What Your Cockpit Gauges Won't Tell You
By David Lane
A respected 3rd gen RX-7 driver took his well-modified car to a race
track. He usually ran it on race gas for track work, but none was available that day. He
"cracked an apex seal" with the car running strongly at 14 psi and no audible
knock. He had ordered an Air/Fuel ratio meter (Halmeter--it was on back-order), and
lamented that its presence in the car at the track would have alerted him to a lean
condition--which would have averted the disaster. Something didn't sound right about that,
so I set out to discover if it was true.
It was more than just an academic question for me. My own car has an
after-market turbo. If some hidden gremlin can kill an engine without leaving a trace, and
without showing itself on any of the usual cockpit gauges, I want to know about it. The
search for the answer involved the sometimes complex relationships between hardware
(gauges), software (fuel), and dufus-ware (that would be the drivers). This essay proceeds
in that order.
Before jumping into it, I need to thank the many people who have
contributed their opinions, guesses, equations and personal experiences. I was also
fortunate to be able to question Corky Bell, Peter Farrell, and John Pizzuto on these
issues. Larry Mizerka's posts covered some of the academic aspects of the subject, and he
has been an excellent source of information. In presenting my findings, I need to
emphasize that some of it is still speculation, bumping into the edges of generally
The challenge is to put what I found into some practical setting
that is actually useable by someone glancing at the car's gauges while flying down a back
road or going for a personal best lap on a race track. Also (as I remind people so often)
I have no scientific background, so I must be able to place what I have learned into some
sort of logical construct, intuitive truth, or practical application before it makes sense
to me. I admire and respect people who recognize truth in numbers and equations.
Unfortunately for me (and my frustrated past math teachers) I am not able to join them.
This post will take the form of an essay. It runs about nine printed
pages. Nevertheless, I hope it makes for interesting reading. There should be some
surprises here for the more sophisticated reader, but I have made a point of including
enough basics so that those new to the subject won't get lost. Confused maybe, but not
lost. Welcome to the club.
All fuel injected RX-7s (and most modern cars of any kind) use a
sensor in the exhaust pipe to help the car's computer adjust how much fuel to inject.
Volvo introduced it in 1976 in conjunction with a three-way catalytic converter under the
name of "Lambda sond." Usually, the computer only consults the sensor
("closed loop" mode) under cruise conditions when it is looking for maximum
efficiency. The sensor--called an oxygen (O2) sensor--puts out a voltage when the oxygen
content of the exhaust gasses falls below the norm for the atmosphere. The voltage range
is from 0 to 1 volt.
Here is how the oxygen sensor functions. Oxygen in the air is
consumed when fuel burns, so increasing the amount of fuel to a given amount of air (a
richer mixture) will deplete a greater part of the available oxygen. The O2 sensor in the
exhaust pipe responds to this by putting out more voltage. Engineers look for 14.7 parts
air to one part fuel (14.7:1) as an ideal mixture for cruising. The O2 sensors in our
cars, therefore, are designed to be most sensitive in that range (14.1:1 to 15.1:1). The
14.7:1 ratio is variously referred to as "Lamda 1" or
"stoichiometric." Roughly 85% of the voltage output range of an O2 sensor occurs
within this limited range of A/F ratios. (This information comes from eyeballing a graph
supplied by the Halmeter people.)
The sensitivity of the O2 sensor in this range is good news for
cruising down the highway because it enables the system to do a great job of producing
maximum mileage and minimum emissions. It is bad news for us performance types because we
need a much greater percentage of fuel in the A/F mixture to extract maximum safe
performance out of our cars. For this reason, our computers ignore the readings from the
O2 sensors (shift to "open loop" mode) and refer instead to pre-programmed
"maps" for fuel delivery when we put our collective feet to the floor.
During acceleration, you need an air/fuel ratio somewhere around
12.5:1. (This came from the Halmeter lit, and may not be exactly right for rotaries.)
Unfortunately, the graph shows that a typical oxygen sensor's voltage variation is very
limited in that range. A secondary flaw is that the sensors on our cars are sensitive to
heat. They don't put out anything meaningful until exhaust temperatures reach 360C (680F),
and anything over about 900C (1650F) becomes problematic. This is interesting since a
modified turbo rotary engine can routinely see exhaust gas temperatures in excess of 900
degrees Celsius. In other words, just when we (dufus-ware) are most interested in finely
tuned information, the O2 sensor is shrugging it's shoulders and saying, "Weeelll, I
think we are possibly dealing with maybe around this amount of oxygen, but it's kind of
hot in here and I wasn't really calibrated for that, so I wouldn't want you to quote
Some manufacturers provide electrically heated O2 sensors to bring
them up to operating temperature sooner. Performance-minded drivers have noted that heated
sensors are much less sensitive to leaded fuel (high octane racing gas). However, above
about 600 degrees Celsius, a heated O2 sensor has the same weaknesses as any other.
NOTE #1: List members have told me that they think the O2 sensors
are heat-sensitive within their operating range; that the reading at a specific air/fuel
ratio will vary with exhaust gas temperatures. I have not come across information to
confirm or discredit this, so I merely pass it on.
NOTE #2: I have heard people state that our cars are in closed loop
mode during idle. This is unlikely to me, since exhaust gas temps are low at that time,
and I doubt that the O2 sensor would be reliably within its operating range.
NOTE #3: Stock RX-7 oxygen sensors can be replaced with less
expensive units. I have heard from Tri-Point that the stock sensors are particularly
consistent over time and temperature variations, and they recommend them when working with
after-market engine management systems, some of which stay in closed loop more of the
It is possible to monitor the output from the O2 sensor using a volt
meter between the sensor and ground. Even people with non-fuel injected engines can
install an O2 sensor in the tail pipe and use the readings for tuning purposes. If memory
serves, most people are looking for about .82 volts under maximum acceleration. Since the
O2 sensor is relatively insensitive in that part of its range, the reading is best used in
conjunction with other indicators to find that perfect mix. A more elegant choice is to
buy an Air/Fuel meter.
The first problem in talking about A/F meters is that there are two
classes of instruments out there. Laboratory grade A/F meters are costly, and not meant
for permanent installation. They are accurate in all ranges, and not bothered by changes
in exhaust gas temperatures. Thus, their sensors can be temporarily placed in the tail
pipe. These instruments have digital read-outs, and are the ONLY reliable way to do fine
tuning on an engine if you are looking for absolute information about high performance A/F
Inexpensive Air/Fuel meters express the output voltage of the car's
O2 sensor as an A/F ratio displayed with LEDs. These meters come in several shapes and
sizes. The primary flaw in their usefulness is that they cannot be more accurate than the
information coming from the O2 sensor. However, they are small enough to mount permanently
on your dashboard, bright enough to glance at when driving assertively, and cheap enough
to be a worthy addition if you start messing with your car. They respond very quickly,
alerting you if your latest "improvement" causes lean running, or if something
in the fuel supply system malfunctions.
What the inexpensive A/F meters won't do is to give you a high
resolution, accurate reading of a performance oriented A/F ratio. The range of most of
these meters is about 16:1 on the lean side, to about 12:1 on the rich side, and most of
the LEDs are unmarked. Two factors become obvious here: First, if you are looking for an
A/F ratio near 12.5:1 an A/F meter of this type will be close to the end of its range
(where the information coming from the O2 sensor is least sensitive). Second, if the
"rich" side of your A/F meter only contains five or ten LEDs above
stoichiometric, you won't have enough resolution to see very small changes. While these
limitations seem to damn inexpensive A/F meters to near uselessness, that is not the case.
The information they give you, when combined with other observations and experience with
the car, can be very valuable--especially if you have some manual control of your fuel
delivery system via the typical add-ons (additional injectors, boost dependent fuel
pressure regulators, fuel computers, etc.) associated with after-market turbos, or with
upgrades for stock systems. The resolution on the meter is consistent with the quality of
information coming to it from the O2 sensor. So, even though the absolute readings may not
be reliable, seeing if the reading stays constant as revs and boost build is very
valuable. As we will see later, the value of this has a lot to do with which generation of
RX-7 you are driving.
There are several excellent sites on the web about gasoline, so I
will limit my comments to octane and its effect on knock. One of the reasons the original
question (less octane equals leaner running?) was so hard to answer is that you are
unlikely to find two fuels that differ only in octane--especially when you are dealing
with racing gas. The notion several people had was that higher octane fuel gives off more
energy. Thus, if (for example) a teaspoon of higher energy fuel explodes, it will burn
more oxygen. This will result in less oxygen in the exhaust. The O2 sensor will develop
more voltage and the A/F meter will give a richer reading. It certainly seems logical.
The problem is that octane is simply a measure of a fuel's
resistance to knock (more on knock later). There is no implication I could find that by
changing octane alone you would also make a fuel release more or less energy. So, if
everything else was left alone, and you put a bottle of octane enhancer (like 104+) into a
tank of 93 octane fuel, your A/F meter would not change its reading.
In the real world, high octane racing fuels are denser and pack more
energy in each "teaspoon." A lab grade A/F meter will display the difference as
a richer reading. Thus, there are at least two advantages to racing fuel. You have access
to more energy at the same fuel flow, and the extra octane will allow you to run at higher
boost levels. An off-the-cuff comment by someone in the know was that you need to raise
the octane rating by three points to accommodate one additional psi of boost. This echoed
another comment--that a 3rd gen running at 14 psi on pump gas could get as high as 18
psi on the highest octane race fuel. Don't try this at home, kids.
The Question: Would our friend have been able to see a lean
condition on his Halmeter if he ran his "race gas tuned" car on pump gas? The
answer I got was most likely not. A lab grade instrument would have shown it (this was
actually observed by Peter Farrell), but stock O2 sensors are pretty numb in that range of
richness. This, combined with the grossness of the LED readouts, makes it highly doubtful
that a dash-mounted A/F meter would have flickered any differently than normal--much less
shown the kind of difference which would have been interpreted as an "alert."
NOTE #4: I came across an interesting factoid from the
"Reference Library" section of www.lubrizol.com which stated that cars need
additional octane as they age due to the build-up of deposits in the combustion chamber.
These deposits take up space, which effectively raises compression. This explains the
knocking I have observed on a number of aging cars I have owned. All responded positively
to higher octane fuel.
Exhaust Gas Temperature Gauges
The next question, of course, is whether or not an Exhaust Gas
Temperature gauge would have helped. Many swear by them because they get an
"absolute" reading of temperature. On the positive side, EGT gauges are not
subject to the non-linearities of an A/F meter. If you can repeat the same scenario, you
should get comparative readings. So, for instance, if you note the EGT reading after doing
a full throttle run from 3000 rpm to red line at 10 psi, then do the same thing at 12
you should be able to see the difference on an EGT gauge. The same can be said for
altering your fuel mix, installing a bigger intercooler, and maybe even changing your
timing (retarding timing results in higher exhaust temps). Further, there are known
parameters out there for exhaust temperatures with rotary engines. Mazdatrix notes in its
catalog that full race engines run between 900 and 954 Celsius (1650-1750F). They also say
that they observed a '89 fuel-injected pro SCCA car that was happiest at 773 Celsius
(1425F). Because of the wide range of "best" exhaust temperatures, anyone who
assumes his or her engine is happy based solely on the number appearing on an EGT gauge is
taking a risk.
This brings us to the "down side" of EGT gauges. The
temperature reading is influenced by the location of the probe: usually on the manifold,
but sometimes aft of the turbos or even further aft than that. A second problem is the
response speed of the unit. A race car, running "full out" on a track, has
plenty of time to develop a stable exhaust temperature. Running on the street, most of us
can't tell whether the gauge is registering the real temperature, or if it was just on its
way up there when we had to let off the gas to keep from ramming the nice person in the
SUV who pulled out in front of us. My point is that for street use, you have to get used
to what the EGT gauge is doing, and be aware of differences when you change something on
your car. A third area of concern is that the EGT gauge (like the A/F meter) will tell you
when something has changed, but neither will tell you exactly what it might be.
NOTE #5: After an autocross run, I checked the "peak-hold"
feature on my EGT gauge, and it was only reading about 775 Celsius. Normally on the street
I see about 825C after spirited driving. The water temperature gauge showed that the car
had heated up by almost 10 degrees (F) during the run, so it is likely that the EGT gauge
did not have time to come up to temperature. Had I leaned my A/F mixture based solely on
what the EGT gauge was telling me I would have been taking a risk.
The Question: Would an EGT gauge have saved our friends engine?
Again, probably not. In real life our track driver (who also used the car on the street
with pump gas) would have been accustomed to those typical readings. Maybe the readings
would have been a little higher when he went to the track, but as long as the gauge was
reading in a reasonable range, he would have had no absolute way of interpreting the added
heat as something which might have damaged the engine--unless he heard knocking--which
brings us to:
Pre-Ignition and Knock
Detonation; Knock; Ping; Pre-ignition. You hear these terms
mentioned all the time, so we might as well straighten them out. Let's get pre-ignition
out of the way first. Nothing mysterious about it. The A/F mixture (intake charge)
explodes before the spark plug fires. You would figure the intake charge would have to get
pretty hot to do that, and you would be right. The pressure from a high compression engine
is enough to generate that kind of heat. (In fact, diesel engines are designed to fire on
the heat from compression alone.) Higher octane fuel is the antidote, so in general, a
higher compression engine will need higher octane fuel. Cramming more intake charge into
the combustion chamber has the same effect as raising compression, so in general, the
higher your boost, the higher the octane requirement to avoid pre-ignition. Finally,
premature inflagration (I just made that up) comes more easily if the intake
charge is hot when it enters the engine. This is why larger intercoolers add a margin of
safety in forced induction engines--at least until you turn up the boost.
Another cause of pre-ignition is a hot spot in the engine. Maybe
some of those carbon deposits are glowing red hot. Maybe the spark plug itself is hot
enough to ignite the mixture before firing. This is almost certainly the case if you have
ever experienced a car that kept trying to run after you turned the key off.
The more tricky term is "knock." Although
most of us prefer to talk about "detonation," it turns out that
"knock" is the correct term as used in automotive texts. "Detonation"
is actually slang, and "ping" is not a well defined term at
all. That having been said, I will stick with the term "detonation" for this
Detonation differs from pre-ignition in that it occurs AFTER the
mixture starts to burn. Normal burning involves a flame front--a relatively slow,
controlled explosion--which marches along in a calculated fashion. As you would expect,
normal burning raises the pressure in the combustion chamber. Sometimes this is enough to
get the last bit of intake charge (called the "end gas") so excited it explodes
before it is supposed to. It is a very hot explosion, on the order of ten times the heat
of controlled combustion.
But there is more to it than that. If you graph the amount of
pressure in a combustion chamber during normal burning, it shows a relatively smooth
event. The occurrence of detonation shows up as a sharp spike on the graph--a sudden shock
wave if you will, with pressures on the order of several thousand psi. The duration and
strength of the explosion is too fast to contribute to the rotational output of the
engine. Like a slap in the face, the full impact must be absorbed within the combustion
chamber itself. Damage is most likely to occur at the weakest points--namely the apex
seals. Piston engines designed for high stress situations can have the piston rings
further away from the crown of the piston. The only choice for a rotary owner is very
expensive apex seals--but even then, there is no such thing as a detonation proof engine.
The shock of repeated detonation will eventually weaken anything it can, and the heat
generated will take care of the rest.
The question: How would this show up on standard gauges?
This is where we get into speculation. I don't know of anyone who
has purposefully run a rotary engine to destruction through prolonged knock with the
intention of seeing what the gauges read during the process. I assume Mazda has
experimented with something similar over the years, but I did not run into any data.
In practice it hardly maters. It all happens so quickly that any
hint of knock must be accompanied by getting off the gas. Yet even this is not simple.
What happens if your exhaust is too loud to hear the knock? And what happens if, say, only
a very small amount of the "end gas" detonates? Would you see the results on a
gauge? Could you hear it? Could it damage your engine if it was allowed to continue? How
about the specter of pre-ignition combining with detonation? Answers to most of these
questions would include so many qualifying statements ("...it's possible that maybe
under certain circumstances...") as to be of little use, but we can still deal with
knock in a direct manner.
Third generation turbo RX-7s and second generation T-IIs come with
knock sensors integrated into the electronics. Knock sensors use a microphone--usually on
the rotor housing or intermediate housing. The mike feeds electronics which are tuned to
recognize knock from the engine. Once identified as knock, the computer intervenes by
retarding the ignition timing. Why does this work?
Gasoline engines are usually set to fire the spark plugs before the
combustion chamber reaches its smallest size (maximum compression). On a piston engine
maximum compression is when the piston is at the top of the compression cycle. The same
happens on a rotary relative to the position of the rotor in the chamber. In other words,
the mechanical compression cycle is not complete at the time the plug fires. Thus, during
combustion, the total pressure in the chamber is a combination of the remaining part of
the mechanical compression cycle plus the pressure from expansion caused by the burning
fuel/air mixture. If you delay the firing of the spark plug, more of the mechanical
compression cycle will have passed at the time the intake charge is lit, so the overall
amount of pressure (and heat) in the combustion chamber reduces. This reduction in
pressure should be enough to ease the tendency for the end gasses to detonate. The more
you retard the spark, the more relief you get from detonation.
Those with after-market turbo kits can add a knock sensor. J&S
makes one that intercepts the firing signal going to the leading plugs and delays it in
proportion to any knock that is sensed. The unit is very sophisticated, and can identify
which rotor face is associated with the detonation. It then retards the spark to that face
only. If more than one face is involved each face is treated independently. Owners of
highly modified factory turbo cars can also benefit from such a device since the range and
capabilities of the stock knock sensor may not be enough to fully protect an engine that
is exceeding factory output.
Just as air/fuel meters fall into two major classifications, so do
RX-7 people. And this fact alone is a major contributor to difficulties when discussing
the value of the gauges and meters which are the subject of this document.
Owners modifying 1st and 2nd generation non-turbo cars are often
working with fairly gross devices for enriching fuel. They might have an adjustment or two
on the fuel pressure regulator, a few knobs and buttons on a controller for additional
injectors, and maybe even some sliders on a gadget that modifies the computer input from
the air flow meter. The engine management computers on these cars have no idea what manner
of gizmos are being bolted on, and while the fuel injected models may be able to sense
greater air flow to the engine, they will either run out of fuel trying to keep up or go
into fail-safe mode. Intrepid power junkies that we are, we immediately try to disable
anything that gets in our way, and we hope to extract as much power as possible without
running into detonation.
Owners of non-fuel injected cars can mess with carburetion--changing
the type and size of the carbs themselves, and also the jetting.
For this group of 1st and 2nd gen owners, the information given by
an inexpensive A/F meter and an EGT gauge can make a huge difference. The resolution of
the instrument is not all that special, but neither is the ability to make very fine
adjustments to the system. We realize that we are playing a dangerous game with engine
life, so we generally try not to get too near the theoretical limits. This "head
room" is our only safeguard to make up for the inherent slop in our ability to
control critical engine functions. Without a full fledged after-market engine management
computer (Haltech, Electromotive, Motec) it is the best we can do. Those of us with T-IIs
have a little more to work with, but bringing a stock T-II to 3rd gen levels of power (and
more) requires many of the same kinds of compromises and risks. Again, headroom is the
best safety solution.
Third gen drivers have an entirely different perspective. The
complex engine management computers and stock knock sensors on these cars make it possible
for them to run safely without much headroom--thus the much higher power output. If
additional power is desired, improvements to hardware are necessary. Unfortunately the
stock fuel maps can only accommodate so much, after which upgrades to the fuel management
system are necessary. These upgrades must stay within the already close tolerances of the
stock computer. Nothing short of a lab grade A/F meter will do the job. Maximum effort
cars run within a very narrow band of safety, so tiny changes in critical systems,
unlikely to be displayed by dashboard meters, can easily account for the difference
between a happy engine and a dead one.
Does that mean that a dash-mounted A/F gauge is useless for a 3rd
gen? It depends on how far you are going with it. Certainly, people doing their own
experiments with intake and exhaust are far better off with an inexpensive A/F meter than
with nothing. It will tell you in a general way if you are exceeding the stock system's
ability to provide enough fuel, and it will be quick enough to indicate whether those
boost spikes you may be seeing are accompanied by a lean mixture. EGT gauges are similarly
useful for reasons already discussed. With both, you may be able to ascertain whether your
after-market chip is keeping up with the latest eight-inch diameter extractor exhaust tip
you bought--the one that has tunable back pressure because it incorporates a modified
Jet-Ski drive unit which sucks the gasses out of the exhaust at a rate synchronized with
the car's engine.
However, if you are trying to push the envelope with one of these
engines, you are going to be working in pretty dangerous territory and, as they say,
without a net. You will need all the help you can get.
It seems clear that a Halmeter and probably an EGT gauge would not
have "sounded enough of an alarm" to barge into our friend's consciousness and
cause him to sense danger. Possibly when our friend couldn't get race gas, it was not
enough to cause anything obvious, but it effectively removed what little headroom he had.
The car was just running too close to the edge. There is a reasonable chance the exhaust
note was too loud for him to hear an occasional "tick tick" under load which,
while not a full-fledged popcorn sound, is still associated with knock. Even more likely,
the knock was not severe enough to be audible under the best conditions. This scary
thought has been confirmed by the number of times people see the indicators light up on
their knock sensors without any other indication that something is being stressed. Maybe
the car had been driven hard previously, and it just picked that particular day to give up
the ghost. We will never know for sure.
I was hoping we would get a report when the engine was torn down,
but the car has been sold and the new owner is putting a Mazda rebuilt engine in it, so
the old engine will be out of sight when dismantled.
The only after-market device I know of which might have alerted him
to an impending problem is a knock sensor. It is very easy to watch the display activate
on pump gas, but fall quiet on race gas. It is equally easy to see the display come alive
at one boost setting, but fall blank again if you back off, or if you add fuel to your
mix. The best news is that you rarely, if ever, hear detonation in the process--and even
then it should fall silent after a single ping. Yes, it is possible to have a little bit
of knock which is not going to reach your ears, but is going to do its work on your
While we might tend to spend time pondering the meaning of a subtle
difference in an A/F meter or a EGT gauge, human nature is to respond quickly to the
character of a knock sensor that is unexpectedly dialing in ten degrees of spark retard.
Driver optimism ("It's probably nothing important."), in one form or another, is
certainly one of the leading causes of engine failures. For those of us looking to push
the envelope in any generation RX-7, investing in an A/F meter, EGT gauge, and Knock
Sensor, is money well spent--especially when taken as a percentage of the cost of the
overall project. We could also talk about water temperature gauges.....but that's for