Return of the diy_EGOmeter (YES, it's finally ALIVE!!)

[garfield at pilgrimhouse.com]

On Tue, 14 Apr 1998 00:22:48 +0100, Sandy Ganz <sganz at wgn.net> wrote:

>Man, you are pumped Garfield! As soon as you get the schematics up, I'll
>see about whipping up a nice PCB if people will be interested. Sounds like
>you have done quite a bit of work to beat the high $$ meters, and that is
>GOOD!

Yeah, I'm pretty jazzed, mainly cuz I want one of these REAL ego meters
to use, meself!

But nah, not really much work (and no, I'm not the modest type); Frank
Parker supplied a tremendously complete set of intel for the whole
operation, so it was just a matter of piecing his SAE info, private
intel, and inputs from several of yous guys together. Him also being in
possession of an NTK sensor and interface box (and NO, he NEVER broke
seals and opened the thing up to look inside; they're toooo eXpensive to
do that!!), and being able to watch it's behavior and run I/O tests on
his end, was another key to fast closure on several issues/questions
that came up in the process.

>Let me know when you get all the kinks worked out of the circuit, and I'll
>take a crack at the PCB for the group.

OK, eXcellent! It'll be real soon. I've already detuned the feedback
path and now it settles tamely, but I now see why the NTK "stoich"
offsets Frank was seeing are seemingly so "large". Like, his one box
seems to always stabilize at VsCell = 0.428V instead of the target
0.450V, and at first glance that seems significant, until you find that
this diff. of 22milliVolts represents a Ip pumping current change around
stoich of a few microAmps. Vewy sensitive! The way we should look at
this is, anything between say 0.420 and 0.480 is still eXtremely,
exquisitely tightly locked in "on stoich", so with the feedback detuned,
I now settle at around 0.435V instead of 0.445V. The difference is still
WAY down in the noise, but it doesn't "twitch" around stoich, like it
did before. So that's one down, and one minor one to go.

The other remaining thing to confirm as working properly is just
basically a switch that stops the Ip pump from operating until the
sensor warms up and it's output has had a chance to cross over to the
lean side of stoich (remember, we're powering up in essentially free
air), before turning on the ion pump to pump us back up to stoich. This
is such a dirt simple part of the circuit, that I expect it will take me
10 mins to confirm proper working of the switch, and that's it. The
hysteresis in the mechanism has already been tested, and works fine;
it's only the current switches I'm using that don't work reliably on 5V
Vdd. The new ones coming from Analog Devices are guaranteed to work on
5V, so I think that's a no-brainer. Once they arrive and that checking
is done, it's off to the Black Forest to put the Elves to work. I can't
imagine the parts costing more than about $30, eXcluding the O2 sensor.
Once I build a couple more to confirm they're repeatably cloneable, we
oughta see what kinda deal we can get with Honda for a group buy of the
sensors. Since anyone can get them in one-zies for $130 from their local
dealer, I'll bet we could easily get them below $100 if we bought a
couple dozen at a wack. That would put the total cost below most of them
El Cheapo meters. I love it!

One thing I do need to point out, tho, for due diligence sake, is that
we haven't yet tested the whole thang all the way from filthy rich to
bone lean, nor will I be able to send my proto back to Frank for a
side-by-side calibration/comparison with his NTK boxNsensor, until I get
them AD switches. Yeah, that's really the very NEXT thing to do, once I
get the parts, is send the proto to Frank for a sanity check &
comparison, and built say two more to confirm repeatibility. If it
passes both THOSE tests, we're truly home free.

Gar