She’s undergone a complete change of character this morning. Same test parameters as before, yet she was climbing at about 1ms-1 – at 3 meters I aborted. Luckily no damage done except to my confidence.
Just look at the graph:
- very noisy vertical accelerometer results shown by faz (cyan)
- ever increasing vertical speed maximum at 2.4ms-1 shown by fvz (magenta)
- height of 2m when I aborted (green)
The target vertical speed was 0.33ms-1, and you can also see in red the desperate attempt the vertical velocity PID was making to turn down the power trying to slow her ascent down, and yet it had no effect. It’s almost as though the ESCs were ignoring the incoming PWM and doing their own thing.
I had changed 2 things overnight:
- I’d decremented the I gain for vertical speed to 10 from 30
- I’d charged the battery
Neither of these provide sufficient explanation for the huge increase in motor power and lift.
It’s as though someone had changed the motors overnight, and the 630us pulse I measured and tested yesterday as the perfect hover speed was today ridiculously high, and having reduced the VV PID gains to align with yesterday’s 630 hover speed, they were unable to reduce the motor rotation speed sufficiently to stop her manic climb.
I just don’t get it – how can there be such a radical change in lift produced by the motors after a battery charge – for a given PWM they should rotate at the same speed and therefore produce the same level of lift, regardless of battery power. I’m flumoxed!
I might have to change back my newer motors – they never did this to me. Before I do though, I will rerun this test with various initial hover speeds to see whether they all converge onto a common hover speed via the VV PID. That won’t explain the difference between yesterday’s and today’s tests, but will help to some extent to rebuild my confidence in these motors.