Megasquirt Support Forum (MSEXTRA)

Hi,

the MAF lookup (mapping MAF voltage to MAF) seems buggy to me.
I have sorted a log file by MAF voltage and got different MAF values for identical voltages.
Then I made a Tuner Studio custom channel "MafCurve" which does a MAF lookup in the
MAF Flow Curve and compared that to the MAF values.
MafCurve = { tableLookup(array.mafflow, array.mafv, maf_volts) }, "g/s";
The max error in this log is 24% !.
FW 1.5.1

I had to cut the log.

Please attach your MSQ.
There could be a number of factors here, ones that spring to mind:
- the datalog points are not synchronised with the actual calculation.
- the curve is linearly interpolated which might not be exactly the same as your comparison curve.

I did a scatter plot in MegaLogViewer of MAF vs. MAFvolts and it give a "fat" line, showing a bit of a spread.

James

Hi James,

thank you for your reply.
Attached you'll find my tune.

Jan

I looked at your log again and your tune. Nothing is jumping out as incorrect. You have a 50% lag enabled on MAF. So the value will rarely match the instantaneous table lookup due to the smoothing. This will be especially visible during transients.

James

Hi James,

thank you for your analysis.
The 50% lag is the default.
Which lag factor do you recommend for MAF ?

Jan

Hi James,

I made a new log with MAF lag factor 100.
Still there are significant deltas between logged and calculated MAF.
The max error is 67%.
A scatter plot MAF vs. MAF Volts shows a lot of spread.
I don't want to claim that this is a bug, I simply want to understand it.

[ I had to cut the log ]

Regards,
Jan

I couldn't quite tell from the msq/msl, but my guess would be this variance is caused by MAT correction or MAT air density table corrections(In 1.4.x/1.5.x firmware I've noticed that MAT air density correction table has an effect on MAF even though it's greyed out in TS and is meant for speed density). Correction would give several MAF values for each voltage, but what I don't know is whether the log value is the raw MAF calculation before corrections or if it is post correction.

I don't think so.
There is no MAT dependency visible in the zoomed scatter plot.
Air correction is between 97% and 103% through the whole log.
There should be no air correction at all with MAF.

also when in maf mode i think it also uses map sampling as maf sampling, that will add a bit of smoothing as well.

for giggles try out map/maf logging in high speed loggers once, it will blow your mind how sensitive it is

interesting, I will make a high speed log.

[quote="W126"]
There should be no air correction at all with MAF.
[/quote]

Right, at least that's the desire anyway. What I had to do to get my MAF to behave somewhat like I expected it to was to zero out all the air correction factors both for MAT air density and MAF/MAT correction. Even though the air density temp corrections shouldn't affect MAF they do for me. You can easily test this by increasing/decreasing the air density correction factor to large/small values and you can see large AFR swings from it.

I did a scatter plot of MAPxRPM vs MAF and all looks fairly normal.

@theturbodog: Air correction cannot be zeroed out, it's disabled in MAF mode.

@whittlebeast: What do you mean by "all looks fairly normal" ?
MAPxRPM and MAF voltage are two different things.

see https://youtu.be/MJ1V0A9uKLE

the question is why I get different MAF values for same voltages.

My guess is this is an issue with data packets may not be synced in the data log. Try logging way faster if on MS3.

Tunerstudio is at the max, I could try SD card logging.
If I understand you correctly, your guess is the logged value is not the value actually used by MS ?

That's my guess

but what's the meaning of a log file who's values are not the values used by MS ?