P4 Altitude Accuracy

[mervn]

Hi All

I have my P4 for about a month now and have an and have previously flown several other drones (yed I flew my Bebop in to the water where it could not be recovered).

I have covered almost 40 Kilometers in seven and a half hours over 53 Flights in this time with the Phantom 4

On a fifteen minute flight this morning covering 2.2 Kilometers at varying heights upto 56 Meters i noticed the Altitude indicator showing over 5 mtrs when less than half a meter from the ground at the same spot it took off from.

Are other experiencing the same issue or is this something I should be having checked out with tech support?

Best regards

Merv

PS I had over 5.5 hours on My Bebop before drowning it have have over an hours flight experience with a Bebop 2

 

[GMack]

Mine varies as well by 15 meters at times. Don't know if it is the drone or the GO software. Last software update seems a bit better, but still annoying.

I did a play with Autopilot flying a large box pattern at the same height throughout and the outgoing leg was at the correct height, but the incoming height was 3-4 meters higher. When it flew out again on the outgoing leg it was back down to the correct height, but went high on the return again.

I don't know why two directions would cause the altitude difference, but it did. Autopilot added some height tolerance settings in the last version last week I need to check out. GO is all over the place in numbers and I don't believe any of them.

 

[DocDiggler]

Yep same problem here no matter if i use litchi or dji go
I think its a hardware issue


Sent from my iPhone using PhantomPilots

 

[GMack]

Yep same problem here no matter if i use litchi or dji go
I think its a hardware issue


Sent from my iPhone using PhantomPilots

Additional info on Autopilot:

I found out that Autopilot hid the new altitude and distance accuracy in their Settings menu way at the bottom. The default was 5 meters/16 feet which is what I observed on the difference between an outgoing flight path and the return even though both were set at 36 feet (Outgoing 36 feet. Return ~ 50 feet.). I reset it to the minimum of 1 meter/3 feet and will try it again and see if it tightens it up.

GO is all over the place a no where to fine tune the altitude settings or know what they are. I don't own Litchi and have no knowledge on their accuracy and if it can be set.

 

[sar104]

Your expectations for barometrically-deterimined altitude are unrealistic for a mobile pressure sensor on a moving object that significantly disturbs the surrounding air. A cubic meter of air at sea level weighs a little over 1 kg, and so the change in pressure per meter of altitude at sea level is of the order of 10 Pa. Atmospheric pressure is 100 kPa at sea level, and since barometric pressure sensors are required to work over a large altitude (and thus pressure) range they measure absolute, rather than gauge, pressure, and so we are looking at a 0.01% change per meter. That's not even resolvable with a typical 12-bit (4096-level) absolute sensor, which can only resolve down to a few meters at sea level. There are sensors available that combine greater bit depth (up to 24 bit) and claim resolution of 20 cm at sea level, but that is only in ideal, undisturbed conditions, which we don't have.

A Phantom weighs about 1.5 kg, and has a wing span of 0.5 m. A very simplistic application of Newton's 2nd law suggests that to support it in flight, the pressure just below the rotors has to be of the order of 60 Pa higher than the pressure above them - equivalent to a 6 m change in altitude. If those were constant states, and fixed relative to actual ambient atmospheric pressure, then it would not be a problem to compensate. But the airflow is not laminar, and the resulting turbulence almost certainly means that all kinds of software signal conditioning is in use to try to back out the real ambient pressure.

Overall it continues to amaze me just how well these sensors perform.

 

[RémyHeli]

I've been flying in tandem at see level with a friend who has a Q500 and we both started from same point and did very similar flight route. After 10 min. we both did stationary at 3m high and mine was showing 4.5 whereas his 3.0.

My friend was surprised about this and told me he never experienced such issue (or noticed I would say). Nevertheless I don't get why such difference in very similar conditions.

I have already posted this experience in this forum but never got explanation why one could achieve such acuracy but the other not.

Wondering also when flying two P4 with same settings and firmware/software and similar flight route if ee shall we expect same deviation or not necessary?


Sent from my iPhone using PhantomPilots

 

[Richard R]

Your expectations for barometrically-deterimined altitude are unrealistic for a mobile pressure sensor on a moving object that significantly disturbs the surrounding air. A cubic meter of air at sea level weighs a little over 1 kg, and so the change in pressure per meter of altitude at sea level is of the order of 10 Pa. Atmospheric pressure is 100 kPa at sea level, and since barometric pressure sensors are required to work over a large altitude (and thus pressure) range they measure absolute, rather than gauge, pressure, and so we are looking at a 0.01% change per meter. That's not even resolvable with a typical 12-bit (4096-level) absolute sensor, which can only resolve down to a few meters at sea level. There are sensors available that combine greater bit depth (up to 24 bit) and claim resolution of 20 cm at sea level, but that is only in ideal, undisturbed conditions, which we don't have.

A Phantom weighs about 1.5 kg, and has a wing span of 0.5 m. A very simplistic application of Newton's 2nd law suggests that to support it in flight, the pressure just below the rotors has to be of the order of 60 Pa higher than the pressure above them - equivalent to a 6 m change in altitude. If those were constant states, and fixed relative to actual ambient atmospheric pressure, then it would not be a problem to compensate. But the airflow is not laminar, and the resulting turbulence almost certainly means that all kinds of software signal conditioning is in use to try to back out the real ambient pressure.

Overall it continues to amaze me just how well these sensors perform.

Don't you just love physics? From your location, I might guess where you work.

 

[Richard R]

I've been flying in tandem at see level with a friend who has a Q500 and we both started from same point and did very similar flight route. After 10 min. we both did stationary at 3m high and mine was showing 4.5 whereas his 3.0.

My friend was surprised about this and told me he never experienced such issue (or noticed I would say). Nevertheless I don't get why such difference in very similar conditions.

I have already posted this experience in this forum but never got explanation why one could achieve such acuracy but the other not.

Wondering also when flying two P4 with same settings and firmware/software and similar flight route if ee shall we expect same deviation or not necessary?


Sent from my iPhone using PhantomPilots

You will always get some variation from system to system. Resistors, caps even the baro sensor all have manufacturing tolerances. Since DJI isn't producing scientific instruments, I would expect that they specify component tolerances tight enough that the circuits can be assembled with a minimum of adjustment, but loose enough to keep the costs down. A net tolerance of +- 10% wouldn't be uncharacteristic. And comparisons between uncompensated systems from 2 different manufacturers could easily have the differences you found in your tests. Reported heights (altitude) should always be considered an approximation. Where it matters, give yourself some margin.

 

[sar104]

Don't you just love physics? From your location, I might guess where you work.

What I love about it is the way that one can apply really basic physics to so many phenomena to get significant insight into how they work. I'm sure that your guess would be correct.

 

[GMack]

You will always get some variation from system to system. Resistors, caps even the baro sensor all have manufacturing tolerances. Since DJI isn't producing scientific instruments, I would expect that they specify component tolerances tight enough that the circuits can be assembled with a minimum of adjustment, but loose enough to keep the costs down. A net tolerance of +- 10% wouldn't be uncharacteristic. And comparisons between uncompensated systems from 2 different manufacturers could easily have the differences you found in your tests. Reported heights (altitude) should always be considered an approximation. Where it matters, give yourself some margin.

I cannot understand why DJI cannot with using all the devices on the drone, e.g. Barometer, GPS, VPS, and its sonar, and still it cannot land and tell me the ground altitude is zero feet as it was on takeoff.

Instead I see sometimes -50 feet upon landing. Really is a badly engineered system against a cheap $20 sonar tape measure that can get within a few inches of it pointing at the ground (Of which the P4 has sonar too, but am mystified why it isn't used in conjunction with everything else and say "Hey I've touched down, and the landing spot AGL is zero feet!" - and GO not displaying -50 feet or flying around and it showing -10 feet underground.).

 

[Frank Pliska III]

Well I'm totally confused now. My son is a commercial pilot and when I fly with him one of the most important pre flight checks is to calibrate his altimeter because the barometric pressure changes day by day. I always thought the Phantom performed its own altitude calibration at take off when the GO app tells you that your the home point is set.
So here's one for you all.!!! What will happen if I hit RTH and the altitude is indicating -50' upon landing. NOT GOOD. P3S

 

[sar104]

Well I'm totally confused now. My son is a commercial pilot and when I fly with him one of the most important pre flight checks is to calibrate his altimeter because the barometric pressure changes day by day. I always thought the Phantom performed its own altitude calibration at take off when the GO app tells you that your the home point is set.
So here's one for you all.!!! What will happen if I hit RTH and the altitude is indicating -50' upon landing. NOT GOOD. P3S

There is no altitude calibration on a Phantom that would be equivalent to setting the altimeter on an aircraft, but it isn't necessary because it only cares about elevation relative to the home point.

In terms of RTH, the only problem that might arise if the reported altitude is wrong is that the RTH altitude could be incorrect enough to cause the aircraft to hit an obstacle that it would otherwise have cleared. But that example requires an opposite error to the one that you mention - it would have to be reporting high, not low. Reporting low has no detrimental effect on RTH - the aircraft will simply return at a higher altitude than programmed and then descend until it lands.

 

[Frank Pliska III]

There is no altitude calibration on a Phantom that would be equivalent to setting the altimeter on an aircraft, but it isn't necessary because it only cares about elevation relative to the home point.

In terms of RTH, the only problem that might arise if the reported altitude is wrong is that the RTH altitude could be incorrect enough to cause the aircraft to hit an obstacle that it would otherwise have cleared. But that example requires an opposite error to the one that you mention - it would have to be reporting high, not low. Reporting low has no detrimental effect on RTH - the aircraft will simply return at a higher altitude than programmed and then descend until it lands.

 

[Frank Pliska III]

So what altitude indication does the AC get telling it is about to land and not to HARD land itself during a RTH?
If the altitude indicates its 10' when the AC is 6" off the ground during a RTH what keeps it from a hard landing?

 

[GMack]

So what altitude indication does the AC get telling it is about to land and not to HARD land itself during a RTH?
If the altitude indicates its 10' when the AC is 6" off the ground during a RTH what keeps it from a hard landing?

Good question.

I manually land as the altitude is always off on my landing spot verses take-off spot even if it is the same spot. I don't trust the numbers or the drone itself at times as it wavers in height too much. Allegedly, the underbelly VPS and sonar should help in softening the landing, but if GO is saying -50 feet who knows how they programmed it with RTH or Auto-Landing? Some have reported the thing hopping up and tipping over on landing too, along with those leg covers popping off.

Really is a crummy design they've made with all the on board tech that can't tell the same approaching and landing spot is zero feet or what it started with.

 

[sar104]

So what altitude indication does the AC get telling it is about to land and not to HARD land itself during a RTH?
If the altitude indicates its 10' when the AC is 6" off the ground during a RTH what keeps it from a hard landing?

VPS.

 

[RémyHeli]

True, VPS. So beware VPS limitations you should know and follow. Main ones are mention in user manual, have a look if unsure.

VPS will not override barometer information, wather correct or wrong, if he's not able to figure out what does the craft (done thru optical sensors scanning and patterns recognition), i.e over frozen surface or during the night.

 

[Crack The Sky]

Your expectations for barometrically-deterimined altitude are unrealistic for a mobile pressure sensor on a moving object that significantly disturbs the surrounding air. A cubic meter of air at sea level weighs a little over 1 kg, and so the change in pressure per meter of altitude at sea level is of the order of 10 Pa. Atmospheric pressure is 100 kPa at sea level, and since barometric pressure sensors are required to work over a large altitude (and thus pressure) range they measure absolute, rather than gauge, pressure, and so we are looking at a 0.01% change per meter. That's not even resolvable with a typical 12-bit (4096-level) absolute sensor, which can only resolve down to a few meters at sea level. There are sensors available that combine greater bit depth (up to 24 bit) and claim resolution of 20 cm at sea level, but that is only in ideal, undisturbed conditions, which we don't have.

A Phantom weighs about 1.5 kg, and has a wing span of 0.5 m. A very simplistic application of Newton's 2nd law suggests that to support it in flight, the pressure just below the rotors has to be of the order of 60 Pa higher than the pressure above them - equivalent to a 6 m change in altitude. If those were constant states, and fixed relative to actual ambient atmospheric pressure, then it would not be a problem to compensate. But the airflow is not laminar, and the resulting turbulence almost certainly means that all kinds of software signal conditioning is in use to try to back out the real ambient pressure.

Overall it continues to amaze me just how well these sensors perform.

Well stated! I could not have said it better, myself.......

 

[Fester]

During an auto landing on RTH, I too have thought the touch down speed a bit high, particularly onto a hard (i.e. concrete, etc.) surface. Not too bad onto grass or landing pad, but a little much when we can do a feather soft manual landing. Is the P4 designed to reduce descent rate as it nears the touchdown? AFAIK thats not a rate we can change. The VPS system is pretty remarkable and I've researched it a bit, but for the most part all I can tell is that the VPS is a "Myriad 2" visual processor made by Movidus. They make the chip, but DJI writes the code. I like to learn as much about my flight hardware as possible, but the details we'd like to know are probably "Proprietary", and if they told us, they'd have to kill us....

 

[wa5tef]

Yep, I don't trust the altitude readings at all. I always land manually because the auto landing is way to fast for me to be comfortable with it.
Jim
WA5TEF