EKF Variance (no kidding) Log assistance requested

Log Analysis -- deprecated Copter Logs -- deprecated
1

[attachment=0]2015-02-27 13-26-48.log[/attachment]Very Smooth takeoff in STAB, switched to ALTHOLD and then to LOITER to verify performance. All was good.
Winds were a bit gusty so I switched back to STAB to fly some slow figure 8’s. Quad tumbled and I heard “EKF variance” through the PC Ground Station. Truthfully I couldn’t be sure if I heard the error message before or after it hit the ground.

I’ve reviewed the attached log several ways, and see the event near the time of the Error message. I would appreciate if anyone could assist to analyze the actual cause.
Ultimately there were vibrations near the event. I don’t know if a prop broke first? My vibrations have been running .25-.35 until this.

Many thanks, in advance,
SC

2

the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.

3

Thanks for the look see. Your log analysis is remarkably accurate in describing what I saw in flight.

As I replay the events again I believe the warning was after the flip.

I’ll definitely enable RCout before venturing up again.

Post flight; the left front prop had 1 blade left and the right front was missing both blades. Hard to determine what parts broke under power when it hit. Considering how well it flies it’s hard to get over the possibility at least 1 blade broke in flight.

Thanks also for your insights on the RCout log value.

BTW: Do you have any knowledge of the PH’s automatic recovery functionality? I see it mentioned, when I search, but I didn’t find any details.

[quote=“RabbitStu”]the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.[/quote]

4

Yeah a tricky one to work out if the props were the cause…were all of the blades near by on the ground where it hit? Not infallible, but a clue. Do all the motors spin ok on the ground after the crash?

I haven’t seen anything about PH’s automatic recovery functionality, what is it?

[quote=“FullScalePilot2”]Thanks for the look see. Your log analysis is remarkably accurate in describing what I saw in flight.

As I replay the events again I believe the warning was after the flip.

I’ll definitely enable RCout before venturing up again.

Post flight; the left front prop had 1 blade left and the right front was missing both blades. Hard to determine what parts broke under power when it hit. Considering how well it flies it’s hard to get over the possibility at least 1 blade broke in flight.

Thanks also for your insights on the RCout log value.

BTW: Do you have any knowledge of the PH’s automatic recovery functionality? I see it mentioned, when I search, but I didn’t find any details.

[quote=“RabbitStu”]the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.[/quote][/quote]

5

I didn’t find all the prop pieces due to 12" grass. I still have some resolder work to do after replacing the arms. But the motors feel/spin good without power.

There is a https://www.youtube.com/watch?v=VJqeyUOM9QQ demonstrating the ability to fly/land when a prop failure or motor out. It is a hex though. I imagine a challenge could be presented if increased vibrations confuse the board. I’ve seen a TED talk that showed a quad/FC doing a great job compensating for various motor/prop issues. Come to think of it, the Pixhawk Auto Recovery may actually be referring to failsafes. I just haven’t run into an explanation, when it’s been referenced.

In theory my event would have seen the quad rotate and/or possibly just adjust motor speed to compensate for a partial prop failure.

[quote=“RabbitStu”]Yeah a tricky one to work out if the props were the cause…were all of the blades near by on the ground where it hit? Not infallible, but a clue. Do all the motors spin ok on the ground after the crash?

I haven’t seen anything about PH’s automatic recovery functionality, what is it?

[quote=“FullScalePilot2”]Thanks for the look see. Your log analysis is remarkably accurate in describing what I saw in flight.

As I replay the events again I believe the warning was after the flip.

I’ll definitely enable RCout before venturing up again.

Post flight; the left front prop had 1 blade left and the right front was missing both blades. Hard to determine what parts broke under power when it hit. Considering how well it flies it’s hard to get over the possibility at least 1 blade broke in flight.

Thanks also for your insights on the RCout log value.

BTW: Do you have any knowledge of the PH’s automatic recovery functionality? I see it mentioned, when I search, but I didn’t find any details.

[quote=“RabbitStu”]the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.[/quote][/quote][/quote]

6

Ah I see what you meant now I think. Your are talking about redundancy/motor loss. I believe at the moment loss of motor/prop on a quad is unrecoverable. Perhaps if apartial blede loss and very lucky iy might land…but that is luck rather than design. I have seen research (non arducopter) quads designed to cope…but they were research with off board processing involved…and the copter yaws very fast during this period…which has o happen with a quad…Its a nature of the beast.

I do recall someone experimenting recently asking for beta testers on DIYDrones…but that was abranch of the code. Pretty sure its not included…if it was people would be shouting about it.

If you want redundancy/ability to cope with motor loss at the moment you need more motors, my X8 had a motor stop and I almost didn’t realise it flew so well…saw the motor not spinning.

[quote=“FullScalePilot2”]I didn’t find all the prop pieces due to 12" grass. I still have some resolder work to do after replacing the arms. But the motors feel/spin good without power.

There is a https://www.youtube.com/watch?v=VJqeyUOM9QQ demonstrating the ability to fly/land when a prop failure or motor out. It is a hex though. I imagine a challenge could be presented if increased vibrations confuse the board. I’ve seen a TED talk that showed a quad/FC doing a great job compensating for various motor/prop issues. Come to think of it, the Pixhawk Auto Recovery may actually be referring to failsafes. I just haven’t run into an explanation, when it’s been referenced.

In theory my event would have seen the quad rotate and/or possibly just adjust motor speed to compensate for a partial prop failure.

[quote=“RabbitStu”]Yeah a tricky one to work out if the props were the cause…were all of the blades near by on the ground where it hit? Not infallible, but a clue. Do all the motors spin ok on the ground after the crash?

I haven’t seen anything about PH’s automatic recovery functionality, what is it?

[quote=“FullScalePilot2”]Thanks for the look see. Your log analysis is remarkably accurate in describing what I saw in flight.

As I replay the events again I believe the warning was after the flip.

I’ll definitely enable RCout before venturing up again.

Post flight; the left front prop had 1 blade left and the right front was missing both blades. Hard to determine what parts broke under power when it hit. Considering how well it flies it’s hard to get over the possibility at least 1 blade broke in flight.

Thanks also for your insights on the RCout log value.

BTW: Do you have any knowledge of the PH’s automatic recovery functionality? I see it mentioned, when I search, but I didn’t find any details.

[quote=“RabbitStu”]the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.[/quote][/quote][/quote][/quote]

7

Thanks for the feedback. I’m close to first flight with a 680 hexa. Perhaps it will provide a touch more reliability if I don’t overload it.

[quote=“RabbitStu”]Ah I see what you meant now I think. Your are talking about redundancy/motor loss. I believe at the moment loss of motor/prop on a quad is unrecoverable. Perhaps if apartial blede loss and very lucky iy might land…but that is luck rather than design. I have seen research (non arducopter) quads designed to cope…but they were research with off board processing involved…and the copter yaws very fast during this period…which has o happen with a quad…Its a nature of the beast.

I do recall someone experimenting recently asking for beta testers on DIYDrones…but that was abranch of the code. Pretty sure its not included…if it was people would be shouting about it.

If you want redundancy/ability to cope with motor loss at the moment you need more motors, my X8 had a motor stop and I almost didn’t realise it flew so well…saw the motor not spinning.

[quote=“FullScalePilot2”]I didn’t find all the prop pieces due to 12" grass. I still have some resolder work to do after replacing the arms. But the motors feel/spin good without power.

There is a https://www.youtube.com/watch?v=VJqeyUOM9QQ demonstrating the ability to fly/land when a prop failure or motor out. It is a hex though. I imagine a challenge could be presented if increased vibrations confuse the board. I’ve seen a TED talk that showed a quad/FC doing a great job compensating for various motor/prop issues. Come to think of it, the Pixhawk Auto Recovery may actually be referring to failsafes. I just haven’t run into an explanation, when it’s been referenced.

In theory my event would have seen the quad rotate and/or possibly just adjust motor speed to compensate for a partial prop failure.

[quote=“RabbitStu”]Yeah a tricky one to work out if the props were the cause…were all of the blades near by on the ground where it hit? Not infallible, but a clue. Do all the motors spin ok on the ground after the crash?

I haven’t seen anything about PH’s automatic recovery functionality, what is it?

[quote=“FullScalePilot2”]Thanks for the look see. Your log analysis is remarkably accurate in describing what I saw in flight.

As I replay the events again I believe the warning was after the flip.

I’ll definitely enable RCout before venturing up again.

Post flight; the left front prop had 1 blade left and the right front was missing both blades. Hard to determine what parts broke under power when it hit. Considering how well it flies it’s hard to get over the possibility at least 1 blade broke in flight.

Thanks also for your insights on the RCout log value.

BTW: Do you have any knowledge of the PH’s automatic recovery functionality? I see it mentioned, when I search, but I didn’t find any details.

[quote=“RabbitStu”]the ekf warning in the log is after it rolling over, can you remember when you heard it compared to the flip?

I have seen ekf errors after an event just because the event confused it, rather than ekf being the cause.

The vibes occur at the same time as the big roll starts. You haven’t got RCout enables so its difficult to see if it lost a motor/prop or something…but it could well be a snapped prop in the air if you have a damaged prop.

Pitch and roll go at about the same rate (although the sign is different…desired for both at zero so not commanded) until about 60 degrees. looks like it went nose down and roll right. was it the front right prop that was broken? If so I think it looks like it went in the air, if not something else did. I think I have the directions right for the polarities of pitch/roll. neg pitch is nose down, pos roll is right. moving my mini quad around in front of me as a visual aid :slight_smile:

If an RC out channel shot up at the same time it might back that up. RCout is probably the one I examine most as it can find all sorts of basic issues with a bit of interpretation. RCout logs aren’t always easy to read…but often problems jump right out and the often give clues ahead of an event that something is up or borderline…often before they show up in the pitch/roll/yaw plots.[/quote][/quote][/quote][/quote][/quote]