• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.

Building my own Quadcopter

#21
Ok... so what’s ur suggestion?
I have no idea on what to do next...

As a pilot:
I would like to learn acro mode since the beginning (and I am training this way with velocidrone sim) but I would like also to limit max roll and pitch to learn some basic 3rd person view flight, so I guess I have to set the equivalent of rate & expo for an rc plane...

As “tuner”:
I lack the pre flight checklist to be sure not to mess everything with a crash.
So far i’ve Checked (other than electrical connections) the motor numbering, the gyro settings the throttle range and the props turn directions
Do i have to control CoG position (compared to what ? The center of the X arms?)
Do I have to trim it like a plane? If yes how? While hovering?
 

FDS

Well-known member
#22
You need Acro Trainer for limiting the bank angle, it’s set in the Modes tab, put it on a 3pos switch with Horizon mode on 1, Acro Trainer on 2 and Acro on 3.
The gyro does the CoG, just don’t put the battery hanging off one end. Calibrate the Accelerometer on a flat surface, it’s in the main screen.
Trim is not used in quads.
You should go into Cleanflight/Betaflight and use the Receivers tab with your TX hooked up, watch the model in the bottom corner and make sure everything moves like it should.
Set the RC rate to about 400 for now, lower rates will help you just like it does in planes.
This video should help you walk through setting everything up-

Works with Cleanflight as well.
 
#23
@FSD : setting Rates and Expo on BetaFlight means I can leave them "disabled" on my transmitter right?

waiting for my 1st maiden I was wondering what is your suggestion for flight controller and pdb...

you've mentioned F4 as better flight controller ... any tips on what to buy next?

F4 Omnibus FC + Matek PDB (F4 FC has a "variant" with an SD card ... is it for telemetry recording? is it necessary?)
or should I buy a FC with integrated PDB like This One
 

FDS

Well-known member
#24
If you set all the rates in Betaflight you don’t need to do any in the TX.
Don’t add too much expo, you don’t really want it in quads, S rate adds softness at the stick centre, expo can make the quad feel very “washy” in the centres, which you don’t want.
I have had great success with the Matek F4 boards. All in One (AOI) boards have the advantage of being low profile, since you only have one stack layer then the ESC’s on the arms. However they have the disadvantage of introducing more “noise“ onto the gyro as all the heavy gauge power wires are on the same board.
SD card is for black box logging, useful for advanced tuning. You don’t have to put a card in there to fly it. Some boards have onboard logging via a memory chip.
A PDB adds another level onto your stack, but moves the wires away from the gyro.
Be aware the Omnibus F4 is widely cloned and many clones have very poor components.
You can get AOI F4 boards under $25 now, the Matek PDB is also very cheap. Which you use will depend on how much space you have. My son is still flying his Matek F4 AOI a year after I built it, it’s been more reliable than my much more expensive build.
 

Ihichi Bolls

Well-known member
#25
NOTHING not even trim get done thru the radio on quads.

The only thing that gets done in the radio is using sub trim menu to make sure end points and channel centers are set to 1000 lowest, 1500 centers, and 2000 highest.

Not sure on flysky gear but spectrum has this redicules thing where throws are only about 70% at default and you have to set their range to 147% to get the proper min and max.

Everything else gets done on the flight controller. For now leave everything in betaflight at default except your rates.

These rates are pretty docile. You will be able to do about a 20ft diameter roll and 30 or so foot loop after learning level flight.

Rc rate set to .42 for pitch and roll.
Super rate to .85 which will give it a nice expo feel. Set yaw rc rate to 1.75 and super rate to .46

ZERO expo ever. Control your curve using super rate only. Other wise you higher throws will be all wonky because you are exo'ing the expo....

Do this in acro mode and youll be stable hovering within 2 maybe 3 packs.

Start by setting on a flat surface and without touching throttle slowly pitch forward and watch where the quad tilts just enough to raise the back arms but not enough to scrape the front props.

Do that in all 4 directions till you get a feel for small movements.

Once you got that down move to adding throttle up slowly just until you see the quad get light and want to lift. Drop throttle and repeat until you get used to where it wants to get light with out bouncing or lurching.

After that work up into a hover slowly adding throttle and keeping the quad as stable as you can with subtle pitch and roll. It will be very rare when you can bring a quad into a hands free hover so undetstanfmd you must pilot the quad at all times.

Practice hovering at least util you can hover and not drift out of a 3 foot circle where you lift off.

Once you can do those steps come back and ask for the next set of practice exercises.

If you do this they way I say you wont be breaking any gear or props as much as you would just trying to pop it in the air and fly.
 
Last edited:
#27
Here the video of my 2nd attempt at hovering...

I need your feedback on some behavior I cannot explain myself:
  1. It seems a little “bouncy” while sitting on the ground armed so I raised to avoid ground turbulence... in this condition I wasn’t able to follow the tips of @Ihichi Bolls .... my thought is it can depend on the low stiffness of the arms and somehow the ground turbulence excites a sort of resonance of the arms... what do you think? Is it normal or maybe I am right and I have to 3d print stiffer arms?
  2. I was flying in acro trainer mode with throttle scaled down to 75% of its full power without air mode. Look at how it was rocking back and forward (it seems a SW issue) until second 36 in which somehow everything turned fine and I had the confidence to start moving it around on my will. Any idea about this one?
also, is it normal that motors become so hot they’ve melted the 3d printed PETG arms? (PETG has a melting temp of around 160deg Celsius !!) it is normal while hovering due to the lack of air flux or do I have to adjust something?

9AD4F916-8A40-4BF7-975E-56E17C63F35A.jpeg
 

FDS

Well-known member
#28
If you are in ground effect it can mess with the gyro’s ability to stay level. Sometimes the D tuning can be a cause for that (I think, @Ihichi Bolls is the man for that.)
The motors getting hot is definitely a tuning issue unless the props are too big for the motors or too steep on pitch, they shouldn’t get that hot. You have 5” props on there?
Theres quite a bit of work to do in the PID tuning. Get the values you have set using
diff all
in the CLI, then post them. Default BF PIDs tend to work better on 5” builds but yours has some non standard stuff, hardly anyone uses 3d printed frames that size since the arms bend too much.
That looked pretty good tho, did you try it in horizon mode? That helps check the gyro function. I wouldn’t hover test where you were, there’s lots of red flags about that space, hard ground, near traffic and pedestrians. Not a safe place at all. If you get a flyaway there you will injure someone or damage property.
 
Last edited:

Ihichi Bolls

Well-known member
#29
Several things could / are happening. The floaty action is probably due to too high min throttle and its right at the edge of lift.

The hot motors can be a tuning issue however I would more suspect the harmonics in the frame driving the fc bonkers and over working the esc. This is why not many frames are made from anything but stiff carbon fiber. Even lesser carbon fiber causes resonance in the frame.

Not knowing how this was put together specifically the bond between frame and flight controller / stack it is probably prudent to not point out a solution until I know for sure. A few decently close pictures of how the FC and motors are mounted as well as stray wiring will give us a better idea for solutions.

Again you should be on stock betaflight settings and only the lower rates or similar I pointed out to get started on. if that was the rates we could go even lower if you think they are to twitchy to start. If so you can drop rate by 5 to 7 points and raise RC rate 10 to 12 points to give an even more expo'd curve to pitch. You seemed stable in roll so leave that as is for now.

All in all good job controlling the lil beastie your first time. I am not sure what acro trainer does but if it does any kind of auto leveling that could also be an issue heating up motors.

This is one of the reasons I say NO betaflight default settings are good for all quads. I am a firm believer in basics first, bells n whistles later if needed.
 
#30
5045 props 3 blades
Horizon mode was tested on a first flight before that one I’ve captured on camera
It was ok but still twitchy at ground. Then I landed switched to acro and tested again.. no issue
Then I switched it off, got the phone and recorded the flight with the aforementioned issue.
In both tests the motor temp was high.. the plastic softened and some motor screws get loose. (No axial load anymore and vibrations did the job)
here my settings:
.
CLI

#
# Building AutoComplete Cache ... Done!
#
# diff all
# Betaflight / SPRACINGF3 (SRF3) 4.0.6 Sep 1 2019 / 01:05:57 (2a64051a2) MSP API: 1.41 / FEATURE CUT LEVEL 6
batch start
defaults nosave
mcu_id 0031001b5734570520383038
name QuaddEo
resource MOTOR 1 A12
resource MOTOR 2 A06
resource MOTOR 4 A07

feature -RX_PPM
feature -AIRMODE
feature RX_SERIAL
feature SOFTSERIAL
beacon RX_LOST
map TAER1234
serial 2 64 115200 57600 0 115200
aux 0 0 1 1450 2100 0 0
aux 1 2 0 1700 2100 0 0
aux 2 47 0 900 1300 0 0
adjrange 0 0 0 900 1200 12 0 0 0
adjrange 1 1 0 1350 1650 12 0 0 0
adjrange 2 2 0 1800 2100 12 0 0 0
rxfail 4 s 900
set acc_calibration = -108,44,-256
set min_check = 1030
set max_check = 2000
set fpv_mix_degrees = 20
set serialrx_provider = SUMD
set motor_pwm_protocol = DSHOT600
set small_angle = 180
set gyro_1_sensor_align = CW180
profile 0

set acro_trainer_angle_limit = 15
set level_limit = 30
profile 1

profile 2

profile 0
rateprofile 0

set roll_rc_rate = 42
set pitch_rc_rate = 42
set yaw_rc_rate = 175
set roll_srate = 85
set pitch_srate = 85
set yaw_srate = 46
set throttle_limit_type = SCALE
set throttle_limit_percent = 75
rateprofile 1

rateprofile 2

set throttle_limit_type = SCALE
set throttle_limit_percent = 75
rateprofile 3

rateprofile 4

rateprofile 5

rateprofile 0
save
 
#32
@Ihichi Bolls :
Acro trainer will only limit max angle to a fixed value to avoid flip and crash
Each motor is screwed with 4 bolt over 5mm of PETG legs (no soft mounts, damper or rubber in between) which are reinforced with ribs to make em as stiff as possible (I ve tried to increase the moment of inertia of the leg section to decrease flexibility) i did not perform any modal analysis or forced response on the legs but I do not think the heat could come as friction dissipation due to resonance phenomena just because the leg with the motor on one end acts like a lollipop and the first bending modal freq should tear apart the leg if resonance occurs.
PDB & fc Are stacked and mounted with plastic screws and spacers (no silicone o-rings, just plastic) and esc are mounted with double side tape and secured with black tape . Each component seems quite cold to me.
lipo battery is warmer but nothin I’ve not already saw on RC planes while pushing them to the limit.

here some picture of my frame design: (cad version.. pictures will come soon)
Red are all the screws (metal)
White all the PETG 3D Printed parts
Brown plastic spacer and screws
Light blue pdb and fc

EBD4BAFD-F762-4977-AD25-12DCBCB1EC80.jpeg 5B012A7F-A6CB-44BB-8D41-4DEDB7B6ABD8.jpeg
2A9D9A5E-6DCA-4451-ADFA-325DD90A3637.jpeg 5E06417E-41D1-4C0E-985D-655385CBD9D1.jpeg CA0E0867-7854-4A73-AF63-0FFF9AF342FF.jpeg
 

Attachments

FDS

Well-known member
#33
How is the FC mounted in the real world? Is it soft mounted on rubber feet? Is it bolted straight through with rigid bolts or stack spacers?
 
#34
Stack spacers ...plastic ones no rubber feet no Orings
Rear bolts (plastic) pack the boards and also hold in position the upper deck
Front bolts (plastic) aren’t connected with upper deck and keep in place the stacked boards

EB3EE3D0-85FC-460F-B3BD-81B6534D36C7.jpeg
 

Ihichi Bolls

Well-known member
#35
With that printed frame I would suggest you go with the rubber bobbin style FC mount. I dont think O rings will be enough mass to cut the vibrations.
Are those the stock pids? That seems funny to have yaw low at 35 but high I gain at 100. Everything I have flown P has been higher then pitch and roll and I usually run my I gains around 40 for P and R and 50 for I on freesytle with the winds I have here to fly in. On my racers its a little higher but not like that.

Last I think 5045 is too big pitch for those motors on tri blades. If it was 2306 it would be liveable but not 2205. I wouldnt go more then 5x4x3 at least until you know how to tune properly and not cookie cutter youtube specs.

Even on my beast quad I dropped back to 5x4x3 again after trying 5x4,3x3 due to heat on 2306 2550k emax motors.
 
#36
@Ihichi Bolls:

yes PID was left untouched as per your suggestion. do you think I have to tune 'em?
about the prop pitch it's a shame, I chose 5045 because it was the propeller used by emax to provide the table with the motor tech info (current, torque, pwr) I did not immagine I will end up with so much heat to handle.. I will look for 5040 instead as per your suggestion
 

FDS

Well-known member
#37
I use rigid carbon frames, I always soft mount my FC. You want to totally isolate the gyro from vibrations.
@Ihichi Bolls Is too much D contributing to the motor heat?
Would using the settings from the BF tuning notes help? To quote those-

This is hypersensitivity to D that has as the underlying cause either flexy arms, high power to weight ratio, faulty gyros or other prop/frame resonance issues.
For 6S quads you should first cut PIDs by about a third. The settings below cut D a lot and P a bit, and filter D a bit more strongly lower down. They may get you in the air without drama, from which point you can sort out the underlying problem/s:

set d_min_roll = 14
set d_roll = 20
set d_min_pitch = 15
set d_pitch = 22
set d_min_boost_gain = 20
set tpa_rate = 75
set tpa_breakpoint = 1400
set tpa_mode = D
set p_roll = 30
set p_pitch = 30
set d_yaw = 0
set dterm_lowpass2_type = PT1
set dterm_lowpass2_hz = 100
 
#38
One curiosity from mechanical point of view about 3D printed arms:

Considering the arm as a cantilever beam loaded at one extremity, if we compare the carbon one to the PETG, to guarantee the same stiffness we have to chose a PETG section in order to ensure a moment of Inertia greater than Carbon one by a factor 20 to 80 times greater (which represent the ratio between the Young modulus of Carbon vs PETG)

assuming the section is unchanged along the lenght of the beam we know that a rectangular section reacts (to the deflection) with the power of 3 of its thickness so this means that the factor 80 under the root square become something around 5.

if my math is right then it means that a PETG section must be thicker 5 time more than a carbon one.
with some optimization (remember we considered a plain rectangular section) I am convinced I can achieve a good stiffness also with PETG

The step 2 will be to optimize the weight of the arm + ESC + Motor to dampen the harmonics coming from the motor (but I need a SW to do this) but again I am confident the design could survive ;)

anyway.... @FSD i didn't get where you've pulled out those numbers... is there any guide I can read to learn about PID fine tuning...
I had no difficulties flying "my beast" with the default PID setting so I am a little bit worried about modifying those values without understanding the effects in details.

anyway i've ordered 5040 props, soft mount for the Boards, and M3 nylon whasher to isolate the engine from the arm (nylon melting point is above 200°C)
 

Ihichi Bolls

Well-known member
#39
@FDS yes too high d gains will hurt a motor. However in this case with betaflight they scale different then other FC's. 35 is actually low to to low middle range. For Kiss I if I raise D to 20 I revert back to 10 and up p gains. I like keeping D on my KISS gear around 12 or 13ish.

I am still pointing to the FC not being soft mounted on a 3d printed frame. 3d prints work fine on 2 or 3 inch quads as the short arms dont allow resonance as much. the long the moment arm is the more area there is to resonate from vibration. Being a one piece unit that vibration transfers to everything and can cancel and multiply where the forces meet.

If set up right the motors should only ever get warm to tight finger pressure. Dropping pitch and soft mounting the fc is gonna help a lot I am fairly certain. Like I said Betaflight has the stock settings pretty dialed in for most builds. They only leave p gains weak as to accommodate a larger range of gear. That tends to loose finishing of fast flips and rolls as well as washing out in tight fast turns, but that is well beyond beginner levels of flight.

In my opinion pids should be dealt with first THEN filters and such used to adjust issues for each specific build. But that requires heavy knowledge of both PIDS and filtering to make work successfully without letting the magic smoke out.


EDIT... Remember.. the motor heat will transfer to the screws as well so be mindful they are not melting internally creating weak spots in the plastic.
 
Last edited:

cranialrectosis

Faster than a speeding faceplant!
Mentor
#40
My old LadyBug quad and my Great Pumpkin Hex. Man these 3D printed toys were fun to build and fly circa 2014!
037_FinalAssembly.JPG

CompleteWithTriBlades.JPG

@Ihichi Bolls is spot on with the heat on the screws ruining the boom from the inside where you least expect it. Check your screws and tighten them before EVERY flight.

Popsicle sticks and tape for splints. Fly em until the parts just won't hang on anymore! :D

I always bought 3D printed frames in pairs. Print a spare and don't sweat the crashes.

Very cool build. Get the PIDs worked out and fly the snott out of it!

Welcome to FliteTest!