Re: [Paparazzi-devel] strange behaviour in lisa m

Hi,

>Cristophe:

>These exceptions are NOT good. You may only call land 1 time while the exception will continuously call it at 4Hz. Option 1: Add a variable : if not yet derouted to land, deroute to land, or add the deroute on low bat in >other flightplan blocks. It will then correctly deroute from the other block instead of constantly reinitializing but never executing landing.

<exception cond="auto2_land && (datalink_time > 180 || (electrical.bat_low && !electrical.bat_critical))" deroute="Standby"/>

We used auto2_land variable, which is TRUE at the beginning and it becomes FALSE when it first goes to Land. Therefore, we expect to go to Land bock for only once.

@Felix:

Our battery voltage varies a lot with throttle. It detects critical voltage and goes to Land, but when it decreases the throttle, the voltage increases and it goes to Standby from Land block. Thats why we put a variable named "auto2_land". We want that, if it goes to Land, it will never go to Standby again.

We will put the second exception to inside of Standby as you recommend.

Best regards,

Refik

From: address@hidden [mailto:address@hidden On Behalf Of Felix Ruess
Sent: Wednesday, September 11, 2013 1:31 AM
To: Paparazzi devel list
Subject: Re: [Paparazzi-devel] strange behaviour in lisa m

Sorry, didn't really pay attention when providing the examples... and just wanted to show how you can check for the battery status.

Should have made it clear that e.g. the deroute to land on bat critical only makes sense as an exception inside the standby block...

So as a rule to follow, don't use a global exception if you want to deroute to a block that you want to leave again (e.g land). That only makes sense for blocks that never end on their own (e.g. Standby, MOB, oval, eight)

On Wed, Sep 11, 2013 at 12:03 AM, Christophe De Wagter <address@hidden> wrote:

These exceptions are NOT good. You may only call land 1 time while the exception will continuously call it at 4Hz. Option 1: Add a variable : if not yet derouted to land, deroute to land, or add the deroute on low bat in other flightplan blocks. It will then correctly deroute from the other block instead of constantly reinitializing but never executing landing.

On Sep 9, 2013 11:20 PM, "Refik Sever" <address@hidden> wrote:

Hi Felix,

We moved to the stable branch. I have added these exceptions to the flight_plan as you recommend:

<exception cond="auto2_land && (datalink_time > 180 || (electrical.bat_low && !electrical.bat_critical))" deroute="Standby"/>

auto2_land variable is TRUE at the beginning, and it becomes FALSE when it goes to "Land Right AF-TD". If we don't use this variable, it continuously switches between Standby and Land blocks.

Today, although we used this variable, it continuosly switched between Standby and Land. Do you have an idea about it?

The bad thing is, in Standby it wanted to turn left, but in Land it wanted to turn right. Because of continuously switching between these blocks, it went in a straight line. We were lucky that the RC worked at that distance.

Using exceptions is not very simple I think. Do we forget something? I dont understand how it went to standby, although auto2_land was FALSE.

By the way, the strange behaviour while uploading the code to Lisa M was due to the ETS airspeed sensor. We did not turn off the power of it during boot.

Regards,

Refik

From: paparazzi-devel-bounces+refiksever=address@hidden [mailto:paparazzi-devel-bounces+refiksever=address@hidden] On Behalf Of Felix Ruess
Sent: Monday, September 9, 2013 12:24 AM
To: Paparazzi devel list
Subject: Re: [Paparazzi-devel] strange behaviour in lisa m

No, we don't add new features to the stable branches, only bugfixes.

On Sun, Sep 8, 2013 at 11:20 PM, Refik Sever <address@hidden> wrote:

Hi Felix,

Thank you for your recommendations.

We upload the code using USB cable. We turn off the 12V battery, and the autopilot takes its voltage from USB. Therefore, some of the peripherals (Airspeed sensor etc) are off during upload. Then, we remove the USB cable and turn on the battery. We always make a power cycle after upload. Sometimes, it does not behave normal, and we need to upload it again.

"in order to make checks for low or critical batter voltage easier and more robust we added this to the electrical subsystem in master."

Did you also add it to stable branch?

Best regards,

Refik

From: paparazzi-devel-bounces+refiksever=address@hidden [mailto:paparazzi-devel-bounces+refiksever=address@hidden] On Behalf Of Felix Ruess
Sent: Sunday, September 8, 2013 10:57 PM
To: Paparazzi devel list
Subject: Re: [Paparazzi-devel] strange behaviour in lisa m

Hi Refik,

in order to make checks for low or critical batter voltage easier and more robust we added this to the electrical subsystem in master.

It will automatically set the electrical.bat_low and electrical.bat_critical flags when the voltage drops below BAT_LOW or BAT_CRITIC for a certain amount of time to avoid setting this on very short voltage drops.

You can use it in your flight plans for example as follows:

About the behavior right after upload new code to the autopilot: depending on how you upload the code (jtag, usb, etc..) it sometimes seems to not correctly reset everything (probably related to external peripherals?).

To be safe you can power cycle once to make sure everything starts up as intended.

@Piotr: do you know why this sometimes happens?

Cheers, Felix

On Sun, Sep 8, 2013 at 9:27 PM, Refik Sever <address@hidden> wrote:

Dear all,

We are using Lisa-M and we had a lot of successful flights. We wanted to go to a slightly farther distance. And we wanted to turn back if the voltage drops below 11V, or the datalink is lost for 180s. We added this line to the perfectly working airframe file:

<exception cond="datalink_time > 180 || (11.0 > PowerVoltage())" deroute="Standby"/>

But, the upload time of the code to Lisa was rather longer. And it behaved very slowly, the PFD screen changed very slowly. I could not moved the ailerons in Auto1 from RC, but it worked in Manual mode. We thought that, it measured the power voltage below 11V at startup, although it was over 12V. And we thought that, it was very slow because of the exception conditon. That was the only change in the airframe file.

Then, we added a variable named "ready" and set it to TRUE only when it first goes to Holding Point. Therefore, we quaranteed that it will not enter exception at power up:

<exception cond="datalink_time > 180 || (11.0 > PowerVoltage() && ready)" deroute="Standby"/>

It was good on the ground. The voltage was 12V. Everything seemed normal. But in takeoff, it behaved again very slowly. It had a loop and crashed.

In the replay, voltage was decrease to 11.2V but it went to Standby before it reached normal altitude. I think that the telemetry rate was low and missed that the power voltage decreased below 11V. But the autopilot entered to exception.

Is this behaviour (slowly reacting) occurs due to this exception condition? Does it freezes the autopilot? What can we do to prevent it?

Sometimes, we upload the code to Lisa M but it does not behave normally. Then, we re-upload the same code and everything becomes good. Is there a possibility that it uploads incorrectly? Why does it take longer time for some uploads? Doesn't it need to be constant?

I am sending the logs. I can also send the video recorded from the wing.

Regards,
Refik

_______________________________________________
Paparazzi-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/paparazzi-devel

From:	Refik Sever
Subject:	Re: [Paparazzi-devel] strange behaviour in lisa m
Date:	Wed, 11 Sep 2013 01:49:27 +0300