High NOx Problem SOLVED (86-89 Accord, others) -- Smog Test

My 1987 Honda Accord LX (carb, auto) was failing the california smog
test for high NOx emissions (a "gross polluter"). After an exhausting
self-education, I finally solved the problem. Here's the digest:

California tests smog on a dyno at 15 mph and 25 mph, with readings for
5 gases. My readings were:

15mph 1960rpm 12.2% CO2 3.3% O2 16ppm HC 0.00% CO 1355
NOx
25mph 2030rpm 12.2% CO2 3.2% O2 15ppm HC 0.00% CO 1188
NOx

These results say a lot. It failed for the NOx (about 6 times higher
than average). The CO2 is very low (should be around 15-17%). The O2
is very high (should be less than 1%). The HC is pretty low (average
is around 30 and 20). CO is very low (average is about 0.10%).
There's too much air in the mixture; mixture too lean.

First, look at CO2 to see how well the engine is running. CO2 is the
biproduct of proper combustion. Ideally it should be 15-17%. The very
low CO2 says that the engine is not running well. It's losing 20-40%
of its combustion efficiency (far below optimal performance). Given
that the CO is low (CO is a product of incomplete combustion), and that
the HC is low (HC/hydrocarbons are uncombusted fuel), it says that the
engine is running lean. Not enough fuel or too much air in the mix.
(If CO and HC were high, it'd be too rich. Either one will cause low
CO2; low efficiency.)

The fuel jetted to the carb is not subject to adjustment (unless you
mess with reboring the jets). The mixture is adjusted by two
vacuum-operated valves (air control valve A & B) which leak air into
the manifold to lean the mixture. Feedback Control Solenoid Valve and
Frequency Solenoid Valve A, B & C are controlled by the "computer", but
all the adjusting is done by comparative vacuum pressure, not by
computer.

First, I tested the O2 sensor. Its function is to tell the computer if
the exhaust gas has too much oxygen, which causes the computer to
reduce the air leaked in, causing the mixture to become richer. The
voltage at the sensor lead varies between 0 and 1 volt; representing
max lean to max rich. Normal operation (I read) has it hovering
between 0 and 0.5 volt, on a cycle of about two seconds. Mine was
staying at 0 volts except during acceleration or fluttering the
throttle. Replacing the O2 sensor made no change. (Wasted $60) My O2
sensor was saying the same thing the smog failure test said.

Too much air in the mix, so I went looking for vacuum leaks. I took
off the air filter assembly (easy) and inspected all the hoses. I
replaced a few that were hardened and that helped a little. I
inspected the vacuum diagram (available at autozone.com--very helpful
free repair manuals online), and indentified all the valves that are
connected to manifold. (I knew the carb gasket was not leaking.) Then
I applied vacuum to each diaphragm to check that they held it--all were
good except for the carb vent bowl diaphragm.

The carb vent valve is located on the front right corner of the carb
(looking from bumper), at the top, and is apparently of very bad
design, with very important consequences of failure. It sits a couple
inches from the carb bowl, and the rubber diaphragm is constantly
subject to fuel vapors and some fuel splash. I've found zero
documentation on the valve, but after much study, I can tell you its
function is to suck the fuel vapor off the bowl at the instant you turn
the car off, and then to seal off the bowl. Whenever the car is
running, the valve is supposed to be activated (open), with constant
suction on line 8. I removed and opened the valve and found that the
rubber in mine was transformed into a tar-like goo. The one I saw at
the junk yard had the same problem as mine. Thoroughly leaking. I
suspect many or most of these diaphragms fail. (This car has extremely
low mileage--less than 50k) The design blunder is made worse by the
fact that the carb rebuilt kits do not include this susceptible piece
of rubber. I was unable to locate this crucial part from anyone (I
didn't dare ask the dealer price).

The failure of this valve causes serious problems:
1) it creates a vacuum leak. Worsening the lean condition.
2) it causes the bowl vent valve to stay closed so the bowl is put
under vacuum. This fights against proper jetting in the carb,
strangling fuel delivery. Causes severe lean.
3) It exposes the whole circuit upstream (line 8) to
rubber-disintegrating vapors.

Aside from failing the smog test, I was having drivability problems.
Forcing the bowl vent valve open did a lot to solve them. (I tried RTV
and gasket-maker to fix the diaphragm but it only held for hours and
soon melted to goo.) The fuel starvation caused stuttering on
acceleration and severe hesitation from a stop. (And I had no idea how
much more power this little car has) The vacuum at the carb bowl was
preventing the jets from sucking in the proper amount of gas. Before,
I had to flutter the pedal (squirting gas) to get going from a stop.
Fixing the leak helped dramatically. But I was still failing smog from
too much NOx.

NOx are caused by excess heat and pressure during combustion. At
excess pressure-temperatures, the oxygen will combine with nitrogen,
forming oxides of nitrogen (NOx; NO1, NO2, etc). EGR problems are a
leading cause of high nox. EGR reduces cylinder temperature by
displacing some incoming oxygen with recirculated (non-combustible;
inert) exhaust gases.

I tested the egr valve by applying vacuum to it at idle and verifying
that the car stumbled hard or dies--the valve worked fine. I also
testing the vacuum applied to it by reading the vacuum on revving the
engine to 3000rpm--it only needs to get up to about 6 in Hg, and mine
was fine.

The final problem was that the timing was too advanced. It was 21
degrees BTDC and should be 15 degrees. Timing that's too advanced
means that the mixture is igniting while the piston is still
compressing, and this causes high pressure and temperature (high NOx)
as the expanding gas is compressed. Retarding the timing to 15 degrees
caused the nox to drop down to sub normal, and I PASSED easily.

Final readings:
15mph 1940rpm 13.8 CO2 1.0% O2 26ppm HC 0.49% CO 168ppm NOx
25mph 2010rpm 13.8 CO2 0.9% O2 29ppm HC 0.49% CO 243ppm NOx

NOx is now about average (passing easily), but CO is five times higher
than average and HC about 50% higher than average at 25 mph. My
exhaust is within limits, and CO2 is better, but still far below
optimum.

For reference, I have old readings from 1994 (when the car was 7 years
old and had only about 20k miles). At that time, the tests didn't
measure NOx, nor did they put the car under load; they just ran at 1000
rpm and 2500 rpm. Here are the like-new readings:

1150rpm 16.2% CO2 0.0% O2 2ppm HC 0.00% CO
2500rpm 17.3% CO2 0.0% O2 0ppm HC 0.00% CO

Now that's CLEAN! So I'm far from optimal. Since O2 and HC are high
now, this means that there's uncombusted fuel and air going out the
exhaust. That suggests that the timing is too retarded, leaving some
air and fuel unignited. And confirming this, in adjusting the
distributor before, I had to swing it almost all the way up.

On further inspection, I found that there's a slight but very effective
vacuum on line 25 (secondary vacuum advance). When the car is hot,
there shouldn't be any vac on line 25. The result is that the
distributor is excessively advanced at idle when it's adjusted to 15
degrees, which means that as the manifold vacuum decreases (wider
throttle) the advance decreases far below what it should be and timing
becomes retarded. The wider the throttle, the more over-retarded the
timing. As suspected, when I disconnect and plug both vacuum lines,
the timing is a few degrees AFTER TDC. This retarded base timing
reduces NOx, but it robs performance.

I'm still tracking down this problem. When I find out why there's vac
on line 25 and fix it, I'm expecting 20% or more improvement in
performance. The car already runs almost like new.

 

Good work, Greg! It sure reminds me why I hate feedback carburetors. Sure,
EFI drives me crazy from time to time but not like that.

BTW, although I don't think you wasted $60 on the O2 sensor - the original
was probably getting pretty wimpy by now - the indications you had suggested
it was responding to the leanness, rather than causing it. If the O2 sensor
output were that sluggish and low (2 seconds is about 1/10th the normal
response rate) the result should drive the mixture richer. But if you had
the $60 to put toward the cause I would have recommended replacing the
sensor on GPs.

Mike

 

wow! that's by far the most detailed post i've ever seen on this group!
good work!

 

Great post..

I'll reread it later for covering the basics is so important and you covered
it well; I can defatilly learn something from this.


--
Stephen W. Hansen
ASE Certified Master Automobile Technician
ASE Automobile Advanced Engine Performance
ASE Undercar Specialist

http://autorepair.about.com/cs/troubleshooting/l/bl_obd_main.htm
http://www.troublecodes.net/technical/