What is AutoTune?

[Rollagti16]

Hello, one quastion. :
What is autotune in megatune??

[efahl]

Rolla,

Auto-tune is an algorithm in MegaTune's 3D tuning dialog that adjusts the VE table based upon feedback computed (usually) from the lambda sensor readings. It can be used with a narrow band sensor to tune the main part of the VE table to stoich, or with a wideband to tune just about all of the VE table.

Eric

[Fastest95PGT]

When using a wideband and AutoTune, does it aim for the target AFR map as opposed to stoich for a narrowband?

[boost junkie]

Its my understanding that autotune just goes off of the EGO correction. If you set EGO correction to use the AFR table then thats what autotune will use...

[Fastest95PGT]

Thanks Shawn.

[efahl]

When using a wideband and AutoTune, does it aim for the target AFR map as opposed to stoich for a narrowband?

As Shawn says, it simply tries to minimize the error in the control variable, which is almost always set to be EGOcorrection. The tuning algorithm looks at the value of the specified output, sees if it's 100 and if it's not it adds/subtracts something from the current table entry to move the output closer to 100.

Code: Select all

   correction     = EGOcorrection_value
   value          = VEtable_value / 100.0
   steps          = gain * value * (correction-100)
   VEtable_value = VEtable_value + steps

There are a bunch of other parameters dictating when this can happen (both in terms of proximity to a node and frequency of updates), which should not be too loose or too fast or the algorithm can become unstable.

Eric

[ochizon]

Im sorry if im a little dense...

There are 2 functions for autotune?

1. Autotuning the VE table: Sets the VE output based on the wideband acheiving the desired output.

2 EGO correction: Once the VE table is set in stone, the wideband functions only for minute alterations to maintain the desired AFR output.

Is that right? Or is it ONLY #2?

[Fastest95PGT]

Thanks for the expanded clarification Eric.

I am wondering if you need to be in closed loop for AutoTune to work with a Wideband since no EGO correction is going on?

[efahl]

There are 2 functions for autotune?

1. Autotuning the VE table: Sets the VE output based on the wideband acheiving the desired output.

2 EGO correction: Once the VE table is set in stone, the wideband functions only for minute alterations to maintain the desired AFR output.

Well... Actually both simultaneously. Your MS will compute EGO correction and apply that internally. MT's auto-tune algorithm is watching that EGO correction and minimizing it by messing with the VE table. Obviously the MS does its EGO correction all the time (assuming closed loop is enabled), but MT can only do its bit when you have MT connected to a running engine.

Eric

[efahl]

I am wondering if you need to be in closed loop for AutoTune to work with a Wideband since no EGO correction is going on?

Now that you mention it, no, you wouldn't necessarily need to be in closed loop mode at all.

You could set the auto-tune up to do something like this, which would work irrespective of whether you were running closed- or open-loop:

Code: Select all

[OutputChannels]
   ...
   lambda    = { 1.5 - 5.0 * egoADC/255.0               }
   afr       = { lambda * 14.7                          }
   targetAFR = { vexInterp(rpm, map, 0, "afrtable.vex") }
   AFRcorr   = { (afr / targetAFR) * 100.0              }
   ...

[AutoTune]
   table = veTableMap
      corrector = AFRcorr
      ...

Just make up a vex file "afrtable.vex" and you're off to the races.

Eric