Tuning the Harmonic Notch

ArduCopter Copter 4.0
,
201

That looks much better!

202

Thanks for all your help

203

Do you know which sensors you have? I thought fmuv5 was BMI88, but it doesn’t look like it from the log.
I’m interested in the very prominent second peak. I think this may be due to frame resonance at the rpm harmonic - it might be worth putting a static notch there to take that out.

204

Here are my Sensors

  • InvenSense ICM20689 accelerometer / gyroscope
  • InvenSense ICM20602 : accelerometer / gyroscope
  • Bosch BMI055 accelerometer / gyroscope
  • MS5611 barometer
  • IST8310 magnetometer
205

What would you recommend for the static notch?

INS_NOTCH_FREQ = resonant peak in Hz to set the notch center frequency
INS_NOTCH_BW = the notch bandwidth, a reasonable default is half of the center frequency
INS_NOTCH_ATT = the notch attenuation, higher attenuation will make the notch deeper and narrower

Do these seem about right?
INS_NOTCH_FREQ = 110
INS_NOTCH_BW = 25
INS_NOTCH_ATT = 30

206

I would do:

INS_NOTCH_FREQ = 130
INS_NOTCH_BW = 40
INS_NOTCH_ATT = 40

that looks closer to what the graphs are showing.

You should set your IMU flags to enable to the BMI055 - doesn’t look like it’s on at the moment and I suspect it might give you higher quality data than the other two.

207

Any idea how to set that flag to enable it?

208

It’s either INS_USE3 or INS_ENABLE_MASK or both

209

Thanks. Do I want to use all three or just that one that you mentioned in particular?

210

I think look at the graphs and then decide. You will get higher sample rates with the BMI55 I believe

211

Thanks. So I will enable all 3 and then see which give the best graph results. I will post this late tonight if I have time to get to the field later

212

Actually I just went to adjust my IMUs and this is what I found…

y4mj5BVS4Z4NYNz_42Bov2wB_N9n5HAZdXKhoii3a__lOaQaQ61GAvuvaFRstbJx1MGLxZJwCEZnqCvR8rQtIdomDEOvkIWOkN7jx8dY42WvbFOThgV3DiL_WWMtz7uxMLWC7lRcXq_n8HU3L60gyLu0vQZAXK0OFHpmePjIXevbgL7iYgB8Qms7JEgMSs3utZTV-h57b2sNkLfGlczOji5bw?width=1155&height=419&cropmode=none1155×419

Then I went to look at INS_Log_Bat_Mask and found that only the first 2 IMUs were being logged.

So I changed it here…

y4msXUrFylFbhJk6Mc6Qcw8WhDdWnijfBGR183khRX3YyOPgwldRmJQnUFhetONyI78Sy1eZixjj8cVGTqIcLRRn6jOvvltVyG0exSUYwhq0u5Ta1yjCK56UzZvlt2D13aEbx-ofPvLqxaXW15K7k0ObkDsPi7V5pNlICV7lYFKpVooep-wSplvwZFQikPPcrGNoGzS681S5Fa8rhPKXUYMpQ?width=762&height=255&cropmode=none762×255

So now the next test run will log all three IMUs.

So it’s not that any were disabled as much as the last one wasn’t being logged

Thanks for all the continued help

213

Ok, logging done on all three IMUs.

I put in the static notch filter that you recommended.
These are the results from tonight

y4mYIgDkic2WrlMmEWm7ZGcU8Ajjq2lvtTM0gwP86f3jclCVJAefcapwHGjX9k5yY6Q72K6rxG0u2SKo6pBs5CCH0HZrmIO-Ym-ZNxNirH83UywjkUCm_zsYgR9P-L6J_jU2JHEQcbD0TYVinFKGi-h5GBPYitIdYSxHoevP-3TQrcdlvJ83q0OD-dgaVdBJo6dxdzutEmjsHG26kZArdCqbQ?width=2702&height=1694&cropmode=none2702×1694

Thoughts?

214

They look about the same once you account for the change in scale, if anything the BMI55 is a bit noiser, so maybe stick with what you have for now - but I think take a further look once you have some logs for higher throttle. The static notch looks like its hitting its mark bang on - you could probably afford to make it a little wider and deeper.

215

@andyp1per Thanks for taking a look

216

Went out and flew for a little while today… Here are the lastest charts… Along with the log files in case you wanted to see anything else

y4macXshEDwPt7SmkFAQsmpu8I5vnGv9oZxcOLKkkh_FtS1lpGjVKtJhQqGsJ07j5Dxs5OGoz_BAcocWMSIs3SfZuqfzOQxf-geYh_sbk7bqLFn3GwPuAR1ALy4Lwmnmf6lCTkMrJROZkQcp4iQvLmdoL0atMepjh2bJe_acl_sXo4G073ksG3m5K3mWbShpi6rKcQjrqql4lx__vgIkf34yA?width=2702&height=1700&cropmode=none2702×1700

https://1drv.ms/u/s!AplYZe-uqPEBguwYbDhHbnsLb1WBWw?e=MLsuYo

217

@andyp1per

I just set up my Big 1300mm Octo

I have just enabled the Dynamic filtering I have the following:

Prefiltering:

y4mm-tuTaDREPq_58t7XqAXjKfIDqJXbqlfCbLXDenRWUnGnaZWVMK-0xrVwY3QmssWd66ug7redYULDiSe4vE79e3LLvZYx3DnGBXt1BeH0KSEwroQ5Ikm99LBYqW-o3MXrEy3dAYtO-sEEzLlbdMsFbFAUVI5zBFUNxC4XVzYsE6NanlY1HTiQazU7J26j0fJlj3dYBwW8DtpGnApjBR46A?width=2700&height=1685&cropmode=none2700×1685

And now Post

y4mw-2hwbhxyqaKF7n7pR9_iSREJmOhdN0sAk3syRvzo_dhXGNuDGdc19oihFKO05r-CDcqu77eWLc3RbiySI_MXBgxuj6aJyK2cyo3VQunOhklaCu1whgwnZqtZkVtM12gtZmJnnym4WRaQOE54hyybvmspz2r4KN84fCuZhkq0c2y89N1a-G6HsWoDl9EYCtLCTsAWZBfidkRaOCV1jq0mQ?width=2701&height=1625&cropmode=none2701×1625

Thoughts?

Static Notch at 100?

218

Can you post your parameters? Looks like you could increase attenuation on the dynamic notch.
I would test a static notch at both about 50 and about 110. The 50hz noise might be causing a harmonic at 100hz. Try both and see what looks best.
A lot of my frames seem to have a resonance around 100hz.

edit - oh you just repled, I’ll reply again…

219

These are my settings:

|INS_NOTCH_ENABLE|0|
|INS_LOG_BAT_CNT|1024|
|INS_LOG_BAT_MASK|7|
|INS_LOG_BAT_OPT|2|
|INS_LOG_BAT_LGIN|20|
|INS_LOG_BAT_LGCT|32|
|INS_ENABLE_MASK|127|
|INS_HNTCH_ENABLE|1|
|INS_HNTCH_FREQ|80|
|INS_HNTCH_BW|20|
|INS_HNTCH_ATT|15|
|INS_HNTCH_HMNCS|3|
|INS_HNTCH_REF|1|
|INS_HNTCH_MODE|3|

Are you thinking something like the following:

INS_NOTCH_FREQ = 50
INS_NOTCH_BW = 40
INS_NOTCH_ATT = 40

220

What does your CTUN, N look like in the logs? I’m curious what frequency you hover at on that size copter. You might want to set the dynamic notch lower