9B Metrics

[Paco Gorina]

Please load a corrected version from the Proves branch. Also has support for showing a map with the track. Look at to right when looking at an opened track.

Map has 2 buttons, cross closes it and the arrows adjust size. 

At bottom buttons select map type.

There was an error in the readong of temperature.

[Chriull]

Your new version just made it to the app store!

[Gil]

I could get the new app on my iPhone. I tested a little bit and generated a zip file that I sent on my PC. I extracted it and typed "gnuplot 9B.dem" but I got an error.

Any help ?

[Paco Gorina]

I need to install GNUplot in osx ?

[Gil]

Don't mind Paco, Paco. Chriull will find the mistake. I can wait. 

[Chriull]

27 minutes ago, Gil said:

I could get the new app on my iPhone. I tested a little bit and generated a zip file that I sent on my PC. I extracted it and typed "gnuplot 9B.dem" but I got an error.

Any help ?

Maybe it works if you change the "%.3f" to "%f" - dont' think so but who knows...

Anyhow - attached my latest running version of 9b.dem

Otherwise it's maybe somthing with the locale settings or there is some difference in the 9BMetrics we have - i just installed the one from the app-store and had no data logged with it so far, i have the 9BMetrics version from github - so there also could be some differences. By tomorow i should be able to take a look at it...

9b.dem

[Chriull]

10 hours ago, Gil said:

Any help ?

Found the problem - divisions by zero are not really benefical...

This Version of 9b.dem now should work fine. If old logs are resent with 9BMetrics 1.1, the script should to fine with the "old" data, too.

Still Todo's:

- i did not find a formating string for sprintf to suppress the trailing zeros after the decimal point

- nicer scaling option. By now i scale with powers of a fixed value (in this script it uses 5 - so scale factors can be .... 0.2,1,5,25,125..., for 10 it would be ... 0.1,1,10,100...) Nice would be factors like ... 0.1,0.2,0.5,1,2,5,10,20,50,100, .... Think i maybe found a way for this just while writing this lines - could be in the next version...;)

- There are imho no procedures in gnuplot? So there is too much code repetition by now... ;( Have too look a bit deeper, what is possible with functions.

- Line style and Line colors: I am not happy with the standard coloring that happens with this version

- Show a grid

9b - V3.01.dem

[Chriull]

22 hours ago, Paco Gorina said:

Please load a corrected version from the Proves branch. Also has support for showing a map with the track. Look at to right when looking at an opened track.

Map has 2 buttons, cross closes it and the arrows adjust size. 

At bottom buttons select map type.

There was an error in the readong of temperature.

Adding the map is a great idea!

(btw - when there is no gpx data for a log, the app crashes...)

The  logs with the error can be exported again wo problems!

I have to look at my version again - temperature is still not included, when i export data from the graphical view?

With some logs the graphical view only shows the first part of the recording - is there some size limitation?

    Edit: Did not find an example - i am not sure if this was just my imagination or it really happens. Or just from time to time? - i will report, once i have      something in detail regarding this topic.

While overhauling the gnuplot script i found some very strange Voltage "hops" in the graphs. With zooming in it seems, that there are from time to time for over 30 secs no new Voltage samples? Looks like this only happens with the Voltage samples - all others have no "sample breaks"

These are 2 logs with some zooms: http://imgur.com/a/f633T

Edit: New gnuplot script version attached

9b - V3.02.dem

[Paco Gorina]

29 minutes ago, Chriull said:

(btw - when there is no gpx data for a log, the app crashes...)

Corrected. Map button is only enabled if there is map data

29 minutes ago, Chriull said:

temperature is still not included, when i export data from the graphical view?

No, I will include it.

 

33 minutes ago, Chriull said:

While overhauling the gnuplot script i found some very strange Voltage "hops" in the graphs. With zooming in it seems, that there are from time to time for over 30 secs no new Voltage samples? Looks like this only happens with the Voltage samples - all others have no "sample breaks"

Do you have the logs. I would like to look at the dates. When there is the same value repeated it puts only two values :

....

first time, value

last time, same value

 

...

 

So we may detect if it is because there are no samples or because the sample is repeated.

[Chriull]

13 minutes ago, Paco Gorina said:

Do you have the logs. I would like to look at the dates. When there is the same value repeated it puts only two values :

....

first time, value

last time, same value

 

...

 

So we may detect if it is because there are no samples or because the sample is repeated.

The sample is repeated, as it seems:

...

958.492	5507
958.545	5508
997.776	5508
997.816	5662
1.028.466	5662
1.028.488	5598
1.050.200	5598
1.050.205	5378

...

There are ~39 secs for sample 5508 and ~21 sec for sample 5598. This is then a Ninebot issue?

Edit: and ~31 sec 5662 - almost missed that repetition

[Paco Gorina]

Added Temperature.Verified that voltage is read in the "fast" loop. Seems more a problem of Nb. than other

958.492 5507

958.545 5508

997.776 5508 Repeated value. 

997.816 5662

1.028.466 5662 Repeated value

1.028.488 5598

1.050.200 5598 Repeated value

1.050.205 5378

It really seems as nb voltage value is updated quite slowly. There is no difference i how the program treats Current and Voltage except that Current has sign

[esaj]

11 hours ago, Chriull said:

Found the problem - divisions by zero are not really benefical...

[Gil]

14 hours ago, Chriull said:

Found the problem - divisions by zero are not really benefical...

This Version of 9b.dem now should work fine. If old logs are resent with 9BMetrics 1.1, the script should to fine with the "old" data, too.

Still Todo's:

- i did not find a formating string for sprintf to suppress the trailing zeros after the decimal point

- nicer scaling option. By now i scale with powers of a fixed value (in this script it uses 5 - so scale factors can be .... 0.2,1,5,25,125..., for 10 it would be ... 0.1,1,10,100...) Nice would be factors like ... 0.1,0.2,0.5,1,2,5,10,20,50,100, .... Think i maybe found a way for this just while writing this lines - could be in the next version...;)

- There are imho no procedures in gnuplot? So there is too much code repetition by now... ;( Have too look a bit deeper, what is possible with functions.

- Line style and Line colors: I am not happy with the standard coloring that happens with this version

- Show a grid

9b - V3.01.dem

We progress. I now reach line number 100 but some problems still happen.

Difficult to understand as scale factor is used several times before line 102.

[Chriull]

Maybe you just send me your data file - should be the easiest to find the fault. Or maybe i already removed the failure with v3.02?

[Gil]

18 minutes ago, Chriull said:

Maybe you just send me your data file - should be the easiest to find the fault. Or maybe i already removed the failure with v3.02?

I am so sorry I did not notice your v3.02. I test it at once.

AND IT WORKS !! Thank you !

[Gil]

Still some issues, but graphical. The legend is inadequate, as you can see. I joined my data file.

 

9B_20160414_234806.zip

[Chriull]

9 hours ago, Gil said:

Still some issues, but graphical. The legend is inadequate, as you can see. I joined my data file.

With V3.02 division by zero was considered, but it was still possible that logarithm of negative values where tried... So it went beyond the already opened vortex

V3.03 should be stable regarding this probs. *knockingonwood*

The graphical issues arose from using head/cut/tr to extract the names for the values. As you see in your graph also the date/time of the log in the title ist 1970-01-01... - so you have no coreutils (package including this three commands) installed...

So i added in V3.03 a switch (use_columnheaders=1) which again uses the internal gnuplot functions instead of head/cut/tr. So you get the legend again, but then there is no possibility for the log date and labels for the left and right y-axis. (You can see how it would look at http://imgur.com/a/f633T)

With this version one also has the possibility to change the line width for each data set (like in the graph here for Voltage, Alt and Temp) and set colors for each dataset. And of course also which datasets are shown at all and if they use the left or right y-axis.

The scaling with the factors ..0.1,0.2,0.5,1,2,5,10,20... should work, too.

The script tries (in the default configuration) to show the current as big as possible, and scales all the other values to fit within this range. This could lead to the "problem", that if one has only negative alt values (going down) and no negative currents (only accelerating while going down ) the alt graph gets scaled to "nothing"... But i assume that wont happen too often - i don't know yet, but maybe i add either the possibilty to "shift zero" and use the whole graph for values as alt (is a relative value anyway), let it squeeze the current graph by ~1/3 and/or give the possibility to set predefined fixed scales for each dataset...

The Roll Angle is not included as shown value in the default configuration - he imho "only disturbs" the graph normaly. If one wants to look at it, a pitch/roll angle graph or similar can be configured quite easily for special purposes.

Ps.: Could it be, that you use the same version as @Blunzn:

I use gnuplot v5.0 rc2 and thats much nicer for zooming around in the graph... There you define with the right/left mouse button a window to be zoomed into and then scoll around with the cursor keys (until it crashes sometimes )

9b-V3.03.dem

[Chriull]

Normally the tiltbacks are about 5° - today i got a "nice" one with ~12°... Max speed measured from 9BMetrics was 23,7 km/h. Think i'm going to accelerate softer in the future - maybe...

Looking at the graph one sees, that the tiltback stopped at about 7°, gave me about half a second to react and then went up to 10-12°... Could be, that it took me this half second to regain balance again and start to fight with the 9Bot - its a bit awkward to lean back to brake once this tiltback angle is reached...

Did i alreay mention that i love this app - it's really great!

@Paco Gorina:

- Once the bluetooth dis- and then reconnects (which works great) all the data till reconnection is lost. At my ride home i stopped at two shops where i turned off the 9Bot and now only have the data since my last stop. Noticed it by the map view that a part is missing and also the Altitude starts at ~6m which can not be.

- Opening a data file (or pushing the info button to send the data) can take quite long - so some animation/progress bar could be nice to show the user that something is happening?

 

And again i hat quite a "voltage jump" without any reason logged - somehow a real strange behaviour of the ninebot. At least all the voltage jumps are not downwards, so that the controller could believe he's out of battery and stops me while driving  Would be interesting if this is just my 9Bot or a design issue that happens with all others, too.

- 

[Paco Gorina]

The data you lost when stopping the 

3 hours ago, Chriull said:

- Once the bluetooth dis- and then reconnects (which works great) all the data till reconnection is lost. At my ride home i stopped at two shops where i turned off the 9Bot and now only have the data since my last stop. Noticed it by the map view that a part is missing and also the Altitude starts at ~6m which can not be.

Found. It reinited when the device connected. Changed to init data just when pushing button .

3 hours ago, Chriull said:

- Opening a data file (or pushing the info button to send the data) can take quite long - so some animation/progress bar could be nice to show the user that something is happening?

Agree. Will do it but need a good idea for the interface.

[Paco Gorina]

Pushed update 

[Skara]

Dang, a rather heavy colleague of mine just killed my 9B! He fell backwards and his weight shot the 9B forwards even though it was trying to get under him... and now it's dead. There's resistance when turning the wheel, so probably something broke in the motor. I already sent an email to the retailer, but I've been riding it every day, and now I'm dreading the coming weeks without it.  And the weather is perfect too!

This is a little off-topic, but my first thought when I realised it was dead: I wish I had had 9B Metrics running. At least the sacrifice would have produced one data point on the limits of 9B.

[SuperSport]

52 minutes ago, Skara said:

Dang, a rather heavy colleague of mine just killed my 9B! He fell backwards and his weight shot the 9B forwards even though it was trying to get under him... and now it's dead. There's resistance when turning the wheel, so probably something broke in the motor. I already sent an email to the retailer, but I've been riding it every day, and now I'm dreading the coming weeks without it.  And the weather is perfect too!

This is a little off-topic, but my first thought when I realised it was dead: I wish I had had 9B Metrics running. At least the sacrifice would have produced one data point on the limits of 9B.

Probably needs a Motherboard. That's a common scenario and symptom of burned Mosfets. 

[Gil]

13 hours ago, Chriull said:

V3.03 should be stable regarding this probs. *knockingonwood*

This version 3.03 works fine. I have to activate the UseColumnHeaders flag and everything is OK. Thank you very much for the nice work.

[Chriull]

6 hours ago, Skara said:

Dang, a rather heavy colleague of mine just killed my 9B! He fell backwards and his weight shot the 9B forwards even though it was trying to get under him... and now it's dead. There's resistance when turning the wheel, so probably something broke in the motor. I already sent an email to the retailer, but I've been riding it every day, and now I'm dreading the coming weeks without it.  And the weather is perfect too!

This is a little off-topic, but my first thought when I realised it was dead: I wish I had had 9B Metrics running. At least the sacrifice would have produced one data point on the limits of 9B.

 

5 hours ago, SuperSport said:

Probably needs a Motherboard. That's a common scenario and symptom of burned Mosfets. 

Thats quite off-topic in this thread (hijacking?), but a breaking cut-off together with increased turning resistance of the wheel afterwards is for 99.99% a dead motherboard (burned mosfets) as @SuperSportstated. And you could not revive the wheel by connecting it to the charger?

Edit: ps: i know thats jot funny in your case, but i assume you would have been number one in regard of logged current, beating @Blunznwith his 45A if you had 9BMetrics running while this incident...

[Gil]

Perhaps could we pin a post at the beginning of this thread with everything needed to use 9BMetrics ?

Short description + link to gnuplot 5.02 for windows + 9B V3.03.dem + link to user guide ?

And I almost forgot link to Paco and Chriull's Paypal accounts ;-)