First plunge to AQ with 80cm hex

[Trnquill]

Since last update my AutoQuad hex has gone through major changes. I switched to lower kv motors and significantly larger props (12" -> 17"). Flight time almost doubled and so the current nearly halved. This got me thinking. Maybe I could move the ESCs from the prop wash under the frame. Maybe even loose the huge heat sinks! I did fast tests by removing the heat sinks and placing the ESC32 inside an enclosure (to isolate it from cooling stream of air from the prop). I ran few tests with high throttle and found out the ESC remained relatively cool. So, I went onward with the plans and designed an integrated ESC32 holder + power distribution board to be mounted under the frame. I already milled it from double sided PCB laminate and soldered first ESC32. Five more to go.



What do you think?

[Trnquill]

ESC32's soldered together with standard caps. Next funny task is the wiring.. I'll have two separate wires for the batteries, three small JST-plug auxiliriaty power leads (one for AQ board etc.) and then of course the motor wires from each ESC. I'm still planning how to do wire routing and some kind of easy mounting clip for ESC signal wires.

[Trnquill]

I designed and 3D printed brackets to hold the ESC32's in place (although the thick copper leads hold them in place quite well already). The bracket also captures caps and offers a holding clip for ESC signal wire.



After this I soldered on all the wires: leads for two batteries, motor wires from each ESC and some JST pigtails for powering miscellaneous stuff. Here's a pic before nice evening doing some braiding..



After tidying up the wiring loom I was finally able to mount the distribution board under the frame.



Not in place in the picture above, there is camera gimbal mounting plate on top (or under, actually) power distribution plate, so it's quite well hidden away under the frame. I did 7min test flight today at around 0°C and all the components felt cool to touch.

My hex is now considerably larger than it was when I started with AutoQuad. Distance from motor to motor went up from 80cm to 95cm when switching to larger props. During test flight today I could visually see and hear arms doing up and down vibration from time to time. I guess I'll need to tune down the tilt rate D and see if things calm down. I'm yet to analyze the log to see what are my vibration and MAG noise levels with this new larger setup and new ESC mounting solution.

[Trnquill]

I took a quick look in the logs and all vibration/noise levels have gone down. Some more than others, but all graphs look cleaner than before. Especially the MAG Y noise has dropped significantly, from around 0,9 (amplitude) down to about 0,08!

RATE noise didnt' drop that much but is now down at 0,3 (amplitude) for X axis, Y is a tad lower 0,2 and RATEZ noise floor has amplitude as low as 0,1.

For ACCs the highest noise floor amplitude is ACCX at ~1,5. ACCZ noise is lower 1,1 and ACCY noise yet at tad lower 0,9.

I guess the noise and vibration is well under control?

LOG_MOT_PITCH varies between -80 and 100, LOG_MOT_ROLL has just slightly larger amplitude (around -100...130). LOG_MOT_YAW bounces between 0 and 50 in a steady hover in very light wind.

[LPR]

You are doing nice work. Good to hear the mags are working well with your design.

To reduce the spark when you plug in the battery I tried using a 1 ohm resister in the circuit of a balance plug that's connected between the AQ power wires and the batteries balance led. I saw no spark when connecting either the balance or the battery power plug. The balance plug connection should also work as a low power input what will keep you safe while flashing new firmware.

[Trnquill]

Thanks! There's actually minimal spark now that total capacitance is way lower. I went from three large caps per ESC to just one small (per ESC).

[Trnquill]

I'm trying to tune the hex now with the new low kv motors and larger props. The booms are few centimeters longer than previously but I made the mounting to frame a tad stiffer, too. What the hex is suffering is really high frequency oscillation. I can see the tips of the booms oscillating really fast up and down like less than 1cm, almost like the machine got shivers. The oscillation/vibration is not constant but comes and goes. Sometimes the whole machine vibrates all the way down to camera gimbal and legs, sometimes just two booms oscillate and at times it's completely still.

I tried tuning tilt rate D as per the wiki troubleshooting, but it had no effect. I went from 3000 to 7500 without seeing any difference in the behaviour. All I did was gentle hover, so I can't say if the changed had any effect, but at least the shivers remained unchanged.

What to do?

EDIT: I thought the vibration isolated camera gimbal hanging under the frame might be a reason for the vibes (it can start oscillating) so I replaced it with static dummy load (three 6S batteries) strapped tightly on to frame. The oscillation/vibration was greatly reduced, but still there. I could see the ends of the booms shake every now and then. I can also hear motors making "tweeting" sound as ESC32's vary the speed very rapidly. So, the problem persists... any help is appreciated!

[chschmid]

Maybe you are a victime of harmonics.
Have a look here:

Cheers
Christof

[Trnquill]

Thanks! I read that article earlier today. I'll have to somehow measure the RPM at hover and determine E value of my booms.

I actually tried to do some rough calculations today for natural frequency of the booms. For 0,025m CF boom with inner diameter of 0,023m, L=0,35m (distance from frame boom block to motor), M=0,16kg (and using E of 20e9Mpa for estimation) I got 38Hz. My motors are 380kv and I run them with 4S battery, which gives me max RPM of roughly 6000. That's 100Hz. Since the hex hovers at about 50% throttle, the frequency should be under 50Hz, which could be really close to natural frequency of the booms. I need to be certain about the actual hover RPM and also measure specific E value for my booms to see if the rough math is even close to reality.

It looks like even if I would convert the hex to Y6 I would still get almost the same natural frequency. Booms would be a tad shorter but mass at the end of the booms double. Using the values above but with L=0,3m and M=0,32kg I would get natural frequency of 35Hz. Which still sits right around the hover RPM (I presume).

[hogster]

Awesome work, oh to have a 3D milling machine!

David