Scan My Tesla official thread (support, suggestions, discussion, feedback)

  • SUPPORT THE SITE AND ENJOY A PREMIUM EXPERIENCE!
    Welcome to Tesla Owners Online, four years young! For a low subscription fee, you will receive access to an ad-free version of TOO. We now offer yearly memberships! You can subscribe via this direct link:
    https://teslaownersonline.com/account/upgrades

    SUBSCRIBE TO OUR YOUTUBE CHANNEL!
    Did you know we have a YouTube channel that's all about Tesla? Lots of Tesla information, fun, vlogs, product reviews, and a weekly Tesla Owners Online Podcast as well!

  • It's OK to discuss software issues here but please report bugs to Tesla directly at servicehelpna@teslamotors.com if you want things fixed.

bwilson4web

Top-Contributor
TOO Supporting Member
Joined
Mar 4, 2019
Messages
532
Location
Huntsville, AL
Tesla Owner
Model 3
Country
Country
#61
This is called "noisy data" and comes from several potential sources of which the ones we have some control:
  • weak signal between OBD and Android - this is handled by testing the signal strength and moving as needed to get a 'less negative' dB signal (i.e., stronger)
  • weak processor on Android - this is handled by recording only the data of interest for an experiment so the processor won't run behind
Once I have the data in a spreadsheet, I make a second data array to the right with continuous data columns. To the right of the data cells, I make column headers for the interesting data where the first data row is set to zero. The second data row either replaces a valid data from the original OR copies the earlier cell. For example, the original column is B and the new column L to the right: "=if (b3="";l2;b3)" Replicate the function for all cells and you'll have columns of continuous data. Copy the new columns and "paste special" to make it raw data. Then you can sort the various columns and 'fix' the outliers.

Sometimes I do a more sophisticated approach. If I know the valid data ranges, anything outside of that range as well as the empty cells are copied from the upper cell ... the last known good data. For example, if a field can not be a negative, "=if(b3="";l2;if(b3<0; l2;b3))"

What we are doing is gleaming the valid data from the out-of-bound values while also making continuous data columns for plotting.

GOOD LUCK!
Bob Wilson
 

Hugh_Jassol

Active member
Joined
Jan 31, 2019
Messages
154
Location
SoCal
Tesla Owner
Model 3
Country
Country
#62
Thanks @bwilson4web - I'm quite familiar with noisy data and how to smooth/fill missing data. My main concern was if that was something unique to my setup or if others see it as well. In this case I am guessing that the BT connection is not using any message integrity checking or other methods to ensure no bit flips/drops/etc. which I guess makes sense in a streaming application. Although I am surprised by how frequently the data gets mangled.

If I had to guess, it's my very slow-n-cheap Kindle Fire 7. As an iPhone/iPad user, I only have this tablet to allow me to use the occasional android app like Scan My Tesla. I'll try reducing the channels, but since the BT connection transmits the CAN data at the same rate no matter what, I'm guess the Fire will get overwhelmed regardless of which channels I am recording.

In any case, thanks for confirming that this is a 'normal' occurrence!
 

amund7

Active member
Joined
Dec 17, 2018
Messages
106
Location
Drammen
Tesla Owner
Model 3
Country
Country
#63
No this is not normal. Well, it is occasionally expected, but of course unwanted.

I will try to make a list:

1. Model S has quite a few hard-coded 'spam filters' that killed most of the corrupt data, made by analyzing commonly recieved corrupted packets. Model 3 does not have this maturity yet, and new packets are added fast.
2. Bluetooth communications for Android does not tell the app how much data is coming, and if there's more. I recieve blocks, these can be 1 character, or 20 and a half line. To get most of the data, I try to stich these together until I get whole lines. This code is super old, made on Model S many years ago. Contains timers to try to detect if the stitched hex actually belongs together, or if there has been a hickup in the data stream, then discard the prevoius piece.
3. Cheap adapters have a hard time keeping up, and more often give crazy values than the (high-grade) OBDLink devices.
4. Some phones are just wierd with their Bluetooth. My Huawei P20 Pro, when I got it early this year, did not work at all. Many firmware updates later it sort of works, but the data comes in very jittery, renders 0-60 timers useless, and also creates occational corrupt data. Even with the same car, same adapter, a lot more corrupt data. Could be amplified by the stitching in #2. I checked some Android developers forums, thinking my code sucked, got the reply that every Android brand has their own bluetooth stack, each worse than the next, Huawei and I think Xiaomi were mentioned as problematic for app developers. I also tried to firmware flash my OBDLink device, with the official OBDLink app - BRICKED IT! Restored it by reflashing with a Lenovo P2, still works today :)
5. The more stuff you're showing on your tab, the more of all the above seems to happen, at least on the ELM327-based adapters.

I have never seen the slower packets such as the min/max cell temperature and volts go corrupt on the model 3. If this happens constantly, please send me some logs (RAW log), and I will see if I can hard-code some filters against it.

I will spend more time on improving the app in this respect, my wish is that the app runs perfectly on all hardware, old, new, expensive or cheap.

But until then, YES better hardware would give you a better experience.
Less data on the same screen could also help, that's free to try.
Try a regular Android phone first, NOT Huawei or Xiaomi.
If that does not help, try an OBDLink LX adapter.
 

bwilson4web

Top-Contributor
TOO Supporting Member
Joined
Mar 4, 2019
Messages
532
Location
Huntsville, AL
Tesla Owner
Model 3
Country
Country
#64
I'm seeing a very low error rate, ~10 out of 30,000 samples which in the big scheme is tolerable. These rare, noise events didn't really bother me. Still I can appreciate the need to share test data with @amund7, the author, via private communication.

My practice is to record just the CSV data, not the RAW log. But I understand the need for the raw data. Would it help if I used the "ALL" tab and recorded a sample, say 5-10 minutes of both CSV and RAW?

To share the data, I would add a 'configuration' file describing the equipment and Tesla along with the CSV and RAW logs stuffed in a ZIP file. Then share a URL to the ZIP file. Does that work for you?

Bob Wilson
 

amund7

Active member
Joined
Dec 17, 2018
Messages
106
Location
Drammen
Tesla Owner
Model 3
Country
Country
#65
I'm seeing a very low error rate, ~10 out of 30,000 samples which in the big scheme is tolerable. These rare, noise events didn't really bother me. Still I can appreciate the need to share test data with @amund7, the author, via private communication.

My practice is to record just the CSV data, not the RAW log. But I understand the need for the raw data. Would it help if I used the "ALL" tab and recorded a sample, say 5-10 minutes of both CSV and RAW?

To share the data, I would add a 'configuration' file describing the equipment and Tesla along with the CSV and RAW logs stuffed in a ZIP file. Then share a URL to the ZIP file. Does that work for you?

Bob Wilson
I need the data that creates the corrupted values, if I am to catch them. So I would need some data from @Hugh_Jassol where he sees those screwed up temperatures.
 

JWardell

TOO Master Member
Joined
May 9, 2016
Messages
4,045
Location
Boston
Tesla Owner
Model 3
Country
Country
#66
It's time to bring up an annoying issue that prevents me from using ScanMyTesla most of the time. It is rarely able to connect to the OBDLink LX. I plugged it in permanently a month or so ago, and I think only a day or two was I able to successfully use the app.
I thought this might be caused by the OBDlink being powered for hours not connected to any phone (while it charges, or parked in sentry), but today I made very sure to wake the car with the phone next to the car and it still would not connect. I can hold down the button to place it in blinking pairing mode, and select it from within the bluetooth devices, but the app still fails to connect. Then sometimes randomly it will work.
I'm kind of leaning toward the being an OBDlink issue, but I welcome your thoughts, does anyone else have trouble connecting?
 

Feathermerchant

Top-Contributor
Joined
Sep 17, 2018
Messages
567
Location
Euless, Tx
Tesla Owner
Model 3
Country
Country
#67
I'd say OBDLink. Which one do you have? I have the one that comes with the wiring harness. It works great with my phone and tablet.
It actually works in the house when the car is in the garage.
 

bwilson4web

Top-Contributor
TOO Supporting Member
Joined
Mar 4, 2019
Messages
532
Location
Huntsville, AL
Tesla Owner
Model 3
Country
Country
#68
I'm kind of leaning toward the being an OBDlink issue, but I welcome your thoughts, does anyone else have trouble connecting?
No trouble but to be sure, I loaded "Bluetooth Pair". This application does a Bluetooth scan reporting signal strength of devices to my Android device.

Bob Wilson
 

Hugh_Jassol

Active member
Joined
Jan 31, 2019
Messages
154
Location
SoCal
Tesla Owner
Model 3
Country
Country
#69
It's time to bring up an annoying issue that prevents me from using ScanMyTesla most of the time. It is rarely able to connect to the OBDLink LX. I plugged it in permanently a month or so ago, and I think only a day or two was I able to successfully use the app.
I thought this might be caused by the OBDlink being powered for hours not connected to any phone (while it charges, or parked in sentry), but today I made very sure to wake the car with the phone next to the car and it still would not connect. I can hold down the button to place it in blinking pairing mode, and select it from within the bluetooth devices, but the app still fails to connect. Then sometimes randomly it will work.
I'm kind of leaning toward the being an OBDlink issue, but I welcome your thoughts, does anyone else have trouble connecting?
I bought an OBDLink LX as well... solved the data "noise" I was getting with the $10 no-name one from Amazon. I've only had it ~10 days and one time, mid-drive, it just stopped. I had to unplug/replug the adapter to get it back. Every other time it worked ok. I'm using it with a Kindle Fire 7.
 

Insaneoctane

Active member
Joined
May 10, 2017
Messages
72
Location
Moorpark, CA
Country
Country
#70
I guess I am paranoid, but all the graphing suggestions seem fairly "unknown" and I am not in the habit of running exe files that aren't known to be harmless.....so I am unsure how to graph my data! I recorded a trip on my way to work today which was over 1 hour of driving and created a 2.5M row CSV and Excel is sometimes choking on it. I was able to plot battery voltage but not cell temp mid (it thinks it's a text field and will only allow count in pivot tables). Definitely struggling here. The first 7 minute file I created seemed to have lots of drops. Still trying to figure out how the data gets written (and when it doesn't), how to graph with no risk of malware and how to automate data logging at some point in the future. All help appreciated.

PS. My temps for "Duct left/right" are all over from -40 to 233...is this temp or angle or something else weird?
 

JWardell

TOO Master Member
Joined
May 9, 2016
Messages
4,045
Location
Boston
Tesla Owner
Model 3
Country
Country
#71
I bought an OBDLink LX as well... solved the data "noise" I was getting with the $10 no-name one from Amazon. I've only had it ~10 days and one time, mid-drive, it just stopped. I had to unplug/replug the adapter to get it back. Every other time it worked ok. I'm using it with a Kindle Fire 7.
Yeah I think unplugging will usually fix it, but that's a pain.
Also suspect issue is due to harness tapping VSec power instead of accessory, might never actually be turning off. Could be some memory leak or timer overflow in the OBDlink.
 

scottf200

Active member
Joined
Sep 21, 2017
Messages
83
Location
Chicagoland
Country
Country
#72
Spec: LR TM3 with 18" wheels

My son's car is down from a high of 308 mile reange to 267. See dark blue line in below graphic.

AFAIK here are the new/perfect values: (72.8 kWh * 1000) / 234 "Wh/mi" is a range of 311.11miles

1) Can other LR TM3 folks using SMT say what their 'Nominal full pack' value is? (his is quite a bit less than 72.8 which is from https://forum.abetterrouteplanner.com/blogs/entry/6-tesla-battery-charging-data-from-801-cars/ ).

2) @amund7, what is "Expected remaining" compared "Nominal remaining"? ie. how determined or calculated. TIA I looked at Scan My Tesla Readings but it is not there: Scan My Tesla Readings Google sheet

3) @amund7, what is "SOC expected" compared "SOC"? ie. how determined or calculated. TIA

SMT pack info image link


Dark blue = est 100% range (This does not appear related to purple temp if you compare last year to this year in the same month or more)


A thru G is cut-n-paste from TeslaFI.COM
 
Last edited:

GregRF

Active member
Joined
Sep 25, 2017
Messages
92
Location
Aptos CA
Country
Country
#73
1) Can other LR TM3 folks using SMT say what their 'Nominal full pack' value is?
As of Today TeslaFi is reporting estimated full range of 293.56. SMT has been reporting Nominal full pack at 69.3 to 69.4 kWh

My mileage has varied:
1578681396009-png.31678

My normal commute has my battery in the 65-80% SOC. After some longer trips with higher charges and ddeper dischargers the battery has some times recalibrated. I'll be curious what SMT reports after the next of these trips.
 

scottf200

Active member
Joined
Sep 21, 2017
Messages
83
Location
Chicagoland
Country
Country
#74
Spec: LR TM3 with 18" wheels
My son's car is down from a high of 308 mile reange to 267. See dark blue line in below graphic.
AFAIK here are the new/perfect values: (72.8 kWh * 1000) / 234 "Wh/mi" is a range of 311.11miles
1) Can other LR TM3 folks using SMT say what their 'Nominal full pack' value is? (his is quite a bit less than 72.8 which is from https://forum.abetterrouteplanner.com/blogs/entry/6-tesla-battery-charging-data-from-801-cars/ ).
2) @amund7, what is "Expected remaining" compared "Nominal remaining"? ie. how determined or calculated. TIA I looked at Scan My Tesla Readings but it is not there: Scan My Tesla Readings Google sheet
3) @amund7, what is "SOC expected" compared "SOC"? ie. how determined or calculated. TIA
SMT pack info image link
Dark blue = est 100% range (This does not appear related to purple temp if you compare last year to this year in the same month or more)
A thru G is cut-n-paste from TeslaFI.COM
UPDATE: Below answers from 'amund7'
1. Mine has around 74.3 these days, at 18 000 km. My previous one from Feb 2019 kept around 77 for the longest time, but then dropped to 74-75 after about 6000 km. Funny / sad observation was that the second car was down to 75 of day 2 of ownership... so either car 2 is *sugar*, or we shouldn't worry too much about this number
But your son's is, well, looks quite sad, I suppose you did try to charge to 100%? How do the cell voltages / cell imbalance look at 100%?

2 + 3
SOC = (nominalRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);
SOC Expected = (expectedRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);

What is expected remaining? Answer is, I have no clue, but it's there for all S and X too, and in this case a user figured it fits better with the SOC % in the car with a cold battery (snow crystal, blue line indicating 'cold' loss)
As of Today TeslaFi is reporting estimated full range of 293.56. SMT has been reporting Nominal full pack at 69.3 to 69.4 kWh
My mileage has varied:
My normal commute has my battery in the 65-80% SOC. After some longer trips with higher charges and deeper dischargers the battery has some times recalibrated. I'll be curious what SMT reports after the next of these trips.
Thanks guys. Here is the data I've collected so far. My son did charge to 100% yesterday and it continued to charge (cell balance) for 48 minutes. Cell Imbalance at 100% was 16 mV but the day before at 80% it was 8 mV.



100% charging. Cell balancing (?) for 48 minutes after it hit 100%. Cell Imbalance at 100% was 16 mV.

 
Last edited:
Joined
Jan 15, 2020
Messages
11
Location
Milano
Tesla Owner
Model 3
Country
Country
#75
UPDATE: Below answers from 'amund7'
1. Mine has around 74.3 these days, at 18 000 km. My previous one from Feb 2019 kept around 77 for the longest time, but then dropped to 74-75 after about 6000 km. Funny / sad observation was that the second car was down to 75 of day 2 of ownership... so either car 2 is *sugar*, or we shouldn't worry too much about this number
But your son's is, well, looks quite sad, I suppose you did try to charge to 100%? How do the cell voltages / cell imbalance look at 100%?

2 + 3
SOC = (nominalRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);
SOC Expected = (expectedRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);

What is expected remaining? Answer is, I have no clue, but it's there for all S and X too, and in this case a user figured it fits better with the SOC % in the car with a cold battery (snow crystal, blue line indicating 'cold' loss)


Thanks guys. Here is the data I've collected so far. My son did charge to 100% yesterday and it continued to charge (cell balance) for 48 minutes. Cell Imbalance at 100% was 16 mV but the day before at 80% it was 8 mV.



100% charging. Cell balancing (?) for 48 minutes after it hit 100%. Cell Imbalance at 100% was 16 mV.

Thank you scottf200 expecially for advices on DatPlot.
I have a problem not being able to plot "only line" and NOT symbols.
If I modify the graph with "Format Data curve " hiding the Symbols but letting show the line the line doesn't appears, SO I have to set to "do not hide symbols to be able to see something.

If you see the SPEED curve it's blue (symbol) and red LINE, but in the graph I can see only the blue symbol, and not the red line.

Another strange thing is related to Scan My tesla. Why I'm loosing so many data from about 7 seconds to 8 seconds (see both Battery Power & Speed).
Yes it seems that I have less datapoints and the values are also NOT following the trend.
It seems I'm loosing the progression in POWER and loosing speed too during this partial data black-out
1579081034488-png.31744
 
Joined
Jan 15, 2020
Messages
11
Location
Milano
Tesla Owner
Model 3
Country
Country
#76
Spec: LR TM3 with 18" wheels

My son's car is down from a high of 308 mile reange to 267. See dark blue line in below graphic.

AFAIK here are the new/perfect values: (72.8 kWh * 1000) / 234 "Wh/mi" is a range of 311.11miles

1) Can other LR TM3 folks using SMT say what their 'Nominal full pack' value is? (his is quite a bit less than 72.8 which is from https://forum.abetterrouteplanner.com/blogs/entry/6-tesla-battery-charging-data-from-801-cars/ ).


Mine it's a Model 3 LRAWD 10 months old and with 20k kilometers
My Nominal Full Pack changed after 2019.40.50.1 and now (2019.40.50.7) it's always 74.9 to 75.0 with 3.40 buffer
Prior to 2019.40.50.1 it was between 73.4 to 73.8 and buffer was 3.30
I'm really interested in this parameter 'Nominal Full pack' and if it's related to battery degradation.
screenshot_20200115_163723-jpg.31749
 

amund7

Active member
Joined
Dec 17, 2018
Messages
106
Location
Drammen
Tesla Owner
Model 3
Country
Country
#77
Thank you scottf200 expecially for advices on DatPlot.
Another strange thing is related to Scan My tesla. Why I'm loosing so many data from about 7 seconds to 8 seconds (see both Battery Power & Speed).
Yes it seems that I have less datapoints and the values are also NOT following the trend.
It seems I'm loosing the progression in POWER and loosing speed too during this partial data black-out
Have you guys tried my UDPLogger for plotting? https://github.com/amund7/UDPLogger/releases
Super easy to use (if you read the instructions on how to zoom and scroll), and is customized to read Scan My Tesla CSV files. Can choose which signals to show.

Over to your missing data issue, this looks like a typical 'hiccup', where the data is correct, but delayed in to the app. Could be the phone OS or bluetooth drivers suddenly were busy doing something in the background, disk writing, or the bluetooth adapter gives a 'buffer full' message and the app has to restart the data stream. Try less signals in your tab. Also try disabling other background apps, especially ones that use bluetooth, music, handsfree, Android power saving settings for Scan My Tesla. You can even try going to flight mode (but turn on bluetooth) or turn off wifi, as some phones have interference issues, or shared chips / bandwidth.

If you can catch this again, if you record both the CSV and a RAW LOG at the same time, send me both files and I can investigate.
On the Speed tab you have a signal called 'accuracy', copy that to your tab and watch/log it. Also packets per second will give you an indication, if it suddenly drops a lot, you are seeing the problem that your graphs are showing.

PS. Some phones are notoriously weird with their bluetooth implementation, I have found Huawei P10 and P20 pro constantly delays and 'jitters' the data, it is even visibly lagging behind on the gauges, then catching up. With those phones, the only thing we can hope for is firmware improvements. My P20 pro didn't work at all with Scan My Tesla when I first got it, but after 6 months it's usable, but still visibly laggy on the gauges, 0-100 timing and drag race logs like yours above will be unusably inaccurate. I dove deep in this, thinking there was a bug in my app, or looking for a workaround, but the experts say all Android phone makers make their own interpretations of the bluetooth standard, some interpretations are very far from the standards, and Huawei was mentioned as one of the worst, causing lots of issues for developers.

If you find some trick that improves the situation, I am all ears, as this question comes up from time to time.
 
Joined
Jan 15, 2020
Messages
11
Location
Milano
Tesla Owner
Model 3
Country
Country
#78
Have you guys tried my UDPLogger for plotting? https://github.com/amund7/UDPLogger/releases
Super easy to use (if you read the instructions on how to zoom and scroll), and is customized to read Scan My Tesla CSV files. Can choose which signals to show.

------>Thanks, I tried to use it and I like it.

Over to your missing data issue, this looks like a typical 'hiccup', where the data is correct, but delayed in to the app. Could be the phone OS or bluetooth drivers suddenly were busy doing something in the background, disk writing, or the bluetooth adapter gives a 'buffer full' message and the app has to restart the data stream. Try less signals in your tab. Also try disabling other background apps, especially ones that use bluetooth, music, handsfree, Android power saving settings for Scan My Tesla. You can even try going to flight mode (but turn on bluetooth) or turn off wifi, as some phones have interference issues, or shared chips / bandwidth.

----> Yes many at least 3 bluetooth processing (SMT, Fitbit watch, and car too), WiFI enabled too .

If you can catch this again, if you record both the CSV and a RAW LOG at the same time, send me both files and I can investigate.
On the Speed tab you have a signal called 'accuracy', copy that to your tab and watch/log it. Also packets per second will give you an indication, if it suddenly drops a lot, you are seeing the problem that your graphs are showing.

----> Just checked and I have only csv log enabled, I'll put raw log too for future references.
One strange thing is that I have only 2 "packets per second" datapoint on over 20,000 samples
----> I'm no more able to find 'accuracy' on Speed tab nor on All tab.


PS. Some phones are notoriously weird with their bluetooth implementation, I have found Huawei P10 and P20 pro constantly delays and 'jitters' the data, it is even visibly lagging behind on the gauges, then catching up. With those phones, the only thing we can hope for is firmware improvements. My P20 pro didn't work at all with Scan My Tesla when I first got it, but after 6 months it's usable, but still visibly laggy on the gauges, 0-100 timing and drag race logs like yours above will be unusably inaccurate. I dove deep in this, thinking there was a bug in my app, or looking for a workaround, but the experts say all Android phone makers make their own interpretations of the bluetooth standard, some interpretations are very far from the standards, and Huawei was mentioned as one of the worst, causing lots of issues for developers.

---> maybe the main reason: I'm using a Huawei P10 phone.

If you find some trick that improves the situation, I am all ears, as this question comes up from time to time.

-----> I'll do some tests using flight mode.
 

amund7

Active member
Joined
Dec 17, 2018
Messages
106
Location
Drammen
Tesla Owner
Model 3
Country
Country
#79
----> I'm no more able to find 'accuracy' on Speed tab nor on All tab.
This signal was meant to live in the Speed tab, and is created by the Speed signal - which model 3's only send when the car is 'on' - meaning after you press the brake pedal. Does that help?
 
Joined
Jan 15, 2020
Messages
11
Location
Milano
Tesla Owner
Model 3
Country
Country
#80
This signal was meant to live in the Speed tab, and is created by the Speed signal - which model 3's only send when the car is 'on' - meaning after you press the brake pedal. Does that help?
Thanks! I only saw it once . I'm away from the car so I can't check now. I'll let you know. But I'm sure that on "Speed"Tab I have only SOC, Cell Temp Mid and Speed (and if reached, the 0-50 0-60 0-100....)