Differences between datalog session and record session?

[creizlein]

I am seeing some behavior that is driving me crazy and I want some comments about it

Whenever I start datalogging as well Start Recording the fuel map overlay is not the same as when I later LOAD the recorder datalogging?
What I mean is, I Start the datalog and the recording process, the fuel map then starts to overlay readings, for instance in the estimate change to target lambda view, then after 30mins of driving i stop both the datalog and thus the recording process

If I compare the results captured during datalog they are somewhat different to what I see if I load the recorded datalog later on? why is that?

Here is the result overlay of the datalog session

snap0417.png

Here is the result overlay after closing the program, starting again and load the recorded session?

snap0421.png

[Spunkster]

Please give more detailed information such as what vehicle/engine, what wideband is being used, the calibraiotn and a datalog showing the issue.

[creizlein]

This is a p28 ECU running s300 v3, Honda Civic B16a2 car/engine with a PLX Wideband connected to D14

The calibration and datalog "I belive" is not dependant as i can reproduce this error at any time, it has been like that always for me, so thats why I ended up asking.

I am using both the stockjdmp30.skl and stockusp30.skl calibrations, Usually to record trips from/to work for about 30minutes or so.

Attached is my current calibration from which i took the screenshots and the mentioned datalog, but i can reproduce this anytime with different settings if needed?

PS: the primary fuel table was slightly adjusted so it has a newer date than the recorded session

[creizlein]

Here is another example I just did to test

1. Opened sManager
2. Loaded my calibration
3. Pressed F9 to start Datalog and Record
4. Drove for about 20minutes
5. Stopped Datalog and Record at the same time (datalog only was NOT running without recording)
6. Captured screenshot of the overlay change
7. Closed Datalog, and opened again the recorded session
8. Captured screenshot of the overlay map, and you can see the differences

Atached is the calibration with the record session as well

snap0422.png

snap0424.png

[creizlein]

Any update on this?

[Hondata]

There was a rounding error with the column when datalogging. The recording was ok.
Fixed for the next release. (Should be this week).
Thanks for pointing this out.

[creizlein]

Cool! Glad you guys were able to fix it.
So the values read/displayed when loading the recorded session were invalid? or the ones showing when live data-logging?
So I know which one I can trust in the mean time

Again, thanks for fixing and keep up the good work! :)

[Hondata]

The trace from the datalog was more accurate.

[creizlein]

I have upgraded to the latest version v2.6.1 and did the same thing, started to trace and record at the very same time (by pressing F9 directly, so it starts both the datalog and the recording) and went for a drive, then come to a full stop and shutdown the car from the key, so both the datalog and recording session stopped at the very same time as well

I must say that upon checking the live tracing and the recording one they look more alike, but there are still some discrepancies, specially with those cells that have very few samples, looks like not all samples are loaded, for instance look at the highlighted cell, the live trace has 74 values and the right cell next to it has 115 samples, on the other hand, upon opening the recorded session the same cell has 86 samples and the next has 105.

So it looks like still the live trace and the recorded one upon loading dont fall 100% on the exact same cell?
The annoying part is which one is more accurate or would be preferable to use

snap0440.png

snap0441.png

[Hondata]

I believe the difference is from the transition delay excluding lambda values after shifting. Can you show what your settings are so that I can test further?
Also, when you perform a test like this you need to start recording without datalogging first (F9 rather than F10) immediately after opening SManager, otherwise previous datalog values will affect the transition delay as above.

[creizlein]

Certainly, which settings do you need? the calibration or the smanager.ini ?
What I do when i am ready to datalog/record is even close the sManager, open it again, load the map and then press F9, that does start the datalog and the record at the same time, once I am done what i do is press F10, which stops datalog and record as well. Record is auto-saved so I do not save the file manually but instead rely on the auto-save feature that i have turned on

[Hondata]

The settings in SManager -> Settings -> Sensor Overlay (Lambda).

[creizlein]

havent touch them... bad idea?

snap0447.png