SD tune

[slows10]

^^ I agree, I ditched the maf to try SD and used the AutoVe tutorial its easy to use. Since you are doing a transplant in a jeep a SD setup may work well for you.

[catman3126]

Hey thanks I found that this morning, so how do I know which custom OS to pic from the bins?

[slows10]

It depends on the os,what o/s are you running now? There is a tutorial on the efilive site that shows you how to switch to a cos and what cos to use depending on what base o/s you are using.

[catman3126]

ok sounds good I guess it will show how to pick the right one?

[Sid447]

With a NA engine there is no power output advantage caused by removing the MAF and running in SD.
Regards,
Taz

You sure about that?

[joecar]

With a NA engine there is no power output advantage caused by removing the MAF and running in SD.
Regards,
Taz

You sure about that?

There's a long thread on LS1Tech on that very topic, it contains evidence that SD and MAF produce the same HP and TQ.

Sid, I believe you have seen that thread.

[Sid447]

Hi Joecar,

There's loads of threads on other public forums about this very thing as most are aware.
The problem is trying to sort out the reality, a person wouldn't know who to believe, as most claims aren't validated or carried out in an un-biased or correct manner.
Maybe in comparison with the later 85mm maf-sensor things would be different.

Using the older 75mm maf, same dyno and a trusted Australian race engineer with the same car, same tune (we had it set just like the HSV GTS single-spark tunes so we could fit or remove the maf at will).
Nothing, and I mean nothing was adjusted other than the physical removal of the maf-sensor; I was there the whole time and there was just the two of us.
It made a difference, the most pronounced was to throttle response.

It convinced me, especially after comparing drivability on the road. The nicest was when open loop, but run CL for sensibility sake.
I don't want to convince or change anyones' preset ideas of anything.
The response to Taz was of a friendly nature with humour intended; not to ridicule or start a game of verbal badminton.

I run SD because I think it's better and have no urge to change other peoples general opinion; if it's the same, then it's the same! (but different).

[Taz]

Hi Guys,I haven’t had much time for the Forum lately … mad rush here to get things done before the snow arrives … its late this year thank heaven !

No malice ever taken from your comments Sid … much like myself, you appear to sincerely wish to assist others. My primary intent in this thread had been to assist the OP with setting up a VSS input to the PCM, which is necessary to get the most out of any street driven tune. The thread then ran head long into switching to a COS prior to addressing pre-existing issues … at which point I was out.

I have not seen the post Joecar referred to, so my apologies if any of the following are repetitive. My first career was in clinical psychology, and I studied research methods and statistical analysis while in university (a very long time ago !). I retired from this field sometime ago, so I may be a little “rusty” on the absolute application of methods and analyses.

I believe Joecar and Marcin (and I’m sure others) have a background in pure mathematics, and would be able to correct any of my misconceptions.

Data collection and subsequent analysis (in this case a chassis dyno) must be valid and repeatable, and must address extraneous and confounding variables. Accuracy and precision are essential elements for scientific instruments (chassis dyno), and establish the internal validity of the measurements made by the instrument.

Accuracy - how closely an analytical result approximates the intended target.

Precision - how closely multiple results resemble each other.

Precision and reliability are similar, in that they are both indicators of repeatability.

An engine dyno has fewer variables to control, as compared to a chassis dyno. With a chassis dyno controls must be in place for any variable that may affect engine power output and also “rolling resistance”, these could include:

• engine oil temperature
• engine coolant temperature
• intake air temperature
• fuel temperature
• transmission oil temperature
• differential oil temperature
• universal joint temperature
• axle and / or axle bearing temperature
• rear brake drag
• tire temperature

After developing controls for all variables, the method of data collection needs to be considered. This may be:

• 5 dyno runs to warm up components
• 5 dyno runs with the MAF in place and enabled
• 5 dyno runs with the MAF in place and disabled (SD mode)
• 5 dyno runs with the MAF removed and disabled (SD mode)

This would be followed up by reversing the order of the dyno runs (with the exception of the warm up runs). The results would then be analyzed. Types of analysis could include an Analysis of Variance (ANOVA), an Analysis of Covariance (ANCOVA), and / or a Multiple Regression Analysis. These analyses would assist in determining if the difference (variance) was statistically significant (i.e. not due to the affect of extraneous / confounding variables).

If the results of this single case were deemed to be statistically significant, the data collection / analysis methodology would then be repeated employing a representative sample of engines … perhaps 10 (i.e. n=10). The results of this group would then be analyzed. This would then yield the ability to state, with a measure of confidence, that the results were or were not statistically significant.

Most people are familiar with a standard distribution curve (“Bell curve”) …. most often associated with IQ scores … mean of 100 … standard deviation of +/- 15. The tails of the curve (high & low ends) are not considered to be statistically significant. This means the difference between an IQ score of 162 versus 176 may not be statistically significant. The “confidence level” of such endeavours is generally established – often in the 2.5% to 5% range – meaning any difference in observed results that are equal to or less than this value, are not considered to be statistically significant.

In the graph you posted the engine made 335.5 HP with the MAF, and 357.5 HP without. This represents an increase of roughly 6.5%. An increase this small could easily be the result of extraneous and / or confounding variables. But, the curves themselves are most telling. The curves run loosely parallel and in good agreement (no statistically significant difference) until approximately 4000 RPM – where they begin to diverge. This coincides with the point in an OEM based tune where the OS switches from SD mode (VE table) to MAF mode (MAF table).

It is probably that the VE table (SD mode) was a reasonably accurate representation of engine airflow, while the MAF table was not. A rework of the MAF table values would have probably narrowed the observed results.In the end, a test using a single engine without controls in place for extraneous / confounding variables, amounts to anecdotal evidence not scientific evidence, and therefore cannot be relied upon.


Regards,
Taz

[Sid447]

Brilliant Taz,

Love the post!

[slows10]

If the air-fuel ratio was optimized during the maf run and it was still down on power to the sd run, you could play with the maf table all day long it would not change the peak power output. A log of the run watching a wideband could confirm it.Not sure how accurate the lamda reading on that sheet really is. Im just assuming that both the maf tune and sd tune were set up correctly. If they were than that data is pretty accurate. The problem is that we probably wont get all the data on the runs or know how the testing was done.