Tuesday, June 30, 2015

Auto-Mission 05

Longer and higher Autonomous way-point auto-mission flight. Heavily over-cast clouds and HDOP was 1.8. This video best viewed in HD (1080p or at least 720), preferably Full-Screen, and with the sound.

With other battery ... I recorded some free-flight footage not included in this YT video.

Sunday, June 14, 2015

My first AutoTune (Testing Only)

Here are the docs for AutoTune , and here is a thread about it. Remember that this procedure is completely optional and that the Default PIDs should be fine for 90% of quads (so say the developers). If you want to try it manually instead, here is the guide for that, and also read this.

And some Auto-Tune videos AutoTune and AutoTrim Video , Video-2, Video-3 .

Now would be a good time to do a Auto-Trim if you haven't already. While not as vital for Auto-Tune, Auto-Trim should be done in windless environment.

How to invoke AutoTune
1. Set up one flight mode switch position to be AltHold.
2. Set the Ch7 Opt or Ch8 Opt to AutoTune to allow you to turn the auto tuning on/off with the ch7 or ch8 switch.
3. Ensure the ch7 or ch8 switch is in the LOW position.
4. Wait for a calm day and go to a large open area.
5. Arm in Stabilize and Take off and put the copter into AltHold mode at a comfortable altitude (7 meters or 20 ft).
6. Set the ch7/ch8 switch to the HIGH position to engage auto tuning:
  •  You will see it twitch about 20 degrees left and right for a few minutes, then it will repeat forward and back.
  •  Use the roll and pitch stick at any time to reposition the copter if it drifts away (it will use the original PID gains during repositioning and between tests).  When you release the sticks it will continue auto tuning where it left off.
  •  Move the ch7/ch8 switch into the LOW position at any time to abandon the autotuning and return to the origin PIDs.
  •  Make sure that you do not have any trim set on your transmitter or the autotune may not get the signal that the sticks are centered.
7. When the tune completes (about 4 minutes) the copter will change back to the original PID gains.
8. Put the ch7/ch8 switch into the LOW position then back to the HIGH position to test the tuned PID gains.
9. Put the ch7/ch8 switch into the LOW position to fly using the original PID gains.
10. If you are happy with the autotuned PID gains, leave the ch7/ch8 switch in the HIGH position, land and disarm to save the PIDs permanently.

If you DO NOT like the new PIDS, switch ch7/ch8 LOW to return to the original PIDs. The gains will not be saved when you disarm.

In summary, the position of the AutoTune switch when you Dis-Arm will determine what PIDs are saved or used for next flight.

First AutoTune Flight:

Ok, so my first attempt of AutoTune worked fine in big field.

SuperSimple-Mode was already assigned to a switch on Taranis. All I did for this setup, was to temporarily swap SuperSimple-Mode on CH7 for AutoTune in Mission-Planner (no other changes were required). So, it turns out that on ArduCopter 3.1.5, there is no indication in MP or Tower that it is actually engaged. To me, it seems more like a Mode-Modifier like Simple-Mode. In a flight, the only indication is observing the aircraft and controls. I was not able to simulate it or test switch settings on the bench.

It was a bit windy and AutoTune took a good 7 minutes to complete. Almost couldn't finish on full-charged 2700 lipo. Battery usually lasts longer ... really stressed it and it was Puffy-McPuffers and hot when finished.

I'm not using these Auto-Tuned PIDs at this time. Instead, I'm still running the last PIDs from the bottom of my v3.1.5 page. I recently found out that gimbal should have been removed (and it's weight added back) or disconnected and "blocked/isolated". Pretty sure it "threw-off" my Auto-Tune process and resulting PIDs. It was also pretty windy, so I don't think that helped.

Monday, June 8, 2015

Flight Session #25 - Autonomous WayPoint Mission

Partly-cloudy with 9mph winds. GPS sats/hdop is good (1.50).

A short autonomous way-point auto-mission. Mobius 1080p/30fps (narrow FOV) with edited sound.

Also did some manual flying in Stabilize and Altitude-Hold that is not included in video. That gimbal footage looks ok also. Practiced my manual landings as well.

Not much has changed hardware-wise on Nova. However, I did shove a couple more (up to 6 now) memory-foam ear-plugs into the gimbal's 4 dampening balls. They seem to help with tiny "whole video shot" vibrations.

Friday, June 5, 2015

Maintenance Procedures

I've done a couple more gimbal test flights ... Where to start? How about I post symptom and then action.

Apparently, the gimbal's extra 185g weight really stresses my little 350-class Nova. It's now approaching it's max. payload weight, so aircraft setup has to be extra good. These new aircraft vibrations were causing gimbal vibrations as well as general quad stability issues (as FC deals with them). It seems the brass-looking metal inserts for the main body screws were not actually fully attached to upper body shell any more. While the screw originally did tighten fully (didn't spin in plastic), the inserts could slide vertically. This was preventing them from doing their job properly. Since body is also frame, this was causing major arm flexing. To repair, insert a screw part-ways and tug to test all of them. I used super-glue and re-glued all 4 main ones (on arms, closest to landing-gear). The rest were ok at this time. If you ever see one that has backed-out ... don't just pop it back in ... it must be re-glued.

While I thought it was the gimbals weight that was originally causing the quad to fly so poorly (and for the motors to sound so labored and stressed), I think this was instead the root cause. This whole "searching for better PIDs quest" might be un-necessary.

Altitude drift:
Altitude holding worse than before. Put packing tape over 80% of new barometer-hole mod. Barometer seems more normal now.

Nova FC parameters:
I got some new (just slightly tweaked) PIDs from Skywarka on RCGroups. With the extra 185g gimbal weight, these seem to work better. See bottom of AC3.1.5 upgrade post for current ones. THR_MID=670 is finally determined and tests good at this weight.

Gimbal dampening:
Shoved memory-foam ear-plugs into gimbals dampening balls. Installed Mobius (and holder-tray) on gimbal arm flat with two 1cm strips of Zeal (instead of nylon 1/4" screw and washer). For extra security, I've tried a small elastic band and/or a tighter nylon tie-strap. End result is more Mobius contact surface area with gimbal arm, and seems to have fixed the "gimbal buzzing".

Gimbal Parameters:
I'm back to using my originally posted SimpleBGC parameters. Apparently, the "whole video shot" vibrations, occasional "jello", and buzzing were all minor hardware crafting problems.

Changed Primary recording mode to 1080p/30fps Narrow. Will decrease "fish-eye effect" when camera is pointed down. While the end-effect looks like a zoom, it's actually a pixel-matched 1080p recording from middle of lens. I also tried 720p/60fps/Narrow ... while it is smoother, it's not as sharp when played back on 1080p monitor. I have it setup as Secondary.

While I think the dev's algorithm is set a bit too aggressively, I had Tower's "Imminent ground Collision Warning" go off a couple of times (still getting used to this gimbaled flying turkey). I've never heard it before so I didn't even know what it was (or if it was coming from Tower or the Taranis). It just sounds like "alternating bells and alarming". I would have preferred a voice-alert (maybe something like "PULL-UP"). A red "alert bar" appears across top of screen, but it's not like this is usually a good time to be looking at tablet.

Monday, June 1, 2015

Flight Session 21 - Gimbal Testing

The Texas weather has cleared-up, so I made it down to the big school field for a couple of batteries thru the Nova. It was partly cloudy and winds were light. GPS signal was good.

Only third flight with gimbal and trying some new tweaked SimpleBGC PIDs today. It's getting better, but I think it can be better still. I wonder if the gimbal mounting (or dampening balls) need to be changed (to help with vibrations and/or jello). Props should still be balanced, but I seem to have picked up some extra vibrations somewhere.

Manual flying was fun but quad acts like it's having a hard time power-wise. In the video, you can hear motors surging, and I'm wondering if that is making the vibrations worse. It wasn't even that windy. Might also be the quads params and PIDs since this is also a fairly new install of AC 3.1.5.

I programmed and ran a couple of Auto-Missions around field with little-to-no problems. I also tried out my new Taranis programming (instead of just using tablet). I just Arm quad and flip the Auto Switch, and give it a bit of throttle. It takes-off by itself, runs mission, and then lands itself. Really happy with how the works now. I suppose, the coolest thing the Nova does so far.

Here's a new video (finally). It's 1080p, so be sure to use gear-icon and full-screen is best. You probably want to turn-down audio a bit (there is no music) but please listen and see if you think quad motor sounds are normal.