Phantom 4 keeps loosing height

[rene van der meer]

My quad looses height just after lift off. I normally go straight to 4-5 m.
The VPS gives the correct height, but the barometric height can be off a few m.
This is just after lift off. The (barometric) height is 4.2 m. The VPS shows 3.8m.

Two minutes later, after moving the quad, the height is still 4.3 m, but the VPS shows the correct 0.8 m.
The difference is almost 3 m/10 feet.

I have calibrated all sensors. Today the weather is stable and there is not much wind.
I have seen other posts about this effect, but this seems worse.

Is this normal?

 

[Tophertello]

I had a similar problem when I first started flying but a full system
calibration seemed to have done the trick for me no more losing height
sorry I can't help other than that..... you could try it a few times as I have found the changes that I make don't always take first time around.

 

[Erised]

The VPS calibration is easy to screw up, I know my first calibration messed up my bird. Here are a few pointers if you wish. Make sure you have no reflection on your monitor, when you hold your bird in front of the screen hold it perfectly horizontal and make sure the cameras are perfectly on the center lines of the screen. Assistant will ask you to move back until the squares line up and are the same size, you are now aligned with your monitor on a XYZ axis. Then assistant will ask you to follow the pattern, DO NOT MOVE the bird on the XYZ axis! only point the cameras to follow the patern, do the same vertical. The first time I did a calibration I had moved the bird arround and boy my bird was all aver the place. I hope this may help.

 

[alokbhargava]

Do not go by the readings shown at low heights. The purpose of VPS is to make drone stable at low heights and if you are getting it, be happy.

 

[rene van der meer]

Erised, thanks for the reply, but the VPS shows the correct value. The quad just keeps descending until it is close to the ground.

 

[rene van der meer]

Do not go by the readings shown at low heights. The purpose of VPS is to make drone stable at low heights and if you are getting it, be happy.

Thanks, but a fews ago I flew a mission with Litchi. The height was 7m, after a while I had to abort because the quad was flying near to the ground. How can I solve that?

 

[GMack]

Thanks, but a fews ago I flew a mission with Litchi. The height was 7m, after a while I had to abort because the quad was flying near to the ground. How can I solve that?

I honestly don't think you can solve that.

I fly with Autopilot and they have a control to set the altitude tighter to about 15 feet at minimum. Default was 45 feet so I tried 15 feet. No telling what the default tolerance altitude is with DJI GO.

I flew around a football field in a box using Autopilot and noticed that the outgoing leg was fine at the top of the goal posts, but the return was about 15 feet higher than the posts. When it flew out again it was correct, but the return it went up again. Puzzling!

Only thing I can think of was maybe as it flew into the low sun it climbed due to the VPS system. VPS into the sun has caused me some odd stair-stepping in the past. That or some barometric irregularities with the direction of flight and maybe wind in the air-frame messing with the barometer. I dunno.

Mine will often show it flying underground too on the screen (-45 feet and landing the same.). I just mark it up to a toy and not some military grade or surveying aircraft. Some days it is better, and other days it is way off. If I think some calibration has fixed it, next day it proves me wrong.

 

[DanMan32]

I was just watching this old house where they used what essentially is an altimeter able to resolve to inches. They used it to determine how level the house was, with a horizontal brick line as the reference. Point being, small scale accuracy is possible with barometric readings. Of course that equipment is probably expensive.
I noticed in the photo that the ground seems to have a slope. Is that screenshot taken at takeoff point? If not then barometric height will be different than VPS since the barometric height is relative to takeoff altitude. VPS is actual height from ground directly below the AC.

 

[MedicFL1]

My quad looses height just after lift off. I normally go straight to 4-5 m.
The VPS gives the correct height, but the barometric height can be off a few m.
This is just after lift off. The (barometric) height is 4.2 m. The VPS shows 3.8m.
View attachment 79571
Two minutes later, after moving the quad, the height is still 4.3 m, but the VPS shows the correct 0.8 m.
The difference is almost 3 m/10 feet.
View attachment 79572
I have calibrated all sensors. Today the weather is stable and there is not much wind.
I have seen other posts about this effect, but this seems worse.

Is this normal?

Some variation I find is normal.
Just a quick, obvious question; Sensors are clean and you're not flying with a gimbal guard on, right?
I find that my Polar Pro Gimbal Guard may interfere with VPS ~ sometimes.

 

[rene van der meer]

Some variation I find is normal.
Just a quick, obvious question; Sensors are clean and you're not flying with a gimbal guard on, right?
I find that my Polar Pro Gimbal Guard may interfere with VPS ~ sometimes.

I don't use a gimbal guard.

 

[Deleted member 62848]

My quad looses height just after lift off. I normally go straight to 4-5 m.
The VPS gives the correct height, but the barometric height can be off a few m.
This is just after lift off. The (barometric) height is 4.2 m. The VPS shows 3.8m.
View attachment 79571
Two minutes later, after moving the quad, the height is still 4.3 m, but the VPS shows the correct 0.8 m.
The difference is almost 3 m/10 feet.
View attachment 79572
I have calibrated all sensors. Today the weather is stable and there is not much wind.
I have seen other posts about this effect, but this seems worse.

Is this normal?

HI not sure but if u turn your noise reduction off those zebra lines will go ,,,,

 

[Richard R]

HI not sure but if u turn your noise reduction off those zebra lines will go ,,,,

No, believe that the zebra lines are overexposure warning. Can be turned off, but shouldn't have any affect on the issue that the OP is dealing with.

 

[rene van der meer]

Yep, I have overexposure (zebra's) turned on.

 

[rene van der meer]

I just wonder how common is a drop of a few for a P4?
Me previous quad was a P3. I remember some variations, but not so much.

 

[Quamera]

I was just watching this old house where they used what essentially is an altimeter able to resolve to inches. They used it to determine how level the house was, with a horizontal brick line as the reference. Point being, small scale accuracy is possible with barometric readings. Of course that equipment is probably expensive.

I find it hard to believe that a barometric altimeter could measure inch variations in height, just normal variations in atmospheric pressure would make that impractical. A conventional surveying level would do the job as well as anything or there maybe an electronic version of a clear hose filled with water available these days if no one wanted to use the original version.

 

[Deleted member 62848]

Yep, I have overexposure (zebra's) turned on.

No, believe that the zebra lines are ove

Yep, I have overexposure (zebra's) turned on.[exposure warning. Can be turned off, but shouldn't have any affect on the issue that the ,,,,,,,,,,hi was saying if u turn off no lines,,,don't know why u losing height so didn't coment

 

[dwallersv]

Erised, thanks for the reply, but the VPS shows the correct value. The quad just keeps descending until it is close to the ground.

With a few very specific exceptions*, VPS altitude will be very accurate, and when it's available (lower than 10m/33ft), you can simply ignore the barometric altimeter.

Variation in baro altitude of 20-40 ft over 20 minutes is well within the expected accuracy of this type of altitude measurement.

What's puzzling to me is why the DJI programming doesn't simply ignore the baro sensor when the ultrasonic sensor is available and in range, then switch back smoothly to the baro sensor with a corrected altitude if it ascends again.

One way to estimate whether you will be having stable, accurate altitudes with the barometric altimeter, or squirrelly, less accurate readings is simply to read the wind for a few minutes.

The windier it is the greater of a pressure gradient you are experiencing in the atmosphere, and this is moving over your geography. As it does, the ambient barometric pressure changes. The steeper the gradient, the faster pressure will change as this weather cell moves over you.

So, expect less accuracy in the barometric altimeter the windier it is. It's not a guarantee, but it's generally a very good correlation. Also, the steadier the wind is, the more certain it's being driven by a pressure gradient that's going to change your situation while you're up in the air.
----------------------------------------

*Soft or uneven surfaces that do not reflect sound waves well will not result in accurate ultrasonic returns. Stone is great; bushes, not so much.

 

[dwallersv]

Thanks, but a fews ago I flew a mission with Litchi. The height was 7m, after a while I had to abort because the quad was flying near to the ground. How can I solve that?

Do the ground elevation vary at all? If more than a meter or two, you'll want to create a ground-relative mission so that the waypoint heights are 7m above the ground where they are located, rather than 7m above the elevation of your starting point.

Even using that feature in Litchi we still can totally rely upon it to get altitudes right that close to the ground. I'd say, based on my experience with ground-following missions (something I've been spending much effort on lately, coincidentally), the error bars for setting waypoint altitudes using the terrain data from Google Earth (which is what Litchi does) is on the order of +/- 15m/50ft. So don't RELY on altitudes from mission planning using the GE data as any more accurate than that, especially where there are a lot of "soft" ground features like trees and foliage.

This means you need to fly the mission FPV first and make any adjustments to waypoints as needed to get them right. Now, that said, my experience is that, although sometimes the altitudes from GE are off as much as 50', that's rare. Most of the time the errors are within +/-10', especially if the terrain is simple, like grass or small brush. Water tends to be pretty accurate too.

 

[dwallersv]

I find it hard to believe that a barometric altimeter could measure inch variations in height, just normal variations in atmospheric pressure would make that impractical. A conventional surveying level would do the job as well as anything or there maybe an electronic version of a clear hose filled with water available these days if no one wanted to use the original version.

Well actually very cheap barometric sensors are capable of accuracy within inches. That's what's providing altitude-hold capability on $25 toy drones. Such measurement is accurate under very specific conditions, however, and is generally unreliable for anything but short-term resolution (no more than a minute or two). Beyond that, barometric pressure variations over such timescales -- and longer -- blow out any accuracy that small because the errors are much larger then.

 

[rene van der meer]

Do the ground elevation vary at all? If more than a meter or two, you'll want to create a ground-relative mission so that the waypoint heights are 7m above the ground where they are located, rather than 7m above the elevation of your starting point.

Even using that feature in Litchi we still can totally rely upon it to get altitudes right that close to the ground. I'd say, based on my experience with ground-following missions (something I've been spending much effort on lately, coincidentally), the error bars for setting waypoint altitudes using the terrain data from Google Earth (which is what Litchi does) is on the order of +/- 15m/50ft. So don't RELY on altitudes from mission planning using the GE data as any more accurate than that, especially where there are a lot of "soft" ground features like trees and foliage.

This means you need to fly the mission FPV first and make any adjustments to waypoints as needed to get them right. Now, that said, my experience is that, although sometimes the altitudes from GE are off as much as 50', that's rare. Most of the time the errors are within +/-10', especially if the terrain is simple, like grass or small brush. Water tends to be pretty accurate too.

I normally explore complex or hazardous missions first. I use the C1 (waypoint) and C2 keys (POI) to plan the mission. Then I can fly the quad along these stored waypoints. I saw that the position is quite accurate, but the height not and had to abort the mission a few time to prevent the quad hitting the ground.