I tried looking at some of my data in the FlySight .csv file format (download off Dekunu cloud) and the vertical velocity looks very suspect and much noisier than North or East Velocity. Overall velocity data is off while Lat/Lon/Altitude is seemingly accurate. I’m working on my swooping, but I know I’m not getting the speeds my Dekunu data reads (If I could actually hit 130 mph on a Velo 103 loaded at 2.3 through the gates, then I should enter some competitions right now! :-p ).
I checked the data recorded in CSV file as well. (using FlySight Viewer, Excel, and MatLab to look at data, so it’s not a units conversion issue when importing data into any of the software I’m using to look at the data) I suspect it might be an issue with whatever conversions are being done either on the cloud or the Dekunu unit to generate the .csv files. (I presume the flysight.csv file you can download off the cloud is being converted/reformatted from whatever the uploaded Dekunu action file is?
Down velocity could certainly use some some smoothing on the data, it looks like North & East velocities are smoothed and Down velocity is not.
I’m an aerospace engineer by day, and work with trajectory data on a regular basis. If possible, I would love the chance to work with Dev Team to try and trouble shoot this.
I’m hoping to just use my Dekunu to track canopy performance, and not need a Flysight in addition to my Dekunu.
Still having issues with this. I just replied on ColOfAbRiX’s thread about his python function as well.
so the Altitude data in the (hMSL) column of the csv file appears to be correct and is being updated at regular 10hz rate. Lat/Long is being held constants for all 10 data points per second, as are North and East Velocity. Down velocity is being reported on the even integer time steps, but reported as 0 m/s on the other 9 data points until the next even second.
I wanted to make a matlab script to calculate my own velocity vector based on lat/long/alt and time, but I can’t even do that if lat/lon data is only being output at 1 hz. Having Lat/Lon data to 6 decimal points should in theory be an accurate location to within 4-6 inches. Simply reporting Lat/Lon as constant for entire 1 second (10 data points) is not helpful.
any word from Dev Team on getting the .csv files to output all vectors at the same data rate? Seriously considering just biting the bullet and buying a flysight if my Dekunu can’t report accurate data
Just got home and put my Dekunu into USB mode, grabbed the corresponding log file, so I have the full data file ‘.csv’ and the flysight formatted ‘.csv’ downloaded from the cloud portal under actions (right click and “save as” the flysight format option). I wanted to compare the same log file off Dekunu vs off cloud.
Confirmed its definitely something in the interface of the cloud portal, presume there’s a file conversion function that takes the full data file uploaded to the cloud from the Dekunu and truncates the plane ride off and converts to flysight format. That converstion must need some work.
Next step for me personally is write a MatLab routine to convert the raw log file directly off the Dekunu (USB mode) to a flysight .csv format, I think I might be able to by pass the cloud portal for now until the flysight file conversion is working.
Interestingly enough, the plot of altitude and speed, appears to be reasonably correct. So whatever function is converting data to display on the cloud plot seems to be working much better than the flysight file conversion.
