voltage: 4.2V Battery used: Turnigy nano-tech 600mah 1s 35/70C
0m:07s - 3.5V (expected initial drop)
0m:13s - 3.4V
0m:16s - 3.3V
0.54s - 3.2V
1m:28s - 3.1V (battery alarm “flickers” once or twice beyond this point)
2m:08s - battery alarm steady
after 20s battery had recovered to 3.7V showing that very little of it’s capacity had actually been used and the quad was able to fly again and followed a similar pattern.
whilst you have already mentioned the Vbat sensor maybe on the fritz (which I hope it’s not on a brand new quad) as the test went on the quad did require more throttle input near the end which is characteristic of a dead battery. but after the 20s battery recovery it went back to flying like it was on a full battery again which sort of proves this isn’t just the board coming up with false Vbat figures.