Knock Sensor Applications

[Johnny]

Not to trying to slam anyone here, but CHT is not real useful with water
cooled heads. You don't get the same kind of responses with it. So for
water cooled, use the EGT alone. It acts exactly the same no matter if air
or water cooled, and it is the one that falls off when you get on the lean
side anyway. In aircraft, with the adjustable mixture being more or less
standard, that's how it's done. Part of flying is engine management and
that includes leaning to peak and for some engines, leaning to xdegrees
cooler than peak on the EGT. 

This was one of the first obstacles I had to tackle when I started to
develop an EFI V8 for aircraft... how do I, or can I run closed loop with
no O2 sensor? No O2 sensor because of leaded fuel. Leaded fuel because not
every airport sells MOGAS. With aircraft being more or less steady state
engines, and because it was easier, at first I decided that I wouldn't do
any closed loop and would run strictly from "table" using MAP, intake air
temp, engine temp, RPM, ambient atmospheric pressure, and throttle
position, with a "trim" pot on the dash for if the mixture needed
correcting, just like a "normal" aircraft engine. Then I thought, why not
have the computer do it instead, using a correction table based on EGT just
like the person would, and achieve WOT closed loop that way. While using a
Haltech E6 box as the ECU, I tried to devise an interface for the EGT
sensor that would give input to the box as an O2 sensor would. A driver or
interpreter if you will. I didn't get it to work as well as I wanted to and
will be doing more with it later on after other more critical issues are
handled.

The main problem I had was getting the EGT sensors I have calibrated to
something useful for the O2 sensor input on the Haltech box. EGT is wide
scale, O2 sensor is narrow.  The other issue was picking an EGT sensor that
I thought I could get in the future as this is to be a production engine. I
ended up doing a (gasp) simple OPamp circuit to do the conversion which was
not satisfactory because of the wide range of the EGT sensor and the
nonlinearity between the two scales. I think I need to go
analog->digital->analog to get enough control of what the output to the O2
sensor input of the box is. Of course if you were building the ECU from
scratch, you could just build all of this in to it. I just wanted to go
flying with an EFI V8 under the cowling, not have to build every little
wheel involved from scratch.

IOW; Yes, it is the way to go for reading the end mixture at high power
settings. The EGT/PowerOutput link are as close a link as you will get. 

-j-

----------
> From: John Faubion <jfaubion at beaches.net>
> To: diy_efi at coulomb.eng.ohio-state.edu
> Subject: Re: Knock Sensor Applications
> Date: Sunday, October 20, 1996 11:00 AM
> 
> > OEMs use an advance determination method called MBT (Minimum spark
> advance
> > for Best Torque), unfortunately knock can sometimes set in before best
> > torque is realized.
> > 
> > So far, things do not look very good. All of the above leads me to a
> > number of conclusions (all mine):
> 
> Everyone is talking about knock sensor problems. Why not do it the way we
> do in kart racing? We only have a single cylinder. With everyone so close
> to each other in terms of weight and power, any loss in power will get
you
> passed! We use cylinder head and exhaust temperature gauges. The CHT
mounts
> under the spark plug and EGT is obviously mounted in the exhaust stream.
We
> tune the engines using low and high speed jets on the carburetors. The
key
> to good tuning and max. power are watching the relationship between the
two
> gauges. As you lean out the carburetor both CHT and EGT will rise. This
> will continue until the point of incipient detenation (can't be heard).
At
> this point the CHT keeps rising and EGT drops. If you let this continue
the
> piston sticks in the bore and your out of the race. 
> 
> There are some drawbacks to this method. The EGT responds almost
> instaneously while the CHT lags behind a bit. We also use aluminum heads
> which would allow the CHT to respond sooner than an iron head. In
addition
> we also only have a single cylinder to deal with instead of 4 to 8. Still
I
> think this might be a better way to tune for max. power since the lamba
> sensor doesn't do a great job of it at WOT. Does this help spark any new
> ideas?
> 
> John Faubion