Anybody running a Butterfly Vario with XCSoar?

I've been using XCSoar on an Android with a CAI-302 for almost 3 years now
and the wind calculation/display works perfectly.

If you tecchies would look outside once in a while instead of at your
whiz-bang panels, you could compare the wind velocity arrow to the cloud
shadows. Oh, wait! That would be too much like pilotage.



"jfitch" wrote in message
...
On Monday, July 15, 2013 3:44:56 AM UTC-7, Max Kellermann wrote:
On Sunday, July 14, 2013 8:08:50 PM UTC+2, jfitch wrote:

Got a question: the Butterfly wind vector and the XCSoar wind vector
(derived from Butterfly information) seem to disagree by 180 degrees.
That is they point toward each other. XCSoar seems to believe it too,
based on glide calculations. I am using the LX format output on the
Butterfly. I am assuming the arrow on each is intended to point the
direction the wind it blowing towards, I.e., a vector. From observation,
the Butterfly appears to be correct, XCSoar 180 off. Has anyone seen
this?



This sounds very much like the Butterfly firmware bug that was reported to
the XCSoar bug tracker 4 months ago:



http://bugs.xcsoar.org/ticket/2660

I will get to the bottom of this. Please tell me, what is the assumption of
XCSoar?

In particular, a CAI 302 !w sentence that looks like this:

!w,180,30,5,,,,,,,,,,*hh

would be interpreted how? Wind blowing from north towards the south, or
south towards the north?
(my interpretation of the description "vector wind direction in degrees"
indicates from north to south)

In an LX mode, an $LXWP0 sentence that looks like this:

$LXWP0,,,,,,,,,,180,3*hh

would be interpreted how? Wind blowing from north towards the south, or
south towards the north?
(description is less clear, "windcourse" would seem to mean same thing as
Cambridge)


How does one get access to the XCSoar forum and bug reporting scheme? I
registered over the weekend, but my registration is being held I guess.

to shed some light in this:

CAI: our fault, this was a bug that has been resolved a while ago.

LX: sorry guys, there is no spec on the LXWP wind direction (at least nobody knows of a specification). Both directions are commonly used depending on software you use. Rumors say that even LX devices exist that send data differently (is this true? i don't know). iGlide e.g. features a switch for that.

We will sure use the most common "interpretation" but the question is which one is the most common?

On Monday, July 15, 2013 8:51:18 AM UTC-7, Dan Marotta wrote:
I've been using XCSoar on an Android with a CAI-302 for almost 3 years now

and the wind calculation/display works perfectly.



If you tecchies would look outside once in a while instead of at your

whiz-bang panels, you could compare the wind velocity arrow to the cloud

shadows. Oh, wait! That would be too much like pilotage.



If you are flying in the flatlands where the clouds are 2000 AGL and likely to be blowing the same way as the air you are flying in, that works great. Otherwise, not so much.....

For example, final glide Saturday, 77 miles Mt. Patterson to Truckee out of 17,500. Cloud drift not discernible from altitude (or forming ahead as fast as drift), wind lines on Topaz Lake indicate calm, wind lines on Lake Tahoe indicate SSW about 10, AWOS at Truckee reporting W at 7, Butterfly showing 25 knots south. Butterfly says tailwind (and it is correct, too), XCSoar says headwind. Drop through about 12,000 ft., wind switches to 7 from the west. Have fun with your pilotage on that one.

Guess I've been flying long enough and without any high tech aids that it
just works for me. And, for the record, at Moriarty we routinely fly at
altitudes up to 18,000 MSL with surface altitudes from about 5,000 up to
around 13,000 MSL.

Personally, I get nervous below about 12,000 MSL unless I'm within gliding
distance of a landable field and, for that reason, my flights aren't nearly
as long as they could be.

"jfitch" wrote in message
...
On Monday, July 15, 2013 8:51:18 AM UTC-7, Dan Marotta wrote:
I've been using XCSoar on an Android with a CAI-302 for almost 3 years now

and the wind calculation/display works perfectly.



If you tecchies would look outside once in a while instead of at your

whiz-bang panels, you could compare the wind velocity arrow to the cloud

shadows. Oh, wait! That would be too much like pilotage.



If you are flying in the flatlands where the clouds are 2000 AGL and likely
to be blowing the same way as the air you are flying in, that works great.
Otherwise, not so much.....

For example, final glide Saturday, 77 miles Mt. Patterson to Truckee out of
17,500. Cloud drift not discernible from altitude (or forming ahead as fast
as drift), wind lines on Topaz Lake indicate calm, wind lines on Lake Tahoe
indicate SSW about 10, AWOS at Truckee reporting W at 7, Butterfly showing
25 knots south. Butterfly says tailwind (and it is correct, too), XCSoar
says headwind. Drop through about 12,000 ft., wind switches to 7 from the
west. Have fun with your pilotage on that one.

On Monday, July 15, 2013 12:03:05 PM UTC-4, Marc - Butterfly Avionics wrote:

LX: sorry guys, there is no spec on the LXWP wind direction (at least nobody knows of a specification). Both directions are commonly used depending on software you use. Rumors say that even LX devices exist that send data differently (is this true? i don't know). iGlide e.g. features a switch for that.

We will sure use the most common "interpretation" but the question is which one is the most common?

Does anyone know if XCSoar correctly interprets the wind direction information that it gets from a V7? Or does XCSoar have the same problem with the V7 that was reported here for the Butterfly/XCSoar combination?

I understand that the problem is that there is no known specification for the windage protocol and so it is not a defect attributable to either vario or XCSoar.

On Wednesday, July 17, 2013 2:28:21 AM UTC+2, son_of_flubber wrote:
Does anyone know if XCSoar correctly interprets the wind direction information that it gets from a V7?

This question cannot be answered, because the V7 does not send a wind estimate over the wire.

On Wednesday, July 17, 2013 5:56:21 AM UTC-4, Max Kellermann wrote:
On Wednesday, July 17, 2013 2:28:21 AM UTC+2, son_of_flubber wrote:

Does anyone know if XCSoar correctly interprets the wind direction information that it gets from a V7?



This question cannot be answered, because the V7 does not send a wind estimate over the wire.

Right, the issue never comes up. The V7 calculates a wind vector for it's internal "get home" function, but it sends airspeed "over the wire" to XCSoar. XCSoar combines the V7 supplied airspeed with the GPS supplied ground speed and direction to calculate a wind vector.

XCSoar can calculate wind based on GPS position (alone) over time but the V7 supplied airspeed makes the wind calculation more accurate and robust.

Did I get that right?

On Wednesday, July 17, 2013 5:13:10 PM UTC-7, son_of_flubber wrote:
On Wednesday, July 17, 2013 5:56:21 AM UTC-4, Max Kellermann wrote:

On Wednesday, July 17, 2013 2:28:21 AM UTC+2, son_of_flubber wrote:



Does anyone know if XCSoar correctly interprets the wind direction information that it gets from a V7?







This question cannot be answered, because the V7 does not send a wind estimate over the wire.



Right, the issue never comes up. The V7 calculates a wind vector for it's internal "get home" function, but it sends airspeed "over the wire" to XCSoar. XCSoar combines the V7 supplied airspeed with the GPS supplied ground speed and direction to calculate a wind vector.



XCSoar can calculate wind based on GPS position (alone) over time but the V7 supplied airspeed makes the wind calculation more accurate and robust.



Did I get that right?

I will say that the inertial derived wind vector from the Butterfly Vario is awesome.

On Thursday, July 18, 2013 2:13:10 AM UTC+2, son_of_flubber wrote:
XCSoar can calculate wind based on GPS position (alone) over time but the V7 supplied airspeed makes the wind calculation more accurate and robust.

Did I get that right?

Mostly right, let me clarify:

With just GPS, XCSoar can only use the "Circling" algorithm, which compares ground speeds in opposite circle positions. Wind calculation occurs only while thermalling.

With airspeed input, XCSoar uses an Extended Kalman Filter (https://en.wikipedia.org/wiki/Extended_Kalman_filter) which continuously compares airspeed and ground speed, and can always derive an accurate wind vector while flying straight.

There is a big difference in quality/accuracy and speed between the two algorithms. According to my tests, the EKF algorithm delivers the exact wind after flying straight for less than 20 seconds, while the circling algorithm takes more than a minute of circling, with varying results.

For those curious, this link has a dump:

http://bugs.xcsoar.org/ticket/2961#comment:4

This is a bug report from a user, he uploaded a NMEA dump, which I fed into both algorithms. Observe the volatility of the CirclingWind results.

TL;DR - feeding airspeed into XCSoar is totally worth it. Get a vario that is capable of doing so.

On Wednesday, July 17, 2013 8:40:41 PM UTC-7, Max Kellermann wrote:
On Thursday, July 18, 2013 2:13:10 AM UTC+2, son_of_flubber wrote:

XCSoar can calculate wind based on GPS position (alone) over time but the V7 supplied airspeed makes the wind calculation more accurate and robust..



Did I get that right?



Mostly right, let me clarify:



With just GPS, XCSoar can only use the "Circling" algorithm, which compares ground speeds in opposite circle positions. Wind calculation occurs only while thermalling.



With airspeed input, XCSoar uses an Extended Kalman Filter (https://en.wikipedia.org/wiki/Extended_Kalman_filter) which continuously compares airspeed and ground speed, and can always derive an accurate wind vector while flying straight.



There is a big difference in quality/accuracy and speed between the two algorithms. According to my tests, the EKF algorithm delivers the exact wind after flying straight for less than 20 seconds, while the circling algorithm takes more than a minute of circling, with varying results.



For those curious, this link has a dump:



http://bugs.xcsoar.org/ticket/2961#comment:4



This is a bug report from a user, he uploaded a NMEA dump, which I fed into both algorithms. Observe the volatility of the CirclingWind results.



TL;DR - feeding airspeed into XCSoar is totally worth it. Get a vario that is capable of doing so.

How do you know a wind vector is exactly accurate? Genuinely interested....

Also I can't claim to understand the EKF filter, is it possible to describe in layman's terms how the vector is derived from straight flight observations? Component velocity obviously, but the vector?