Building my own Quadcopter

Tammeo

Member
Following ur suggestion I've ordered soft mounts and propellers with decreased pitch... I swear i'll leave PIDs untouched ;)
Meantime I started designing the 2nd version of the frame with shorter arms and increased inertia section to counter vibration issues... Same size of the current version.

the only thing that scares me the most is the position of the battery: too aligned with propellers... I have to think about some sort of shield but I do not want to add unnecessary weight ...

Design2.JPG 1571736871654_Design1.JPG
 
Last edited:

Ihichi Bolls

Well-known member
The only thing I noticed with a batteries position is how it reacts in high G maneuvers. Other then that the fc takes care of adjustments.

If you are worried over prop strikes in a crash just be sure to have a good battery pad like ummagrip and use two battery straps. That or bottom mount the battery. The Fc wont care either way like I said.
 

Tammeo

Member
Time for another update.

hw upgrades:
  1. rubber spacer for the stack of pdb + fc boards
  2. Stiffer arms (increased thickness, same design)
  3. Yellow tape & Nylon spacer between motors and arms
  4. 5040 propellers
sw upgrades:
  1. Esc frequency from 24k to 48kHz
  2. motor phasing from 16 to 22deg
  3. No changes to PID & Rate values
  4. Throttle still limited to 75%
  5. Max roll & pitch angle set to 15deg
Result of the second flight test in acro trainer mode (around 5min of flight time)
  • No arm vibration at all
  • No arms melting cuz All components are super cold (yeeeeeh!!!)
  • Motor screws slightly untightened but no one found loose (no issue or control instability found)
  • Quad Was super stable: hovering was easy and I’ve got confidence to Hoover also facing it at 90deg.
The only issue was during the first 20 sec of hovering: no issue at doing that maneuver but the motor sound was different (somehow louder) then suddenly it stopped and become more like a buzz (like commercial drone)
I didn’t notice strange behavior (power loss or twitch or similar) but it seemed to me like a sort of recalibration of the esc or fc board.
next time I will try to start in acro mode (instead of acro trainer) to check if it is somehow related to that particular mode.
Any suggestion about the issue?

Overall feeling was super fun.. the set up is ok for me and I will train my LOS Skills with those values coz my skill as a pilot is far beyond the limit of my frame and current setup. Here under 2 pics of my new setup

C85A34CE-1380-4DA7-A91B-3033EDF8578D.jpeg 32EE1097-C6FE-47CA-9838-289730643AE5.jpeg
 

Ihichi Bolls

Well-known member
Congratatz on a successful and what sounds like an enjoyable flight. I gotta ask though because I am not understanding how you are new to quads n such but are changing things like motor timing, throttle percentages.

I can almost guarantee the weird sound the first 20% of the flight was the change in motor timing. In 5 years of flying I have never had to do that.

Where are you getting all this information to make changes like that before the quad has even flown?

Once again... STOCK betaflight settings STOCK esc settings and lower rates. NOTHING else is necessary for learning how to fly. If it is too twitchy you can lower the rates further then I suggested and build back up over time. No trainer modes are necessary no limited angle.. That is all going to teach you one way to fly. ALL bad things you will have to unlearn later as you progress.

The method of learning I will and have told you will train the muscle memory and get you used to fine movements and build upon each step with minimal time and damage to your gear..
 

Tammeo

Member
@Ihichi Bolls, I am new to the hobby but I am involved (at work) in the design & development of things... so basically i spend my days solving issues and designing stuff....

The mindset I use at work pushes me to understand each setting I do not know and each implication it will have on the issue I am facing even in the hobby ;)

Searching for rs2205 issues I discover many ppl experienced overheating problems somehow related to torque, propeller or damaged motor due to fixing screws too long that will damage wirings ..

reading some articles on esc and blheli32 I discovered that
  1. Raising timing will reduce torque (somehow related to my issue) and it is ok for high performance motors like RS2205 with ton of torque
  2. Increasing esc frequency will increase the smoothness of the controls limiting the peak of current through esc & motors (and power directly related to motor heat)
so I decided to try both :)
I agree w u the chosen value probably will be suboptimal for a skilled pilot but I am not so maybe my poor skill makes the suboptimal fine for me :D lucky me !!

Also,
At the end I followed ur suggestion keeping PID unchanged, adding soft mount for boards & tuning down pitch.. that’s the reason why I have to thank you so much for your help. ;)

Tonight I’ll try acro and I’ll post feedback ;)
 

Ihichi Bolls

Well-known member
Im not rippin on ya mate.

I just see sooo many people who are new watch the top guys and cookie cutter things when they should be concentrating on getting in the air not optimizing gear out of the gate. Most times they end up here going "I followed so and so's video and my quad wont fly or it flys like crap or I fried gear".

Its best even if experienced to start at the beginning with just the basics then use the tools available IF there are problems or if something will enhance the style you want to fly. Adding things just because someone else did will only compound issues as there is not one quad on this planet that is exactly like another even with the exact same components and software.

My biggest let down is with what Joshua Bardwell does . He is a great technical analyst and is very knowledgeable about many things. The way he presents things however is where we run into issues when people take what he did on his quad and automatically apply it to their own. The Rotor Riot crew or even Flite Test do the same at times. Teaching is one thing. Presenting things as the be all end all for all situations is just not viable specially in our hobby where there are so many options for everything.
 

Tammeo

Member
I agree with you.
Let’s keep things simple. The quad is a quite complicate thing to understand and touch so many concept related to electronics, aerodynamics, controls, design which is almost impossible to master in a short time. Little step and a lot of post processing analysis to understand setup, mistakes, bad habit ....

Anyway I’ve just finished the 3rd flight and I’ve tested it both in acro and acro trainer .. it still does the same for the first 20 sec of flight each time I unplug and plug the battery... it seems a sort of auto calibration... at the end of the first 20 sec the motor buzz comes ... everything sound smoother and if I land unarm and rearm motor the “auto calibration” won’t appear until the next battery unplug/replug... I’m searching the forum for similar behavior... I need to understand :)
 

Ihichi Bolls

Well-known member
In the features box is the item "inflight calibration" checked by chance? Maybe they tagged that when you use the trainer thingy.
 

Tammeo

Member
No inflight calibration triggered or enabled.
anyway dunno what I did but I downloaded the blackbox data and opened with the viewer... I cannot use it (and probably my telemetry is lacking sensors and signals) but on the bottom line of the screen you could see the behaviour I am talking about... from 40sec on you can see where all the things became smooth pointed by the red arrow. I made a video out of it check how it "behave" from second 9 to second 11 on

Screen.JPG

 

Tammeo

Member
I ran another test. A catastrophic one... overheating come and the peak was so high it unsolder a wire of the esc... you can immagine how hard was the landing...

I switched back to 3s battery to check if somehow my hw is not suitable for 4s and the overheating is related to high current of 4s... unfortunately I’ve got the Same loud buzz motor behavior for the first 30sec ... but I ve noticed almost all the heat comes from this louder initial phase (motor almost untouchable after 10sec of hovering)

So I have re flashed all the firmware of both escs and fc and re set up everything...
Then I ran multiple test to check if the heat comes from esc protocols. unfortunately not but I can confirm the heat disappear when motor start to buzz like “normal” ones.

Last test I did I changed the approach: instead of plugging the battery, waiting for the beeps to stop, arming motors and flying I plugged the battery and waited with 0 throttle armed motor and I trained my control sensitivity acting on roll & pitch until I was able to barely lift 2 arms at time (as suggested by @Ihichi Bolls ). After a minute or so I pulled the throttle and went hovering and I’ve noticed the motors buzz normally and heat was gone...

so i was wondering: is there a sort of procedure to arm & takeoff I am not following which can cause the issue I am experiencing?
 

Ihichi Bolls

Well-known member
those yaw pids still jump out at me as an issue...

Lets detune a bit and get back to some basics.

Set pitch and Roll P gains to 3.5 Set I gains to 40 Dgains to 25
Set yaw P to 50 yaw I to 40 Yaw D to 20

Set feed forward to 0.

Now all I want you to do is arm the quad count to 5 disarm and check motor heat. If cool then arm and hover for 5 seconds land dis arm and check heat. If cool re arm hover for 10 seconds land and disarm. Repeat this adding 5 seconds each hover until you can hover for 30 seconds.

Things we are looking for is does that initial weirdness go away. Second is heat on the motors. Third is grinding noises from the motors.

Edit just for giggles... what esc's are you using. Its possible older escs are not compatible with Dshot but still run on it but not correctly. We may have to drop back to multishot to smooth them out if we dont see marked improvement with these pid changes.
 
Last edited:

Ihichi Bolls

Well-known member
Just read the manual for the FC.

The 5th bullet under the features states optimized for one shot. I would say revert back to oneshot for the esc protocol. I know on my gremlin with the Femto FC when I was researching best settings I learned that multishot was actually faster then dshot 300 which is the best that FC could use. Then we may be able to revert back to default pids and try the 5045 props again if we get the motor heat issue resolved.
 

Tammeo

Member
Is the wiring connection the same between dshot & oneshot? Or do I have to solder wirings in another way?
 

Tammeo

Member
Ok did what you suggested...
I’ve depleted 2 full 3s 2200mah l’iPod batteries arming disarming hovering and measuring motor temp.
the root cause for the issue :
  1. is not confined to the first 30sec of flight (more later)
  2. Is not dependent by ESC protocol (tested both multi, Dshot, Oneshot)
  3. It seems not dependent on vibration (at least not the ones detectable by uman sight)
  4. It happens every time I unarm and then rearm the quad
Point 4 requires more explanation:
  1. Zero throttle. Neutral pitch neutral roll.. motor Unarmed.. battery plugged. Drone ready to go
  2. Switch to armed position, props start spinning “buzz” is regular no overheating (if I stop and measure it.) I call this state now on “idle position”
  3. when I increase throttle (not enough to take off or move the drone on XY plane) and then put it back to zero then motors set point decreases without reaching idle position > props start to spin faster than idle and motors rev seems unbalanced (different rpms) cos the drone starts moving on XY plane (mind that the throttle input was so low that it wasn’t able to move it so the movement is probably caused by the rpm imbalance)
  4. Heat start to increase along with motor temperature. (30 seconds are enough to unsolder wires)
  5. From now on 2 ways to solve the issue: solA unarm the motors and let everything cool down.. solB input some pitch or roll to lift only one side of the drone from XY plane and usually this re-balance all the rpms, buzz becomes normal and flight is stable again.. then if you give throttle it starts hovering and motors temperature start to drop without issues.

At the end of the test I took off while unbalanced and I strafed left and right using roll stick and this solved the issue (risky maneuver... I was lucky not to crash)

So that’s all I’ve collected today
 

Ihichi Bolls

Well-known member
Ok so far except the heat.

What happens is the FC specially in a self leveling or pitch limiting mode will usually NEVER be level on a solid surface. There fore the FC makes a correction. When no results on being level it corrects again. It will continue to hunt and motor levels can drop to zero or runaway to max as it does this. ANY input will reset the seeking of the new level.

If in acro mode it will hold what ever position it gets set at. So once you got two arms off the ground it worked to keep them there as it should. It was achieving that goal so therefore no runaway rpm.

At this point if you are confident on muscle memory to be very sensitive to small movements you can start to hover.

Lift the quad off the ground between knee and waist level (usually good minimum height to get out of ground effect yet not be at face level if you panic) hold for 5 then land. Check motor temps. Repeat until you can get the quad in a controlled hover within a 3 to 5 foot diameter circle for 30 seconds or more. Keep track of motor heating and any wobbles or grinding type motor noise.
 

Tammeo

Member
Ok,

up to now I've used 4 batteries training my hovering skills at waist level.
so far I am confident with the following maneuvers:
  1. Hoovering with tail pointed at me (100% of the time i can keep it levelled inside a circle of 1.5 meter of diameter)
  2. Hoovering with nose pointed at me (75% confident - the circle is somehow "bigger" let's say from 1.5 to 2.0m ideally)
  3. Hoovering with RH side pointed at me (40% confident - the circle is around 4mt ... I found it very difficult)
  4. Strafing with tail ponted at me (100% confident of the control in Acro Mode with 30°max bank I can keep Z position almost unchanged)
  5. Strafing (fwd / bckwd / left / right) with nose pointed at me (75% - Z will oscillate from knee to head level depending on how quick I can react)
  6. Strafing with side pointed at me - I don not even attempt this confidence 0%
  7. Oval pattern confidence 25%
  8. "8" pattern confidence 10% (few attempts not properly "in control")
 

Ihichi Bolls

Well-known member
Your on the right track as how I would direct you to train. But moving forward before each next level is 100% is where everyone goes wrong and learns bad habits. Ideally you want to do all the different hovers and keep it within a meter of lift off position.

At this point you can turn off that trainer mode limitation as 30 degree tilt at this point would not benefit you for recovery flying under 2m height. I think you have managed to get skills to control full on accro enough to finish out the learning process. you can always turn it back on when moving to faster forward flight where the quad is away from you and you may lose orientation. Get rid of the safety net and get the confidence in your new found skills.

How is the heating issue? any funny noises or shaking?

Keep rockin mate your doing well next is gettin us some video of your progress.
 

Tammeo

Member
white smoke from motor #4
hard landing and a lot of disappointment...
i cannot focus on my flying skills if I cannot obtain a reliable quad...

I was reading THIS article... do you think adding capacitors on both esc and PDB could somehow help me with my overheating issues?