RE: alpha N vs MAF (long)

[albert jenab <jenab@xxxxxxx>]

Land Shark <lndshrk@domain.elided> wrote:

<snip>

 >You CANNOT tune with a stock Oxygen Sensor.. the cheapest REAL meter 
 >is well over $1000 dollars by itself.. this "systems" are being sold
 >with BS air fuel meters that are dependent on stock-type Oxygen Sensors
 >which CANNOT and DO NOT yield reproducable results from run to run.. 

 >Don't believe me.. then call Robert BOSCH.. they developed the sensor,
 >I think they might know ;)

 >ALPHA-N is NOT for street cars.. it doesn't belong on a daily driver, it's
 >for race cars.. my system is an improvement over that on the Gruppe A
 >touring cars.. NOT FOR STREET USE!

 >And yes, I know some people ARE running it on the street, but it's not
 >my suggestion, nor is it my wish!

 >As far as any external box with knobs and a MAF...  well.. if'n it were
 >THAT easy to drop a MAF on any car, you'd see a lot more BOSCH cars with
 >MAFs! It isn't.. it requires a COMPLETE code rewrite!

 >Jim

 >PS: The whole thing comes down to this.. (for you math heads)

 >    Try to match a LogE curve to a 5th order polynomial using 3 points
>     and then tell me how much error you get.. (hint: lots)

OK, I'll bite.  Maybe this is beyond the scope of digest chat and into trade secrets.  But I'm a math head and a digital controls weenie and would like to understand this.  Would you mind expanding on your remarks a bit?  I guess you are saying that a stock O2 sensor has too much drift, noise, error etc. to accurately use it as a calibration sensor for precise performance tuning.  Then how can it be used for real-time AF mixture control in an engine?  Said another way, if I were to buy a very accurate calibrated O2 sensor and plug it in, would I see a lot better performance out of my car?  Or does the chip sort of do most of the work open-loop and then the O2 sensor only makes minor corrections for gross out-of-range conditions?  This would mean the stock O2 sensor doesn't really do much under normal operating conditions, and your remarks about unsuitablity for tuning would make good sense.  Would also explain why certain people's chips are way better than others....

So ignore the O2 sensor for a while.  These add-on boxes use three knobs and these knobs control some scalar quantity maybe over roughly 1/3 the rpm or 1/3 HFM output signal range each, or are these the gain parameters of a PID or similar controller?  If it's gain parameters for a closed loop controller, I don't see how the shape of the ideal curve matters, because you are only at one point on it at any given time.  However, if the O2 sensor truly sucks as a real-time feedback control sensor, then obviously 3 points is not enough to keep the chip operating in mostly open-loop mode.

I don't think anybody in their right mind would claim that such a system would outperform a purpose-designed chip which had used a high-priced O2 sensor to calibrate for a fixed set of mods.  But is it utterly unreasonable to expect improved performance, or is this also total BS?  For example, in addition to an E36 M3 (with one of your chips), I have a supercharged Miata with a J&S knock sensor.  You can't get chips for it, as they are not removable.  There are a couple of different HFM systems out there, one which replaces the entire EM computer with a programmable unit, and then there's the split second device.  Based on your comments I would lean toward the completely programmable unit even though it's more expensive.

Finallly, suppose someone built an aftermarket add-on system which used a very high-quality O2 sensor and some mondo blazing fast DSP chip and tried to do real-time highly accurate & very tight closed loop control of just AF mixture.  Would this work?  I understand Honda has an O2 sensor system on some street cars now which is accurate enough to "see" the exhaust pulse coming out of each cylinder and adjust its mixture in real time.

Any comments would be greatly appreciated.  This whole engine management area is voodoo to me.
- -Al

------------------------------