THIS ISSUE WAS RESOLVED BY REPLACING THE STOCK ESCs with ARRIS 30A OPTO SIMONK!
Also bought a programmable OSD and use the Android version of Qgroundcontrol on a cellular data enabled 10" tablet. Removed the OSD as it is redundant.
Goodbye stock DJI and your limitations - Hello to having complete control of my unit!
The rig works perfectly now!
Fellow pilots. Forgive me if I am not posting this in the right place.
I have recently completed a near factory looking conversion of a DJI Matrice 100 N1FC to Px4 and if using only one battery an one gimbal (a SteadyGo3 modified to house a FLIR Vue Pro 30Hz) the unit operates fantastic!
I am running the latest Mission Planner and the Latest quad FW. I decided on PosHold Mode as primary, Loiter as secondary and RTL for my flight modes on a 3 position switch. Stabilize Mode did not work well for me at all and caused some VERY hard landings and erratic flight behavior.
The top speed must be somewhere around 65 to 70 MPH (no kidding - followed my wife in her car down a road - off to the side of course and had no trouble keeping up) and the control is amazing.
Added gimbal number 2 without the second battery which is a Tarot T4-3D and Hero 3+ for standard FPV - still flew well. A little front heavy since it was designed to balance with the second battery in place but still smooth and stable in the air.
Added the second battery (the bay was already in place) and the copter vibrates - not too aggressively but certainly not as stable as I would like. At hover you can really tell though. The oscillation is about 1/2" left to right from center when hovering . Switched from the plastic props to carbon fiber and no change. Even used my CNC to fabricate a balancing rig and balanced the props (which were near dead-on from the factory anyway).
On a side note - also having an issue with the OSD pointing opposite of what I am doing on the receiver display (mine has a little arrow that I assume points in the direction of travel) the OSD is from Amazon and is labeled as Micro OSD but I noticed it runs MinimOSD firmware. I don't believe it is directly programmable but I haven't tried - I will test that later as the OSD is the least of my issues at the moment.
The COG is near perfect with both batteries and gimbals after moving the top battery as far back as possible and the bottom back a little less - I am using the stock motors and stock DJI 620D ESCs (I guess this is the same as the 620S on the DJI site since the specs are the same and there is no data sheet on the 620D) and are rated for 5Kg and CNC milled an anti-vibration standoff for the Px4 where the stock N1 mounts. Gimbals are mounted to a CNC milled 3mm carbon fiber extension placing one in the DJI stock location but slightly lower (the heavier FLIR) and the other just after it and high enough to clear the FLIR if both gimbals are facing forward.
NOTE: I am not sure the stock DJI ESCs respond to ESC tuning since they are tuned from the N1 and DJI wraps everything in a proprietary blanket. The unit flew perfect prior to the Px4 install and the Px4 operates with one battery fine so I am not sure this is relevant. Just curious if going thought ESC tuning has any effect.
I had to set the THR_DZ to 125, THR_MID to 600 and THR_MIN to 250 (thinking about changing it to 300 after the crash described below) to prevent stalls when aggressively pitching or rolling. I also had to reverse the pitch by changing the R2_REV to -1 otherwise the quad flew reverse when I moved the pitch stick forward. The other controls, yaw and roll, were fine (and yes, the little white arrow is pointing forward).
Other than a few tweaks for the 6S battery to display correctly, the OSD settings and other minor sensitivity settings this is all I changed.
NOTE: I am using a FrSky X9D remote with an X8R receiver. 8 channels mapped to the Px4 and 8 on the receiver so I can control gimbal 2, the video switcher and turning the nav LEDs on or off.
The other night I decided to do a full test and even though I had vibration in the rig, again, pretty noticeable at hover. I did a 20 minute flight test. On returning to the LZ manually I hovered at about 60' before starting my descent. The quad started rapidly descending (without my input) and then took a nosedive about 5 ' from my roof and dove into my roof. Obviously it wasn't at full speed because the damage was minimal. My Tarot broke apart on impact and the antenna covers on the arms broke (good thing I have spares of each) and now I am hesitant to fly it.
I know the unit should handle the weight. I did a test of all components mounted with the stock N1FC (although not powered since DJI doesn't support multi gimbal setups and switching of video sources without major hacking that I am not willing to do) and on the N1 the machine flew great with everything on it including a 1/2Lb dummy load so I could assure myself the rig would handle everything including any small add-ons I may want in the future.
What am I missing here? How is one battery (that actually balances the rig) causing so much trouble?
I wanted to note that he batteries were still at 70% when the crash happened and prior to the crash, excluding the vibration it did fly nearly acceptable.
Also any suggestions on getting the OSD to display properly would be appreciated. I followed some guide on the internet that had me change parameters in SR2 but there are no descriptions in Mission Planner as to what the settings are for nor the values or range of values. If someone can explain SR2_EXT_STAT through SR2_RAW_SENS (8 of the 9 parameters since 9 is obvious) I would really appreciate it. I am assuming this is controlling what my OSD is displaying.
If additional information from me is needed please let me know - I can provide configs and photos if it would be helpful.
Thanks in advance
In running the individual motor tests I noticed a grinding sound in motor B (M4 on the Matrice - had to physically remap it via the interface header to match the Pixhawk). I have spare motors and the spare makes the same noise. Could this be a noisy ESC? I have a spare ESC but before I change it (the Matrice has so many screws on the top plate) I wanted an opinion on this. It would appear others have this issue but I have not read where it has been isolated to one motor; usually it is all of them. The wiring to it is solid and I used the stock DJI header to interface with the ESCs.
Also bought factory 8 pin cables to interface to the M-100 main board for powering the ESCs, accepting the batteries and so on.
Could this "grinding motor" have been the culprit? I am not sure it would explain the nose dive since that would take a loss of the front motors I would think, but it may explain the vibration. Any ideas?