[Diy_efi] DIY-WB question

[Garfield Willis]

On Fri, 18 Oct 2002 09:53:46 -0700 (PDT), allen brown
<abrown02000 at yahoo.com> wrote:

>Can't make up your mind? Last time you said that the
>variences were due to the CalR resistor being selected
>from just a few possible values. IOW too course of an
>adjustment. So what is it?

Nope. If you'll look more carefully at archived posts, you'll see it's
not a matter of making up anyone's mind, but BOTH factors conspire to
make the CalR treatment in the DIY-WB seriously *faulty*. I have
reported BOTH phenomenon, on several occasions. The REASON for the CalR
circuitry being in error is because the designer of this bit (Bruce Roe)
decided to interpolate from just two sensors (I have both his
handwritten notes and e-mail on what was done, BTW, obtained from a
friend of mine :), and since the CalR values are lumpy, apparently was
seriously misled by using such a small lot (two sensors is hardly a
'lot', but you see what I mean) in figuring out the voltage divider that
uses the CalR to trim the sensor's "gain".

Because the 'accuracy' of the CalR for any given sensor hops around a
fair amount due to the lumpiness, in order to get the compensation
circuitry right, you would have had to measure thru quite a number of
sensors to get a clue as to what was required in the gain adjustment
circuitry, to get IT right.

I'll give you a real whirld example: I have two sensors in front of
mean, representing the widest spread of sensor gain we've seen to date,
in calibrating 280+ sensors by now. I have over time culled out the ones
on either ends of the bell curve, as well as logged the distribution of
gain amongst that growing sample.

If you were to apply the DIY-WBs "circuit to use it", to these two
sensors, the 'lower gain' sensor would be miscalibrated by the DIY-WB
and off by over 25%, and the higher gain sensor would be off by over
16%. By these figures I mean, that the DIY-WB's "circuit to use it",
would have trimmed the former sensor's gain by an amount that's 25% too
low, and the latter sensor would have been trimmed by an amount that's
16% too high.This range of disparity from the circuit design flaw in the
DIY-WB (and it's pirated Oz counterpart), are the reason you see people
getting such wide variations on what the DIY-WB reads in open/free air.
The end result is, that altho with the DIY-WB circuit, the free-air
output should be nominally 4.00V, the actual value one sees even AFTER
applying this erroneous CalR 'correction circuit', will be off as much
as +-20-25% of either rich or lean F.S. (ie, in either direction), or
upwards of 0.5-1.0AFR static errors. This correlates well with the
variations in 'free air' Vout I've seen reported, and the static errors
in the two Bosch comparison tests reported, on various lists.

It's bad enough when the CalR 'correction' is sloppy/lumpy by design;
getting the circuitry wrong which then uses this lumpy CalR, only
compounds the problem.

The designers of that circuit really should fix it, assuming they're
able. However, the first step would be in testing enough sensors to get
a better idea of what the mfg. variation/spread those CalR's actually
represent.

Gar Willis
EGOR Technology
www.egortech.com

_______________________________________________
Diy_efi mailing list
Diy_efi at diy-efi.org
http://www.diy-efi.org/mailman/listinfo/diy_efi