[idea] quick calculation for lpf setting

[jy0933]

i'm trying to figure out which freq lpf is the best go with for best result.

then here's what I got.. I dont know if it is a right way to do it..


assume the motor is out of balance..

1. the first thing is to find the motor rpm at hovering, assuming N

2. consider this is the hovering, we still want to eliminate the possible vibration on descending>>> take 80% of hovering as the lowest frequency we want to elinimate>> 0.8*N

3. convert rpm to hz>>> freq= 0.8*N/60 (hz)

so the biggest problem is finding the rpm at hovering
best solution is use an rpm sensor/tachometer
secondary is to use ecalc or thrust calculator (like this one http://personal.osi.hu/fuzesisz/strc_eng/index.htm)
and take a factor of 0.9 to eliminate the drag (10% is huge already)

then choose the closest lpf in mwc for best result

I wonder if this is going to work?(from the code aspect ?!)

also, it seems to me there is only gyro lpf... is there also acc lpf and mag lpf?.. or the lpf is generally applied to all 9 axis?

[furex]

i'm trying to figure out which freq lpf is the best go with for best result.

well balanced motor and props is the best way.

like you, for my knowledge , i wondering about methods to measure frequencies.

with motor KV, we can approximate RPM:
ex=> 1200KV Lipo 3S
=> 1200 X 12.6=> 15.000 rpm
=> 250Hz motor freq

if we use a 2blade props, the frequencies is multiplied by 2.No?

IMO, the best way is to log data from sensors ,and to make some graph.
Vibrations from propulsion kit ( motor & prop) must be studied. but we must calculted the real view from sensor beacause the frame can decrease or increase this phenomena.

[jy0933]

yeah.. of course the balance process is a must.. but sometimes it is just not that ideal... lpf will still do some help after all..

about sensor data log .. i tried to grad the raw data.. but it was not significant even i push throttle above hover point.. it jumps between +/-5... i was thinking if we can get more accurate log.. say 100x the raw data.. then the small number becomes significant... should add some accuracy on analyzing the freq

[furex]

about sensor data log .. i tried to grad the raw data.. but it was not significant even i push throttle above hover point.. it jumps between +/-5... i was thinking if we can get more accurate log.. say 100x the raw data.. then the small number becomes significant... should add some accuracy on analyzing the freq

+1K

[furex]

what we need is there : click-me and me too