The thing is that I've never changed max cell voltage, and I've never seen this issue before. Have flow many HV packs on the Volador without seeing the issue.

Max cell voltage was set to 4.3, and it's showing 7 cells. I increased it to 4.4, unplugged an plugged in again, and now it's showing 6 cells. Cool.

Now let me look at my AOS. Max cell voltage set to 4.3, and it shows 7 cells. The thing is that in the field it doesn't show the abnormally low average cell voltage like the Volador did. Never has.

...

Somehow I never posted this earlier today, but I flew today after writing the above, and the Volador did not give me low battery warnings. I had already removed the average battery reading in the OSD, and just watched for 21 V pack voltage, and sure enough, that's when I started getting low battery warnings, as expected.

Once again, the AOS did not have the issue of getting confused about number of cells with an HV pack, as usual. On a different note, the AOS motors were running very hot today for some reason, so I stopped flying it until I can investigate. On the good side, I installed a diversity rx with the antennas oriented at 90 degrees to each other, and the reception was much improved.

When we’re talking about tenths of a volt per cell (although really, we’re talking about hundredths- 0.05v if we’re looking at the default BF max cell voltage of 4.3v vs a fully charged HV @ 4.35v/cell), the voltage of a lipo can vary wildly depending on how long ago it was charged, its health, your charger calibration- on top of that we’ve got your voltage sensor calibration in BF, temperature, the amp draw of the load, etc.

But, as I’ve said many times before, I’m often wrong. I could be completely off base, and rather than a simple setting, it could be that the voltage sensor on your esc is faulty.

I’m glad that you haven’t run into this problem before- but for your future builds, I would suggest noting that it’s a good idea to increase your max cell voltage in BF to 4.4v if you’re planning on running HV lipos.

Either that, or you may have to occasionally plug a battery in, let it sit and drain for a bit, maybe arm and hover for a second to help drain off that excess voltage, then unplug it and plug it back in to get your average cell voltage to read correctly. Or, just charge your HV lipos to 4.2v per cell. Or maybe 4.25. Not above or approaching 4.3v.

Personally, if I was running batteries on two different quads with the correct max cell voltage set, and one never gave me a problem with cell count, and one occasionally did- I’d think that the one that never gave me a problem probably had a poorly calibrated voltage sensor, maybe set a little too low, or, the one that occasionally read the wrong cell count was set a little too high.

https://betaflight.com/docs/wiki/configurator/power-tab

Looking at my blackbox readings in PIDtoolbox from my test flight yesterday moving master multiplier from 0.80 to 1.80 in steps of 0.20, I determined that 1.40 was the sweet spot. Anything more than that resulted in oscillations, and anything less than that overshot the setpoint on the pitch axis.

Actually a MM setting of 1.00 was fine for the roll axis, although 1.40 had slightly less latency. It was the pitch axis that looked better at 1.40.

So, I set the master multiplier to 1.40, and flew a couple of packs today--took the day off from tuning. It flew great, and importantly, the motors were quite cool after each flight.

(I wrote this yesterday, but once again neglected to click Post Reply before getting distracted. In this case, it was to go pick up my wife from the airport, so good excuse. At any rate, "yesterday" = Sunday and "today" = Monday).

(08-Aug-2023, 07:25 AM)Lemonyleprosy Wrote: When we’re talking about tenths of a volt per cell (although really, we’re talking about hundredths- 0.05v if we’re looking at the default BF max cell voltage of 4.3v vs a fully charged HV @ 4.35v/cell), the voltage of a lipo can vary wildly depending on how long ago it was charged, its health, your charger calibration- on top of that we’ve got your voltage sensor calibration in BF, temperature, the amp draw of the load, etc.

<snip>

Thanks.

Actually I recently calibrated the voltage in the power tab, and for both, a Scale of 111 is required, vs. the default of 110. I wonder if this calibration on the Vola caused whatever underlying factor is causing the bad cell count to surface?

Also, I've noticed that my charger always shows a higher voltage when I first plug in than that that was displayed on the OSD at the end of the flight. Right now, for example, I just read 22.1 V on two HV batteries that I had run down to 21.0 V or less according to the OSD readout. My charger showed 22.16 and 22.17, so pretty close to the multimeter reading.

Re-reading your post, I see that I'm neglecting the load while flying compared to sitting on the bench.

I'm going to experiment some more, and compare voltage in BF to the OSD voltage displayed while not armed.

My highest priority is to figure out why the AOS motors were running so hot yesterday, and also to replace the VTX antenna on the AOS, which has gotten pretty beat up. Still need working on a mounting solution--I might take you up on your offer to print me a better VTX antenna mount. I'll post some pictures in my AOS 3.5 build thread.

Regarding osd voltage vs charger voltage- After resting, lipo voltage bounces back a bit. That’s why I usually try to land at around 3.5v per cell- after it sits for a minute, it comes back up to around 3.7v a cell, which is a good enough storage voltage for me.

If your aos motors started running hot and you didn’t change anything prior- if your tune is a bit aggressive, I’d look to see if a prop etc. is damaged, maybe a screw loose, something that would take it from a “clean” build to a less than clean build. That said, I’m new to the whole tuning thing, so I might be off base there.

As mentioned in my 3.5 inch build thread, I loaded the Betaflight 4.4 tune and filter presets on my AOS35, after which ]the motors ran very cool yesterday. I then notched up the master multiplier in increments of 0.20, and when I hit 1.6 or 1.8, the motors started making a very noticeable stuttering sound just after takeoff, so I landed and notched it back down to the next lower setting, and the motors sounded good, and were cool after the test flight.

I did replace one prop that had all of it's tips amputated or shredded.



But I don't think that was bad enough to make the motors run so hot. It was in the 90s that day, but the Volador30 motors were not hot after its flights that day.

I took a blackbox reading at each MM setting, doing a number of snap rolls and flips for each test flight. I'll load the blackbox files a little later and see what I can learn from them. Right now it's getting close to sunrise, and I like to walk at this time of day. I've been strapping my fly backpack on, and walking over to the closest high school field to fly as part of my morning walk, but I want to do some more prep before I fly again, so I won't do that this morning.

Oh, and the AOS35 did that same thing where it thought the battery voltage was low--for the first time. So, using the OSD PID editing interface, I changed the max battery voltage to 4.4, and after that, the problem went away. That was after I completed the first flight, where I just ignored the low battery warning, and flew till the pack voltage hit 21 V.

I've been getting significant vibration on my Volador 5" with throttle between 30 and 40 Hz. I see it in the goggles when flying. This can be seen in the blackbox log three different ways. First, looking at the main display:



The markers are set between about 30 and 40% throttle, and we see the noise mainly in the roll axis (yaw not shown, but no issues with yaw). I set the in and out points (using I and O keys) at about 30% and 40% throttle for the frequency analysis.

Looking at frequency in the analyzer display:



We see the noise over quite a broad range, but with much higher magnitude at lower frequencies, with a peak at about 35 Hz.

Looking at freq vs throttle:



This shows us basically the same thing.

One thing I tried is using different props. This seemed to help the noise in the pitch axis, but not the roll axis. Here's the main screen from a previous log before the change in props:



After noticing this initially, I checked my screws, and many of them weren't very tight, so I tightened them all up. I think this might also have helped with the pitch axis noise.

I've checked the motors looking at the accelerometer in the betaflight motors tab, and they're squeaky clean.

I've tried various filter and PID tuning changes. One thing that seemed to help some was setting a notch filter at about 30 Hz with a cutoff of 20 Hz (or something like that). I had to dial the master multiplier down to 0.8 for this to kind of work. But we aren't supposed to set notch filters below about 100 Hz, so I wasn't happy with this, and it only filtered out the most intense frequency range of noise, not all of it.

The popup help for  Gyro Notch Filter 1 says, "To be useful, the resonant frequency must be constant across the throttle range", which it is not. It also says, "Avoid setting notch filters below 100 Hz except in special circumstances", and "Static notch filters are rarely required in 4.3".

I've watched hours of PID tuning videos looking for a solution. One short by CR discusses "mid-throttle noise", which he characterizes as "the bane of FPV pilots", and which he attributes to flexing of the frame in the roll or pitch axis, with the solution being to add stringers between the front and rear arms. That's beyond my pay grade for now.

For now I'm just living with it, but it's very frustrating to have the quad vibrate quite a bit as I move through the problem throttle range. I'm starting to think about building my next 5" quad so hopefully I'll have something to fly without this problem. I'm considering the Apex frame, since this seems to be top of the line.

I'm guessing that this is a mechanical issue, but I haven't been able to find anything loose or that otherwise could obviously be causing this.

Any thoughts?

I solved it.

In a UAV Tech filter tuning video, Mark Spatz set the ESC PWM frequency to 48 kHz to eliminate some low frequency noise. My AM60A ESC was set to variable PWM (24-48 kHz), so I changed it to 48kHz, and it eliminated the troublesome low frequency noise.

Here's the gyro (filtered) frequency plot for a throttle ramp:



There's still some low frequency noise, but the magnitude is much smaller, and it was not noticeable in flying.

The improvement is even more obvious in the frequency vs. throttle plot:



Note how the intense horizontal band of noise between 30 and 40% throttle is gone.

Interestingly, there appears to be more pre-filtered (gyro scaled) noise, which we can see in the main graph:



But there's nothing special happening in the 30-40% throttle range on the roll axis, as before. Here are the freq and freq vs. throttle plots for Gyro scaled roll:





But as we see in the filtered noise plots, the noise is adequately handled by filtering.

Here are freq plots for gyro scaled pitch (pre filtered) and gyro pitch (filtered):





Crushed!

And same for freq vs. throttle:





So apparently the lower PWM frequencies at 30-40% throttle caused some sort of resonant vibration. Setting it to a fixed 48 kHz eliminated this issue.