One prop, two chunks of GPS mast so far!

[crayfellow]

Inline with @SchrodingersCar 's 'Humbled' thread, I'm struggling with the basics of getting a 1000mm-ish craft in the air.

I'm now pretty comfortable flying a 250-class quad (Blackout mini H with a nice FPV setup), and a FliteTest Electrohub wooden tricopter. I started with the 250 and cc3d so I wouldn't get distracted with fancy GPS, alt-hold, or any of that. Also built a Pixhawk rover to get used to its capabilities. Moved on last Sunday to building the tri (APC 9x4.5 MR, X2212 KV980, 30A Lumenier ESC, Pixhawk, telemetry, M8N GPS, FrSky X8R, simple 250mW vtx and cam) to try out Pixhawk in the air. My 7 year old son was able to fly it (with supervision in a big field) thanks to Loiter mode!

All along my plan has been to build a larger hex as a research platform for various experiments into applications of sUAS. Here's the setup:

• QuadFrame foldable pro SIXcopter frame with aerialpixels M325 motor mounts
• KDE 3520 400kv motors
• KDE 55A ESC's
• APC 15x5.5 MR props
• FrSky X8R
• Flytron strobon NEO LED strips x 4
• Pixhawk with telemetry and attopilot 180A voltage/current sensing
• 5V 2.5A pololu regulator to power Pixhawk (in place of the 3DR power module since I am using attopilot and 6S)
• 5V 2.5A pololu regulator to power servo rail (KDE opto isolators, rx, and LED's)

I double and triple checked the motor order (by viewing this page regarding ESC's and motors for Pixhawk) as well as motor direction (first with a servo tester, then a final test when armed but no props), then checked that the right props were on the right direction motors.

Upon the first step, the poor thing promptly flipped to the left and broke its GPS mast. So I pulled it out, stuck in what was left, tightened the set screw, and went back to reading.

I am well aware there are specific things one must do to get KDE ESC's happy with Pixhawk. I had gone through the "all at once" Pixhawk ESC calibration. I'm not sure whether setting RCx_MIN to 1100 and RCx_MAX to 1900us should happen before or after ESC calibration. It seems like after I adjusted the MIN and MAX the props spun upon arm, but then seemed to spin up much faster upon throttle. So then I did another ESC calibration, and they no longer spun when armed, but I double checked and my 1100 and 1900 numbers were still there.

Anyway, I tried 3-4 more times after reading I might try another accelerometer calibration, tried fiddling with the roll/pitch with little throttle bursts to make sure it was even orientated properly (it seemed to be), but all of this culminated in 3-4 more flips. The final flip broke a prop so I'm clearly missing a few things.

I am using 450mm legs, and especially with no payload, am I just asking too much for it to happily pop in the air with little to no pitch/roll intervention like the mini or tri? This guy's payload is an InfinityMR gimbal, which is all set to go, but I'm obviously going to make sure the rig is happy before bolting that on for flight. I know I should have some weight in the place of the gimbal; is that in fact mandatory?

Would I be better off switching to shorter legs (I have some 315mm), and adding some weight, or would it still be so tricky to get this guy in the air?

Any other thoughts?

View attachment 25519

 

[fltundra]

No weight on bottom with those lipos on top. Would bet CG is way off causing flips.

 

[crayfellow]

No weight on bottom with those lipos on top. Would bet CG is way off causing flips.

right - makes sense. I forgot to mention that is 2 x 6S 10Ah Multistars.

Just to get the thing in the air for testing, should I put one of the batteries on the bottom, or just add an equivalent payload on the bottom to balance it out?

 

[fltundra]

Just mount up one multistars to the bottom. Then take two chairs back to back and lay each motor arm combo. Balance just inside the motor mounts on all three and see where the CG lies, and balance from there. Then fly it, and add the other lipo and re-balance. You will need ballast to replace exact gimbal/camera weight on the bottom, if your going to remount them on the top. still checking CG, needs to be a tad bottom heavy.

 

[Mactadpole]

I think something else may be going on that is causing the flips. I first flew my nearly identical bird with 2x 6s 8Ah Multistars on top with no payload and it popped in the air just fine on default pid's but was certainly a bit unstable. Nowhere near flipping. Make sure you've done radio calibration and all the other required calibrations. When you arm the copter the props should be spinning fairly slowly, if that's not happening then we need to figure out why not.

Have you used the motor test feature to check that the correct motor is spinning and in the right direction?

I don't know the story behind changing RCx_MIN and MAX. Never had to mess with it. I'll read up on it a little later.

 

[fltundra]

He has 2.37 kg of lipos on top with nothing to counterbalance.

 

[Old Man]

I'd guess a lateral balance would be more of an issue than vertical. Are you trying to ease up with the throttle for lift off or "pop" it off the ground? You've used the term "popping up" a couple times but I'm not clear how that was referencing. Try literally popping the throttle quickly to get the copter off the ground. I was having a similar problem with a Vector until Av8Chuck provided some advise.

The stabilizing features of some controllers do not become active until throttle position rises above a certain point. Below that a system can be "wobbly" as the throttle slowly comes up. Long gear legs and a lot of low RPM motor thrust amplify any thrust imbalance and cause the copter to tip, generally fore or aft.

Assure a good lateral balance and give it a go over deep grass. From experience, the grass really saves on props

 

[crayfellow]

I'd guess a lateral balance would be more of an issue than vertical. Are you trying to ease up with the throttle for lift off or "pop" it off the ground? You've used the term "popping up" a couple times but I'm not clear how that was referencing. Try literally popping the throttle quickly to get the copter off the ground.

After the first flip I was pretty intimidated, so I wasn't too aggressive on the throttle in case it would do something up in the air and do more damage when it came down.

The stabilizing features of some controllers do not become active until throttle position rises above a certain point. Below that a system can be "wobbly" as the throttle slowly comes up. Long gear legs and a lot of low RPM motor thrust amplify any thrust imbalance and cause the copter to tip, generally fore or aft.

Assure a good lateral balance and give it a go over deep grass. From experience, the grass really saves on props

And can you believe I didn't order any extra??

 

[Old Man]

Most don't so don't berate yourself. I learned that lesson the hard way too.

 

[crayfellow]

Sounds like balance and the FC/ESC calibration are likely both at issue. I am certain about prop CW/CCW and motor direction, and tested nose out by (for example) holding roll to the left a bit, giving a bit of throttle, then same with roll to the right, then same with pitch to ensure bias would be in the specified direction. But even the little test tipped it over, I'm definitely making things more difficult on myself with it so top heavy.

Here's the situation with balance. According to the Pixhawk manual:

The KDE ESCs have fixed PWM ranges so you must manually set the output range of each PWM signal so that RCx_MIN is 1100 and RCx_MAX is 1900us using the Advanced Parameter or Full Parameter Settings Page in the planner.

The first time I flipped I had done the Pixhawk ESC calibration (throttle up, connect batt, confirm LED pattern, disconnect and reconnect batt, hear tone, press Pixhawk button, throttle down, hear ESC tones). However at that time I had NOT set RCx_MIN and RCx_MAX as honestly I was curious what would happen (this is research, after all). Upon arming, the props did not spin, but seemed to smoothly spin up upon adding throttle.

Then, I went through RC1_MIN/MAX to RC6_MIN/MAX under advanced parameters in Mission Planner and set them to 1100-1900. But I did NOT alter trim or any other setting, and thought that might be an issue. The next time I armed, the props DID spin, but seemed to be too sensitive to adding throttle, suggesting to my intuition an altered curve on the PWM signal. Flipped again

Then I thought, why not do the ESC calibration again, so I did that.

Oddly, after that even though Mission Planner still showed the 1100-1900 I had configured, the motors did not spin up upon arming, suggesting the calibration reset the values I had defined (which makes sense, right?). But this time, although they were not spinning upon arming, they still seemed too sensitive to the throttle stick.

So maybe with the ESC's the manual configuration supersedes the "automatic" ESC calibration? Any of you guys using KDE ESC's with Pixhawk?

Another strange thing, two of the times when it tipped, at throttle 0 on the stick, the props continued spinning and I even heard the "disarm" Pixhawk tones. So it seems like it automatically disarms upon tipping (did not know it did that), but didn't kill the motors. That was extra intimidating to me, since I'd risk overloading a motor/ESC if I didn't get to it quickly enough and tip it upright, or would have to reach over spinning props to disconnect the battery. I'm going to need to remedy that before the next test, maybe by setting a temporary failsafe to kill the motors or something so I can just turn off the tx.

Seems like when I get some more props I should put the shorter legs on there no matter what so I can get some confidence hovering this guy.

Thanks so much for all your help thus far guys, I'll keep you posted.

 

[crayfellow]

Most don't so don't berate yourself. I learned that lesson the hard way too.

I know what you mean - thanks. These heavy craft are definitely a whole other ballgame. I am reminded of building a drag racing car and picturing all the bolts and things I had carefully torqued coming undone when I would test-drive it. I think it adds to the pressure when you build it yourself, you "know too much".

While I was doing this test, I could see it WOULD fly if I was on the pitch/roll keeping it tweaked, but I wasn't doing myself any favors with all that top weight. @Mactadpole Shawn maybe the 10Ah's made just enough of a difference beyond the weight of the 8Ah, plus I have the ridiculously long landing gear.

My tendency is to add weight to the bottom instead of moving a batt, so I have an experience more like I will have once the camera is there. Wish the local hobby shop stocked big APC props, I may have to call a few.

for a bit of fun, here's my 1.5yr-old very interested in 'woodie' the tricopter's view:
View attachment 25529

 

[MadMonkey]

Is it possible that you have an IMU turned in the wrong direction physically or in the software settings?

I had an IMU backwards once and my first takeoff was an immediate flip to the rear.

 

[JoeBob]

Popping up into the air is intimidating. You don't want to hurt your baby.... I sometimes use a hand launch to clear obstructions on the ground (tall grass that grabs a leg) or to get a feel for motors when I'm inside. Be sure your face is well away from any prop and you have a good grip on the bottom of your MR, and have someone turn up the throttle slowly. You can feel what the bird is trying to do, and decide if you want to let it go, or turn it off and re-do.

For the sake of all the lawyers reading this: Don't do anything that I just wrote - it's too dangerous.

All of my Pixhawks are on quads, so I can't verify my suspicion on a hex; but that connection diagram makes no sense.
1, 3, and 6 are Clockwise
2, 4, and 5 are Counterclockwise?
It seems like 1,3,5 should rotate one way and 2,4,6 should spin the other.

You can verify spin direction using 'Motor Test' (Engage the safety switch to get autopilot approval; bump the throttle spin rate up, mine has to be at 30%). On a quad, the 'Motor Test' sequence goes around the quad as opposed to following the numbered sequence. Pay attention.

Anyone on the forum have a Hex on a Pixhawk to verify the wiring diagram sequence?

 

[crayfellow]

Is it possible that you have an IMU turned in the wrong direction physically or in the software settings?

I had an IMU backwards once and my first takeoff was an immediate flip to the rear.

I had done a VERY careful accel calibration, then another, and based on my "little bit of roll/pitch with a bit of throttle" tests, the orientation was OK. Good thought, I will keep track of orientation and make sure I am recalibrating with any changes to avoid an easily avoidable goof-up.

 

[crayfellow]

@JoeBob I used the diagram from this APM page. Then I have the frame in Mission Planner set up as a "plus" as opposed to an "x".

I used a servo tester on the individual ESC's to check direction, then all at once when armed but without props. I didn't know there was a Pixhawk "motor test" that would go though the sequence; I will try that too. I see now it's on the same page I linked to above.

 

[Mactadpole]

Anyone on the forum have a Hex on a Pixhawk to verify the wiring diagram sequence?

I can verify based on my build that is very similar.

Have you done the motor number test procedure from the wiki? I have made the mistake of having the wrong motor plugged into the wrong slot on the Pixhawk and it did cause it to tip.

When you arm the copter the motors need to spin, you can adjust their speed using the MOT_SPIN_ARMED parameter.

I still don't think its the weight on top and the long landing gear. My landing gear give it 350mm of clearance, so pretty tall too. Like I said, I had the lipo's on top to start with too. Can't rule it out though so strap on some dummy weight underneath, it won't hurt anything for sure. I would eventually get those lipos more in plane with the rest of vertical COG by mounting them like I did or some other way that moves them down.

In stabilize mode I can take off very gently with no issues but it certainly helps to get a few meters in the air. Also make sure the copter is not shaken or disturbed when plugging in the batteries because this can cause calibration issues. Ask me how I know. It just got me because I was careless. Crashes are going to happen unfortunately. I decided to launch the copter from the cover on the bed of my truck because I was in a field of tall grass. I climbed up on the tailgate, plugged it in, and then jumped off the tailgate. I ignored a BAD AHRS warning by just turning off the Kalman filter process and sent it on its mapping mission. First leg went perfect but a little more than halfway through the second leg it turned north and lost altitude into a pine tree at about 24 meters. Pretty sure it was able to compensate for the gyro drift at the beginning and then and could no longer cope with differences on the way back. I'm rebuilding now at a cost of about $400. Luckily I'm a tree climber and was able to go home and get gear to get it out of the tree.

 

[Old Man]

Pixhawk is supposed to disarm after tipping but the new Pixhawk-2 seems to be failing to do that with the Solo it's currently fitted to. I've never had a fail to disarm upon tip issue with Pixhawk-1, but I've only rolled one a few times so the experience is a bit limited.

What are the throttle end points set to in the transmitter? I found with a different FC that Tx end points can be as important as the ESC calibration. Without useful low throttle end points no amount of ESC calibration would permit the FC to arm the ESC's. Your 1100-1900 values may have that covered in the Pixhawk but you might want to check the Tx as having similar settings. I'm not familiar with the FRSky system.

I'd need to go back to the Arducopter site and read up more on throttle configuration before going any further on the subject.

 

[crayfellow]

@Mactadpole that picture is hard to look at! Good to know it wasn't even more catastrophic, and I'll be extra careful about power-on calibration.

 

[crayfellow]

Pixhawk is supposed to disarm after tipping but the new Pixhawk-2 seems to be failing to do that with the Solo it's currently fitted to. I've never had a fail to disarm upon tip issue with Pixhawk-1, but I've only rolled one a few times so the experience is a bit limited.

yup - I recalled videos of Solo laying there trying to fly on its side when this was happening. I am certain at least once, and maybe twice, that I heard the disarm tones yet the motors continued spinning. That reminds me, I can't believe more people aren't talking about the solo using a fork of APM specific to Solo. As an engineer with a long history in open source this just feels so totally dirty and messy. Yes, it is a commercial product with a somewhat unique version of Pixhawk. But that seems like more of a "branch" to me, and not for Solo, but for what they are calling "Pixhawk 2". Just bad practice all around, which will lead to weird confusion about feature sets and maintainability problems going forward, especially once they have a "Pixhawk 2" on the market. We really need a separate thread for that discussion.

What are the throttle end points set to in the transmitter? I found with a different FC that Tx end points can be as important as the ESC calibration. Without useful low throttle end points no amount of ESC calibration would permit the FC to arm the ESC's. Your 1100-1900 values may have that covered in the Pixhawk but you might want to check the Tx as having similar settings. I'm not familiar with the FRSky system.

I'd need to go back to the Arducopter site and read up more on throttle configuration before going any further on the subject.

Yeah, I haven't checked tx endpoints, just did the ESC calibrations before/after tweaking the advanced params for the PWM min/max, but it makes sense tx range would have an impact as well.

 

[crayfellow]

I did find one mistake I made due to a misunderstanding of what the RC1...RCn parameters mean in APM. I understand this corresponds to input channels, so I shouldn't have set 1100-1900 as endpoints for RC1-RC6, only RC3_MAX and RC3_MIN, for the throttle channel.

However, I still wonder about RC3_TRIM. Following radio calibration it will say 982 for min and 2006 for max with 982 for trim. When I change min to 1100 and max to 1900, but leave trim unchanged, RC3_TRIM should correspond to the min value, 1100. Maybe the APM doc writer assumes this is obvious, but we might as well be pedantic and say 1100 for min, 1900 for max, and 1100 (same as min) for trim on whatever channel corresponds to throttle (which in my case is 3).

So @Old Man it seems like tx min/max and/or trim shouldn't matter with these set in advanced params, I'd just need to make sure they always get reset manually if I ever do an RC calibration again. Am I correct in understanding Pixhawk will just take whatever range the tx will give, and map it to the overridden range I set? It would seem strange if I also had to set the 1100-1900 range on the tx, unless of course I wanted those values set "automatically" when doing an RC calibration.

Also @Mactadpole I went through the motor test and all is well.