TS (autotune) vs MLV different results

[angus99]

After a few recent mods to fix an idle issue, I’ve noticed a lean spike when pulling away from very low speeds or standing start. The odd thing is that auto tune keeps trying to lean out this area of the VE table. However, if run the VE analyser in MLV, it richens up this area as you would expect.

Is there any reason why the two programs could be treating this area of the map so differently. The results for the rest of the table is very similar.

In the meantime it may be best to tune the low load area of the map with MLV and then apply filters so autotune doesn’t change the table in this area.

[LT401Vette]

You would need to post log files and MSQ's to see what it going on. There is likely some condition on your setup that may need a filter.

[angus99]

Sorry for the lack of Log files, but I'm struggling to get my windows tablet to connect to wifi. I'll see if I can transfer the files to my PC with an OTG cable and memory stick.

In the meantime, I've figured out that I'm using completely different lambda delays between the two Apps.

I've tried to work out the correct lambda delay from my logs and it seems that I'm experiencing a very large delay of about 9 data points in MLV at low revs/load, and a little lower at higher revs. I'll try and update the lambda delay table in tunerstudio to compensate.

One other thing I've noticed is that I'm using the default averaging lag factors of 50, under basic settings / general lags. Would increasing these to say 90, reduce the lambda delays I'm seeing in my log files?

[LT401Vette]

One other thing I've noticed is that I'm using the default averaging lag factors of 50, under basic settings / general lags. Would increasing these to say 90, reduce the lambda delays I'm seeing in my log files?

very little. 50 should be fine for this.

[angus99]

Here's my tune and log from earlier today.

I seem to be struggling to iron out the lean areas on cruise and also some rich spikes on initial throttle. Tried both veal and auto tune with different lambda delays, but get very different result with each. In fact autotune seems to try and lean out the areas where it is already lean.

Also noticed that ae is kicking in on occasion below the tpsdot threshold.

[LT401Vette]

I'm not seeing anything that should make much difference between the 2. However, your set up is a bit off the norm as you are running alpha-n and it is a very old firmware.

More specifically where are you seeing the deviation from a rpm and TPS% bin perspective?

[Paul_VR6]

If you are analyzing in mlv make sure you have the load axis selected correctly. Threw me off a few times on some ancient fw.

[angus99]

I'm not seeing anything that should make much difference between the 2. However, your set up is a bit off the norm as you are running alpha-n and it is a very old firmware.

More specifically where are you seeing the deviation from a rpm and TPS% bin perspective?

Is there an issue using alpha-n? The engine is a 5.0l rover v8 in a TVR chimaera, most of these running megasquirt seem to use Alpha-n rather than map based tables. I should really get round to updating the firmware, just scared that I would have more issues and maybe struggle to get the car running again.

The main areas I was seeing differences were transient areas, e.g pulling away at junctions. Which is why i thought the lambda delay might be an issues. I've since updated the delay table in tuner studio and the differences between autotune and mlv are smaller, although autotune does still try and lean out the table in some areas where the logs show the AFR in the 15-16 range. Maybe I need to tweak the table a little more.

[angus99]

If you are analyzing in mlv make sure you have the load axis selected correctly. Threw me off a few times on some ancient fw.

Do you mapean the 'default y axis field' under the options tab? Mine is set to TP.

[LT401Vette]

Yes... I was going to question the same thing, but then figured MLV is where you said it was working, so it was probably right.