Have a good day !

I would like to ask more experienced pilots to help me to understand the reasons of DJI Spark strange behavior when it take off.
And strange difference between Altitude given by barometer and VPS sensor on Spark telemetry.



It all started with the fact that the spark, neatly packed in his DJI bag, was dropped from a height of less than a meter.
After that, he started jumping "like a frog" as shown on https://www.youtube.com/watch?v=_9Vo_1nBUQk
Firmwares are ok, I check 3 last versions with minor improvement on 0.9 and update again to last 1.0 firmware.
RC checked with WIFI and OTG connections, no differrence, 100% RC signak bothway. Compass and IMU both are carefully calibrated.

So as I understand problem most likely in a bottom VPS sensor
I check this method https://www.youtube.com/watch?v=aD2megb5_yc and receive an huge improvement, but result was preserved for only a few flights and Spark start to jump again. If I take off from the ground with manual engines start and vertically upward at full throttle-up (left stick up to 100%) after 2-3 meteres of altitude drone will start to feel more contollable but whatever have a 20-30 cm drift on position and height. What was unacceptable to me because I fly in the mountains and on rocky terrain. So I continue my investignation.

And found a note about rerouting ESC wires https://www.youtube.com/watch?time_continue=30&v=pBa8sXuBtes&feature=emb_logo

This will be a reasonable idea because gymbal connector are made from a thin coaxial cables and bottom VPS camera have a unshielded ribon cabble that goes in touch with front-right engine ESC wire. So I reroute ESC wires so they go flat on the drone shell bottom and don't touch a bottom mainboard shield. Please see:

ESC Rerouting

Jumping stopped, but SPARK, having risen through the autotakeoff, when all sticks are at the center untouched - try to fly-away in the same direction.When drone is faced on me at 11 hour (right-back and rise up) For the purity of the experiment, I ran the launch indoors i.e. without GPS signal. As a result, over and over again the drone goes to the ceiling.

DJIFlightRecord_2020-08-06_[19-40-24].zip (17.71 KB, Down times: 1)

2020-8-6 Upload

Click on the file name to download attach

In log I see strange: difference between an [ALT] value, reported as I understand by combination of a IMU and barometer and [ALT VPS]. At the same time [ALT] changes sequentially, although what is incomprehensible to me goes below zero... and [ALT VPS] by abrupt jumps. And I don't have a enough experience to take clear decison: do this cause (like I think) by some kind of uncompensated interference between ESC wire and VPS cable or may be VPS sensor are dead and simly must be replaced.
Or this is cause by broken IMU (calibrated w/o error at 0.01 !) or compass (also calibrated to the green state)  In one of 6 flights I recive in the end compass error on my log. So too many options (



If someone experienced in this theme can help to determine reasons of this Spark bug I'll be very thankful !

Best Regards, Ivan InSky.studio

In the absence of any other advice, my suggestion is that there must be something wrong with it.
And if you can get better advice from DJI themselves, then good luck to you.

Thanks for the recommendation! This is undoubtedly the most correct decision, but now up to the autumn I am on an expedition in small village deep in Siberia and there is simply no opportunity to send the copter to DJI service center before I goes back to one of the big cities. So I'm trying to fix it on the place. Or at least understand what exactly failed in order to replace the module.

Here a example of copter behaviour https://drive.google.com/file/d/1n5S_2Izi9weHtPQPmOaxdXOBjhjeUR_s/view

At 01.8" motor are started and up to 07.6" no RC actions. After this time only RC Dn for force landing w/o result - copter are compleetely uncontrollable by RC
so all this strange drifting is a strange copter behaviour w/o pilot interaction...

DJIFlightRecord_2020-08-07_[16-36-21].zip (22.69 KB, Down times: 0)

2020-8-8 Upload

Click on the file name to download attach
Flight log

If someone experienced in flightlog reading are here - please help me to understand:

1. Is it normal what [ALT-VPS] transmitted intermittently as shown on graph?
2. Why [ALT] goes under 0 is it a realative or absolute values ? What is the source of [ALT] value IMU + barometer, right ? It seems that [ALT-VPS] in total give much more precise values...
3. Can we, based on the logs and videos, determine which of these reasons was most likely to cause this failure:
    3a  Some kind of interference between ESC wires and VPS ribon cable? IMU cable?
    3b  Broken IMU (even when it can be calibrated up to 0.01 readings in the DJI.Go on precisely leveled surface)?
    3с  Broken VPS module?
    3d  Broken navigation module (PCB with big DJI marked flight controller and compass)(even when compass can be calibrated slo-o-owly up to start of green scale on DJI.Go)?

Yes, PCB's bottom schield are insulated very vell
accoding to a dji.forum user's suggestions )

See my theory, this: https://forum.dji.com/forum.php?mod=viewthread&tid=220364

Jakab Gipsz Posted at 8-8 14:10
See my theory, this: https://forum.dji.com/forum.php?mod=viewthread&tid=220364

Thank you a lot for sharing this method!

But in my case I carefully check and clean any contact and clean it with ethanol many times, problem with "jumpy frog" spark goes but strange drifting imediatly after take-off hapens...

But can confirm - oxidation is an issue, because contact pairs are covered by some kind of hotglue only from the sides. So humidity can access contacts.

Jakab Gipsz Posted at 8-8 14:10
See my theory, this: https://forum.dji.com/forum.php?mod=viewthread&tid=220364

Thanks for sharing

Hi people!!!
Привет Ivan03Rus Та-же самая история с моим Spark )
-плохой долгий поиск gps
-появлялась ошибка инфрокрасного датчика
-прыгает при взлете и не держит высоту , не возможна посадка
-была проблема потеря сигнала видео на телефон и долгий отклик пульта и дрона ( OTG кабель решил проблему)
-сегодня понял что не работает передние  сенсоры (в приложении нету границ препятствий и не реагирует на движения перед сенсором)

пробывал много методов:
-замена всевозможного по
-замена двигателей
-изоляция нижней стороны платы( использовал экранированую ленту и изоленту и медную ленту вставки на шлей нижних датчиков ему пох/)
-перепаивал антенну и чип GPS с mavic air у них одинаковые микросхемы ( на какоето время все хорошо стало с жпс /брал полную антену и 12 спутников)
-пропаял микросхему  LC1160 power management IC ниче не дало

Обнаружено интересное поведение , если оставить один пропеллер на переднем правом моторе и запустить моторы стиками ( не взлетать) то значения  датчика высоты начинает меняться до 12м /при этом дрон стоит на полу/ на других моторах такого не было замечено
Где-то разгадка рядом, связано с наводкой возможно кварц помехи ловит или мертвый какой-то из датчиков

Добрый день !

Да, очень похоже - в моём случае прыжки при старте прекратились после того как опустил на дно корпуса кабели правого переднего двигателя.
Но при этом я снял чёрную прокладку из пенорезины в которой проходил правый коаксиальный кабель антены. И сразу после этого началось рысканье и сваливание коптера при старте. Прокладка была сильно приклеена и снималась с усилием и я думаю, может ленточный кабель от VPS сеносоров такой нежный что где-то повредил.

В любом случае прихожу пока к такому решению:

Для диагностики отключить нижние сенсоры - заклеив их изолентой или через конфиг.

g_config_avoid_obstacle_limit_cfg_safe_dis = 2
g_config_control_avoid_atti_range        = 10

Чтобы исключить VPS со всеми его наводками и интерференцией от силовых проводов. А есть опыт такого отключения VPS ? )

* * *

Have a good day !

Very close to my experience - in my case Spark stop to "jumping" on take-off when I rearange ESC wire from front right motor.
But for doing this I remove a black sponge that hold coaxial cable from right WiFi antena. And right after this copter sart this drifting and staling on take-off.
Think that sponge was glued very well and may be VPS ribbon cable to that it a glued can be damaged. Even if it is not visible.

So I come to decision to temoraly disable VPS sensors.
By ducktape to the sensors or by changing config:

g_config_avoid_obstacle_limit_cfg_safe_dis = 2
g_config_control_avoid_atti_range        = 10

And then see how this will affect copter stability. Do someone have practical expereience with this ?

Here I see example of VPS sensors blocking on MM: https://www.youtube.com/watch?time_continue=282&v=KmfZKL_hkRg&feature=emb_logo