This thread is privately moderated by billpa, who may elect to delete unwanted replies.
Aug 02, 2012, 06:40 PM
Rick
United States, CA, Oceanside
Joined Mar 2011
2,814 Posts
Quote:
 Originally Posted by Galand Ofcourse it has an ABSOLUTE way of telling which way is up. Gravity. Same as your iPad or iPhone turning the picture right side up. Differential accelerometer sensing. No need to "remember" it senses continuously.
From Silverexpress' first reference:
Quote:
 Accelerometer As named,an accelerometer is used to measure the acceleration. But what we are interested in is the tilt angle of a robot and what has it to do with the acceleration? The answer is gravity. The earth’s gravity is a constant acceleration where the force is always pointing down to the centre of the Earth. When the accelerometer is parallel with the gravity, the measured acceleration will be 1G, when the accelerometer is perpendicular with the gravity, it will measure 0G. In short, the tilt angle can be calculated from the measured acceleration by using this equation: θ = sin-1 (Measured Acceleration / Gravity Acceleration) Unfortunately, this theory can only be applied when the robot is completely static. If the robot is moving, there will be other components of acceleration acting on the robot and causing the calculated tilt angle to be inaccurate. The solution for this problem is low-pass filter the data from the accelerometer. Typically, the acceleration components causing by the dynamic movement of the robot only happens in a short period of time, while the gravity acceleration is acting on the robot permanently. By low-pass filtering the data from the accelerometer, we can filter out the unwanted high frequency acceleration components and we are left with only the gravity acceleration which can be used to calculate the tilt angle of the robot. However, the low-pass filter will increase the latency and slow down the response time of the measurement as shown in the graph below. That’s where the gyro comes in.
My interpretation; an algorithm is needed to blend and filter sensor data to calculate "level". An accelerometer (nor any other existing sensor) cannot detect up or down directly when placed in a moving object.

Here's a thought experiment:

You have a Guardian in a plane circling in a stable, coordinated turn at constant altitude and 60 deg bank angle (2Gs). The Guardian is off. Not just off "mode", it is powered off. Plane is still circling. You flip a switch to turn a relay on and power the Guardian. Plane is still circling while Guardian initializes in off mode. Now activate 2D mode and center the sticks. What happens?

I would expect Guardian to level the pitch (already level), and straighten out the path with the rudder, but I expect it to be confused on bank angle at least for a moment.

Of course you can't actually do this because Guardian has to be powered to pass through the servo commands while establishing the circle flight. That's why it is a thought experiment.
Last edited by choochoo22; Aug 02, 2012 at 06:47 PM.
Aug 02, 2012, 07:06 PM
Registered User
Chicago, USA and Manila, Philippines
Joined May 2007
193 Posts
Melnic, did you have a chance to do a microscopic check which chip they're using? If Invensense that is, it is possible the 6axis accelor/mems combo which is cost effective and efficient.. I think the bulk price is around \$7 each and probably lower at this time due to increase usage of it thru our smarthphone,pads,etc. I believe the Guardian 2d/3d only cost less than \$10 bucks to manufactured and that includes labor. Of course R and D is not included ..

Btw, back to Guardian, they could have package BETTER for consumers just like how the \$20 lipos that is package in a box with foam/bubble wrap (Mr. Bill Pa, are you with us?)

Quote:
 Originally Posted by Melnic Floater,thx for the link but as a person highly interested in gyros, I Indeed care about the brand of chip. I"ll agree that most others won't need to know. Invensense makes it easier on the developer and thus to date, the majority of those boards that pick one of them become highly successful. The BeastX Microbeast was one of the first 3axis heli gyros to use 2 of the earlier ones and the results were very impressive. There are some other chips that are mems but not as developer friendly as the Invensense.
Aug 02, 2012, 07:52 PM
ancora imparo
Melbourne, Australia
Joined Jul 2005
9,871 Posts
Silver agree with that. There is a big difference between assembling some components on a board and writing decent software to make it work in a sophisticated manner.

Mainly wanted to say a big thanks for the links to reliable references on how the various components in this little wonder work. I found more useful info in the two year old Invensense presentation and the other stuff on their web site than most of the (mis)informed speculation on the forums. I learned a lot of good stuff.

Thanks again.
Quote:
 Originally Posted by Silverexpress I agree that Eagletree has done an awesome job on this Guardian. For \$80 we can spend more time enjoying and flying our models than sitting in front of a laptop crunching code.
Aug 02, 2012, 07:56 PM
Registered User
Canton, Michigan USA
Joined Jul 2007
17,025 Posts
2nd Guardian maiden tonight mounted to scratchbuilt delta elevon plane with a big rudder. This plane has KFm airfoil and it's made out of Dow fanfold foam. It's a good flyer without the Guardian. Anyway, enough on the plane. What's important is the CG location was a viable location for mounting the Guadian so I mounted it 5.5 inches behind CG. The Guardian is mounted upside down and 5.5 inches behind the CG. See pictures of the blue bottom of the plane. Not being on the CG seems to be no problem. I also added a piece of 3mm CF to the back of the slot to stiffen the area which now is home for the Guardian.

Set up was the same as my other delta plane which had a Guardian installed. However, this time I turned on the coordinated turn function. More later.

The maiden (no video, sorry) went well for a first flight of a plane that's a son of gun to set level. In 2D mode, the plane kept climbing so I screwed up setting up level flight so I landed and set it up again. 2nd flight it was better but still it tended to climb. Throws were to low because I hadn't flown this one for 3 years so I dialed them back. Anyway, the flight went well but I didn't like the coordinated turn mode. It's shut off now. It felt clumsy to me.

# Images

 Aug 02, 2012, 08:02 PM Southern Pride Haralson County GA. USA Joined Oct 2004 35,074 Posts It is so much easier to trim in flight and then just Reset Trims on the ground thasn to reset Level Flight and Trim but some seem to enjoy the challange opf getting the aircraft at the exact same attitude on the ground that it in fact flys level at.
 Aug 02, 2012, 08:19 PM Registered User Joined May 2010 1,297 Posts I now own three of these things. Set up one on a 60 sized electric slab 3D model, and would have sworn up and down that it was set up correctly. Took off and it went near vertical and started screwing itself course thread, right into a park bench. Pretty much totaled the model. Dunno why I even took off with it on, on others first flight, I would take it up and hit 2D mode and go from there. Dang nabbit!
Aug 02, 2012, 08:32 PM
Registered User
Canton, Michigan USA
Joined Jul 2007
17,025 Posts
Quote:
 Originally Posted by everydayflyer It is so much easier to trim in flight and then just Reset Trims on the ground thasn to reset Level Flight and Trim but some seem to enjoy the challange opf getting the aircraft at the exact same attitude on the ground that it in fact flys level at.
Charles,

I am sensing an air of condescension in your post but I am OK with it.. You make a good point though. The notion of trimming the plane to fly level and then reset trims is not really called out in the 14 page manual. It just says how to reset trims not why to reset trims.

Bill
Aug 02, 2012, 08:40 PM
Registered User
Canton, Michigan USA
Joined Jul 2007
17,025 Posts
Quote:
 Originally Posted by Deadstick 8409 I now own three of these things. Set up one on a 60 sized electric slab 3D model, and would have sworn up and down that it was set up correctly. Took off and it went near vertical and started screwing itself course thread, right into a park bench. Pretty much totaled the model. Dunno why I even took off with it on, on others first flight, I would take it up and hit 2D mode and go from there. Dang nabbit!

Bummer. Sorry about the crash. Can you check your setup after the crash or is it too damaged?

Bill
 Aug 02, 2012, 08:46 PM Team Wack-a-Mole Maryland Joined Oct 2008 8,441 Posts deadstick, you think 2D mode was not level and that messed it up? I'm wondering if one or more of your gyro axis were reversed. Remember control surface moves in the direction you push the wing/tail. that is: Lift right wing up, right aileron moves up. Lift tail up, elevator moves up. Push tail right, rudder moves right.
 Aug 02, 2012, 09:34 PM Registered User Joined May 2010 1,297 Posts Roll was reversed. It will fly again, but not look as pretty. Oh well. Been doing it long enough to know it happens.
Aug 02, 2012, 09:38 PM
Registered User
Canton, Michigan USA
Joined Jul 2007
17,025 Posts
Quote:
 Originally Posted by Deadstick 8409 Roll was reversed. It will fly again, but not look as pretty. Oh well. Been doing it long enough to know it happens.

At least you know why. Next time it will be fine.

Bill
Aug 02, 2012, 11:39 PM
Registered User
Joined Jul 2009
333 Posts
Quote:
 Originally Posted by Silverexpress It sounds like the unit may have come loose from its mounting or it is also possible that it was not getting enough juice (power). You have a 15 second window right after the Guardian comes "On" to make changes to 2D level flight, and trims.
Even after the minor crash landing, the Guardian was still affixed to its Velcro bed right on top of the CG in the fuse, but the power thought may need looking into. I think the built in BEC in the 30amp ESC that PZ puts in the T-28 is a 3amp one but there are also 4 micro servos (non-digital I assume) so I thought this would be sufficient - - but maybe not which would explain why the ground check would be fine with a basically static A/C.

The manual reference to turning the Guardian "on", & yours as well, I have understood to mean when it first receives power from either the onboard RX battery, or in my case, from the 3S2200 Lipo through the ESC. Not when I flip my 3 position Mode switch from off to either 2D or 3D. Right?

Jed
Aug 02, 2012, 11:48 PM
Registered User
Joined Jul 2009
333 Posts
Quote:
 Originally Posted by Prof100 I would check and see if the control surfaces are correcting in the right direction. I have flown planes with gyros and accidentily reversed their direction. Boy was that a surprise when flying.
Prof100 - I'm kind of thinking that somehow, even after I ground checked them, that this may be an issue, because that would maybe explain the really erratic control inputs from the stab unit. I also had the Guardian setting right next to my Hitec Optima 7 RX, & maybe that isn't good, but I'll have to read the manual's fine print again to see if I missed this. Thanks for your help,

Jed
Aug 03, 2012, 12:04 AM
Registered User
Joined Jul 2009
333 Posts
Quote:
 Originally Posted by choochoo22 Try setting the plane on its wheels on a flat surface and setting level. Safest to work on this with prop removed or motor disabled. Turn 2D mode on, the surfaces should stay neutral. Don't touch the transmitter again. Now move the plane carefully on each axis individually and make sure Guardian is correcting each in the proper direction. EG: lift the tail and see if Guardian applies up elevator. Pick the plane up and wave it around to some "unusual" attitude. Now "fly" the plane by manually turning it in the direction each surface is deflected until that surface neutralizes. EG: If you are seeing left aileron, rotate the plane left until the aileron is neutral. Repeat for each axis. The plane should be back in the same level orientation it was when you started.
ChooChoo22 - Many thanks for your insightful observations & suggestions. But for certain, the next time I attempt to set up the Guardian in this T-28, I will not only have the manual right next to the A/C, but I'm also going to follow your recommendations re-stated above. These are so good they should be part of the manual! And the T-28 is all repaired & after a test hop to ensure its airworthiness, I'll be reinstalling the Guardian for another attempt. After all, where would we be if Wilbur & Orville hadn't kept at it! Cheers,

Jed
 Aug 03, 2012, 12:32 AM Electric baptism 1975 Vernon, BC, Canada Joined Dec 2000 3,440 Posts [QUOTE=jedorme;22342190]Even after the minor crash landing, the Guardian was still affixed to its Velcro bed right on top of the CG in the fuse, but the power thought may need looking into. I think the built in BEC in the 30amp ESC that PZ puts in the T-28 is a 3amp one but there are also 4 micro servos (non-digital I assume) so I thought this would be sufficient - - but maybe not which would explain why the ground check would be fine with a basically static A/C. Hmm, I thought 3 amps would be plenty. I have run 5 HS91 on a UBEC for maybe 100 flights no problem. UBEC has 3 amp output.