Posted to Technical Discussion Forum on 2/28/2010
Mechanical Misfire Diagnosis - Let's Talk
Some of you will remember my previous post alluding to
mechanical misfires. This post is an offshoot of that post
with a little different twist. If you don't like long posts,
then you can go ahead and close this one now. I'm not sure,
but this could possibly be the longest post in iATN history.
As I started typing, I just felt compelled to include all
the pertinent data in one place so as to support my point as
well as possible. I can assure you I have not given you even
half of it.
My goal in this post is two-fold. The first approach is to
present a case that there are misfires out there caused by
mechanical operation of the engine that cannot be easily
diagnosed. I will go so far as to say that you will NOT
diagnose (or prove) them using normal non-intrusive
diagnostic procedures. They require teardown sometimes in
order to even confirm their presence. Some can be confirmed
with the advent of some more stringent failure criteria for
the current mechanical testing. The advent of pressure
transducers will make some of this easier, but is not for
the unlearned. The ultimate goal here is to elicit some
evidence from those who may have fine tuned their procedures
that can help eliminate common shortfalls in current
mechanical testing methods, by providing new insight,
methods, or tolerances for testing.
My second goal is to elicit some responses on new potential
uses of these higher end sensors and how they can be used to
"prove" where the problem exists, if that is possible. Or to
shine the light on common misuses of "normal" diagnostic
processes in isolating mechanical misfires.
For those of you who refuse to believe that they do exist,
that is your prerogative. If you have run into one of these
and you have developed a way to prove their source, then by
all means chime in and help us all learn how to do it. If
you haven't run into one of these, and your only response is
that they do not exist, then please save the bandwidth and
do not reply. I won't negate the potential that difference
in tooling will affect the results of testing procedures as
well. Please respond in kind with any useful information.
If you had told me two years ago that it took longer than 30
minutes to isolate and prove a dead miss, I would have told
you that you were an idiot (well, not TOLD you, but I would
have thought it anyway) and didn't know what you were doing.
But since we have seen 3 of these in 2 years, my thoughts
have changed. Now when I see those posts or tech-mail
regarding diagnosing a misfire, I have a lot more respect or
at least empathy for some of them.
The type of misfire that I am referring to is NOT electrical
in nature. It is NOT the intermittent bump or jerk on the
highway or under load. The type of miss that I am talking
about is the one that misses in the bay. They will miss
mostly at idle and may, or may not diminish or disappear (or
appear to) at higher rpm. The miss will be nearly consistent
in nature and may be called a "dead" miss. The affected
cylinder may have zero power contribution or low power
contribution. Isolating this misfire to a specific cylinder
is not hard. Either misfire monitors or cylinder kill,
preferably both, can be used to isolate the exact cylinder
causing the concern.
I will state that while I can and do use a scope to help
diagnose this type of problem, secondary analysis is not my
strong suit. I use it mainly to identify a cylinder and
potential electrical problems, and then resort to mechanical
tests. I have begun using pressure transducers, but my
diagnostic ability on analyzing the waves is very new and
extremely wanting. I was hoping to be able to use them to
verify what is causing the problem, or at least prove a
mechanical issue to warrant teardown. I also have never used
a gas analyzer. Mark Warren has written about them recently,
and I am interested in them, but have zero experience or
availability of their use.
So let's get started in diagnosing this misfire at idle. We
have verified which cylinder it is. We pull the plug and
move it to another cylinder, as well as the wire. The
misfire stays the same on the same cylinder.
You install a vacuum gauge on the engine. The hose should be
centrally located and be as short as possible, with no
restriction in the hose, such as a cone on the end of the
hose. The cone is a restriction that can dampen any
oscillations that might appear. The vacuum gauge is rock
solid. It might float slightly/slowly if rpm is not steady,
but it does not pulse at all. You snap the throttle and
watch the gauge drop to very low, then very high on
deceleration, then settle back to where it was before and no
pulsing is present. You rev the engine to about 2500 rpm for
a minute or so, then bring it back to idle. The vacuum gauge
They might have a distributor or a coil pack. You perform a
cranking compression test and compression is good, within
10-15 psi on all cylinders (that you checked anyway). I
never check all the cylinders, just the bad one and a couple
of good ones for comparison. I like to choose the one next
to the cylinder that is bad, and choose another easier one.
This is performed with a mechanical gauge that has the check
valve in the quick coupler, not at the plug hole. The test
is performed with a closed throttle, not an open one. I
always thought that it was a waste of time as I am not
looking for a specific number, but a comparison between good
and bad. Naturally the peak should be over 100 psi and all
cylinders should be close. I know this is not "book"
procedure. I am giving you my "real life" procedure. I
normally see the 150 -180psi range.
You can perform a relative compression test using an amp
probe, or just monitoring battery voltage. Just remember
that the pattern will be backwards, and which direction you
current is going depending on which test you are using. This
may vary depending on your scope settings.
You perform a running compression test. With the engine at
idle, using the same mechanical gauge configuration as
before, you burp the gauge and watch the needle bounce and
build. You count the number of bounces until it quits moving
and record the peak pressure as well as comparing individual
bounces. You burp the gauge and do this several times while
maintaining the same rpm on the engine or letting the
computer do it naturally. You compare a known good cylinder
to the bad one. I have been told that as little as 5 psi
difference in this test reveals a valve sealing problem, but
my anecdotal evidence reveals that this is not enough to
warrant a misfire.
Then you install your leak down tester. Mine is home made.
It has two gauges with quick couplers. The secondary gauge
adapter has a restrictor with a small hole that allows only
a small amount of air volume to pass. You find TDC
compression so that the engine won't turn and apply air to
the cylinder. One trick I have found here is that if you
just put 100 psi on the cylinder, it can push marginal
valves closed and they test good. So what I do is to start
at less than 10 psi. I actually apply the 10 psi to the
cylinder in order to find TDC. With the small amount of air
applied, I can turn the engine over with a ratchet on the
alternator pulley, or some other way of turning the belt.
When you move the belt, as the piston begins to reach
compression stroke, you will notice that each turn of the
belt results in a slight increase in the second gauge
needle. As long as the piston is travelling upward you will
get a slight tick in the second gauge. The first time that
you don't get the tick when turning the belt means that you
have reached TDC and you can now pressurize the cylinder.
Apply the air slowly, rising from 10 psi up slowly to the
100 psi mark. This mark just makes it easy to do the
percentages. You can take it as high as you want, but I
normally only go to 100. I figure if the prior compression
readings are good, then any more is not beneficial. Normal
leakdown that I see is 10-15%. Compare known good to the bad
cylinder. Anybody know a spec for acceptable variation here?
You already have good compression. It is not unusual for me
to see as much as 20% leak down on good cylinders.
You test for vacuum leaks using carburetor cleaner, brake
cleaner, propane, or whatever you have available. You can
even use a smoke machine. None are found. You can just
listen to the engine to see any immediate response, and you
can also watch your 02 sensors to see if there is any
response at all. No change in the engines operation is not
conclusive. No change in the 02 sensor is conclusive, to me.
Just make sure whatever type of fuel you are using is
combustible. Brake clean is not recommended, I'm told, due
to the toxic gas it emits when it burns. But since I am
being honest, I use it more than anything else, as it is
always within easy reach.
You also block the PCV hose and check for internal vacuum
leaks. By removing the vacuum hose or the PCV valve and
blocking it off, you then access the PCV fresh air intake,
and block it off as well. Now take your reading at the
dipstick tube. If you take your reading at one of the PCV
ports, just make sure you pull the dipstick and put a finger
over it or block it as well to prevent any potential of
missing any vacuum leaks. If pressure does not build
slightly, then you are probably not blocked off completely.
Either that or you have an internal vacuum leak that is
amazingly exactly equal to the engines blowby. Any vacuum
detected during this test reveals an internal vacuum leak.
Scoping the coil current confirms proper firing time, and
swapping parts makes no difference. You scope injector
current to confirm proper amperage and control.
Fuel trims are off, but not excessively. They show a
commanding a little rich, which you attribute to the
misfire. You reset fuel trims, or artificially force them
near zero, just to make sure you don't have a false input,
and the miss remains. You perform an injector flow test. You
install your fuel pressure gauge and inspect for proper
pressure and residual pressure. You bleed all of the air out
of your gauge so that you can get an accurate reading then
using either the scan tool or a separate injector driver
tool perform an individual flow test. All of the injector
drops are the same within 1 psi or less difference.
Another fuel test that I have never used, but was mentioned
previously was to shut off the fuel pump and use propane to
run the engine. With a large tank and the proper adapters,
you can run the engine on this alternate fuel delivery
method to see if the misfire is gone or not. If so, then you
must have missed something previously to my way of thinking.
EGR? If the car has not been driven, and only idled, reach
back and feel that EGR valve. If you burn your fingers, you
found your problem. If not then it might still be leaking.
There are several ways to check this, from vacuum testing,
liquid testing, etc. Worst case scenario, remove it and
fabricate a block off gasket.
So, we have eliminated spark, fuel delivery, compression,
leak down, and vacuum leaks from the equation. What is left?
We must have a mechanical problem. We just haven't proven it
yet. My recommendation at this point is to pull the valve
covers and loosen the rockers to see if the miss goes away.
It has been my experiences that valves that are too tight
will not seal properly allowing the cylinder to miss. But
with this evidence you can see that it is difficult to prove
that existence by external means. Then you say "but it can't
be a tight valve (or valve sealing problem) causing the
problem because my leakdown test showed good" I am here to
say that is not true all of the time.
The first one we had was a Dodge truck, 180k miles on the
odometer. All of the above tests were good, but this thing
continued to miss at idle. Not intermittent. It was all of
the time. After my tech's run the basic tests and cannot
find anything, they come to me. I normally don't work on
vehicles unless it is a problem child or a case study. I put
him on something else and I proceeded to run test after test
with what I knew at the time. Everything was good. This was
before my intro to pressure transducers, so I don't have any
fancy waves for you. What I can tell you is that the problem
was mechanical. Because when I work on these types of
things, there is no time clock, I can spend as much or as
little time as necessary to come to a conclusion. My shop
continues to run with minimal input from me, so I can play
when we get these.
After running and rerunning numerous tests with no
definitive conclusions, I finally caught it. I had 60%
leakdown on that particular cylinder. We had a valve sealing
problem that passed every running and cranking compression
test I threw at it. It even passed the leak down test every
time. I think it was like the 4th or 5th time that I finally
caught it. The customer chose a new engine. We installed a
new engine with the old ignition and fuel delivery system.
Ran great. The only thing replaced was mechanical.
Here is a picture of the valve sealing surface of that
[File47615 ¦ Worn Seats
The second one that I have was discussed in the previous
It is a 2004 Impala with a 3.4 litre. The cranking
compression and leakdown tests were all good. While the
running compression test on this one provided a clue to a
mechanical problem, I have never seen a spec as to how much
of a difference can cause a problem. I have been told
(afterwards) that 5 psi difference (with rpm maintained
within 50 rpm) is enough to prove a valve sealing problem. I
have seen this exact issue and did not have any types of
misfires. This one had 10 psi difference. I would be lying
to you if I told you I checked it after the fix.
On this one, the running compression on the bad cylinder had
pulses to 75 psi but the needle would settle at 60. On a
good cylinder the gauge would pulse to 75 also but the
needle settled at 70 psi. A different good cylinder had
gauge pulses to 85 psi and settled at 70. Our decision was
that the only problem was running compression of 10 psi
difference (using a mechanical gauge with a check valve) was
enough to warrant a valve job. After the valve job, the car
ran the same. Only then did we have the idea to loosen the
rockers. After loosening the intake rocker less than ¼
of a turn, the misfire went away. If this had been on the
rear or on another type of engine, loosening the rockers
while running could have been impossible. The lifter was
stuck. A new lifter fixed this problem. The intake valve was
effectively too tight when the engine was running.
The vacuum gauge never pulsed on this one. Here are some
pictures of the vacuum pattern on that engine. This engine
had a consistent miss at idle that appeared to quit with
just a 100 rpm increase or less. Remember, this thing passed
compression and leakdown testing.
I also have running compression tests with transducers on
this one. I will leave that for another post.
After fixing that Impala with new lifters, later on we had
the Blazer come in that was mentioned in the previous post.
After my tech ran some preliminary testing, I had them
perform a running compression test. When my tech reported
the 10 psi difference in running compression, I was
ecstatic. I finally had one that I could analyze to death,
using my newfound knowledge (however small it was) of
pressure transducers and the 5 psi running compression spec.
Before the last couple of years, I had not been doing
running compression tests because I have never needed it.
Using cranking compression and leak down tests were all that
was necessary to prove a problem.
Because of my schedule, I did not get to follow this thing
to a final conclusion on that day, but I did get a chance to
rerun the running compression test myself while collecting
some data and carefully documenting the problem. My running
compression test showed a difference of only 5 psi. It was
less than my tech's results, but the cylinder was still
missing. Since the engine miraculously fixed itself before I
could perform multiple duplicates of the same types of
compression and leak down tests, I don't have an exact
answer on that one. I have enough proof to say it was
mechanical, and I suspect a stuck lifter, same as the
Impala, just no proof. The problem with the 5 psi running
compression spec is that after repairs, I was able to verify
that same 5 psi less on the running compression test, but it
was on a different cylinder, and it was running perfectly.
Shoots that one down doesn't it.
The point of my post is to lend credence of the existence of
these type of misfires, as well as to give a centralized
place where beginners can go to glean some more info. I will
start another post about the pressure transducers and
focusing on the one particular vehicle that I know what was
wrong with. I only hope others will add some useful info to
this post to help us all see additional potentials in
testing in the future.
I am sure I missed something, so you guys can help me out
What I am hoping is that someone has run one of these down
completely, and can help all of us if/when we get one of
these in our bays.
Scott from Missouri