It will take some time for me to add SBN or SBP support. At the same time it's easy for me to add the NMEA (i.e. the file with NMEA sentences). Would it be helpful for someone?

Select to expand quote

arseny said..
>>



If you will get time then please write the GPS speeds of the WindRace's 3 points for best 2 sec.
Thank you for your feedback.

Sorry I can't figure out how to do that, the report only tells you the average, not each point?

Select to expand quote

arseny said..
It will take some time for me to add SBN or SBP support. At the same time it's easy for me to add the NMEA (i.e. the file with NMEA sentences). Would it be helpful for someone?

Yoyo was saying today, that when you save an SBN file to GPX in realspeed, the GPX file has the doppler data in it.
But yes if the NMEA has the doppler speed, I'm sure that would be an advantage to somebody.

The NMEA contains Doppler's speed at least if GPS chipset supports Doppler's data (speed and direction). The official GPX format doesn't contain the speed field (see the set of possible fields for the GPX's way point is at www.topografix.com/GPX/1/1/#type_wptType - no speed field in GPX format at all according to that resource).

Possibly some manufacturers of GPS devices write the speed value via the fields of their own GPX extensions. If so then I can add the speed in the same way. I just need an example of such file to do that (email support@windrace.info)

Select to expand quote

decrepit said..

arseny said..
>>



If you will get time then please write the GPS speeds of the WindRace's 3 points for best 2 sec.
Thank you for your feedback.

Sorry I can't figure out how to do that, the report only tells you the average, not each point?

There is a link "Build GPS Track and charts" right in WindRace's report. Turn off the layers except "GPS Points" and best "best 2 sec GPS". Use the layers button (at the right top corner of the trace area: ). After that please zoom the trace to the "best 2 sec" area (highlighted in red) and click on the 3 points of that area to see their details (the GPS speed is among the displayed data).

I prefer to use browser on a computer for such actions (screen wider and browser works faster). I use the report's "send" link on Android and google drive to copy the report on a computer:


Sorry for troubling you

Yes, very nice, the more I see of this app the more I like it.

1 point details:
moment: 00:04.35
speed: [GPS] 16 knots; 15.817 knots
pos: -32.58805°, 115.6444233°
direction: [GPS] 92.41°; 90°
sattelites: 10
HDOP: 0.86, VDOP: 1.45, PDOP: 1.68

2 point details:
moment: 00:04.36
speed: [GPS] 15.941 knots; 16.136 knots
pos: -32.5880533°, 115.6445117°
direction: [GPS] 92.26°; 92.552°
sattelites: 10
HDOP: 0.83, VDOP: 0.82, PDOP: 1.17

3point details:
moment: 00:04.37
speed: [GPS] 16.47 knots; 15.817 knots
pos: -32.5880533°, 115.6445983°
direction: [GPS] 89.11°; 90°
sattelites: 10
HDOP: 0.86, VDOP: 1.44, PDOP: 1.68

And here's the GPSResults 2 points
gps results 2sc

#286 083257 16.017

#287 083258 15.959
#288 083259 16.484

#289 083300 15.784


Very strange, the numbers are almost the same?
There's no where else on the file, that has numbers close to this, are the 2 programs are getting different numbers from the phone?





Looks the same to me

Thank you for the details. Now I see what the "Trapeze" calculation means

The WindRace's calculations are in meters per second (behind the scene). And the report's data is in the same units. Possibly I missed the tail of the number somewhere in calculations. May be one of the reasons declared before... Will check the code and write back soon.

Q: Is the "Trapeze" calculation a kind of a standard. I.e. should I use that way of calculation for best 2 sec (10sec, 100m and etc)?

decrepit, can I also ask you to write the coordinates of those three points from GPSLogit (it can help to understand the reason of the difference)

great to see someone going to such lenghts !! Thanks Arseny (and thanks Michael too for the testing !!)

Select to expand quote

arseny said..
>>> If so then I can add the speed in the same way. I just need an example of such file to do that (email support@windrace.info)

Ok sent yesterday.

Select to expand quote

arseny said..
decrepit, can I also ask you to write the coordinates of those three points from GPSLogit (it can help to understand the reason of the difference)

No worries, realspeed is easier so I've temporarily skipped into windows here's what it says.
Best 2sec 16.178kts
#287, 01:32:58, 115.6445117, -32.5880532, dop speed 15.959
#288, """""""59, """"""""""983, """"""""""""", """"""""" 16.484
#289, """""3:00, """".64468 , """""""""499, """"""""" 15.784



I don't think you need worry too much about "Trapeze" there's only one of our programs, "realspeed" that uses it by default and that hasn't been maintained this century.
It seems current thinking is just to use 2 points, but I can't tell you why.
The maths involved is way over this old head!
Some people may find it a useful option though, so I won't discourage you from pursuing it.

The differences seen could be just lack of resolution in the files, but also could be the difference method of integration used.

GPS-Results has a number of options. It used cubic spline by default for the 2 sec and 10 sec but not sure if it is Trapezoid or Rectangular for the others. If you run that version of RealSpeed I sent you Mike you should be able to choose either Rect or Trap and work it out.

Select to expand quote

decrepit said..

arseny said..
>>> If so then I can add the speed in the same way. I just need an example of such file to do that (email support@windrace.info)

Ok sent yesterday.

Hi Michael

Thank you for letting me know (found the letter in my spam folder).

Bingo, now I see the the speed in GPX! :).

WindRace generates GPX file according to the GPX 1.1. It is the latest format version which has no speed field actually. The file you have sent uses GPX 1.0 (the previous format version, released in 2002). As I found the GPX 1.0 format supports the optional speed field www.topografix.com/gpx_manual.asp . I'll add the GPX 1.0 format support and will made it selected by default.

I'm planing to prepare an update at the next week. The update will include:
1) calculations improvement
2) the stats for 2 nearest points (like another one kind of stats)
3) GPX 1.0 output

any suggestions are welcome

Select to expand quote

seanhogan said..
great to see someone going to such lenghts !! Thanks Arseny (and thanks Michael too for the testing !!)

If you really found that the app is useful then please rate it in Google Play (the asterisks at the bottom of the WindRace's listing). It affects to the amount of users. I'll receive more feedbacks from users and reports about bugs (if any) to fix. (and thanks Michael too for the testing !!)

Arseny, I note when looking at the Windrace generated gpx files in notepad or wordpad the speed (and other readings) have only 1 decimal place whereas gpsBabel has about 6 or 7 (?? way overkill). I guess this is probably different from your Windrace Report calculations but maybe an extra decimal point may make things more accurate when running the tracks on 3rd party software. ie(Realspeed, GPSResults or GPSActionReplay).
Some software loggers also put out COG info and if the tracks were calculated from SOG and COG rather than positional data then I'm sure they will be far less "noisy" .
HDOP has been bandied about a lot (and some of the above 3rd party programs wont run if you don't have it) but really it is a hangover from positional trackpoint and as what most of us are interested in is speed, then the SDOP is probably more important than HDOP.

This is what I find very strange. the same data from the same phone is simultaneously sent to two programs, shouldn't this data be identical??????

Data in WindRace HTML report
1 point details: moment: 00:04.35 speed: [GPS] 16 knots; 15.817 knots pos: -32.58805°, 115.6444233°

2 point details: moment: 00:04.36 speed: [GPS] 15.941 knots; 16.136 knots pos: -32.5880533°, 115.6445117°

3 point details: moment: 00:04.37 speed: [GPS] 16.47 knots; 15.817 knots pos: -32.5880533°, 115.6445983°

Data from GPSLogit .sbp file via reelspeed.
#287,01:32:58, 115.6445117, -32.5880532, dop speed 15.959
#288, """""""59, """"""""""983, """"""""""""", """"""""" 16.484
#289, """""3:00, """".64468 , """""""""499, """"""""" 15.784

So windrace "2 point" is logit # #287 but the co-ords are slightly different, -32.5880533° for WR and -32.5880532 for logit, and speed also different 15.941 kts for WR and 15.959 for logit.

"3 Point" WR and #288 logit have the same co-ords difference and speeds 16.47 WR and 16.484 for logit.

So are the two programs interpreting the raw data differently, or is the phone sending different data to each app???

If it's two lots of data, what's the time difference? Can we combine multiple files to get a multiple Hz result? (I bet we can't)

Do we really want the same results as GPSlogit knowing the accuracy issues.

IMO we should not be comparing any of the Windrace results against logit or having it copy logit (apart from some of the functionality) less it ends up relegated as another "live reference only" program but not much good for final results.

I'll use the Pro version and run it up against the GT11 next session to compare results

I think you miss the point Alby, there's no issues with logit at all. Logit isn't the source of the accuracy issues, it's the phones raw data.
The issue is with the strange things some of the phones do.
At the moment WindRace stops the display off syndrome, but there's other problems besides that, and your phone doesn't have that problem anyway.

My Question is about how 2 apps apparently get different raw data from the same source, not about how the data is processed.

Ahhhh you must have typed that slower, I now understand

So arseny, in the comparison above, I suspect WindRace chooses which 3 points to analyse from the positional data, whereas realspeed is choosing them from the doppler data. There's a disparity in where the fastest speed is between them.





Red trace is doppler yellow positional, this is with the positional 2sec selected, when doppler 2s is selected it moves to the right.

Select to expand quote

elmo said..
Ahhhh you must have typed that slower, I now understand

It's all that red fur it has to get through

Select to expand quote

decrepit said..
So arseny, in the comparison above, I suspect WindRace chooses which 3 points to analyse from the positional data, whereas realspeed is choosing them from the doppler data. There's a disparity in where the fastest speed is between them.





Red trace is doppler yellow positional, this is with the positional 2sec selected, when doppler 2s is selected it moves to the right.

Michael, the 'best 2 sec' uses positional data and the 'best 2 sec GPS' uses the speed from the GPS chipset (Doppler). If you suspect that the 'best 2 sec GPS' uses the values of the positional speed then it is easy for you to check the suspicion: Use the "speed chart" in the WindRace's report :



The chart displays the both speeds (positional and GPS). Just zoom it to the range of moments 00:04.34 - 00:04.38 or around it (the sliders are at the top of the chart). I think now that yoyo is closer (yoyo, I need some time for SDOP, thanks for the proposal) - the extra decimal points and trimming instead of rounding could be the real reason. Because the same location data should arrive on both apps installed on the same device. But the Android itself provides two sources for the locations: NMEA sentences (coming from GPS chipset) and the positional data on the Android's level. Android's level possibly corrects the data (then we have another possible reason).

As I wrote I'll check the calculations soon.

Arseny and Mike, when I look at the Windrace gpx files with say notepad it seems to me that the <desc> gps speed o.o m/s ( km knot miles) </desc> etc is just the speed calculated from the position Lat and Long data. This is different than when you look at the track report on the Android unit itself and you can see the cleaner Doppler (ie gps unit calculated) tachyograph VS the "noisy" trackpoint data.
I removed the speed info on the exported windrace gpx file and it made no difference to the tachiograph speed (which was basically the trackpoint data with no Doppler info when looked at with Real speed.

Bit rushed at work .
Will make point clearer when have more time.

Select to expand quote

arseny said..
>>>>

Michael, the 'best 2 sec' uses positional data and the 'best 2 sec GPS' uses the speed from the GPS chipset (Doppler). If you suspect that the 'best 2 sec GPS' uses the values of the positional speed then it is easy for you to check the suspicion: Use the "speed chart" in the WindRace's report :

>>>>>

The chart displays the both speeds (positional and GPS). Just zoom it to the range of moments 00:04.34 - 00:04.38 or around it (the sliders are at the top of the chart). I think now that yoyo is closer (yoyo, I need some time for SDOP, thanks for the proposal) - the extra decimal points and trimming instead of rounding could be the real reason. Because the same location data should arrive on both apps installed on the same device. But the Android itself provides two sources for the locations: NMEA sentences (coming from GPS chipset) and the positional data on the Android's level. Android's level possibly corrects the data (then we have another possible reason).

As I wrote I'll check the calculations soon.

OK, best doppler are at 36, 37 and 38sec, best positional are at 35, 36 and 37.
So it's as I said, WindRace is using the positional data to determine the best doppler speeds, that's probably the cause of biggest difference in logit and windrace speed results.





The peak of the red line is 37s and the peak of the blue line is 36s The WindRace gps speed is taken from 35, 36 and 37. The three programs I have use 36, 37 and 38 from the logit data.

Although looking at the graph more closely, I think WindRace may have picked the best three, again I have no idea what is going on here????????????????

Select to expand quote

yoyo said..
Some software loggers also put out COG info and if the tracks were calculated from SOG and COG rather than positional data then I'm sure they will be far less "noisy" .
HDOP has been bandied about a lot (and some of the above 3rd party programs wont run if you don't have it) but really it is a hangover from positional trackpoint and as what most of us are interested in is speed, then the SDOP is probably more important than HDOP.

You was right. I missed some decimal data in the report and GPX file. The update will be soon.
Thank you.

If the "NMEA 0183" option is selected then WindRace uses $GPRMC sentence of NMEA protocol to extract the doppler's data. The speed over ground, Knots (SOG) an Course Made Good, True i.e. the true course over ground (COG) are in the $GPRMC sentence according to the aprs.gids.nl/nmea/#rmc . Those values are considered like "GPS speed" and "GPS direction" in the app (will rename "direction" to "course"). Those values will be added to the corresponding fields of the GPX 1.0 protocol (www.topografix.com/gpx_manual.asp and www.topografix.com/gpx_manual.asp#course).

As I found the SDOP value is provided by the SiRF chipsets (bioresonant.com/dl/download.htm?name=SDOP.pdf). I think that the SDOP data should be reflected in the NMEA by the SiRF specific sentence but i cant find it in their NMEA description at www.sparkfun.com/datasheets/GPS/NMEA%20Reference%20Manual-Rev2.1-Dec07.pdf .

Select to expand quote

Although looking at the graph more closely, I think WindRace may have picked the best three, again I have no idea what is going on here????????????????

WindRace have picked 3 points (the two seconds duration is usually between 3 GPS points). As I wrote I'll also add the data about "best 2 pts" and "best 2 pts" to the report.

Yesterday I've managed to record a report which has different areas of "best 2 sec" and "best 2 sec GPS". The speed chart corresponds to the track data in the report. Avg speed is evaluated during the track recording but the report and GPX file contains the speed and coordinates with the trimmed data in the decimal part of the number. I'm working on the fix. Will update the app soon.

Thank you for the investigation.

Thank you for app, Arseny.

It is great to find someone interested in our gps needs who has the skills to actually do something about it.

Whilst SDOP is "needed" supposedly for record performances personally I find the elevation/altitude more illuminating. I have a Canmore which always starts at about -400 to -250 metres elevation and the trackpoints bounce around all over the place but over about 10-15 seconds the elevation heads to near zero (+- a few metres sea level) and the trackpoints gravitate to the correct location. If I get a speed spike usually it is obvious as the elevation changes quite a bit so it is easy to tell the points are not valid. However if the SDOP info is there and easy to extract you might as well put it in.

Select to expand quote

yoyo said..

Whilst SDOP is "needed" supposedly for record performances personally I find the elevation/altitude more illuminating. I have a Canmore which always starts at about -400 to -250 metres elevation and the trackpoints bounce around all over the place but over about 10-15 seconds the elevation heads to near zero (+- a few metres sea level) and the trackpoints gravitate to the correct location. If I get a speed spike usually it is obvious as the elevation changes quite a bit so it is easy to tell the points are not valid. However if the SDOP info is there and easy to extract you might as well put it in.

Will add the altitude via www.topografix.com/gpx_manual.asp into GPX 1.0 according to its description.

Yes I also saw an hight elevation at the start or during radiointerference too. To add a kind altitude GPS filter (optional) is not a bad idea. Seems it will work for us.

Let me know if the chart (time/elevation) is needed in report?

Here's an example of why sdop values are important to us.




Yesterday I wore 2 GPSs, their results were very similar apart from the Alphas, the GT31 gave me a 23kt top score, but this alpha was the 2nd best with the GW52 and only a 22.9. The sdop values (the white dots alongside the speed graphs) show which GPS to to trust.
The GT31 has a big error margin at the gybe entry, but the GW52 has hardly any error at all.

BUT, I doubt you'll find any phones that have this data available.
If you come across any, please let us know, it would increase the value of phone data to the GTC

SDOP values are pretty meaningless for the the GPSTC. Nobody ever checks them and there's no mechanism in the posting system for rejecting tracks that are out of range or have errors. If you are going to use the values then you need to rebuild the system/software to take advantage of them. They may be great for record attempts but that's not what the challenge is about. I doubt even 1% of GPSTC speedsailors even know what they are. For the ones that spend hours poring over the data they are a bit of a giggle but making things simple and getting out on the water are more important to the majority of sailors.

Roo