MG4 Review: new Carscanner Android Auto OBD app Test with OBDLink CX BLE adapter (also new on Carplay)

neerav

Established Member
Joined
Oct 15, 2023
Messages
71
Reaction score
83
Points
27
Location
Sydney, Australia
I use the Carscanner Pro Android app on my phone to see OBD data from my EV

Today I plugged my phone into my MG4 77 and was pleasantly surprised to see that Carscanner has a new Android Auto OBD app (also new on Carplay)

In this video I tested this new Carscanner Android Auto app with my MG4 77 and an OBDLink CX BLE adapter

 
Got a written synopsis?
The Car Scanner app (which reads data from an OBD port scanner so you can view data on the car not otherwise shown by the regular car systems) has been added to Android Auto (and Apple CarPlay I believe). Previously it was viewable only on the phone app, now you can see it on the car's display screen.
 
I can confirm it also works with Apple CarPlay.

The major limitation, at least in its present guise, is the data displays on the car's screen are limited to text only, no gauges like you can set up on the phone's screen. If it were able to display gauges on the car's screen then it would be much better.

Navigation is pretty basic, no swiping between screens just a layered menu. Also partly an issue with the lack of responsiveness of the car's touch screen. Too often the back button would not work.

Another issue is validating it is reading the correct data channel for what is displayed, while often the data from the correct data channel is nonsensical.
 
I can confirm it also works with Apple CarPlay.

The major limitation, at least in its present guise, is the data displays on the car's screen are limited to text only, no gauges like you can set up on the phone's screen. If it were able to display gauges on the car's screen then it would be much better.

Navigation is pretty basic, no swiping between screens just a layered menu. Also partly an issue with the lack of responsiveness of the car's touch screen. Too often the back button would not work.

Another issue is validating it is reading the correct data channel for what is displayed, while often the data from the correct data channel is nonsensical.
The data I’m getting seems less non-sensical since I changed this setting:

IMG_4277.jpeg
 
As this is an adaptor setting it helps with the Android version too.

For any values that are still always nonsensical, I delete them from the dashboards to improve the refresh time.
 
The data I’m getting seems less non-sensical since I changed this setting:
Thanks. I did that but still getting weird data.

e.g. here's one screen with battery SoC flipping between plausible and stupid:


and another which is different again, shows both SoC and SOC with an order of magnitude less stupid before flipping to something plausible but neither of which match the car's reported SOC:

 
Something a bit different for fun... (credit to Prof. Robert Chung for this concept).

Here's a plot I created from OBD data captured on my short drive from home this morning to a local town:

Screen Shot 2024-09-10 at 3.43.45 pm.png


It's a calculated estimate of the elevation profile of the drive using the relevant equations of motion using the power, speed and distance data the OBD CarScanner app supplies.

First of all it's wrong. The calculated elevation changes are way over exaggerated.

I'm yet to sort through the reasons. The raw data format from the CarScanner app doesn't help and I'm assuming my initial attempt to massage the data into a useable format likely resulted in virtual elevation calculation errors, but really it's just an example of a visual diagnostic tool.

Essentially the principle is balance sheet accounting for energy.

For each time interval, given the speed data you can estimate how much power would be required for each of:
i. the change in kinetic energy from the start to the end of the interval,
ii. overcoming air resistance,
iii. overcoming rolling resistance, and
iv. drivetrain losses.

You add those up and the difference from actual recorded power is then assigned to changes in elevation (gravitational potential energy).

You enter in some starting guesstimates for the coefficients of aerodynamic drag and rolling resistance and then adjust those until they match the actual elevation profile.

It's easiest if you do a loop circuit a few times, then the "virtual elevation" should return to the same level at the same point in the loop on each lap. If they don't then you adjust the CdA and Crr estimated values until they do.

In the above I used the following inputs / assumptions:
Drivetrain Efficiency: 85%
Air density: 1.205 kg.m⁻³
Mass: 1,770 kg
Crr: 0.017
CdA: 0.80 m²

I'll try to find a suitable loop and also see what I can do to correct my data massaging of the carScanner OBD data, and then report back.

It can be a nifty way to estimate the aerodynamics of the car.
 
Thanks. I did that but still getting weird data.
The link below has a few ideas gleaned from when I initially tried to get it working.


The bit that may be applicable to your dashboard page is that some pairs of values will work OK on a page or graph of their own but not on the same page or graph. The easiest way I found to look at update rates and corruptions of the different cases was to use the graphs on the live data page.

Doing it for that one page's values shouldn't take long. If you end up wanting to go further, I'll gather more of my notes and saved configurations together and send them to you, probably by PM is most appropriate.
 
The bit that may be applicable to your dashboard page is that some pairs of values will work OK on a page or graph of their own but not on the same page or graph.
I have a custom page with just battery power and speed, which seemed to work OK.

But since speed is already available, I'm more interested in dynamic data the car doesn't show, with drive battery power being the most obvious.

I note from the data file I experimented on creating the virtual elevation profile that the peak power recorded was 62.9 kW and peak regen was -38.8 kW.

I've managed to clean the data a little for the plot but I forgot about braking which messes up virtual elevation calculations. Regen can be accounted for but would need an assumed efficiency number. Ideally I am looking for a loop to be able to do laps without need of braking and for regen to be low/disabled.
 
So how did you capture it???? I can't see any way to do this with carscanner. I've copied the .brc files to the computer, but I can't find anything that will process them
All this is for the Android interface on the phone.
I can't remember the exact details, but I think what I did was:
From the main screen go to data recording
The record data kickboxing should be ticked already.
Any data files already recorded should be listed.
(It may start a new file on each connect ??)
Short press a file to view graphs in the app - I've not tried this.
Long press a file to get rename, export and remove options.
Choose export
Choose file type (one of the CSV file types was of no use to me)
Choose save to file
For Android you then get normal share type options
Transfer saved file to PC somehow.
Open CSV file in excel.
Do calls or graphs as normal.

Note if you swap dashboards or live data values while running, some cells on a row will be blank if that column value is not used on the current dashboard page.
 
Thanks mate, guess I should have read the manual!!!!

Think it's going to be a long learning curve to be able to do anything with the spread sheet.
But Bonus, the .txt file is actually a misnamed .ubx file, that my GPSSpeadreader can process nicely and give a google earth overlay,
thusly.
1725966361867.jpeg


And here's the speadreader output.
1725966657082.png


This gives me a lot of confidence in the car's gps, it has typical ublox accuracy +/- 0.04knots for a ten second measurement. It's using doppler data and sampling at 10hz. The track is very smooth, (unlike our tracks very sawtooth from our jolting around in the chop), so it's well mounted somewhere so it has a good sky view.
I guess it's running in vehicle mode, that will extrapolate, speeds when signal is lost, going in tunnels etc. (not much good to us as it shows us still sailing when we crash).

So we can have confidence in its data, nice to know something is good.
 
Last edited:
I just bought another Bluetooth dongle, this time cheaper and using BLE.
And I get zero nonsensical values. These are my settings in car scanner:
IMG_4308.jpeg

IMG_4309.png

IMG_4310.png
IMG_4311.png
IMG_4312.png
IMG_4313.png
IMG_4314.png
IMG_4315.png
IMG_4316.png
 
That's good news. I'd given up on using my OBDlink unit until I had more time, and even if I had got it working better I'm sure there would have been limitations.

I'll report back when I see how well it works under Android.

I guess should I have the need, it'll work with ABRP too ?.
 

Are you enjoying your MG4?

  • Yes

    Votes: 908 77.7%
  • I'm in the middle

    Votes: 171 14.6%
  • No

    Votes: 90 7.7%
Support us by becoming a Premium Member

Latest MG EVs video

MG Hybrid+ EVs OVER-REVVING & more owner feedback
Subscribe to our YouTube channel
Back
Top Bottom