Jump to content

Gil

Recommended Posts

  • Replies 220
  • Created
  • Last Reply

nice, now i can ride with your great app & watch vee73's videos at the same time, can't get better than this :)

maybe a feature for an future update ? :D:D

will try your gps version as soon as i get access to the .app file

IMG_4745b.pngIMG_4744b.png

IMG_4749b.pngIMG_4750b.pngIMG_4752b.png

Link to comment
Share on other sites

On 2.4.2016 at 3:51 PM, Paco Gorina said:

By the way i updated GitHub to have last modifications including GPS support

 

Thanks! Just tried it - works great! The gpx files can be easily imported in google maps ;)

Sampling got nicely faster - i summed up all the time differences between the current samples. There are two big accumulations arount 0,035 and at 0,07 secs sampling time. These values are about 70% of all samples. Up to 0,07 max sampling time covered with version 1.0 just 37% of all samples.

I thought that this should mean, that there are mostly about 14 to 28 samples per second (1/0,035 and 1/0,07) but looking at the graph it is still very edgy ;(

The average sampling time was still 0,1 sec - which correspond to the graph with something about 10 points per second. Looking at the data file it also looks like there are many fast samples, but then there are everywhere inbetween some quite "slow" samples which destroy the average... ;(

Will make some more logs while driving tomorow - maybe it was a bad evening for bluetooth connections ;)

Link to graph with the distribution of sample rates Version 1.0 compared to Version 1.1 http://imgur.com/bUpAoUr

Edit:

Just a link fyi, why i look at the sampling rates: 

@Jason McNeilshowed here the difference between 10Hz and 50Hz sampling - with 10Hz it is like the 9BMetrics log by now, just very "edgy". With 50Hz it starts to get smooth and looks about sufficient to capture all the spikes/changes. Unfortionatelly the used Wheel is not specified in the post - there could be theoretically differences with the needed sampling rate?

(Just an explanation to the numbers in Jason's Graph, which took me some time to understand: He stated 0,021 sec interval with 13 Datapoints which would lead to a much higher sampling rate than the 10 Hz. Here he had an typing error - the time difference is 0,021 Minutes and not seconds...)

Link to comment
Share on other sites

I have made a small histogram. First sample times if they are for the same value because they may be very large if the value doesn't change.

Results here 

histograma_1

Average is about 0.07 s, stdev 0.05s

Most of the samples are  between 0.02s and 0.06s

 

 

 

To get 50hz means a 0.02 s sampling rate, going to 0.03 only a 20% of the samples are from this frequency. Must see if it is caused by the wheel not giving more or just values not changing too much.

Link to comment
Share on other sites

Made some modification to try to get better sampling rate. Remembered Nyqvist theorem (https://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem) which essentially says that to reconstruct a signal with maximum frequency f needs a sampling frequency of 2f. So if we were having a sampling rate of 0.07 s then 2f = 14 hz and our limit frequency is about 7 Hz, not very good. Will see what happens with new data with the modifications in the sampling algorithm. Now the limit should be the Ninebot or the BLE Interface. Exploring the possibility of configuring haw many values we want to record fast and the rest may go slower. 

Link to comment
Share on other sites

On 5.4.2016 at 4:53 PM, Paco Gorina said:

Better sampling rates. Today in a real commuting of 2.2 Km and 11'29", about 14.000 current samples :

https://www.flickr.com/gp/56267272@N06/0sS06R

Mean 0.04s -> 25Hz sampling. Getting better.

Wow - great improvement!

9BMetrics records still nicely and stable (did not have time to try out your last commit ;( ), but today i noticed my first problem with it: I tried to send 6 logs from the last to days per email and some where send as txt file (2 files like in the old version) and some as zip file (3 files) and inbetween log files vanished from the table view. So i lost one log by this...

Link to comment
Share on other sites

Bug located and destroyed :D.

In some cases when the zip file wasn't created by some obscure reason, it presented the .txt file to send and then deleted it.

If you shared the .txt file it is possible to recover it. Just resend it to the iPhone (mail or message or anything) and select openIn.. 9BMetrics .

Pushed new version to github. If anyone uses the old, please upgrade

Link to comment
Share on other sites

I am not a Pyhon crack but as I have installed in iPhone and iPad <a href=https://appsto.re/es/P0xGF.i">Pythonista</a>, a  very good implementation of Python (2.7) for iOS and it supports MatPlotLib, I have developed two small scripts to show selections from the graph window i 9BMetrics.

First, create a folder in Pythonista to contain the scripts (let's call it 9BMetrics) and a sufolder called "Logs" for the logs. It should be named "Logs".

Second load the following scripts :

Plot

Import

They should be named Import.py and Plot.py.

Now when in 9BMetrics graph window sharing sheet be sure to enable pythonista 

https://www.flickr.com/photos/56267272@N06/shares/49462i

Now select Pythonista and navigate to the Import.py script and execute it

https://www.flickr.com/photos/56267272@N06/shares/8n62C1

It will copy the file to the Logs directory of Pythonista.

Then running Plot allows to select a file and which variables you want to display

All may be done in the iPhone, etc..

And here is an example of the result:

 

https://www.flickr.com/photos/56267272@N06/shares/3B72Kt

 

 

 

Link to comment
Share on other sites

On 5.4.2016 at 4:53 PM, Paco Gorina said:

Better sampling rates. Today in a real commuting of 2.2 Km and 11'29", about 14.000 current samples :

https://www.flickr.com/gp/56267272@N06/0sS06R

Mean 0.04s -> 25Hz sampling. Getting better.

I got with my first short drive 8657 samples within 274 seconds -> 31,5 samples a second! http://imgur.com/oh3jYx4 looks quite smooth!

On 8.4.2016 at 0:05 PM, Paco Gorina said:

Pushed new version to AppStore

With the new version you changed speed from "38" to "181". 181 only records absolute speed - so there is not directional information in it: here i did my first test with just pushing the wheel forward and backward with my hand: http://imgur.com/ogGkkYA

Edit: Attached find the new gnuplot skript for Version 3 to come...

 

9b.dem

Link to comment
Share on other sites

15 hours ago, Chriull said:

I got with my first short drive 8657 samples within 274 seconds -> 31,5 samples a second! http://imgur.com/oh3jYx4 looks quite smooth!

With the new version you changed speed from "38" to "181". 181 only records absolute speed - so there is not directional information in it: here i did my first test with just pushing the wheel forward and backward with my hand: http://imgur.com/ogGkkYA

Edit: Attached find the new gnuplot skript for Version 3 to come...

 

9b.dem

As fas a I know, wither 38 or 181 are absolute values and have no sign (I have never treated the sign ). Anyway I will add the 2 values to the program and make some tests.

It is possible that another variables has the direction.

Link to comment
Share on other sites

9 hours ago, Lukasz said:

Dear Paco,

Any chance for the Android version maybe?

Luk

+1

 

2 hours ago, Paco Gorina said:

Sorry, not Android expertise :(

Fantastic app! Would be great to get android version... Oh well.. ( 

Can't for the life of me force myself to like the apple watch... Have to admit - I've no reason to get any smartwatch, but I could easily buy an pleasant androidwear watch just for 9b1 use

Link to comment
Share on other sites

4 hours ago, Vik's said:

Can't for the life of me force myself to like the apple watch... 

The apple watch is too expensive.  But, it does give a very good experience, and it is VERY well built.

Unfortunately, without an iPhone, it doesn't do much good, other than a watch and calculator.  That's the boat I'm currently in.  I've not been wearing it since I upgraded to the Galaxy S7 Edge.  Will likely go back to apple due to this and touch screen issues on the S7 Edge.

Link to comment
Share on other sites

I wonder if a selfmade ios app like 9BMetris can send data to a pebble/gear s2 watch though their ios app(for gear s2 yet to come)- or one has to choose between a 9Bmetrics - watch connection or the brand specific app to watch connection...

Link to comment
Share on other sites

@SuperSport ..too expensive for what it is. But it's not the price that holding me back. As a owner of multiple fine swiss timepieces, I have to say that apple "watch" design is just [----] hmm... not my cup of tea... That, and the fact that I am not using an iphone (or any smartphone, other than at work) is a dealbreaker for me.

But that' is just off-topic :)


I really appreciate the Paco Gorina's work.

Link to comment
Share on other sites

I have found a Peeble SDK for iOS. Will give it a look the weekend

 

https://developer.pebble.com/tutorials/ios-tutorial/part1

Respect AppleWatch, it is a problem of what must and must'n do a watch.

Pactically, it is much better at presenting data from iOS (no big battery load) than at doing things itself. 

Ex. I have an application for mountain running / skimo. It works very well if it presents HR etc. from iOS and the band is connected to the iPhone.

If you use the watch HR it is OK for short workouts but it kills battery in long ones kill the battery.

An alternative is connect the Bluetooth band to the watch. It is OK but if you take the phone it is better to connect it to the phone.

I tryed to connect the Watch directly to the Ninebot but the SDK doesn't allows it for the moment.

 

Link to comment
Share on other sites

@Paco Gorina

Some new points while testing 9BMetrics for your information:

- two out of three logs could not be exported. Now if i click on the corresponding info button in the table view i get an Error: "Error when tprocessing files",0x00000...). But i still can enter the graphical view and export the data from there.

- exporting the data from the grafical view: temperature is not included

 

Link to comment
Share on other sites

  • esaj pinned this topic
  • Chriull unpinned this topic

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...