[batata003]

Hi,

When I pitch or roll my quad remains altitude but when I yaw it loses too much altitude. Anyway to solve this? I already checked the option 1/cos but yaw still makes my quad lose altitude.

[proctoidpaul]

I'm not an expert on all the options and their impact in LP, but I know some altitude drop is fairly normal for most quads.  Perhaps it can be improved with well calibrated GPS mode assistance (or some other settings), but like so many others I'm struggling to get GPS modes working well.

Perhaps a more thorough description of your FC used and GSC flight mode settings would help others pinpoint more specific recommendations.

Paul

[batata003]

Hi Paul, thanks again for helping. As I understand, when a quad YAWs, 2 motors will spin faster and 2 will spin slower than before in such a way that the total amount of thrut should be kept the same so the altitude, right?

[DocHardinger]

Normally quad goes up when you stop yawing because standard escs dont use damped light or active braking...so in your case do you use active braking or damped light on your escs? Its just a thought...

[batata003]

No, my ESCs have the default configuration of factory. They are Emax SimonK ESCs and I think the break is not active cause when I shut down the motors the props keep spinning for a while.

[DocHardinger]

Or maybe if you yaw very fast your quad is too weak to hold height with two motors...or you just calibrated escs not properly...maybe post your config file here File>export uav settings

[TheOtherCliff]

Are you flying LibrePilot? 

You mention 1/cos which is the math in CruiseControl.  CruiseControl is not affected by yaw, only by bank angle change from roll/pitch..

The only time I loose altitude during a yaw it is because I bump throttle at the same time since they are on the same stick.

A real test to confirm this would be to have Altitude Hold throttle mode on your switch and switch to it, then try some yaws.    Of course AH is not available on CC3D.

[batata003]

Yeap, you are right, CruiseControl should help me avoid this problem. But if I use CruiseControl another problem happens: if you read https://librepilot.atlassian.net/wiki/display/LPDOC/Cruise+Control you will see cruise control also turns off motors when quad is inverted.

In another thread I reported this as a bug: if I fly very high and turn off my motors for 5 seconds (for example) and if my quad gets inverted after that 5 seconds, if I try to apply thrust my quad will keep falling and falling cause the motors will not be allowed to spin while inverted. Last week I crashed my quad and almost destroyed it because of this buggy behaviour.

If would be nice to only have an option to activate 1/cos but desactivate this buggy problem of cruisecontrol which may cause serious injuries.

[TheOtherCliff]

CruiseControl does not turn off the motors.  It reduces them to 5% (unless you have throttle stick at zero).  If you have your motor neutrals set too low, this can look like the motors are off, but they are at 5%.  Librepilot will increase this 5% to whatever is necessary in order to achieve stabilization or stick control.  Also, 5% is only the default value.  You can set it as high as you want.  Configuration -> Stabilization -> Expert -> MinThrust

If your quad is small or if you have a battery with less cells, you can put on a coat and glove for arm protection and test just how strong it will correct from inverted in Attitude mode with CruiseControl with current LibrePilot (but much less with old / OpenPilot).  Stop the motors by using zero throttle, flip the quad inverted with your hand, hold quad very tight upside down and add a small amount of throttle.  It twists very hard to get back to right side up.  It sounds like you had some other problem (receiver reboot, out of range, using old firmware, kept stick at zero throttle the whole time, etc.).

I just did this with a 330 quad that uses low RPM (1000KV) 2212 motors and 3s battery but this test with a 2s battery.  It was set up with Rattitude and CruiseControl.  It acted just as described.  You really need the reduced cells battery to do this test safely.

Also, past a certain falling speed, props are stalled out even at full throttle and don't generate thrust.  That happens sometime around 3 or 4 seconds of free fall.  Theory seems to suggest that it can even get stuck upside down because of control reversal due to a spinning prop causing more drag than a stopped prop.

[TheOtherCliff]

If you need to be able to recover at zero throttle stick, there is Always Stabilize When Armed, or the easy way that can be changed on the transmitter is to just increase the throttle trim enough so that the motors always run (even when pulling down hard on the throttle stick).

[QC102]

No, my ESCs have the default configuration of factory. They are Emax SimonK ESCs and I think the break is not active cause when I shut down the motors the props keep spinning for a while.

Default on Emax SimonK ESC is Brake - OFF.
If you can - get an Emax programme card for  SimonK... they are  good value and easy to  use.

[batata003]

@TheOtherCliff I will do exactly what you said, I will test it and check if my dronw will recover from upside down using my hands. I will also make sure always stabilized is activated in order to make sure my quad is always stabilized.

The trim idea is great cause I can get throttle to zero and the trim will still keep some throttle in the bacgroudn preventing problems.

Thank you so much guys, you are awesome!

[TheOtherCliff]

AlwaysStabilizeWhenArmed can be dangerous in that zero throttle does not stop the motors.  I don't use ASWA for that reason.  There is a sticky thread you should read...  I like to be able to stop the motors with just a reflex, without searching for a switch, and have my fingers on the sticks to restart if I think I can recover.

I would test it as described, without ASWA, to see that it does indeed recover if you simply have some throttle stick and neutral everything else.  You might have another issue that needs to be found.

Also, be aware that a stalled out prop does not generate lift.  A prop is probably stalled out if it is falling half as fast as the 3/4ths tip speed (upright) and a bit faster inverted, but these assume full RPM.

A 10000 RPM (full throttle, windmilling KV1000 on 3s) prop turns 10000/60=167 RPS.  At 3/4ths of the way to the tip, a 10 inch (25.4cm) prop travels about 60cm per revolution.  That is about 100m/s at full power so the prop is probably stalled out at full throttle at 50m/s.  At half throttle, the prop is stalled out at 25m/s.  25m/s is reached in free fall by a little more than 2.5 seconds, so falling with motors off for 3 seconds would require more than half power just to get the props past stall speed.

There is another problem with stalled props and that is there may be a thrust control reversal due to the fact that a spinning prop generates more drag than a stopped/slow prop so when Attitude mode stabilization spins up the low side it generates less drag and falls faster instead of slower.  Once upside down, the lower motor speeds up to try to pull it through to upright, but that generates more drag and the low arm rises.  Stalled props (quad falling at high speed) make attitude mode completely unstable work backwards when inverted.  In Rate mode you might see through your FPV that left roll actually rolls right.

[batata003]

You are correct, it's not good to have a switch to turn off the motors, it's not safe and I already almost had an acident cause I lowered throttle and forgot the motors would not turn off so I had to move the switch and while that the drone was kicking in the ground dangerously.

But I think AlwaysStabilizeWhenArmed is good cause when I try to do a veeery fast descent I can lower my throttle to the minimum and it still keeps my quad stable, avoiding the problem that I had that the motors would not turn on anymore. I think what I had was hard to reproduce but surelly CRUISECONTROL turned off the motors while my drone was upside down and because it was not doing a flip, it kept upside down and I could not turn on the motors anymore. Maybe a but or something but I am really sure I could not turn on the motors after my drone got upside down for a few second in the air. I am playing with AlwaysStabilizeWhenArmed  and so far I had no problem, I did at least 8 flights in the last days and  my descents were all perfectly, that's amazing how AlwaysStabilizeWhenArmed is working fine!

I liked to read your calculations and learned a lot with them but I think I didnt had a stall problem, I moved my throttle stick all way up and the motors visible didnt spin fast, they had a very slow spin and fell.

[TheOtherCliff]

I ask that you (safely) try to re-create this (inverted, throttle above 5%, but motors slow and not trying to turn it right side up) in hand and with no changes from your crashed setup.  When upside down CruiseControl will set throttle to 5% (unless the stick is at zero) but then the Attitude mode stabilization will try to make it right side up and will increase the 5% a lot to do it.

I made this test myself and proved that at least my setup works as I think it should.  I'm just afraid that you will add something and it will work (maybe only till you get the same scenario) and you will continue to think that CC is broken, or that CC is actually broken and because it now works for you, I won't know that it needs to be fixed, or that you have some other issue (receiver issue, etc.) that will cause another crash, or that there is something in the setup (like stalled props, but not stalled props) that neither of us knows, but will cause other people to have crashes till we understand it.

At this point, the closest thing that matches the facts is an oscillation when pointed straight down that could for instance be caused by stalled out props and the control reversal from a spinning prop having more drag.  This could act to make it stay completely upside down and as soon as a low motor starts to speed up (to pull it lower as the first part of flipping it upright), it generates more drag and goes higher (instead of being pulled lower as stabilization wants), so a different motor is now the lowest and it starts to speed up (while original motor slows down) ... repeat ... it stays perfectly upside down (perhaps with some oscillations) and no motor is running what you would call "fast".

I sent a message to another developer to look at this thread and this reasoning.  Let's see what he thinks.