Warm Start idle issues

[sloth4urluv]

I've noticed that on warm starts the car idles lean and has to trim a good bit positive then goes negative after a minute then finally stabilizes close to zero.

Attached is the calibration and datalogs
(Recorded a cold start and let it warm up, then turned the car off and reuploaded the calibration to clear the long term trim and recorded a warm start)

I have the table setup so that idle falls squarely in the middle of one cell.
I was thinking of rescaling the table to add another column at around -23" to add a little buffer between idle and the -26.2 column.

Also on the cold start log, why does the long term trim trend negative while the short term trim trends positive?

Cold Start.png

Warm Start.png

[sloth4urluv]

Any ideas on this?
I’m kinda at a loss on what to adjust to solve this.
After looking at the log closely I noticed that the MAP value towards the beginning is actually slightly higher, but I know the next load cell up is correct because if I turn on the AC and headlights it falls squarely in that cell.

Obviously I could increase the amount of fuel in the cell but then when I come to a stop the long term trim crawls even further negative. The ECT doesn’t change much between when it first starts and when it stabilizes, the battery voltage is staying constant, IAT stays pretty constant, Fuel pressure constant, oil pressure constant.

Since it’s at operating temperature I can’t really do anything with the fuel compensation. I’m going to try and change out the Wideband sensor but I’m not that hopeful as the car seems to idle happier when closed loop drives the AFR down to 14.7:1 and last time I checked the compression on all cylinders was pretty close to 250. I don’t see any obvious cracks on the header either and don’t seem to have any leaks.

Any suggestions would be appreciated.

[Hondata]

Injector dead times look very low.

[sloth4urluv]

What should they be for ID1050Xs (50psi)?
I just followed the dead times provided by ID for the listed pressure.

[sloth4urluv]

Also, does the reported injector time in the datalog include the dead time or is it the time in addition to the dead time?

[Hondata]

Datalogged injector opening time is the total time.

ID provides numbers but they are not measured using a Honda ECU. Each ECU will be different in the current profile it uses to open and hold the injector. So the ID numbers tend to be 0.15 to 0.25 ms too low.

[sloth4urluv]

Thank you for the info!

Is that a constant offset for all voltages or is that centered around 12V or 14V?

Seems like I should hook a scope up and measure the voltage across the injector pins (at the injector) and measure the voltage across them as well as the battery voltage.
I could then calculate the resistance in series with the injector (measuring the resistance of the injector as well).

With that I should be able to calculate the voltage at the injector for each voltage in the dead time table and then look up the deadtime from the table in the ID1050x data sheet.

Does that sound sane?

If so, then to get me in the ballpark with my fuel map could I do something like this?
Example:
In a WOT pull it took 10.68ms of injector to hit the target AFR, Voltage was 13.4V so the interpolated dead time was 1.01ms), the interpolated Fuel Counts was 6930.
Subtract the dead time from the injector time to get the time commanded by the Fuel Counts (9.67ms).
Divide the Fuel Counts by the commanded time to get counts/ms (716.6).

Now assuming I had to add 0.25ms to the dead time I could multiply it by the counts/ms constant to calculate how much I should offset all of my cells (-179 Fuel Counts).
(I made a graph of the of a pull that compared the dead time offset injector time vs interpolated fuel counts over the pull and got a pretty linear graph, but noticed that the slope is different on the low fuel table so I would need to calculate a different offset for that table)

I don’t expect it to be perfect but I think it would get me in the ballpark.

[sloth4urluv]

Never mind on that second slope, I was an idiot and the tool I made was referencing the high fuel table even though it wasn’t in VTEC. Referencing the correct table I get the same slope.