H520E RTK From the Box Up

chascoadmin · Oct 25, 2021

Good morning from Texas! Just wanted to post an update that I think after about 70 flights we have the H520E RTK dialed in. The v2 DataPilot solved the turning behaviour but we have a load of other bugs and enhancements to deal with and are pushing as hard as we can to get some movement.

When we started we were getting average camera location RMSE's of about 30cm horizontally and 20cm vertically which doesn't make sense for the vertical to be lower unless it was the flight behaviour affecting the horizontal positioning. It's a ton better than without RTK but a load of GCP's. Unfortunately it was still not acceptable but I think something happened in v2 and the way we are configuring flights because we are now getting 16-18cm horizontal and 5-7cm vertical.

The flight changes involved making sure to manually set the speed at 10-14mph depending on the overlaps we were running which puts the capture interval at about 2 seconds. Unfortunately the algorithm they are using for setting the rest of the parameters based on overlap and speed is now overestimating the battery and it is usually about 20% more required than what the time estimate shows. If you do not set the speed manually the aircraft always defaults to 22mph which is ridiculous so it is one of the items on our list.

chascoadmin · Oct 26, 2021

Today's schedule

Locating tract corners and setting checkpoints for the Yuneec H520E RTK. Looking a little ominous… … but it was worth it… … and then lunch.

community.emlid.com

chascoadmin · Jan 13, 2022

So we are still working with Yuneec to get a long list of issues we have with the H520E RTK... One by one we're knocking it out but this is taking forever. I'm sure the engineer in China has plenty to do though, lol.

Still taking images out of turns regardless of where the turning point is set.
“Disconnecting” on battery swaps. You either have to wait a couple of minutes to see if the unfinished portion of the flight is repopulated and sometimes it never does. I then have to rotate the starting point to come from the other direction and then try to match up with where the previous flight left off. This makes flying a 3-battery mission impossible.
Need to be able to manually define the start point to alleviate #2.
Auto speed calc for flight planning still runs at 22mph regardless of settings which can put the shot interval well below the cameras tolerance. Manually reducing the speed of the flight corrupts the anticipated values for flight duration and shutter interval and then causes overestimation the battery life.
Need to know the specs req’d for the GNSS to report a fixed condition. It is taking too long to fix and wasting battery.
Terrain following is hit or miss as to whether or not it will allow it in the field. We might want to consider offline caching.
There are still two different locations for PPK data in the mSD card.

zephyrus · Feb 20, 2022

Hello! Are there any news with solving mentioned problems? A light in the tunnel for next weeks, months?

chascoadmin · Feb 20, 2022

zephyrus said:
Hello! Are there any news with solving mentioned problems? A light in the tunnel for next weeks, months?

We just had a good meeting last week with the US reps and a couple of other users. I need to go back and recompile the list of issues as a few of these have been resolved or a resolution is in progress and a few new ones have been added.

chascoadmin · Feb 20, 2022

Existing

Still taking images out of turns regardless of where the turning point is set.
- I have found that this is due to the method of turning and when images are captured. With most drones they either stop-turn-stop-turn and then shoot or they have a constant rotation and at the end of the rotation the camera is already pointed straight forward whereas the H520E is trying the rotation method but is short of the full turn when it is coming out and has initiated a capture. This is why they have the setting for a turnaround distance in order to give the drone more time to adjust its heading to the next waypoint before the capture. Many times there are constraints which we must fly in and extending the turn beyond the perimeter is not desirable. I would prefer to have a more consistent and complete rotation but that will require a reconfiguration of programming.
“Disconnecting” on battery swaps. You either have to wait a couple of minutes to see if the unfinished portion of the flight is repopulated and sometimes it never does. I then have to rotate the starting point to come from the other direction and then try to match up with where the previous flight left off. This makes flying a 3-battery mission impossible.
- This is still an issue but I have been able to successfully complete the last several multi-battery missions. It's definitely broken but if I wait a couple of minutes it does finally load the remaining pattern. This lag along with the time it takes to reacquire a fix in low data areas wastes a lot of battery. This is marked as a priority.
- I have seen one instance of even after waiting I finally received a "mission transfer failed" message and "IGN Request List: Busy" and the flight plan disappeared.
- I have also seen one recent instance of DataPilot crashing during this process but I haven't seen this for a couple of versions and have not been able to repeat it.
Need to be able to manually define the start point to alleviate #2.
- In consideration.
Auto speed calc for flight planning still runs at 22mph regardless of settings which can put the shot interval well below the cameras tolerance. Manually reducing the speed of the flight corrupts the anticipated values for flight duration and shutter interval and then causes overestimation the battery life.
- This has been reproduced by the other pilots and acknowledged but I don't expect it to be an easy fix.
Need to know the specs req’d for the GNSS to report a fixed condition. It is taking too long to fix and wasting battery.
- Acknowledged and waiting on Yuneec Engineering for the specs. Proposed access to more telemetry.
Terrain following is hit or miss as to whether or not it will allow it in the field. We might want to consider offline caching.
- This has been attributed to low data availability. If there is enough data it will work but it may take a while to recalculate if there is a change to the flight. Offline caching will be considered.
There are still two different locations for PPK data in the mSD card.
- Acknowledged and to be fixed. We have found that the PPK data in the same directory as the images is the correct one. The files in the root of the card are unexplainable. They are similar but I have not figured out exactly what the are reporting.

Added

Incorrect date/times in image Exif data.
Camera optimization reports showing increasingly worse departure from camera intrinsics which gets worse as the camera pitch becomes more oblique, IE 0.5% deviation at nadir and 3% deviation at -65deg. Thereabouts.
Batteries immediately reporting as low as 29% as mission starts and then slowly rises to a point of convergence of the actual correct percentage.
- I had 7 batteries which I was positive were fully charged and three of them in a row showed 100%, started the mission and they then dropped immediately between 29-33% and climbed to about 43% and then depleted normally. I stuck my neck out by letting it continue but I have seen this in the original H520 and was able to verify it is a systems reporting issue. Not proven but I'm pretty sure it was because it was 32F.
- Possibly part of an overall systems power/reporting/configuration issue I am only getting about 13-14 minutes of useable flight time between 95% and 25%. I am almost positive we use to get 22 minutes on average.

If anyone else has any issues that are persistent, even if they are not consistent, please report them here and I will do my best to get them looked into.

Pawel_(PL) · Feb 20, 2022

chascoadmin said:
Existing

Still taking images out of turns regardless of where the turning point is set.

I have found that this is due to the method of turning and when images are captured. With most drones they either stop-turn-stop-turn and then shoot or they have a constant rotation and at the end of the rotation the camera is already pointed straight forward whereas the H520E is trying the rotation method but is short of the full turn when it is coming out and has initiated a capture. This is why they have the setting for a turnaround distance in order to give the drone more time to adjust its heading to the next waypoint before the capture. Many times there are constraints which we must fly in and extending the turn beyond the perimeter is not desirable. I would prefer to have a more consistent and complete rotation but that will require a reconfiguration of programming.

“Disconnecting” on battery swaps. You either have to wait a couple of minutes to see if the unfinished portion of the flight is repopulated and sometimes it never does. I then have to rotate the starting point to come from the other direction and then try to match up with where the previous flight left off. This makes flying a 3-battery mission impossible.

This is still an issue but I have been able to successfully complete the last several multi-battery missions. It's definitely broken but if I wait a couple of minutes it does finally load the remaining pattern. This lag along with the time it takes to reacquire a fix in low data areas wastes a lot of battery. This is marked as a priority.

I have seen one instance of even after waiting I finally received a "mission transfer failed" message and "IGN Request List: Busy" and the flight plan disappeared.

I have also seen one recent instance of DataPilot crashing during this process but I haven't seen this for a couple of versions and have not been able to repeat it.

Need to be able to manually define the start point to alleviate #2.

In consideration.

Auto speed calc for flight planning still runs at 22mph regardless of settings which can put the shot interval well below the cameras tolerance. Manually reducing the speed of the flight corrupts the anticipated values for flight duration and shutter interval and then causes overestimation the battery life.

This has been reproduced by the other pilots and acknowledged but I don't expect it to be an easy fix.

Need to know the specs req’d for the GNSS to report a fixed condition. It is taking too long to fix and wasting battery.

Acknowledged and waiting on Yuneec Engineering for the specs. Proposed access to more telemetry.

Terrain following is hit or miss as to whether or not it will allow it in the field. We might want to consider offline caching.

This has been attributed to low data availability. If there is enough data it will work but it may take a while to recalculate if there is a change to the flight. Offline caching will be considered.

There are still two different locations for PPK data in the mSD card.

Acknowledged and to be fixed. We have found that the PPK data in the same directory as the images is the correct one. The files in the root of the card are unexplainable. They are similar but I have not figured out exactly what the are reporting.

Added

Incorrect date/times in image Exif data.

Camera optimization reports showing increasingly worse departure from camera intrinsics which gets worse as the camera pitch becomes more oblique, IE 0.5% deviation at nadir and 3% deviation at -65deg. Thereabouts.

Batteries immediately reporting as low as 29% as mission starts and then slowly rises to a point of convergence of the actual correct percentage.

I had 7 batteries which I was positive were fully charged and three of them in a row showed 100%, started the mission and they then dropped immediately between 29-33% and climbed to about 43% and then depleted normally. I stuck my neck out by letting it continue but I have seen this in the original H520 and was able to verify it is a systems reporting issue. Not proven but I'm pretty sure it was because it was 32F.

Possibly part of an overall systems power/reporting/configuration issue I am only getting about 13-14 minutes of useable flight time between 95% and 25%. I am almost positive we use to get 22 minutes on average.

If anyone else has any issues that are persistent, even if they are not consistent, please report them here and I will do my best to get them looked into.

Looks like there are a lot of issues to be fixed/resolved. It is strange that so much is to be improved yet for a platform which was launched over a year ago.
It does not look good at all

chascoadmin · Feb 20, 2022

Pawel_(PL) said:
Looks like there are a lot of issues to be fixed/resolved. It is strange that so much is to be improved yet for a platform which was launched over a year ago.
It does not look good at all

From my side of the fence it is not as bad as it seems but it does seem intrinsic to the old Yuneec way of development. We had a really good thing going with the beta of the original H520 and were able to get it to a stable point within a couple of months but this was how it should be a month before release. I am not trying to be ugly but if I owned Yuneec the dev team would have a fair amount of questions to answer. If you recall the retrofit RTK to the H520 was supposed to be released early 2019 and that didn't happen so after we all thought it was a dead horse the H520E and E-RTK are released over a year later. The RTK retrofit flop is at the same time our US beta team was disbanded. What's history is history and from what I am hearing it sounds like they are making a genuine effort to assure a reachable US presence but also much more focus on the commercial industry while deprecating the hobbyist side.

We are 2 months from having our unit for a year and I have a full library of communications ready to either help or to take whatever action is necessary to get us off the solution and into something else. They understand this as well.

chascoadmin · Mar 29, 2022

Well guys it looks like we might be at the end of our Yuneec adventure (gauntlet). I received a new replacement for the unit that we have been testing and it failed as well with an "Unstable battery connection" error. Our Ownership is now involved and we will be pursuing a complete refund and will move on to another brand. Hopefully they will eventually figure it out but we have taken too many losses to not move on. I'll check in every once and a while but for now fly safe and enjoy what you do!

Steve Carr · Mar 29, 2022

I'm very sorry to hear this. Yuneec makes some outstanding models yet they can't complete the final tuneups. I think they change engineers more often than I change underwear. Their initial designs were brilliantly planned only to be relegated to fumbling engineering missteps. It's a failure of leadership. And we can all understand how that ends.

chascoadmin · Mar 29, 2022

Steve Carr said:
I'm very sorry to hear this. Yuneec makes some outstanding models yet they can't complete the final tuneups. I think they change engineers more often than I change underwear. Their initial designs were brilliantly planned only to be relegated to fumbling engineering missteps. It's a failure of leadership. And we can all understand how that ends.

Well said. That has been my experience for the last 5+ years. They have got to have the worst Engineering QC program I have ever dealt with and I am in an IT and Engineering driven role. Or it could be that (please excuse me) it's a bunch of 14 year-olds coding. Actually that would probably be better than what we have been getting.

Steve Carr · Mar 29, 2022

Yuneec should have partnered with this Forum for tech support a long time ago. I think many of the members here spend more time on troubleshooting than their technicians have. And the skill level here is most assuredly better. What we lack is access to the firmware tools to modify and test. It's a case where arrogance trumps rationality.

chascoadmin · Mar 29, 2022

Steve Carr said:
where arrogance trumps rationality.

...and success.

Pawel_(PL) · Mar 30, 2022

chascoadmin said:
Well guys it looks like we might be at the end of our Yuneec adventure (gauntlet). I received a new replacement for the unit that we have been testing and it failed as well with an "Unstable battery connection" error. Our Ownership is now involved and we will be pursuing a complete refund and will move on to another brand. Hopefully they will eventually figure it out but we have taken too many losses to not move on. I'll check in every once and a while but for now fly safe and enjoy what you do!

Bye, bye ... we are going to miss your valuable comments and inputs.
Good luck.

terrence davidson · Mar 30, 2022

I had that unstable battery error before the second battery head meltdown but that error only came up after fully charged up gifi h3 battery was placed in the h520e. check your flight battery head for gaps between metal plates that connect to the drone power head. any that are larger than normal will throw that error.

terrence davidson · Mar 30, 2022

now that I have 2 h520e one of which is rtk verson and the other is just the h520e. the non rtk h520e is using first gen h520 power connector while the h520e rtk is on second gen power source with new internal frame. this weekend going to setup for 6 flights each back to back this weekend and see what the results are.

chascoadmin · Mar 30, 2022

terrence davidson said:
I had that unstable battery error before the second battery head meltdown but that error only came up after fully charged up gifi h3 battery was placed in the h520e. check your flight battery head for gaps between metal plates that connect to the drone power head. any that are larger than normal will throw that error.

I thought the same thing but only one out of my 10 batteries has any kind of a gap between the orange case and the connection insulator. I only tried five different batteries because at that point I knew something was wrong above and beyond what was wrong with the other one and I've wasted enough company resources on this.

terrence davidson · Mar 30, 2022

I am in contact with vertigo now they have mention about felt like material battery grips that are on the inside are missing were not installed by yuneec. they are also missing on my h520ertk but first gen non rtk h520e has them. I trying to get some clarification on that.

chascoadmin · Mar 30, 2022

terrence davidson said:
I am in contact with vertigo now they have mention about felt like material battery grips that are on the inside are missing were not installed by yuneec. they are also missing on my h520ertk but first gen non rtk h520e has them. I trying to get some clarification on that.

100% what I have experienced. Bottom line is that the aircraft has a major power distribution systems design flaw and we can no longer float their R&D bill. I have kept track of all my time spent not being productive, lost contracts and the costs of other hardware we have had to purchase because of the unreliability and we are at just over $14,000 at this point including the drone itself. Ownership was not happy at $10k.