dTPS/dT noise level

[fphil]

Hello, happy new year all,

What is the lowest dTPS/dT level that you configure for the enrichment?
In my case,  I cannot set less that 9(20mV/100ms). Under that level Vems run most of the time Acc Enr on, I guess because of noise.
I would like to understand if the noise comes from the line or from the computation.

Philippe

[fphil]

This is a plot of TP and tpsDOT as given by the.csv file



TP noise and tpsDOT noise do not seem to be correlated which let me think that tpsDOT noise is numerical.
I believe that tpsDOT is what is called dTPS/dt by VT. So VT appears to smooth out the plot and does to show all the dTPS/dt = 8 points and the acc enr is triggered  apparently without reason if reading the VT plot.

What throttle speed is given by 8? If the TP range is 0-5V, I get 160mV/100ms or 3.2%/100ms. This is what one gets when opening the throttle slowly by 32% in 1 sec. Too bad I cannot use that acc enr level 8 in my case.

Philippe

[MWfire]

i use 5 or 6 with new firmware.
Do you have log?

[fphil]

5 or 6 lowest level does not work in my case for sure. Do you have a coil type trigger?

fw is 1.2.30. Engine is Maserati 6 cyl odd fire. 6 prim crank Trig, 3 taken into account by the fw (so a 120° gap and probably not many points to compute a derivative). Firing 90, 150, 90,.. obtained by 30° delay put on every 2 firings.

Acc Enr as underneath.

[MWfire]

please give me a log.

[fphil]

The matter is that the noise seems to depend on the fw version.
dTPS/dtnoise level (from the .csv log) is less than 2 with that config:



that is with fw 1.2.15, "RPM filtre, CAM base troigger" disable and acc enr not used.
Are there any other config parameter which may has to do with this noise level?

[mattias]

You run 1.2.15,  with 1.2.20 there is much improved acceleration enrichment, try that. Watch out for CLT/MAP ignition adjust table in more recent firmwares.

[fphil]

The first plot with dTPS/dt noise level = 8 was with fw 1.2.30. Thanks for the tip about the CLT/MAP table, the entries are 0.

This is the log plot where the out of sync that I have reported (http://www.vems.hu/wiki/index.php?page=MembersPage%2FFPhil%2FSomeOddFireConcerns) has occurred:



One can see that the engine is most of the time in accenr mode (no ego correction).
The accenr config is as above (first) that is



I am curious to know how the dTPS/dt values are computed in case of coil type oddfire.

[VEMS]

Hello fPhil,


from 1.2.20 and up the acceleration enrichment calc and application is done more frequently and the measured dv/dt is normalized to previous sample speed range, hence the higher noise level; however this is easily compensated by retuning the dv/dt colums of your accel enrichment table. Using fuel film is generally not recommended (unless absolutely certain what you are doing,) config validate should have warned accordingly.

Than the matter of the trigger error, it is detected hence the flagging of the "less sec trig error" in your log.

Best regards, Dave

[fphil]

ok, when I compare the dTPS/dt with the calcdTPS/dt from here



I find a factor 2 on dTPS/dt (360%/s for dTPS/dt compare with 180% for calc dTPS/dt if I am not wrong), which puts back the noise level at 4. (I already reported the issue through VT). You may check in your case.

[fphil]

Hello Dave,

... the measured dv/dt is normalized to previous sample speed range, hence the higher noise level; however this is easily compensated by retuning the dv/dt colums of your accel enrichment table

allright better one knows about, so the scale factor of the ACC MAP enrich table is wrong.

Than the matter of the trigger error, it is detected hence the flagging of the "less sec trig error" in your log.

Indeed at 23:06, the outof sync occurs at 23:04
So 2 full seconds to damage the engine or having a car crash

Regards

Philippe

[VEMS]

allright better one knows about, so the scale factor of the ACC MAP enrich table is wrong.

Actually no, scaling of the table has not changed. Axis unit still needs updating, good catch!

Indeed at 23:06, the outof sync occurs at 23:04
So 2 full seconds to damage the engine or having a car crash

I will have to take a look at your log; will review and update your report accordingly, nice that you were able to catch a trigger log of the exact event thats very helpfull.

Best regards, Dave

[fphil]

Actually no, scaling of the table has not changed. Axis unit still needs updating, good catch!

Since I believe what I see  I did the histogram of the ratio : my_dTP/dt / tpsDOT for relevant values, "TP" and "tpsDOT" being the .csv data from the genboard.
The ratio is 2 (factor 1.8 ), other max of the histo are 1(factor 0.5) and 4 (factor 0.25)
That is 1%xtpsDOT (named dTPS on the VT window) = 2%xmy_dTP/dt = 10mV/100ms (throttle Volt range is 0-5Volt) instead of 20mV/100ms before.

dTPS units seemed to have been divided by 2 with the new fw, and indeed, I get the same level of noise than before 4x20mV/100ms or 8x10mV/100ms which is 16%/s (total opening of the throttle= 100%)

from 1.2.20 and up the acceleration enrichment calc and application is done more frequently and the measured dv/dt is normalized to previous sample speed range, hence the higher noise level

hmm, at first sight my dTP/dt values simply calculated from TP.csv, ((TP(t)-TP(t-dt))/dt, as good/precise than the tpsDOT.csv values sent by genboard, or even better!

Philippe

[VEMS]

Hi Philippe,

Thanks for the analysis, its always nice to see some detailed log number crunching going on.

Best regards, Dave