ITB tuning on a street legal car

[roulli]

Great info, thanks Andy.

With my motor on SD setup, I found that things remained rather linear until 85kPa (MAP) , from there on I had to increase resolution. So that principle is the same as with yours.

The major difference is, that your engine is rather linear over the rpm range (from left to right)
The TR6 engine will peak out at 3500 and then VE values will drop until it redlines at 5800.
The reason is that the V. efficiency of this old iron, long stroke lump with it's relatively small valves (non-crossflow setup) maximized at 3500 to 4000 rpm.

But still the engine came originally with mechanical fuel injection from Lucas with 6 ITBs, which I converted to efi. So I didn't put on any Jenvey stuff paired with wishful thinking...

Patrick

[roulli]

You can set the load algorithm to %baro and filter on "Load"

2swe is correct. You were right except it is not 90 kPa but 90% baro.

I was about to follow this advice, but setting the load algo to "%baro" means actually that I have to leave the "ITB" load algo.
How does this fit, as I'm looking into ITB mode?
Or is there another place than "Basic/Load Settings"--> "General Settings", where I have have to set the "% baro% load algo?

[Yves]

Hm, if your normal baro is around 100 kPa where you normally live, it shouldn't make much difference.

I just searched for at what point I reached 90 kPa at a certain rpm point and started noting them out of the different logs and compiled a curve with that. Hasn't failed me yet.

[muythaibxr]

You can just gently drive it around in % baro mode for a while then switch back to ITB.

[2swe]

or you can use your current datalogs, divide MAP by ambient pressure and multiply result by 100.... simpler, faster..

[roulli]

or you can use your current datalogs, divide MAP by ambient pressure and multiply result by 100.... simpler, faster..

That's actually, what I did last night.

I fettled around with the "PROCBARO" filter formulas explained above and it worked a treat.
The (real) baro switchpoint curve is now lower and flatter. The residual points, after filtering in the Scatter diagram, were also more defined in a line (I mean the residual cloud was more defined as a line)

Now i believe to understand that the reason I had many cases of constant VE values of 60% (@ the "at switch point line") between MAP values of 85kPa and 90 90kPa (which corresponds to BARO 90 to 94 kPa at log time)
As the "TPS baro switch point line" (left diagram) was too high and too steep there are many cases were both conditions were not fulfilled to switch from SD to AN
--> Load jams at 60%

I'll make a test drive with the correct left ITB curve this afternoon.

Thanks a lot 2swe for your advice

Patrick

[Peter Florance]

I have actually stopped using closed loop while tuning. I find it gets in the way and makes it harder to see how close the tune really is.

I need it when tuning cams.
Otherwise it takes too many pulls.
Up to 7% or so with good PID seems to do OK in the dyno.

Sent from my XT1080 using Tapatalk

[roulli]

Testdrive went great.
MLV data-log shows now the swap from SD to AN, where it should occur: at the defined 60% ITB Load

I upgraded MLV to HD version to use the Histogram function. This allowed me to show the AFR yielded for each cell.
Copy-pasted the AFR histogram, AFR target and the VE table into Excel in order to generate an adapted VE table.

The car drives now great, with a smooth transition from SD to AN and the requested kick into the redline.
I just need to tweak a little at very light throttle openings, where it is a bit lean.

So at the end ITB turned out to be the right algo for my TR6.

Thanks again to the forum for the great support,

Patrick

[whittlebeast]

Post up a big log. We always like to see how people are coming along.

Andy

[roulli]

Ok, will do so, but with a file size limitation of 750kB we are rougly looking into 20min drive, if zipped

Patrick

[whittlebeast]

PM me for an email. I can get really big emails.

Or I bet you could guess my Gmail address...

[roulli]

Well, there's always room for improvement….
What I don't understand is the following :
Why does the Histogram of the VE1 values from the data log not render the same values, than the VE1 table in the .MSQ file?
For a specific rpm/ load combination, the VE value should be given by the .MSQ and represented in the histogram , shouldn't it?

VE1_MSL vs- VE1_MSQ.jpg

Same for the AFR Target value in the MLV histogram. Should also be given by the AFR table in the .msq, even-though the table size is different for .msl and for .msq
The values should be similar for the sam rpm/load cell...

AFR Target_MSL_vs AFR_MSQ.jpg

I'm puzzled

[roulli]

Hi,
sorry to insist…
but can anybody explain, please, the divergence between the target values in the .msq and the logging of these target values in MLV.
See screenshots above.

(The .msl is too big to upload, but I can trim it if required)

Well, there's always room for improvement….
What I don't understand is the following :
Why does the Histogram of the VE1 values from the data log not render the same values, than the VE1 table in the .MSQ file?
For a specific rpm/ load combination, the VE value should be given by the .MSQ and represented in the histogram , shouldn't it?

VE1_MSL vs- VE1_MSQ.jpg

Same for the AFR Target value in the MLV histogram. Should also be given by the AFR table in the .msq, even-though the table size is different for .msl and for .msq
The values should be similar for the sam rpm/load cell...

AFR Target_MSL_vs AFR_MSQ.jpg

I'm puzzled

[whittlebeast]

The answer to that is somewhat involved. That number is generated statistically. The more random data you have to use for the statistics, the better the probability will be that the two plots match. But the bottom line is the two will never match perfectly.

Andy

[roulli]

ok, I kind of understand that measured z-data won't be constant for a given x,y
But VE and AFRtarget are given values. For a given rpm and given Load, they should always correspond to the value ordered by the .msq.

My understanding is that the VE value is an ordered percentage of the pulse width defined in the .msq. as a function of (rpm and load)

So statitically MLV should always read the same predefined VE value for a given rpm and load, thus averaging to the same VE value.

What is wrong with my consideration?

Patrick

[hector]

Kinda late to the party but since I found the thread very informative I would hate to see it die without an answer to your question. I don't have an answer but I do have a guess as to what might be going on.

The data log numbers are probably actual numbers being used for fueling. In other words, it includes all the other factors that are used to determine final fuel numbers for VE. You know, things like Baro, Air Density, Coolant temp, etc. All these things determine the final calculation not just the VE table numbers. The VE table is actually a reference, not a command as you stated.

Now, the AFR table being different is odd but I would suspect a bit of the same here.