Maximum advance

[Axel Rietschin]

> On Thu, 12 Jul 2001, Axel Rietschin wrote:
> > > That's the plan! I figure catching detonation would at least be step
one
> > > :)
> >
> > For less money you put your engine on a dyno and you tune it the old
> > fashioned way, using your ears as sensors and your brain as signal
> > processor. Quite effective. Expensive transducers won't really help you
tune
> > on the road anyway and, IMHO, are not really required unless you are
> > developing, say, a new head of your own or do serious engine research,
or
> > maybe if you ECU can actually use them.
>
> True, but my major drawback is that the nearest chassis dyno is about an
> hour away, and they all seem to charge $75-$125/hr for strap down time. I
> could easily spend over $1000 on dyno time alone. With the pressure sensor
> you can see what the results are during combustion. The PV diagram should
> let you get the timing and fuel spot on. I get frustrated on
> the dyno sometimes since I don't know if I have too much timing, or not
> enough, etc.

The true solution is to use an engine dyno. Chassis dyno is probably a bit
better than open roads to make a ignition map, but it is still far too noisy
to do a good job IMO. Also you may well melt a few engines before you figure
out what is knock and what is not on the pressure diagrams, or while you
debug your filtering software.

With a copper pipe bolted on the head and connected to a "stethoscope" on
the other end, across a fully soundproof wall, you will hear even isolated,
spurious knock bursts and be able to back off timing very quickly. For this
process to work well you need to stabilize the engine at every load/rpm
points and adjust them individually, and you need silence and concentration.

Knock is very random in nature and you may only have one burst of 10 or 15
out of 200 or 300 burns that pings, I'm not sure how you'll catch and
display that in real time with a PC, but your ears definitely will, every
time, provided you are in a silent environment.

When advancing timing small steps at a time you will feel an ever so subtle
change in the absolutely chaotic internal engine noise that will tell you
knock is just about to show his ugly face and soon thereafter you'll hear an
erratic burst of noises somewhat similar to what you hear when you lit a
match, only much shorter, slightly standing out of the huge background
noise. When you hear that, quickly back off 2-3 degrees and you are done
with that map point. If you back off timing but still hear more matches
lighting up, reach that big red pushbutton (the one saying "emergency stop")
as fast as you can but chances are it is already too late.

When I mapped my engine we were up to three in the booth, the person driving
the dyno, myself on the laptop and the third person with his hand on the red
button. We had one expensive piezo sensor connected to an oscilloscope
(useless), one standard Bosch knock sensor connected to a Nagra tape
recorder with me on the headphones and the copper pipe for the dyno
operator; the only words allowed were "OK" to move to the next rpm site and
"Stop" to back off immediately.

We spent about 5 minutes on every row on the map including maximum load,
releasing the brake in small steps to stabilize the engine at every rpm
sites on the current row; it took me about 10-12 seconds per site to reach
the premise of knock and back off two degrees or so; we let the engine idle
at 3000 rpm for 2-3 minutes between rows to allow the turbo to cool down
from bright yellow to dark red; the ECU was taking care of the mixture
details using its own NTK sensor, precisely tracking the target lambda set
in the appropriate map (we did a rough fuel map before using a very
conservative ignition map).

The whole process took several hours and take my word for it, it takes a lot
of nerves to play with the knock limit at high load for 5 minutes flat at a
time with the whole exhaust header and turbo yellow-white, with up to 1100C
on the temp dial toward the end of the rpm range. The ECU was connected to
all the usual engine sensors and we had additional sensors for almost
everything, pre and post intercooler temp, per cylinder exhaust temp, oil
pressure, fuel pressure, oil and water temp in and out of the engine, oil,
water and fuel flow, manifold pressure, crankcase pressure and another WB
O2 - nearly 20 extra sensors  - with computer controlled normal range, alarm
and emergency stop values for every of them (the dyno would shut down
automatically if one of the value went out of the allowed range) - I
describe all this just to emphasize you can't really do a proper ignition
mapping job on a chassis dyno, much less on the road, and that you need more
than a plug sensor, no matter how good, to do it right.

Also I don't want to discourage anyone, but the only ECUs I know that are
capable of using those pressure sensors for closed loop ignition are using a
dedicated DSP chip just to process the sensor's signal, and they _are_ the
ECUs, I mean they already knows where the crank is and when they fired the
cylinder they are monitoring. I believe the signal is extremely noisy and
somewhat hard to interpret properly in real time. The software has so many
adjustable parameters in the knock strategy section it makes me wonder how
long it takes to adapt it effectively to a particular engine. Note that I'm
not against progress (I believe I was the first to post references to the
NTK/NGK spark plug sensors here a while ago) it is also well known that Saab
and others pioneered innovative ways of sensing knock but for most of us,
the old way is still by far the best IMHO.

--Axel


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