InspektorGadjet Posted at 2-10 10:22
The Hassleblad concept, but using full 1" sensor + zoom, seems the ultimate combo

I've been telling them that but they don't seem to care.

InspektorGadjet Posted at 2-10 10:29
That is cool to know, I haven't bothered with that but also felt sometimes coming down was too slow.
Will definitely investigate more.

I think it's more fun to fly an FPV quad but it doesn't have RTH capabilities like the DJI drones have but flying w/ DJI assist for non-commercial work or some sort of personal investigation is boring.

grahamjohnson10 Posted at 2-10 13:53
Please disable the autosync before the next test, that way you will capture and keep the DAT file.

Only enable the autosyn once you have exported the DAT file.

I will check the sync from now onwards, thanks for the info!
The graph seem a steady curve and hold of the value.

AntDX316 Posted at 2-10 14:20
I think it's more fun to fly an FPV quad but it doesn't have RTH capabilities like the DJI drones have but flying w/ DJI assist for non-commercial work or some sort of personal investigation is boring.

Yeah, I plan building a racing quad, and buy some small fpv ones, just for fun,  but my main interest is photo and video, not boring at all for me

InspektorGadjet Posted at 2-10 10:29
That is cool to know, I haven't bothered with that but also felt sometimes coming down was too slow.
Will definitely investigate more.

Some people have lost their M2 drones I meant.  As long as you know what to do nothing would go wrong.

InspektorGadjet Posted at 2-10 15:18
Yeah, I plan building a racing quad, and buy some small fpv ones, just for fun,  but my main interest is photo and video, not boring at all for me

Yeah, you'd be amazed at how small an 1108 is from a 2208.  I heard the 2208 is the best for overall balance.

grahamjohnson10 Posted at 2-10 13:53
Please disable the autosync before the next test, that way you will capture and keep the DAT file.

Only enable the autosyn once you have exported the DAT file.

What program do you analyze DAT files with?

Dirty Bird Posted at 2-10 17:24
I'm not sure?  Could an airflow-induced pressure change confuses the flight computer, leading it to falsely conclude it is maintaining altitude when it is actually ascending or descending?

I have never seen a log on this forum showing baro reading wildly off what is visually observed. For low flights reported it is mostly consistent with IR sensor reading. You are right the baro sensor is very sensitive so the drone can tell 1'' of difference in height. Upside of the mainboard houses the baro sensor. It is the circled black square with a tiny hole on the top left corner. Being right next to the battery and microusb port during flight, it is in a very well-shielded position. The heat from the battery makes it very unlikely to be accumulate condensation.

However, it may still be susceptible to airflow caused by outside environment due to the big opening to the battery bay behind the gimbal assembly. The ventilated design was not seen before the mini in the DJI product line I think. The baro sensor in the mini may face "Venturi effect" when an air current try to squeeze past the narrowing gap around the baro sensor, causing reduced air pressure and in turn interpreted by the FC as unwanted climb. The result, drop for altitude compensation. I am merely taking an educated guess but if it is true, some FW tweak to the FC could alleviate the issue.
The climb that is observed in fast forward flight has been explained here as the "transitional lift". The props generate more lift at set RPM once it goes against wind due to leaving turbulent air behind. The FC does not considerer this effect to the mini climbs goes fast against wind. I think it is a very good reason and again something can be addressed in a FW update of the FC.

InspektorGadjet Posted at 2-10 15:18
Yeah, I plan building a racing quad, and buy some small fpv ones, just for fun,  but my main interest is photo and video, not boring at all for me

That IMU error on flip is notmal, it does that whenever the drone is flipped over.

Guorium Posted at 2-11 05:59
What program do you analyze DAT files with?

I use CsvView.

Guorium Posted at 2-11 06:21
I have never seen a log on this forum showing baro reading wildly off what is visually observed. For low flights reported it is mostly consistent with IR sensor reading. You are right the baro sensor is very sensitive so the drone can tell 1'' of difference in height. Upside of the mainboard houses the baro sensor. It is the circled black square with a tiny hole on the top left corner. Being right next to the battery and microusb port during flight, it is in a very well-shielded position. The heat from the battery makes it very unlikely to be accumulate condensation. [view_image]

However, it may still be susceptible to airflow caused by outside environment due to the big opening to the battery bay behind the gimbal assembly. The ventilated design was not seen before the mini in the DJI product line I think. The baro sensor in the mini may face "Venturi Effect" when a air current try to squeeze pass the narrowing gap around the baro sensor, causing reduced air pressure and in turn interpreted by the FC as unwanted climb. The result, drop for altitude compensation. I am merely taking an educated guess but if it is true, some FW tweak to the FC could alleviate the issue.

Very very interesting, thanks for sharing!

Guorium Posted at 2-11 07:07
That IMU error on flip is notmal, it does that whenever the drone is flipped over.

Cool to know, I had that feling but one never know, and wanted to be as clear and truthful about what happened during my flight, I wanted to report that too

grahamjohnson10 Posted at 2-11 07:31
I use CsvView.

Same. I thought you could have something different.

InspektorGadjet Posted at 2-11 07:58
Very very interesting, thanks for sharing!

Or this is the baro sensor just under the top lid located on the top side of ESC board. This is more open than our old candidate being pretty close to the vent holes on the sides of the mini. It says '6407 BA9S13' on it. BA could mean barometer but google does not tell what it is. Sometimes chip makers put an indentation on the chip which could get mixed up with the air inlet of a baro sensor. Without prying open one myself for close inspection I cannot be sure which is the baro exactly but it will have to be one of the two. But still, it is in that grey body shell with many holes that could be susceptible to wind.

Guorium Posted at 2-11 06:21
I have never seen a log on this forum showing baro reading wildly off what is visually observed. For low flights reported it is mostly consistent with IR sensor reading. You are right the baro sensor is very sensitive so the drone can tell 1'' of difference in height. Upside of the mainboard houses the baro sensor. It is the circled black square with a tiny hole on the top left corner. Being right next to the battery and microusb port during flight, it is in a very well-shielded position. The heat from the battery makes it very unlikely to be accumulate condensation. [view_image]

However, it may still be susceptible to airflow caused by outside environment due to the big opening to the battery bay behind the gimbal assembly. The ventilated design was not seen before the mini in the DJI product line I think. The baro sensor in the mini may face "Venturi effect" when an air current try to squeeze past the narrowing gap around the baro sensor, causing reduced air pressure and in turn interpreted by the FC as unwanted climb. The result, drop for altitude compensation. I am merely taking an educated guess but if it is true, some FW tweak to the FC could alleviate the issue.

Love you detailed post.

I totally agreed that air movement around the drone can cause incorrect readings.

So we see lot's of uncommanded height changes with no stick input when there are plenty of GPS at the time which also very accurately sense height.

So why haven't DJI written in the firmware code to detect this.      

grahamjohnson10 Posted at 2-11 17:43
Love you detailed post.

I totally agreed that air movement around the drone can cause incorrect readings.

Maybe because its not cvlear whats the cause. And when you try to solve an problem thats not exactly analyzed things goes worst.

Its not plausible that this is an baro issue, then the readings would be wrong.
For me it happens after 12mins in sport mode, i think its an thermal issue.

Christian U Posted at 2-11 20:03
Maybe because its not cvlear whats the cause. And when you try to solve an problem thats not exactly analyzed things goes worst.

Its not plausible that this is an baro issue, then the readings would be wrong.

Were you able to replicate the issue with extensive flight time in S mode each flight?

grahamjohnson10 Posted at 2-11 17:43
Love you detailed post.

I totally agreed that air movement around the drone can cause incorrect readings.

They probably haven't tested this enough to give them a solution considering how random this issue is. I am not aware of anyone reliably replicating this issue. If you know please show me.

Guorium Posted at 2-11 16:54
Or this is the baro sensor just under the top lid located on the top side of ESC board. This is more open than our old candidate being pretty close to the vent holes on the sides of the mini. It says '6407 BA9S13' on it. BA could mean barometer but google does not tell what it is. Sometimes chip makers put an indentation on the chip which could get mixed up with the air inlet of a baro sensor. Without prying open one myself for close inspection I cannot be sure which is the baro exactly but it will have to be one of the two. But still, it is in that grey body shell with many holes that could be susceptible to wind.

Chip with red square around it.  From picture, dimple looks like a pin-1 marker more than hole.


"It says '6407 BA9S13' on it."  Could be a proprietary IC, custom sourced for DJI.  Or just not well know.



Amphenol makes a Air Pressure transducer IC, which does not have any port (air hole).  Guessing chip messures pressure on top / bottom of IC case.  Whereas other IC chips for measuring air pressure changes do have one or two air ports.

Guorium Posted at 2-11 20:57
Were you able to replicate the issue with extensive flight time in S mode each flight?

yes ive maded ~20 test flights over 4 days and after i had the methodic ive reproduced every time with wind and without wind in the middle of an flown line and after stops the only thing that not changes was that it happens 10-12 mins of time full flight in sport mode

the first time ive seen it ive flown only 100m in sport mode before the whole flight in mode-p so it seems that it can happen too after an long flight in mode p and then during an short amount of time in sport mode but ive never get this securely reproduced

Ive described that here: https://forum.dji.com/forum.php?mod=viewthread&tid=207966

Ive replicated the issue on every flight ive teste so:

Switch to sport mode

Fly ~100m away from you in an straight line Full trottle forward (ive done that in ~3-15m height but i thing it dont matter i just not wanted that the dron can fall deep)
the same full trottle backward back

Do this ~10-15mins and you should have the issue around minute 12
I can confirm this issue only on 1 out of 3 drones i had in my hands until now, so dont expect it from every one.

Guorium Posted at 2-11 16:54
Or this is the baro sensor just under the top lid located on the top side of ESC board. This is more open than our old candidate being pretty close to the vent holes on the sides of the mini. It says '6407 BA9S13' on it. BA could mean barometer but google does not tell what it is. Sometimes chip makers put an indentation on the chip which could get mixed up with the air inlet of a baro sensor. Without prying open one myself for close inspection I cannot be sure which is the baro exactly but it will have to be one of the two. But still, it is in that grey body shell with many holes that could be susceptible to wind.

Yeah I cannot find a Datasheet for that IC part number either...

The markers shown in one of the IC corners indicate PIN number one, so they are propperly positioned for soldering.

However by searching a bit I found a few datasheets and baro sensors... I guess DJI would be using something similar... only opening would give more answers...

Christian U Posted at 2-12 00:17
Ive replicated the issue on every flight ive teste so:

Switch to sport mode

Hey Christian, what do you mean by in "1 out of 3 drones", minis or other models?

InspektorGadjet Posted at 2-12 00:43
Hey Christian, what do you mean by in "1 out of 3 drones", minis or other models?

Minis, my original one had another issue (see "strange rotating in sport mode") and in meantime i have buyed an second one that has the drop issue. And dji has exchanged my original one so i had 3 pcs until now to test. (And maybe i have soon an 4th cause the exchanged one had the rotating issue again)

Christian U Posted at 2-12 00:17
Ive replicated the issue on every flight ive teste so:

Switch to sport mode

Hi Christian,

Nice tests, if flown within VPS range, did the craft actual climb (compare with VPS height) or is it just the baro height indicate higher values?

I will test as well, after the Ciara and Dennis storm left my area.

cheers
JJB

Christian U Posted at 2-12 01:39
Minis, my original one had another issue (see "strange rotating in sport mode") and in meantime i have buyed an second one that has the drop issue. And dji has exchanged my original one so i had 3 pcs until now to test. (And maybe i have soon an 4th cause the exchanged one had the rotating issue again)

Oh dear... Does the rotating happens only S mode?
I wold expect some consistency between 3 models...

Christian U Posted at 2-12 00:17
Ive replicated the issue on every flight ive teste so:

Switch to sport mode

Can you post your log, it should show all this happening in time .

Christian U Posted at 2-12 00:13
yes ive maded ~20 test flights over 4 days and after i had the methodic ive reproduced every time with wind and without wind in the middle of an flown line and after stops the only thing that not changes was that it happens 10-12 mins of time full flight in sport mode

the first time ive seen it ive flown only 100m in sport mode before the whole flight in mode-p so it seems that it can happen too after an long flight in mode p and then during an short amount of time in sport mode but ive never get this securely reproduced

That one where you lose 1.7m in altitude from a 2m height? Ok I might try to replicate that with your method. Just back and forth in S mode non-stop for 12min? May I ask you what the temperature was like during your tests.
Some climb in fast forward motion (or into the wind) is to be expected as the props are more efficient with faster airflow. It is called "translational lift". Happens in all rotary flight systems. DJI FC seems to not consider this effect so you can see climb like this even in M2P.

All this seems a bit scattered, it might be better if all those with problems posted to one thread with one form of log, I suggest phantom help, it also seems that some had big drops but continued to fly, where these drops or just drops in logs, it would seem odd that people would continue to fly after seeing 50ft drop.
So post log but most important give explanation of what happened and what you done.

I had posted to say this could be a barometer problem, but many who are on this thread said this was impossible, so maybe let us know what you think caused it and if it can be fixed in FW.

I have tried to recreate some of this, only two things, tried in very cold weather and found drops and gains, but I could control these at any time during flight, and sport mode, fly fast forward sport mode, let off sticks craft drops but rises back to gps position it had when it dropped, so I know I have no problem.

Guorium Posted at 2-12 03:32
That one where you lose 1.7m in altitude from a 2m height? Ok I might try to replicate that with your method. Just back and forth in S mode non-stop for 12min? May I ask you what the temperature was like during your tests.
Some climb in fast forward motion (or into the wind) is to be expected as the props are more efficient with faster airflow. It is called "translational lift". Happens in all rotary flight systems. DJI FC seems to not consider this effect so you can see climb like this even in M2P.

It was around 10° these days. The climbing was most from ~5 to 15m over the forward flight. Back from 15 to 18 or so. Your right that was expected and i mostly have commanded it down when iam at my position before the next round.

nteresting is that my first min dont had this climbing. Ive make the same test with it and it stays nearly at the same height.

JJB* Posted at 2-12 01:56
Hi Christian,

Nice tests, if flown within VPS range, did the craft actual climb (compare with VPS height) or is it just the baro height indicate higher values?

I will search the flight logs (hopefully today evening) and add them. Then you can look yourself.
Here are siome Screenshots in the other thread where the VPS and Baro Alt is shown:
https://forum.dji.com/forum.php? ... 966&pid=2074637

Pretty comprehensive testing and explanations done here.

Christian U Posted at 2-12 04:38
I will search the flight logs (hopefully today evening) and add them. Then you can look yourself.
Here are siome Screenshots in the other thread where the VPS and Baro Alt is shown:
https://forum.dji.com/forum.php?mod=redirect&goto=findpost&ptid=207966&pid=2074637

Ype, have seen them already. That`s "normal" for a MM  ;-)  in the file we discussed here craft climbed 7 meters with no vertical speed, so it did not climbed at all (confirmed by the 'pilot'), after that yes  some climb/drop beacuse of fwd stick 100% > 0%.

I have the intention to fly above solid surface withing vps range; full speed to and fro in sporyt mode, see what baro indicates versus vps height.

Stupid is that a did a flight week ago with too much wind (MA could not hover and was blown backwards) and my MM did a great job. And no unwanted climb or dropping height!

Mayby the first built MM has problems....i have no clue yet.

Same for baro height  indications, after 20 mins of flying (up to 120 meter) my baro height at landon differs the most at 2 meters, often less. (wich i think is pretty good)

cheers
JJB

Christian U Posted at 2-12 04:34
It was around 10° these days. The climbing was most from ~5 to 15m over the forward flight. Back from 15 to 18 or so. Your right that was expected and i mostly have commanded it down when iam at my position before the next round.

nteresting is that my first min dont had this climbing. Ive make the same test with it and it stays nearly at the same height.

Sport mode makes it more pronounced. It is all about what the wind is like around the mini. Best to test this indoor. Wind is hard to control outside even on a calm day. You measure temp in celcius right?

Christian U Posted at 2-12 01:39
Minis, my original one had another issue (see "strange rotating in sport mode") and in meantime i have buyed an second one that has the drop issue. And dji has exchanged my original one so i had 3 pcs until now to test. (And maybe i have soon an 4th cause the exchanged one had the rotating issue again)

Quite possible 1 out 3 with mass produced commodity product.  Cheapest sub-component source is used, and maker can change due to availability, volume discounts, foot-n-door pricing.  
Seen flaky IC problems even with same manufacture due to bad batch run.

Guorium Posted at 2-11 06:21
I have never seen a log on this forum showing baro reading wildly off what is visually observed. For low flights reported it is mostly consistent with IR sensor reading. You are right the baro sensor is very sensitive so the drone can tell 1'' of difference in height. Upside of the mainboard houses the baro sensor. It is the circled black square with a tiny hole on the top left corner. Being right next to the battery and microusb port during flight, it is in a very well-shielded position. The heat from the battery makes it very unlikely to be accumulate condensation. [view_image]

However, it may still be susceptible to airflow caused by outside environment due to the big opening to the battery bay behind the gimbal assembly. The ventilated design was not seen before the mini in the DJI product line I think. The baro sensor in the mini may face "Venturi effect" when an air current try to squeeze past the narrowing gap around the baro sensor, causing reduced air pressure and in turn interpreted by the FC as unwanted climb. The result, drop for altitude compensation. I am merely taking an educated guess but if it is true, some FW tweak to the FC could alleviate the issue.

I’ve stated in a number of posts that my Mini climbs when flying into the wind, it’s less noticeable at lower speeds and far more noticeable in S mode compared to P and C.  I’m tired of people suggesting calibrate your controller or simply stating it doesn’t climb, It’s not a super big issue for me and I’m sure they will address this moving forward with a FW mode.  

TDZHDTV Posted at 2-12 06:49
I’ve stated in a number of posts that my Mini climbs when flying into the wind, it’s less noticeable at lower speeds and far more noticeable in S mode compared to P and C.  I’m tired of people suggesting calibrate your controller or simply stating it doesn’t climb, It’s not a super big issue for me and I’m sure they will address this moving forward with a FW mode.

It is a result of translational lift. The FC of the drone does not anticipate the increased prop thrust at speed when heading into the wind so it climbs. RC is not likely to be off but it is an easy option to eliminate first.

TDZHDTV Posted at 2-12 06:49
I’ve stated in a number of posts that my Mini climbs when flying into the wind, it’s less noticeable at lower speeds and far more noticeable in S mode compared to P and C.  I’m tired of people suggesting calibrate your controller or simply stating it doesn’t climb, It’s not a super big issue for me and I’m sure they will address this moving forward with a FW mode.

Yes that is what I thought, my latest post of PANO+AEB  I flew very slow, mainly P mode and had no noticeable problem at all, the problem is, I saw some bird approaching and I have to change to S mode and return and that makes me anxious cause I'm not sure if it will do some climb or drop.

I love the mini, so when I encounter something weird I like to report so, hopefully DJI will address it and I will have less and less anxious moments. I'm sure nobody wants that.

Guorium Posted at 2-12 07:40
It is a result of translational lift. The FC of the drone does not anticipate the increased prop thrust at speed when heading into the wind so it climbs. RC is not likely to be off but it is an easy option to eliminate first.

The transitional lift makes a lot of sense, I noticed than flying backwards doesn't happen so often?

Also if flying forward while ascending seem to be less noticeable... I tend to do this when you eye is in line with the mini and the mini only seems to decrease in size, like a straight line between your head and the drone. Is a technique I saw  of some super experienced user flying backwards thru tight things or objects.

Guorium Posted at 2-12 05:35
Sport mode makes it more pronounced. It is all about what the wind is like around the mini. Best to test this indoor. Wind is hard to control outside even on a calm day. You measure temp in celcius right?

Yes 10°C, ive flown with very different wind conditions and the behavour was always the same.