SkEye version 6.8.0l

[Harshad RJ]

All,

I am happy to announce a new release of SkEye. This went through a lot of testing; thanks to all the beta testers!

Release notes:

New feature:
Support for remote sensors. See https://www.reddit.com/r/skeye/wiki/remote-sensors for more details.
Currently, Wiced Sense is supported. Other devices will be supported soon. I need to update the docs.

Other improvements:
* Quick settings for ecliptic opacity, highlight opacity.
* Time machine interface can be minimized.
* Ability to jump to a month / year in the time machine.

​About remote sensors​

I had a brief but good run while observing with a remote sensor (Wiced Sense) and latest SkEye. I will write a report in more detail shortly, along with detailed documentation.

​About calibration:​

I have learnt a few things in the previous week, which is applicable to both remote sensors and local Android sensors:

• You need to spend a couple of minutes in calibrating the sensors, before using the device as a GoTo guide.
  
• To calibrate the accelerometer you need to move the device around very gently. The device needs to be tumbled and rolled on its sides. There is no need to rotate it around the azimuth. This is best done with the device held in hand.
  
• To calibrate the magnetometer, the device needs to be in the exact spot where it will be used, and then needs to be rotated such that it points in as many directions as possible on the celestial sphere. This is best done with the device attached to the telescope / binoculars / etc.
• The figure-8 calibration procedure that I used to recommend was very crude. It is good enough for quick hand-held use but not for GoTo purposes.
  

The above ritual of calibration is a little cumbersome, but it makes a huge difference in accuracy.

​cheers,​

--

Harshad RJ

[Sami Jumppanen]

Hi,

this is great news!

I've followed some of the discussion, but haven't had time to do anything, not going to get external sensor,  maybe not even spend time with sky at all. But the magnetometer calibration mentioned here in the end is something I've recently done, for my  quadrocopter DJI Phantom 1.

I've experienced the importance of proper compass calibration. There's always been only one calibration method (as the firmware supports it, but it also gives good results): rotate the drone horizontally, around its Z axis 360 degrees once. Then hold it nose down (vertically) and rotate it 360 degrees. The indicator light tells you when you've turned it enough.

There is no detailed instructions on how gently you need to do that, but the environment counts. No metal structures nearby, electric cables in the ground, and no phones close to the compass! A strong magnet can even make the compass not calibrating at all, in which case it needs to be demagnetized - with a strong magnet, by moving the magnet all around the magnetometer several rounds.

There are no numbers available for calibration success, it's only seen when you fly the drone. If it doesn't hover in place but does elliptical circling, calibration should be done again. You can even lose the control with bad compass behavior.

Anyway, though to write in case the two axis calibration could work. The figure 8 is actually trying to achieve that, but it's inaccurate.

Thanks.

Sami

--
You received this message because you are subscribed to the Google Groups "SkEye Discussions + Announcements" group.
Visit this group at https://groups.google.com/group/skeye.

[Harshad RJ]

Hi Sami,

Indeed, the quadcopter / drone community is very active in making use of IMU sensors. I found a number of ideas for calibration from their forums!

Based on your description, it appears that your quadcopter (DJI Phantom 1) has partial calibration done at the factory. This probably includes accelerometer and hard-iron calibration for magnetometer. Only some part of the calibration (soft-iron calibration) is apparently done by user.

This is fine with an integrated product like the quadcopter. Android devices also have something similar for the local sensors, but we can't know for sure. It is not documented and might vary for each Android model.

In case of SkEye+remote sensors we have more control and it is better to perform all calibrations by the user.

The pros and cons of SkEye's remote sensor approach are:

• Pro: Is more accurate
  
• Pro: Does not require demagnetizing
• Con: Requires slightly more time to do the complete calibration

cheers,

-- 

Harshad RJ

[Sami Jumppanen]

Hi,

thanks - the situation is similar; the compass unit is extrenal, located in the plastic landing gear, to be out of the electronic interference. Demagnetizing is not needed, unless the unit is badly saturated. Never happened yet.

The assistant software on computer performs basic or advanced (longer) calibration for accelerometer upon request, and shows the values and their modulo for judging if the values are ok. It's funny how the assistant software pop-ups tell actual recovery procedures and detailed info, while the user manuals / wiki do not :D

I've crashed once badly (on my first flight, because the compass was not calibrated, I just forgot it!), and the IMU went "bad". After a long examination and web searching the only thing I could try is to take out the flight controller module and hit it hard sideways onto the table. I hit it twice. Then installed back, checked the assistant software, it could calibrate now and I got perfect readings! I'm guessing that the crash can magnetize the accelerometer (or there is just electrostatic binding to the "wall" in the micromechanic sensor), so it doesn't give valid readings. The accelerometer chip is not really replaceable, as it's cast into resin or epoxy within a lead cube. The lead cube and foam cushioning helps to absorb vibrations from the drone, and thus give more stable readings. But this is off topic here, sorry :)

The manual compass calibration takes approximately one minute. The point must be getting the min/max on every axis. The firmware records those extremes and knows the scale. What I don't know is how linear the compass is, does it "pull" towards certain number of sectors or not, and how quickly it will settle (how slowly must I turn around). Turning around just once feels a bit sketchy, IMHO, it cannot rule out unexpected fluctuations in the magnetic field. So if you've found good methods, it's great! Too bad the drone doesn't give choises for the calibration.

Sami

--

You received this message because you are subscribed to the Google Groups "SkEye Discussions + Announcements" group.
Visit this group at https://groups.google.com/group/skeye.

--

Sami Jumppanen
http://netti.nic.fi/some-e

[Harshad RJ]

On Thu, Sep 15, 2016 at 1:53 PM, Sami Jumppanen <sami.ju...@gmail.com> wrote:

After a long examination and web searching the only thing I could try is to take out the flight controller module and hit it hard sideways onto the table. I hit it twice. Then installed back, checked the assistant software, it could calibrate now and I got perfect readings!

​Ha ha.. Next time I see a bug I will remember to hit my Android on the side till it behaves.​

 

The manual compass calibration takes approximately one minute. The point must be getting the min/max on every axis. The firmware records those extremes and knows the scale. What I don't know is how linear the compass is, does it "pull" towards certain number of sectors or not, and how quickly it will settle (how slowly must I turn around).

​The gentle movement is for the accelerometer calibration.. If you turn too fast, the accelerometer will record the acceleration due to movement. We want it to only record the acceleration due to gravity.

For both accelerometer and the magnetometer​

​, the distortion in readings is ellipsoidal in shape. It is possible, but risky, to estimate the shape from a few readings only. I am assuming that the drone maker is estimating and storing some parameters during assembly. ​

 

--

Harshad RJ

[Sami Jumppanen]

​The gentle movement is for the accelerometer calibration.. If you turn too fast, the accelerometer will record the acceleration due to movement. We want it to only record the acceleration due to gravity.

For drones, there is no accelerometer involved in compass calibration. And so it should be in every application. In the final use you have compass and accelerometer / gyro on co-operation to get accurate and smooth output regardless of the movement. Accelerometer is calibrated on level surface, not touching the drone.

There is one thing: the drone does not need to know the true north, just the field strength around X and Y axis. It does record the nose direction on start-up, and can use it for determining where the home is, but doesn't know about the real world, and it's not used if you fly in normal control mode (relative to the nose). Maybe auto-return uses it, I haven't made observations regarding that situation.

In telescope guidance use you need accurate absolute directions. But it's maybe not a key point, in drones also you need evenly distributed values (but not on Z axis?).

[Harshad RJ]

Sami,

Yeah, what you say makes perfect sense for drones. Thanks.

I won't reply further on the list since in this thread I would like to focus on droids rather than drones :)

cheers,

--

Harshad RJ