[Diy_efi] DIY-WB question

[Garfield Willis]

On Fri, 18 Oct 2002 22:10:31 -0500, bcroe at juno.com wrote:

>Anyone mistrusting the Rcal circuit could put in about a=20
>10K ohm pot instead of the sensor resistor, and adjust=20
>it to get the specified open air voltage. =20
>
>If quite a few users would compile their J9 to J8 voltage=20
>in free air, along with the ohms of their sensor Rcal, we=20
>could compile some kind of statistical error record for=20
>the design. Make adjustments if indicated.  Or perhaps=20
>just show its doing the job. =20
>
>Bruce Roe

Clearly you have a chickenNegg problem with this approach. The 4.00V
nominal Vout that you DIY-WB guys spec'd (actually *assumed*) is BASED
on your own presumptions on both how the CalR & sensor calibration
worked, and what the actual range of Ip were for these sensors in
free-air, both of which assumptions are seriously in error. NOW you're
telling people to substitute a pot for the actual sensor resistor (CalR)
value, and adjust it until ... until WHAT? Until they see 4.00V on the
output of the DIY-WB? This is exactly the same bogus attempt that the
Aussie guys have recommended! You're both chasing your tails on this
one, in locked cadence. The fact that many builders report wide
variation on hitting that 4.00V is proof that there is something
seriously wrong with the CalR adjustments, but it doesn't say ANYTHING
about whether that 4.00V value on Vout is the *correct* value to trim to
in free-air.

Look, the key here is that the Ip (the oxygen pump current) level varies
by quite a bit, over the range of sensor mfg. variations, even if
they're all placed in free-air, at the same temp, pressure,
moon-position, you name it. The 'free-air' oxygen concentration is the
same, but the *sensors* are NOT. You DIY-WB 'designer' guys essentially
run that current (Ip) thru a resistor to produce the output Vout of your
'design'. Without an understanding of the nominal/centroid of the Ip
current distribution, you haven't a CLUE (not even a tiny one) as to
what the value of that resistor should be, to establish a 'standard'
Vout, against which all other sensors must be trimmed, either upRdown
from there. In short, you don't even know what a nominal Ip for these
sensors actually IS, in free-air, do you!? Of course you don't; it's
never BEEN published for the Honda/NTK variant, and you guys certainly
have made no attempt to gather a decent statistical sample to determine
it on your own.

You've got two variables working there. One is the mfg. variations, the
other is WHAT is your overall calibration curve!? The curve that's
posted on the DIY ftp site, is BASED on your CalR circuitry and the
original NTK cal curves, which in both cases, is just plain
presumptiously wrong. So now you're saying to fix that error, one should
adjust the CalR value!!!?? Noooooooo, no, no. That's completely bogus.
Because you don't know WHAT to adjust it to! You have no calibration
standard!

The curve of Ip vrs. AFR that NTK published in their SAE article, and
from which you guys finagled your 'calibration curve', says that in
free-air, the Ip will be centered on suchNsuch a value, and then the
CalR value is to be used to trim sensors that fall on either side of
that nominal free-air Ip. But you guys, without measuring a large lot of
sensors, haven't a CLUE as to what that nominal free-air Ip value IS. I
know for a FACT you don't, based on the 250+ sensors we've measured so
far. I'll give ya a hint, it's most definitely NOT what was published in
the original NTK articles, because the Honda/NTK sensor is considerably
different from the sensor in that article. In fact, anyone with an NTK
blue-box and one of NTK's original sensors knows that the original NTK
sensor is quite distinct from the Honda sensor, by appearance *alone*.
And most definitely, the Ip vrs AFR curves are not at all the same.
They're off by as much as 2mA in free-air, just as an example. Put that
in you bogus CalR circuitry and smoke it.

What that 'genetic' difference means is that you've gotten just about
every feature of the calibration issue for these sensors WRONG! You
messed up the CalR circuitry itself, you missed the lumpiness of the
CalR values, and their effect on the CalR calibration spread if you
tried to plot the correction from just a couple lumpy values, and you've
also PRESUMED that you could just lift the Ip vrs. AFR curve from that
SAE NTK article and didn't even check to see if you were dealing with
the same sensor family. Well, I've got news for you bunkie, you're NOT.

I'm not sure if you could have gotten anything else mangled, but perhaps
I've overlooked some further ripe possibilities. Y'argh.

Here's how this needs-be works. FIRST, you get a curve for the
nominal/centroid Ip vrs. AFR for the sensor family & mfg. spread you're
dealing with (make that the Honda/NTK sensors we're ALL using, instead
of cutNpaste'ing from some SAE paper on that sensor's *forebears*). THEN
you apply the *correct* trim CalR algorithm to each sensor (completely
different, BTW, than the CalR used in the original NTK sensors; those
resistors were in the hundreds of ohms, as again anyone with an NTK
blue-box will confirm, NOT in the kilo-ohms like the Honda/NTKs; another
hint that something is different betwixt the two). The DIY-WB circuitry
& design erroneously PRESUMES the first (the calibration curve), and
ineptly screws up the second (the CalR method). It's only a wonder of
statistics that some of the users of the DIY-WB actually get AFR
readings that are even close to accurate. And that's just the STATIC
calibration/error issue, not to mention the heater control and transient
Ip servo-loop design flaws that cause the device to either lose Ip lock
because the heater either shuts down or allows the sensor to cool too
much, OR the gross indicated AFR overshooting actual, by several AFR
routinely.

Now, I haven't actually said in chapterNverse and schematic/parts-values
how to fix that bloody kludge, but I've certainly given out some pretty
direct hints on how/why/where the necessary changes must be done, before
you have anything even *approaching* what the rest of the trade has
accomplished so far. Nor do I harbor any illusions on your fine
'moderator' Mr. Rivethead's likelyhood in publishing this, but at LEAST
I've stated the facts REPEATEDLY about these issues, ALL confirmed with
data already posted on DIY-EFI itself, and by OTHERS, not meself.

So as I've said before, you need to raze the house and start with a new
foundation, if you hope to get something approaching *measured* AFRs.
Christ, why don't you guys just reverse-engineer an actual properly
designed circuit from somewhere instead of foisting this travesty upon
the sport of proformance automotive. This endless bullshit is just the
most disgusting evidence of totally corrupt humanity I have personally
ever witnessed. The people involved in promoting this sham should be
grossly ashamed and shamed out of the sport; instead what do we have, a
PC moderated EFI list that supports and perpetuates their scam!

Christ, it's bloody despicable. And you want to crucify me for just
telling it like it is. Shit, doods, get a CLUE.


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