Sunday, January 18, 2015

Nova Rebuild - Small problems - THR_MID

Here are a few small problems I noticed. Since quad is flying so good now, these are easy to spot.

THR_MID Seems Off
Still working on why THR_MID value has to be so high. 700 seems high for this minimal payload. I have a basic stock Nova (with stock 8045 props) with the following installed:
FrSky X8R Receiver
Stock camera Mount
Mobius camera
Prop Guards (stock - 4)
HK 915Mhz radio transceiver (a small radio module)

Can 700 be right? That's currently what it needs to be for me to hover at 50% stick. Is thrust low?



Also, I'm getting errors like this, each time I run a Auto Analysis on a log file (other lines Pass Good):

Test: Thrust = FAIL - Avg climb rate 38.36 cm/s for throttle avg 712
Test: Thrust = FAIL - Avg climb rate -0.30 cm/s for throttle avg 771


This parameter is used to set the throttle (between manual-control and other Flight Mode transitions) and the stick's 50% mid-point. Any time the quad's weight or payloads change, you should probably re-set it.

To get an accurate reading, you do a very simple and basic flight (only):
  • Arm quad and gently hover at 2-meters.
  • Land, dis-Arm and read THROTTLE_TRIM Parameter
  • Set THR_MID parameter the same and Write params back


I can see on the Taranis display that my Throttle Ch-3 goes from -100 to +100 when I move the stick through it's full range (just like the other sticks).
I can see in Mission Planner/FailSafe that my (calibrated) Throttle CH-3 goes from 990 to 2015 pwm, and 50% mid-point travel is around 1502 (again, similar to the other sticks).

I have gone through the Parameters and they look ok (mostly stock). The numbers here are proportions based as percentages (x10) of the PWM values.  I use WinMerge Compare. It works for both text files (nice for parameter files) and comparing directories. LiPos are not at fault either.

My Logs all look ok (I think) . This log is without props, running at 50% and then 100% throttle.
- Left side is CH-3 RC-IN/RC-OUT. This appears to be in PWM values. RCIN from 1000-2000 is correct with 50% being 1500.
- Right side is ThrIn and ThrOut and appears to be ESC values. ThrIn is 1-1000 with 500 being 50% and 1000 is 100%. ThrOut 50% (hover stick) is 700 because that was my THR_MID at the time (also makes sense why 500 is default value ... being half of 1000). Notice ThrOut is hitting 1000 or 100% which is what I was wondering about. Now, what ESCs do with that control-data is another story.

To the untrained eye, RCOUT is what looks to be "off" but it must be ok because it's hitting the proper targets on the right-side graph?

Of course my ESCs are calibrated, but I was wondering if I could see the table that matches PWM values to ESC range. ModellerMark (from says that the calibration of each ESC itself is held within the firmware of each ESC rather than as a parameter within APM .. and is not easy to view. I suppose it might be an APM algorithm and not really a "table" at all.

For Logs, I discovered that MOTORS is now RCOUT (so that's on, the PWM values). I think RAW is now IMU and/or COMPASS (again, both currently Enabled). In fact, pretty much everything else is Enabled for logging as well. This is a good intro to APM Logging (also here and here).

I just wanted to make sure that nothing was broken or mis-set, causing thrust issues.

Resolution/ Conclusion:

By removing the Prop-Guards, I got THR_MID down to 630 and no more Auto-Analyze Fail messages. It seems the 17g * 4 = 68grams weight is not the only penalty to be paid for installing them. They appear to be disturbing the air-flow around the props and further affecting the thrust of the quad. This makes the "weight penalty" for the Prop-Guards higher ... around 90 it seems. It looks like my idea of keeping them installed always as standard equipment (to avoid occasional prop-strikes due to tipping on landing) is not realistic .... not with a fully payload-loaded quad anyway. And finally, since they affect the airflow around the props/ overall thrust ... basically performance of the propulsions system ... they seem to make the aircraft more unstable (therefore counter-productive).

Other Thoughts:

Comparing my payload and THR_MID to other in the RCGroups forums, there still seems to be a discrepancy of sorts. While I have v1.0 ESCs, we have seen gradual newly revisioned ESCs being released from assembly line (v1.4, 1.6, and now even 2.1). We all know the ESCs aren't changed for no reason. While I'm sure they are making dependability changes, I wonder if the newer ones can handle more Amps, and are therefore delivering more power to stock motors (resulting in more thrust).

The quad is still mostly default with 8 inch 8045 stock props, stock motors, and revision v1.0 ESCs. I think (even slightly) different props will change thrust as well. Might be time to experiment with some alternate ones.

Edit Feb.01,2015 ... been working on it. Not sure what I changed (Parameters should be back to default) but this is now at a reasonably low level. Got it down to THR_MID=570 (still with no Prop-Guards). The Nova All-Up-Weight is 1002g.

Saturday, January 17, 2015

Nova Rebuild - Flight #10 Test Success (Winter 2015)

Flight test went great. Nova is flying better than ever before.

It hovers great with no WobbleOfDeath (WoD) in Stabilize, Loiter, or Alt-Hold. I had good GPS, so I took it up to about 50 ft. While I have had the Nova for about 6 months now, today's flight session was really just like number 10 or so. See last few posts for my progress on this rebuild.

In summary, the FC is now mounted using Zeal, and placed so that accelerometer/gyro chip is centrally located, and perpendicular to designated axis. Check-out these low vibrations. X/Y are -1 ~ +1 . Z is -12 ~ -8.

Also, X/Y originate at 0 and Z originates at -10 now ... which should be correct if you analyze the mid-points counts of the established Acceptable Ranges of -3/+3 and -15/-5 .

And I think these numbers (generated by calibrations) show just how close the GYRO chip is to "center" or 0.00 OFFSet. Apparently, the prime ACCelerometer CALibration number must be 1.0.


Finally, on my Nova PnF model ... if I examine where the FC was originally factory mounted (when I first received it), it was much closer to this now determined optimal location. It turns-out, it wasn't as randomly placed as I first thought. While not perfect, it was closer to optimal than the off-set location I originally picked for it (and used for first 10 flights). I think the fact that it was incorrectly mounted non-perpendicular is what threw me off.

I did research this in forums a while back, but ended up getting some inaccurate or maybe just un-conclusive data. This post is the result of my Winter 2015 rebuild (which starts here). I suppose I'll never know if Zeal or exact FC placement is more important than the other, but you can be sure that I plan to install any future FCs in other quads using this "hard learned" lesson and method. Man, talk about details being important. It should also be noted that excessive vibrations had nothing to do with the body/frame being cracked and repaired (that arm is no-more weaker than the rest of body).

A whole Multi-Mode flight (with PropGuards on still). Altitude is graphed on the right.

Thursday, January 15, 2015

Nova Rebuild - APM & Radio Setup (Winter 2015)

This is the other half of the rebuild procedure.

The first time MP was installed, the required Arduino USB drivers should have installed in Windows.Updated MissionPlanner to version 1.3.17 Build 1.1.5472.17896 via online updater. After it finished, closed and restarted it.

Be sure props are removed. Everything should be connected (even Compass). Connected Nova LiPo and then connect USB cable to laptop. After Windows "USB device detection chime" ... clicked Connect in top right-hand corner.

MissionPlanner/Config-Tuning/Full Parameter List/ . Clicked Save button to backup current Parameters. Took screen-shots of Basic and Extended Tuning screens. I use WinMerge to compare parameter files. Remember that calibration values are saved along with other parameters (so those fields are always a little different).

MissionPlanner/Config-Tuning/GeoFence/ . Disabled since I have no GPS inside house. Clicked Write Parameters. In HUD, getting occasional "PreArm: Bad GPS Position" messages which is normal at this time. Remember that if GeoFence is enabled, it will not Arm motors (in any mode), so best to Disable it for now.

MissionPlanner/Terminal/ . Click green Connect button next to APM drop-down (towards the left) for special Terminal mode connection.

ArduCopter v3.1.2
FW (firmware): 120

Exterior Red and Green LEDs appear to turn-off while in Terminal mode.

According to forum reports, I have to keep this HK v252_v2.0 FlightController running at stock (special ?) AC-3.1.2 because I'm using Telemetry from connector soldered into proper PCB pads.  This is why (below) we can only load default parameters from a file. Do not do anything that loads a different firmware. This tweaked ArduCopter version is NOT available for download anywhere.

Click "Log Download" button. Download and then delete any available logs. A good diagnostic set of Logs to have turned on is: ATTITUDE_MED GPS PM CTUN NTUN RCIN IMU CMD CURRENT RCOUT COMPASS.

The above numbers are where each Motor/ESC-Set should be connected at FlightController (M1-4). There are ClockWise and CounterClockWise propellers, and must be installed right-side-up (but don't install them yet).

In Terminal, you can type "test", then "motors". They should all spin in their proper CW & CCW directions (as indicated on the arms themselves) and should all sound about the same. It will start at Motor#1 [M1] (aka Forward-Starboard Motor) but instead of following the true X4 diagonal-like M# pattern, it will instead ... proceed Clockwise around quad (so it's test order is M1, M4, M2, M3).

You can type Exit and Help to navigate menu structure. Some other good tests are baro (barometer) and ins (gyro/ accelerometer). Move quad and watch values change. They also preform a calibration test at the beginning that likely would not pass if module is bad. When finished in Terminal, click Disconnect button in top right.

Start Full/Clean Setup

Connected Nova LiPo and then connect USB cable to laptop. After Windows "USB device detection chime" ... clicked Connect in top right-hand corner.

MissionPlanner/Config-Tuning/Full Parameter List/ . Since this is a PnF model, there is no telling what parameters and PIDs were pre-loaded in FC (or if they were even set or checked). NO NOT run Setup Wizard. Clicked Load button to load Factory Defaults parameter file (must have been previously saved or downloaded from trusted source). Since I am running AC3.1.2, that's the file I got. Click Write Params button and give it a minute or so for params to be saved to FC. Click back to main HUD screen. Notice we are doing this first, before any Nova calibrations.

If you need to load Default Parameters for a newer version of ArduCopter for which an existing parameter file does not exist, you can Default them yourself ... on Full Parameters List with "Reset to Default" button in the right column.

Click MP's "Disconnect" button. Power-Cycle LiPo battery power on Nova (reboot FC) via XT60 connector. Reconnect USB cable, re-connect MP ... the RX, FC, and GPS should all be lit and/or flashing.
MissionPlanner/Config-Tuning/Full Parameter List/ . Click Save button to backup current Parameters as new file. As you favorably progress, save your params to a file.

Turn on radio TX and select proper model number (ie "Model-1"). The 2.4ghz TX radio itself (in my case FrSky Taranis) should be calibrated by this point (if required ... only done on radio TX itself). If not already, the TX should be properly Binding to RX inside Nova quad by now. The radio's trims should be centered aka "at neutral".

Initial Setup / Mandatory Hardware / Frame Type
-   Be sure Type: X-Y6A is selected

Initial Setup / Mandatory Hardware / Radio Calibration
-   Use this PC screen to observe. Make the changes on radio itself (when checking or programming these).
Reassignable Channels on Taranis TX Radio: Channels 1-5 working test setup
MIXER (5/64 assigned) - (Screen 6/13)
1. AIL 100 0-8 (Aileron - Roll)
2. ELE -100 0-8 (Elevator- Pitch)
3. THR 100 0-8 (Throttle - Power)
4. RUD 100 0-8 (Rudder - Yaw)
5. SA 100 0-8 Switch_A (Flight Modes 1-3)

Depending on the orientation of the signal only cables, your main MIXER map might be slightly different.

Now, click on "Calibrate Radio" button and follow directions. It will say "Completed" on a green button when finished. The 4 basic channels should now be setup properly. Note, when you pull back on Elev/Pitch, the green-bar should go up.

Flight Modes Quick Setup (3 main ones on one switch)
Initial Setup / Mandatory Hardware / Flight Modes
Flight Mode 1 - Loiter
Flight Mode 4 - Stabilize
Flight Mode 6 - Altitude Hold
I set as above and then clicked "Save Modes" button. You can double-check it by using the SA switch on Taranis TX. The 5th "Flight Mode Channel" should now also be setup properly.

MissionPlanner / Initial Setup / Mandatory Hardware / Compass
Don't change or disturb these, but they should be
Settings: Enable, AutoDec, Orientation = Manual (Rotation_None)
Also, in Full Parameters COMPASS_EXTERNAL,0 is set (trust me, it works in this APM_v252_v2.0 FC). DKemXR's explanation is "the stock FC does not have an on-board compass (some APM FC's do) so the "external" one is electrically in the same place as the "internal" one would be".

Since I have no GPS fix (indoors) for FC to use for looking up Magnetic Declination, look it up at website, and enter it Manually.
Live-Calibrate Compass (point each of 6 sides straight toward ground, and slowly spin around a couple of times - aka. Compass Dance). I just hold mine and spin-around slowly (CW and then once CCW), but here is a similar way using a chair. Just remember to do all 6 sides, including top. It will draw 6 circles on the globe. It will beep and finish about the time you complete all 6 sides. My current offsets are -100, 17, 16 and are within acceptable limits.
Set Auto Dec back on. As long as you have GPS Fix at flying field (where-ever that might be), the Magnetic Declination will be looked-up from compressed file-table stored inside FC memory.
If you have Telemery radios, you can use them for this step.

MissionPlanner / Initial Setup / Mandatory Hardware / Accelerometer Calibration.
Use level table and perpendicular angles. Prop it up or hold very still when clicking the button on each step. Complete each step before MP times-out and auto-advances (if it does, you have to start over).

On the main MP screen (hud), be sure quad is level, the Artificial Horizon is responding properly (remember that it will look like a flight simulator in hud). Also, that the Compass points in the proper direction (even when pitching up/down).

ESCs calibrated (All at Once) in case end-points changed (center Trims first). It's a good idea to do this at least once, especially if you have been working on ESCs or motors. I also recommend re-doing it if Radio has been re-calibrated (RC stick end-points have likely changed).

FailSafe doesn't need to be changed but should be tested (see previous posts).

In MissionPlanner/Config & Tuning/Extended Tuning ... give Opt-7 a function (I set to SuperSimple Mode). CH-7 is already assigned to the proper 3-Position switch at Taranis. In MissionPlanner, changed CH-7 to be Super-Simple Mode. Middle position (or first click) ends up being Simple-Mode (non-GPS but directional) and Down should be Super-Simple-Mode (GPS required but even works "behind" pilot).

So far, I've successfully tested Simple-Mode with Stabilize and Alt-Hold (switching it on, on-the-fly). I also used it once when Nova was far away and I lost orientation. When bench testing radio, there is no indication in MissionPlanner or DroidPlanner that S-M or S-S-M is engaged. Even though it's just a Mode-Modifier, I find it strange that the only way to know if it's setup properly is to fly quad and see what happens (especially for a Mode that changes what the sticks do).

I also set Opt-8 (I set to RTL). I like having that on it's own (emergency) switch.

Click Write Params button. Don't change anything else on this screen unless you really know what you are doing. In Taranis (Mixer), assign both to their own dedicated switches (if not already). Check all Modes and sticks on FailSafe screen.

Program the GeoFence parameters, but I'll probably leave it Disabled for now.
Finally, change these modified Parameters, and then Write them.
Save Parameter file as final backup.
  • Install rest of screws.
  • Balance and then install props (double check CW and CCW).
  • Apply a little cellophane tape to compass puck cover.
  • Charge all batteries (Radio, Nova LiPos, Mobius cam, DroidPlanner tablet).
  • Carefully, do low-throttle "Hold by Landing-Gear Test" to check that no sticks are reversed.
  • Go fly. Be safe and have fun.
Edit: The results are here . Also, see posts about recent flights.

Nova Rebuild - APM Internal Component Placement (Winter 2015)

Methodically going thru aircraft to try to cure WobbleOfDeath. Here are some pics of my work. I think this series of posts will also be a good example of setting up a Nova/CX20 from scratch.

Removed everything from PowerBoard plate and cleaned away old foam-tape.
Since the FC boards are not actually screwed down inside white plastic case, there is some movement. I'm not sure, but I think the PCBs (and attached sensors) can vibrate inside case. I added 2 small pieces of gyro (single-sided and plush) foam-tape to inside of case bottom (not on top where barometer-foam piece is). Seems to work fine.
Zeal gel-tape is highly recommended to mount FC and isolate it from aircraft vibrations. It's not that costly when you consider how far it goes. It seems that while the cheaper alternative "plush gyro mounting tape" also isolates FC from vibrations, it allows the FC to micro-rock-and-sway, which ends up making it an inferior solution ... the Zeal is confirmed to be much better suited for this task. I used 4 cubed 11mm x 11mm pieces as recommended (one in each corner).

FC is again perpendicular to front of aircraft, but now mounted with gyro in exact middle of aircraft (distance between motor shafts). I am mounting in the Default Flight-Controller Orientation (aka AHRS_ORIENTATION,0 ). The RX radio now (foam taped) on it's side, in only other place it will fit.

 Accelerometer/Gyro chip (on top PCB, inside FC) should be right under edge of red "O" in NOVA label, specifically market by dot on nylon tie-strap. Another way to measure is ... between Pins 2 & 3, and just past the middle (going in other direction).  

Final placement. Nothing touches each other or any Power-Board component. Everything is loosely tied-down to deter any un-wanted vibrations. Power-Board layout and USB cable further limit possible GPS-Module locations. Specially-cut a gray hard-foam piece to extend platform for GPS-Module, and wedged and foam-tapped in place level. No more RX antenna on outside but one is close to it's old location (but now inside and on opposite side). Near front area is left for FC-RX wiring, and any future upgrades like maybe a SBUS-CPPM converter. Re-mounted 915mhz Telemetry radio back underneath quad where it was before.

Re-mounted Compass/ Magnetometer in a new way with black foam tape and nylon tie-strap (but no hot-glue). Minimum metal screws, and use non-magnetic screw-driver. Simple is good. Red wire is Positive (VCC +3.0v - 3.3v) and Black is GND - verified with meter.

Flight Controller and Radio Wiring
Still using connection method #2 from HobbyKing PnF Setup Video (fairly standard but connecting "Signal Only" as allowed by bridged power rails).
RC Cable 1CH-1 VerticalCH-1 Vertical
RC Cable 2CH 2-4 Horizontal (signals only)CH 2-4 Horizontal (signals only)
RC Cable 3CH 5-7 Horizontal (signals only)CH 5-7 (FC Aux 1-3) Horiz
A-JumperCH 8CH 8 (FC Aux 4)

  • Power from BEC-5v goes to RX first, and then to Flight Controller. FC really gets its 5volts from CH-1 connection.
  • Vertical connections are the standard way, with standard RC-Cable radio plug (Gnd, Power, and Signal).
  • I used "black to the right" when plugging in horizontal "servo-like" RC Cables.
  • A-Jumper is a standard (single-pin) 3inch Arduino Jumper cable
  • While Channel-Signal-Rail is on FC bottom, it's on X8R's top (opposite).
  • By using previously unused SBUS socket for power, frees a standard channel. It shares the same power-rails.
  • At this point, I can't even see a reason to need BEC-5v routed to quad's under-carriage connection panel. If need arises in the future, I can just use a RC Y-cable.
  • This simple wiring works fine again (for now). When and if I get my gimbal, I will switch over to a FrSky SBUS-to-CPPM Converter wiring setup.
Good pic that someone made.
The white boxes are where you connect Dupont (servo like) RC-Cables. Mine is mostly like this pic. Only difference is that I went ahead and connected CH-6 also to RX (in case needed later). Red VCC and Black GND drawn boxes are just informational.
Another good pic someone drew
This is post 1 of 3 for the rebuild. See next post for APM and radio setup and programming.

Sunday, January 4, 2015

Nova Flight - Unstable and WoD

Seems that WobbleOfDeath is back. This flight was at big field and took it up high. Clear day with Winds 5-8mph. Stock GPS is giving unbelievable sats and hdop.

Seems like the Nova has only flown without WoD a couple of times, since it crash-landed in a tree a couple of months ago. It flew so badly today that I only flew one battery through it ... I was afraid it was going to flip over and crash-down hard.

It flies now (only fairly, seems unstable) if I'm in control. It will hover at 4 feet but that and consecutive AutoTrims continue to take a good amount of stick correction. If I try to let it control itself (loiter or RTL) it starts wobbling pretty bad. It's like it doesn't know where level is or is slow to respond. Also, sometimes when it RTLs (wobbling but finally touches the ground) it can't sense it landed and skates across the ground (thank goodness for prop-guards).

I suppose another symptom is that the quad seems over-sensitive to "within limits" vibrations. I have also noticed that "level" slightly drifts away after a time (on the same desk, it's not perfectly level any more a few days later).

In Terminal/ Tests (AC 3.1.x only) I did test GPS, barometer, compass, and ins (accelerometer/gyro). It really hammers the sensors and you can view real-time data. I have a feeling it might also reset them in some way.  I am NOT getting any "Bad Gyro or Accelerometer Health" messages in MP-HUD, so that's good.

Calibrations work fine. When calibrating the Accelerometer/Gyro be sure quad is perpendicular/level and very still (propped-up, not being touched) at each stage. GPS and Compass seem to be working fine.

I just checked the logs from this flight. Auto-Analysis is mostly ok except for some Pitch/Roll/Thrust Fails due to extreme rocking. There was also a Compass Warning about 34% mag-field change. In normal logs, ATT clearly shows WoD happening. If PerformanceMonitor shows high or spiking LongLoops it could be a sign that the processor is being over-burdened. PM shows no NLon over 4 (out of 1000) which is very good. Curr/VCC (5v) is 5.10v~ 5.18v. However, IMU is showing vibrations again. While I previously got X,Y down very low, they are now reaching the Acceptable Limits of  -3~ +3, . Same with Z axis, which is now again over limit of -15~ -5, at -17~ -3.

Edit 2015-01: If you look closely at above graph, you will see that AccX is also beyond acceptable range. X is really -4 ~ +1. I think this is because (at logging time) the FC's gyro/accelerometer chip is not in true middle of quad.

It currently flies so unstable that if I continue to fly it (even with gradual trouble-shooting changes) I'm pretty sure it's going to flip and crash. A couple of times today already, it Rolled and Pitched very close to 90 degrees. I have come to the conclusion that either the Flight Controller is bad, some component is installed fundamentally wrong, or the frame is just to weak to run the motors within acceptable vibration limits. I'm also wondering if that after a bad crash (and the accelerometer and/or gyro gets jolted to a new limit or g-rating) they become more sensitive? Speaking of the body and/or frame ... it's no secret that to keep the aircraft light, the designers selected a body shell that also acts as the frame. It's purposely barely adequate and is in no-way over-built.

I would like to have a stable flying quad so I can move on to autonomous missions, add a gimbal and better AP camera, and other cool things. I have decided to strip the aircraft down ... and rebuild it (in a very simple and basic config) one last time. I need to verify the Flight Controller is good. If it needs a FC or a stronger after-market frame, I'm not sure if I will try to fix it or just strip it for parts and move on to the next model. All that said, it's been a good first GPS-capable quad. I have learned a lot about APM and quads in general this summer and fall.

Edit 2015-03: The rebuild worked great. Here are the results.


Friday, January 2, 2015

Hubsan H107d X4 FPV Mini-Quad received

I got this H107d to fly indoors during the winter. It comes with a fairly nice radio (built-in LCD, SD-card recording, and other cool features). Control is 2.4ghz and Video over 5.8ghz.

It flies good inside with included prop-guard installed. Fairly stable and hovers good. Lots of fun but challenging is such confined spaces (I'm used to flying outside).  I also ordered 5 extra 1s 380mAh Tenergy lipos for it. It comes with a USB charger cable. Battery connector is small and rare, so I ordered a special balance-cable for BangGood to allow me to charge these lipos on my real charger.

See Video-1 and Video-2 and Video-3 and Video-4.