PI closed loop algorithm for 1227747 ECU?

[Scott Lundahl]

I've out of town and away from my e-mail, but I wanted to say thank you for 
the information.  I've burned an EEPROM as follows:

Prop tables at $4a8 & $4c7 have been reduced by 2/3 (since the 
displacement, and hence fueling requirement, of my engine is about 2/3s of 
the GM unit).  I have suspected the PI overshoot you mentioned for some 
time (the O2 cross count frequency has been well over 10 Hz).

Increased $48f int time delay table by 1.5 (since the the distance from the 
TBI to the O2 sensor in about 50% longer than a stock GM V8 - those 4-2-1 
headers are quite long)

Set $2f9 to 1

Set IAC table at $5c7 up + 1

Enlean table at $3fa to $80 (go big block)

So far it seems to run perfectly.  No trace of surging or other 
badness.  Tailpipe backfire virtually eliminated (pleasant little burble 
now)!  I haven't put on many miles, so things may converge to other values 
where things don't work as well as time goes on (this has happened 
before).  The idle still hunts a bit, so I'll work on that.  But it looks 
good so far.

Thanks again,

Scott

PS - As for the distributor, I'm working on that too.  It turns out that an 
early 70's large cap GM HEI unit will fit this engine if you change the 
drive gear.  I have one with the correct gear.  I've also figured out that 
the ECU-controlled ignition module from a 90 Corevette will fit in this 
distributor (with a little re-tapping of mounting holes).  As near as I can 
tell, this module should talk to the '747 (alternatively I could mount the 
"correct" module outboard of the distributor but that would look 
messy).  The ESC and knock sensor modules are a different story.  I know 
that each set is "tuned" for the engine block application.  The aluminum 
block on my engine will have totally different characteristics from that of 
a stock GM unit.  Is it possible to use the ECU controlled spark without a 
knock sensor (assuming one is careful about advance curves, etc).


At 11:01 AM 7/4/00 -0400, you wrote:

>For the cl idle, reduce the prop gain at 0x2f9 to 1. Might also
>want to go to a larger vac line to the map. A 1/4" line helps
>when engine is cam'd.
>
>If the idle is osc due to iac movement, increase (slightly!), the timer
>vals at 0x5C7 through 0x5CA.
>
>Should have a GM O2 sensor, the response time matches the
>the PID loop better.
>
>Prop gains. May need to be reduced, tables 0x4A8 & 0x4C7. Prop
>gains are interesting, they always work to bring the O2 val in the
>opposite direction, it lean, prop adds fuel, if rich, removes fuel.
>Sometimes it can be too drastic, causing over and under shoot.
>
>If backfire is through exhaust, there is a decel enleanment table
>at 0x3FA, big blocks have some huge numbers there. Don't
>worry about the int and blm's moving a bit, stock engines are
>the same way.
>
>BobR.
>
>Might be worth getting an EST distrib in there. The other
>half of the drivability is in the SA.
>
>Scott Lundahl wrote:
>
> > Hello everyone.  I'm new to the list and this is my first post.  I'm doing
> > an FI addition to a 1980 Triumph TR8.  This is the car with the GM 215 (3.5
> > liter) cid aluminum block V8 derived Rover engine.  OEM fueling for this
> > (49 State) car was via twin Zenith carbs in the usual British style.  The
> > California version had Lucas (really Bosch L Jetronic) port fuel injection.
> >
> > On my car, the original Zeniths were later replaced by a good old low tech
> > Holley 4 barrel and Edelbrock Performer manifold.  I added a tame to medium
> > cam and headers to perk things up a little.  This setup ran OK, but lacked
> > many of the usual amenities such as a choke that worked properly, idle
> > speed control (yes this car has A/C) and so on.  It was also OPEC's 
> dream car.
> >
> > So I decided to add non-eom FI to the car (the eom systems have no
> > performance potential and are incredibly expensive).  I chose a TBI system
> > using a 1227747 (C3 series) ECU because it interfaced easily to the current
> > setup in the car.  The C3s are also easy to manipulate via EPROM (which is
> > good because I have burned many so far).  This system is driven by the
> > stock (non-ESC) distributor via a buffer.  A heated O2 sensor is on the
> > header collector (4-2-1).
> >
> > Anyway, I have the system in and dialed in to the point where it will run
> > with perfection when I set the ECU to run in Open Loop mode.  It will run
> > well with the fuel map set for slightly lean or slightly rich.  However,
> > when I set it to run Closed Loop, the system develops a light throttle
> > cruise surging (very annoying) and backfire on decel (I can almost
> > eliminated the backfire via the decel params).  It also wants to oscillate
> > at idle.  The BLMs tend to wander all over the place.
> >
> > Since things seem to run OK in Open Loop when fueling is slightly rich or
> > slightly lean, running at stochiometric does not seem to be the problem
> > (some engines just don't like to run there).   There must be something
> > about how C.L. operation is accomplished that is causing an issue (check my
> > logic here).  Therefore, I suspect that the PI parameters (gain factors,
> > time delays, etc) may not be correct for this engine (this is somewhat
> > plausible when you consider that this ECU was originally setup to operate
> > with an engine 50% larger or more than mine).  Thus it operates at a point
> > that is close to instability in C.L.  So, I have two questions.
> >
> > Does anyone have any suggestions / experience / solutions with a similar
> > symptoms?
> >
> > Does anyone have information on how the PI algorithm operates on this
> > ECU?  I have seen such details of how the spark control algorithm works.  I
> > also have the detailed hack of the type $42 (this is an incredible piece of
> > good work) but the comments in the PI section aren't quite detailed enough
> > to reverse engineer what's going on.
> >
> > Any help/suggestions would be greatly appreciated.
> >
> > Thanks,
> >
> > Scott
> >
> > 
> ----------------------------------------------------------------------------
> > To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
> > in the body of a message (not the subject) to majordomo at lists.diy-efi.org
>
>----------------------------------------------------------------------------
>To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
>in the body of a message (not the subject) to majordomo at lists.diy-efi.org

----------------------------------------------------------------------------
To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org