[Gmecm] Re: autotuning

William Lucke william.lucke at highspeedlink.net
Tue Aug 30 02:25:32 UTC 2005


VEMaster sounds like it could be implemented almost as effectively in an 
Excel spreadsheet...

That's exactly the reason it's so simple... there's no real art to the 
algorithm of updating the VE table. Just simple math... over.. and over.. 
and over... the perfect kind of work for a computer program.

I DO think a sanity check is required, but instead of VE values, it ought 
to be something like "Is my engine ready to tune yet?"
Time saving algorithms or not, tuning an engine with a huge vacuum leak is 
a waste of time either way.

I do think it ought to have an adjustable minimum BLM difference before it 
starts working on the VE table, though. Initially set the min diff to 5, 
and it will leave BLM's from 123 to 133 alone. Once it stops making changes 
(IE, all the BLM's are within 5), then drop it to 3, then 1 if you're 
really picky.



Will


>From: charles at taildragger.info
>Subject: Re: [Gmecm] Auto-tune algorithm for real-time tuning programs
>
>I suggested that to Craig Moates about a year ago and he said he had
>already been thinking about it.  Seems like the obvious think to do once
>you have a chip emulator (I have Ostrich) and Tunerpro RT running.  All
>the data is on the laptop, why put a person in the loop when the person is
>just mindlessly implementing an algorithm anyway?  Just drive around until
>all the BLMs are 128!
>
>There would be some subtleties though, it seems.  You'd need some sanity
>checking so the VE table doesn't get totally trashed if you have some
>major problem with your engine.  First thing that comes to mind is a
>max/min VE table that you can't exceed on either end.  Plus maybe a
>max/min BLM to accept into the update routine...
>
>Charles
>
> > The latest generation of AEM EMS's have an "autotune" feature that allows
> > the computer to dial in its own fuel map based on the readings of a wide
> > band O2 sensor.
> >
> > The only portion of that algorithm that is not currently available for DIY
> > tuning of GM computers is a feedback loop linking BLM's to VE table
> > values.
> > This still has to be done manually. I wouldn't think it would be terribly
> > difficult to add that algorithm to a real-time tuning suite like TunerPro
> > RT or TunerCAT RT. The software could look at the datastream and after the
> > BLM value in a given cell has stabilized, would modify the corresponding
> > VE
> > table entries to reflect the value of that BLM cell. I know that each BLM
> > cell accounts for multiple VE cells, so some smoothing and conditioning
> > and
> > interpolation would be involved, but the problem is pretty simple
> > computationally.
> >
> > That way, you just drive around with an emulator in your computer and a
> > laptop in your passenger seat and the fuel map basically dials itself in
> > as
> > the BLM's change. It will still require tuning of the spark table and
> > final
> > manual adjustments/smoothing to the VE tables, but it would save a whole
> > lot of time and effort.
> >
> > Has anyone else been thinking of this?
> >
> >
> > Will



-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.344 / Virus Database: 267.10.16/83 - Release Date: 8/26/2005


_______________________________________________
Gmecm mailing list
Gmecm at diy-efi.org
http://lists.diy-efi.org/mailman/listinfo/gmecm



More information about the Diy_efi mailing list