Long-Term Fuel Trim "Valleys"

[bigbadspoon]

Lately, I have had this weird thing going on in my data logs, where the LTFT will slowly dip down to -10% (takes about 10 miles), then rise back up to 0% (after about another 10 miles).

The weird thing about this is that say, after I flash in a new calibration, it will hover around 0 for well over 60 miles, then do one of these dips, then hover at 0 for ~60 miles, then do another dip. It's been doing this for about 4 days now, and I am completely dumbfounded as to what is going on. It's even more strange to me that this seems to occur during mostly highway driving where I pretty much set the cruise control at 65 for 20 mins.

My car is a 2012 Si using FP 1.4.5.0 with an AMF-based tune. The tune for these datalogs has been on the car for about 250 miles, so I would have thought the LTFT would settle by now.

Thank you for your time.

[Hondata]

If you look at the period where you are on cruise control at 65 mph the short term trim is around -9%. This is why the long term trim is heading to -9% as well. Then when you slow to 45mph the short term trim is positive, and the long term trim follows as well.

[bigbadspoon]

Right, I understand the LTFT follows the STFT. What I do not understand is why they are going low like that. If you check out the "to work" datalog, on the same road, same calibration, going 65, my STFT is hovering around 0%. LTFT is as well.

[Hondata]

Because you had your headlights on to work and off from work (battery voltage 14.1V & 12.2V respectively), and the average loads are a little different too (52 kPa & 45 kPa).

[bigbadspoon]

I could see that having an effect with a MAP-based tune if the fuel curve were not smoothed out, but I'm still using the AFM, so there shouldn't be any huge disparities between cells like that.

Changes I have made from the AFM values included in the reflash that came with the software (and those are not terribly different than stock, if changed at all) are only 1 or 2 percent and the dip seems to happen independent of those values. For example, if I plot AFMv. vs STRIM for the "to work" log, the STFT for the highway loads (1.9v-2.2v) is +/-1%, but if I plot that for the "from work" log, the same range is between -7% and -5%.

Also, I just loaded in my logs from today and it did the exact opposite. To work, battery is 14.1V, AFM is hovering between 2v and 2.1v, MAF looks like it's floating around 50, but the spikes are +/- 10kPa, STFT and LTFT are within 1% for the whole highway portion. Coming back from work, the battery voltage is about 12.2, AFM is between 1.9 and 2.1, and LTFT dips from 0 to -9, but doesn't go back up. STFTs are in the -5% area for the most part during cruise. Even when I get off the highway and go 45, they are still -9.

Do you really think a 7kPa difference in load could throw off the fueling by 10%? Have you seen this happen with the previous Civic or with the TSX?

I forgot to say in the previous posts, but my car is bone stock besides the tune.

[bigbadspoon]

Also, I forgot to say in the last post, I really appreciate you taking the time to look at this with me. I'm totally stumped, and I've mentioned it to some people at work and they said they haven't seen this before. I'm sure you guys get inundated with stuff like this all day, and from the posts I have seen, not enough people say thank you.

[Hondata]

I think there are two factors. If you graph the AFM flow vs short term fuel trim (using the XY graph) then you'll see that the engine runs rich (STFT is negative) below about 22 g/s flow. This means that it will pull fuel at lower load. The second effect is from the lower battery voltage, which changes the injector dead time (and at high flow rates, the fuel pressure). It could be that the dead times from the factory are not 100%, but that would be unusual.

What I would do is to limit the long term fuel trim to +=0% (effectively disabling it). Then start tuning the AFM based on the XY graph with AFM vs STFT (as above). It looks like it needs about 5% reduction in flow below 2.0V

[bigbadspoon]

That makes sense. I wouldn't have thought to disable the LTFT for monitoring these changes, but I see how it should help. I'll post back when I get some data on this.

[bigbadspoon]

So, over the last few days I ran a tune with the LTFT turned off, and made some adjustments to the AFM map to compensate for the peaks and valleys. It looked like a large part of the problem was that the fuel trim varies wildly at idle, and it's so inconsistent, there doesn't seem to be much to be done about it. I'm guessing this is due to the intake temp changing before reaching the cylinders...

Anyway, the logs seemed to be leveling out at 0 until today. On the way to work everything was within 2%, which is pretty good, but on the way home, everything skewed WAY down low compared to the to work log.

For example: to work at highway speeds the AFM was reading 2-2.2 and that range said +2% on the STFT vs AFMv graph. On the way back, the same range was -5%. I'm not sure how I'm supposed to tune it when the average varies by 7% or more for any given value.

I have attached the latest logs below. Any insight would be appreciated.