ECU Affect on Rebuilt Engines: Knowing the Basics Can Prevent Costly Comebacks
When a customer purchases a rebuilt engine they expect to get many miles from their investment. But there are many things which must be done right to assure that this happens. One area that is increasingly important in this regard pertains to the engine’s electronic control system.
The electronic module that controls the engine goes by many names, e.g., ECA, ECM, EEC, EFI, etc. All of these acronyms commonly refer to the vehicle’s computer. By true definition, this computer is actually a micro-controller (MC). Early MCs were eight-bit systems. Eight bits equal a byte, and a byte is the basic unit of measurement in microprocessor-based systems.
A byte has 256 binary combinations, and each combination can be an instruction code. It is this code which lets you manipulate data and perform mathematical functions. The code is stored in a Read-Only-Memory (ROM) chip and it is permanently burned in, although "flash" ROMs are becoming increasingly popular where you can change the code via a modem.
The instructions stored in the ROM consist of code arranged in, hopefully, a logical manner called a program. There are also look-up tables and sensor linearization data stored in the ROM for the particular engine for which the ROM was encoded. The module also contains driver circuits for the injectors, the ignition module, the idle control valve, the fuel pump and EGR valve.
The module includes an analog-to-digital converter for reading the various analog-type sensors and converting this data to digital form so the MC can understand it. Some modules also contain the hardware for the cruise control, automatic transmission and alternator control. Where this hardware is located varies considerably from vehicle to vehicle. For example, Ford likes to jam everything into one box where Chrysler uses a separate power driver module.
Due to the many variations, attempting to change an engine without the precise vehicle service manual, or a least the less expensive Electrical and Vacuum Manual, is foolhardy to say the least; these systems are just too complicated. The component locator is the key source of information for finding where these modules are hidden in the vehicle.
In my experience, Honda prints the best manual and Ford the worst. Honda explains exactly the input and output characteristics of each sensor where Ford goes through literally hundreds of pages of indexless programmed troubleshooting. If you are fairly sure the ECT sensor on a Ford is not working, you may never find what resistance this simple sensor should have at a certain temperature.
Besides the engine controller, there are similar size modules for the ABS, digital key entry, suspension, transmission, cruise control, lighting circuits, etc. None of these modules is identified in any understandable language. They are hidden under the hood, under the seat, under the dash, behind kick panels, in the trunk, and side panels in SUVs. You can spend hours just trying to locate the correct module.
The function of the engine control module (ECM) is to run programs just like your home computer. Like your home computer, ECMs can run more than one program. The program size in the ECM is generally very short as it is stored in ROM.
A lot of hype exists from both test equipment and vehicle manufacturers regarding the troubleshooting ease of even the latest OBD II and III systems. However, what is done compared to what could be done is still very minimal. In general, most vehicles are consumer based products and, as such, are very cost conscious. Don’t expect miracles from any scanner or diagnostic tool, the information provided is usually very limited.
The power supply
Before looking at the programs, all MCs are subject to errors caused by glitches. These cause the MC to jump any number of positions in the program counter and the next instruction could be some weird data stored in the RAM section instead of the ROM, causing very strange engine behavior.
Most vehicle programs are written in loops to be self correcting, compared to your home computer where you would crash, generating the infamous reboot solution. But even at that, periodic glitches can drive the troubleshooter insane. How can the MC tell you what is wrong via a scanner if it can’t properly run its own diagnostic program? The very first step is ensuring that the MC is getting the correct voltage.
Most MCs have a five-volt integrated circuit power regulator that is fired from the vehicle’s battery. The vehicle’s battery, in turn, is charged and has its operating voltage controlled by the alternator. Without a good battery/alternator combination, you could spend hours and big bucks changing every module in the vehicle and get nowhere.
This five-volt supply also fires all of the sensors, so if it is off, every sensor would be, too. You must first check the vehicle supply voltage and the MC operating voltage to be sure these lines are clean and at the correct voltage. For this you will need a good oscilloscope and an accurate digital voltmeter. At high loads, where you also expect the MC to operate properly, you should measure no more than 300 millivolts of ripple across the battery terminals. The actual voltage across the battery can vary from about 13.2 to 15.7 volts depending on testing temperature.
The five-volt line should have no more than 20 millivolts of ripple and be within 0.1 volt of the five volt mean. You measure directly across the module’s connector to prevent ground loop voltages in the readings. For one example, when cranking, if the battery is weak, it may only supply 7.5 volts to the MC, but the MC requires at least nine volts at its input connector so that the IC regulator can output five volts.
With the low input, the computer cannot operate properly and the car may never start. You could spend hours checking the fuel pump, spark plugs, coil, etc., when your only problem could be a low starting battery voltage.
Another major cause of MC malfunction is electro-magnetic interference (EMI). The easiest rule to follow is to place every wiring harness back with the proper clamps exactly the way you found it. If you don’t, you may be in for a lot of headaches. For example, routing a spark plug cable near the MAP sensor can inject pulses into this sensitive output lead that will drive the MC crazy, fouling up both ignition timing and injector dwell time. No engine can run right with poor MAP data. Worse yet is getting a vehicle that has aftermarket accessories added or was tampered with by an inept mechanic.
Ground is not ground
While all modules eventually return to ground, how they return to ground makes a huge difference. Sensors have their own ground system that feeds directly back to the MC. If someone adds a stereo power amplifier and finds his ground at one of the sensor terminals, this can play havoc with the MC. In a five-volt digital system, logic levels are precisely defined; less than 0.8 volts is a logical zero, greater than 3.3 volts is a logical one – only a difference of 2.5 volts. If just 2.5 volts of noise is getting into the MC on one of the data lines, it will run the wrong instruction!
The MC is also powered via the fuse box, the ignition switch, some with a relay, and all with a multitude of harness connectors. All of these connection points must be checked from the battery terminals to the final module including the key ground points. The vehicle’s Electrical and Vacuum Manual lists the vehicle’s power distribution key hot and grounding points and is a good reference to follow.
The wonderful world of vehicle electronics has yielded many horror stories. For example, the car cranks but won’t start. After much checking, the injector manifold pressure was found to be low. Checking the MC fuel pump circuit showed an output, and the fuel pump relay was sending 12 volts to the pump. It was determined that the pump was bad. This required removal of the gas tank to change it, a rather expensive labor operation.
However, the car still would not start. The problem was that the fuel pump ground under the driver’s seat had a loose bolt. If the mechanic would have checked the voltage to the pump, putting his ground lead at the pump and not just any handy ground, he would have discovered the problem immediately.
The first step is always making sure that both ground and hot are getting directly into the appliance. The second step is making sure this voltage has low ripple, is free of glitches, and is of the correct value. If the test lamp is out, you need a good meter and a scope. Corroded cable harness connectors, poor grounds, corroded fuse holders, and dirty switch contacts are major causes of MC related problems.
Using a poor rebuilt alternator without the proper avalanche diodes and transient protection, an undersized or weak battery, and/or corroded battery terminals can fry the MC beyond recognition. Alternators can kick out more than 100 volts that will make the inside of the MC module look like burnt toast.
The MC runs different programs stored in the ROM. These vary somewhat between different model years and makes, but are generally all the same.
Some cars, upon turning on the ignition switch to the run position, will turn on the check engine lamp for a second or two, then turn it off. On other cars, the check engine lamp will not go off until the engine is started. There is a simple self diagnostic test you can run. This test involves just checking the interconnections to the basic sensors and perhaps a quick RAM test.
RAM is used like a blackboard to store the MC’s computed data such as injector dwell time, ignition timing, EGR operation, fuel pump on, error codes, and on some MCs, the modified look-up tables to correct for sensor variations. The RAM is reset by disconnecting either the MC fuse or the battery.
There is only one engine coolant temperature (ECT) thermistor type sensor used, as there is only one of each of the other sensors. This boils down to the fact that the MC has to accept that the sensor readings are correct, i.e., if the engine temperature is 160°F, but the sensor says it’s 130°F, the MC accepts the 130°F as correct.
To state that there are sophisticated algorithms to correct this error is nothing short of a lie. Another factor is that the sensor may be good, but due to other circumstances it can provide the wrong information. If the ETC is in a filthy, plugged, corroded cooling system that has plenty of hot and cold spots, the ETC will feed the wrong information. Likewise the manifold absolute pressure (MAP) sensor can’t provide correct data if the engine has a major vacuum leak.
The zirconium oxygen sensor is the most prone to generating error codes when, in fact, this sensor is actually working fine. If one spark plug were of the wrong type, or not properly torqued into a cleaned spark plug hole permitting blow by, this gust of air can actually blow out the spark causing misfiring. This misfiring may only occur at a relatively low engine speed range.
I had one engine that only missed between 3,000 to 4,000 rpm; faster or slower than this speed and the engine ran fine. As there is only one or two oxygen sensors for the entire engine, the excess oxygen makes the MC think the engine is running lean, so this "dumb" sensor leaned out the entire engine! Individual cylinder monitoring has been talked about to prevent this, but for an eight cylinder engine, the electronics costs would be eight times as much.
What this means to the engine rebuilder is that if one or two spark plugs are not sealing properly, the engine will run lean and hot. Many of today’s engines are made paper thin, so hot spots can burn out a head gasket, warp a head or cause other damage. Guess who gets the blame?
Another false oxygen error code can be generated by a loose or corroded injector connection. In this case the oxygen sensor thinks the engine is lean, so it enriches the fuel causing the other cylinders to carbon up, besides frying the catalytic converter. You could get either a spark plug or an injector miss that the very slow response time oxygen sensor will not pick up. In such a case, the engine will last longer, but probably not long enough to reach expiration of the warranty period.
Most vacuum leaks that cause erroneous MAP readings will not generate an error code. This can lead to false ignition timing that can cause detonation, another engine damaging parameter. The positive feedback exhaust (PFE) sensor used in Ford cars can be completely melted without generating an error code. The EGR valve can likewise be plugged, causing a dangerous combustion temperature rise resulting in burnt out head gaskets, valves and even holes in pistons.
In such cases the standard explanation goes, "But the engine electronics are okay because I wasn’t getting any codes." It takes about a 10 Mbyte program to properly test the operation of the micro-controller. I can guarantee that you won’t find such a program burned in your ROM chips. At most, vehicle manufacturers are just providing a short form, quick diagnostic test. What error codes are generated may be very misleading.
Based on the fact that the original engine failure could have been very easily caused by a non-code system problem, I would recommend that all the inexpensive sensors be replaced and the more expensive ones be carefully tested for proper operation as part of an engine rebuild installation. Such sensors to replace would include the ECT, TPS, ACT and the oxygen sensor.
The MAP, PFE, high altitude sensors and the air mass sensor (AMS) should be well tested. Some engines use an anti-knock piezoelectric sensor that retards spark advance. With the proper compression ratio and octane fuel, this sensor will never have to work. But other programs intentionally advance the spark to where detonation occurs, then retard it for maximum fuel economy.
Improper octane fuel causing engine damaging detonation cannot be corrected by the anti-knock sensor. If the fuel pre-ignites on the compression stroke, no amount of spark retard is going to correct it. An engine damaged this way is very hard to prove. Worse yet, fuel octane ratings aren’t always reliable. Engine knock is also the result of a rich running engine with excessive carbon build-up.
Fuel-injected engines are really no different than their carburetor counterparts. The darn things still need a very rich mixture to start up. Here data from the air charge temperature (ACT) and the ECT sensors tell the MC to increase the injector dwell time. Air/fuel ratios of less than 12:1 may be encountered on real cold startup days. The throttle position sensor (TPS), basically a radio like volume control knob attached to the throttle valve, also inputs data to the MC. The harder you step on the pedal, the richer the mixture becomes.
Not all vehicles use a warm-up program as this is very similar in operation to the open-loop mode program to follow. The vehicles that use the warm-up program depend on open loop sensors, but have an additional function to increase engine speed when cold. The speed is then reduced back to normal as the engine warms.
The controlled element is the idle bypass valve (IBV) or idle control speed valves. As always, each manufacturer has its own acronym for the same functional device. Earlier vehicles used a motor controlled throttle valve stop that simply emulated the function of stepping on the gas pedal. Where the valve is normally pulse width modulated by the MC to control engine idle speed, this pulses a solenoid activated valve that lets air bypass the main throttle valve.
The IBV action is also controlled by such things as the field duty cycle of the alternator, the air conditioner compressor, a power steering sensor, and the neutral safety switch indicating whether the transmission is engaged or not. Ford electrically heated windshields also jack up the idle to about 1,500 rpm when engaged, if the outside air temperature is below 32° F.
Engine stalling problems with idle control can be caused by any of these idle control sensors failing. However, they are not that difficult to trace. Simply stepping on the gas pedal would correct any of these sensor failures. So, if you engage the transmission, for example, and that causes the engine to stall, but is corrected with a light touch on the gas pedal, you can expect a bad or misadjusted neutral safety switch. None of the idle control sensors will cause engine stalling due to misfiring. When these sensors go, the engine still runs smooth, but simply dies due to an overload.
Open-loop mode program
Another good way of looking at an EFI engine, is that the driver controls the amount of air that enters the engine. This is true with the exception of the idle control valve. But the driver could do that function as well, a little tedious perhaps, but possible. It is then the job of the MC to add the correct amount of fuel to maintain the stoichiometric air/fuel ratio of 14.7:1
With direct fuel injection, this will change to air/fuel ratios of 50:1 or more, but that is another subject. The MC accomplishes this feat by measuring the amount of air entering the engine and adding the correct amount of fuel. Most EFI engines have an injector manifold pressure of about 40 psi that must remain constant. So the only other variable is the length of time each injector remains open.
Regarding the injectors in open-loop mode, this pressure/time relationship is at best an assumption. You open the injector valve for so many milliseconds at a constant pressure and should therefore get so much fuel. If an injector is skipping beats due to a bad connection, is dirty, or is just worn out, it is not doing its job.
If one injector is starving its combustion chamber, the mixture for that cylinder would be lean, causing that cylinder to run extra hot. Likewise, a leaky injector can "carbon-up" a cylinder leading to piston ring lock causing the engine to burn oil. This oil is redirected via the PCV back into the intake manifold, normally upstream from the injectors. This leads to carbon build-up in the remaining injectors resulting in one cylinder running rich and the others running lean.
Running an injector leakdown test, i.e., pressure gauge on the fuel manifold, fuel pump engaged, but then turned off to build up pressure, then ignition off, lets you know there is a problem if the pressure reading falls off before the specified limit. But it does not tell you where this problem is located. It could be a minor leak in the fuel pressure regulator that is a spring-operated valve returning fuel back to the tank, a leak in the fuel lines, a weak check valve in the pump itself, or leaky injectors.
The point is that poor injectors can cause premature engine failure, as can an out-of-tolerance pressure regulator, or a weak fuel pump causing lean running conditions. While some systems use a pressure sensor to inform the MC that the pressure is off in order to make adjustments for the proper air/fuel ratio, most systems do not. And the argument can also be used, "what if this pressure sensor is out of tolerance?"
Getting back to the argument, assuming good driving habits and oil changes, something caused the original engine to fry in the first place. If these same components are tacked into your rebuilt engine, won’t the same thing happen again? Yes, except this time you will be liable.
To learn how much fuel to add, the MC needs to know both the volume and the density of the air entering the engine; air/fuel ratio is done by weight, not by volume. This is accomplished by either the direct or the indirect method. The direct method uses a MAP sensor or MAF sensor located somewhere between the throttle body and the air cleaner.
The MAF is essentially a pressure transducer where the pressure is somewhat related to the volume of the air passing over it due to the venturi effect. The MC takes this reading and normalizes it via the look-up tables stored in its ROM. The density of the air is measured by the ACT, a simple temperature sensor that is also fed into the MC, where, in conjunction with the MAF again, it is referred to look-up tables to kind of get an idea of how much air weight is entering the system.
Another variable is the barometric pressure that affects the oxygen level in the air. This can vary with both weather and altitude. Our key interest is learning how much oxygen is entering the engine by weight in order to determine the amount of fuel to add.
Yet another variable is the amount of exhaust gas that is recirculated. This is also indirectly measured by either a potentiometer mounted on the stem of the EGR, or a pressure sensor in the EGR feedback path. However, there’s nothing to tell if the EGR passages are plugged.
The direct method is not so direct after all. Sloppy installation or a crack in the air circuit plumbing can really screw up the MAF data. In the open loop mode, we are dealing with essentially theoretical values for both the "oxygen" and the fuel entering the engine. In some MC systems, a learn program is used to compare the look-up table values of the open-loop sensors when the engine is in closed-loop mode. This is an attempt to make some adjustments that are stored in RAM.
However, if the vehicle has voltage problems, these corrected values will be constantly erratic and really cause some strange engine behavior. Sometimes an erratic running engine can be simply corrected by disconnecting the battery for 10 minutes so the MC can relearn from scratch. It may take 10 miles of driving for the relearn cycle to complete.
In indirect open-loop mode systems, there is no MAF, but the engine speed, TPS, MAP, ACT, altitude sensor and, assumed EGR feedback, are used to kind of guess at how much air is entering the system. A lean or rich running engine can be caused by any of these sensors being out of tolerance, or the combined tolerance being added in such a way to lean out the engine.
A carboned engine can be cleaned, but a lean running engine has to be rebuilt. Fuel control was never that precise in automotive history; in aircraft, it was and still is the pilot’s responsibility to set the proper air/fuel ratio by using an air/fuel meter in larger aircraft, or the all important Exhaust Gas Temperature (EGT) gauge in smaller aircraft.
I often wondered why the EGT gauge, or at least a sensor, was not included in today’s MC systems. If the engine exhaust gets excessively hot, time to add more fuel. The older 500 cid engines kicking out maybe 0.3 hp per cid had plenty of reserve for poor air/fuel ratios. Today’s engines kicking out one to three hp per cid made of lightweight alloys have very little room for error; a lean mixture would soon fry them.
The major limitation is the oxygen sensor itself; it can only operate when it is almost red hot or at least, say the books, at 350° F. Some oxygen sensors have an electric heater element added to speed up the heating process, but this also manages to more than double the price of these sensors.
Those short city hops many drivers do never lets the engine get warm enough to reach close-loop mode. The oxygen sensor is keyed to set the air/fuel ratio to 14.7:1. Any driving condition that requires a greater air/fuel ratio, particularly acceleration, trailer pulling, climbing hills, etc., where the TPS demands more gas, switches the system back to open-loop mode.
City stop-and-go driving practically cancels out the benefits of closed-loop mode. The response time of the current generation of oxygen sensors is very slow, some as long as one pulse per second. This means that an exact 14.7:1 ratio is never met, however, the oxygen sensor is constantly hunting. The actual air/fuel ratio could vary between 13.7:1 and 15.7:1. Over a one second time interval, many combustion cycles can occur within this time span.
Another way of looking at it is that the engine is running lean 50% of the time and rich the other 50%. The MC is constantly looking for a change in output from the oxygen sensor. If this change ceases, it is back to open-loop mode. If the TPS output remains fairly constant, and the ECT says, hey buddy, you should be in closed-loop mode, but the MC does not see an output change for a predetermined time limit from the oxygen sensor, an error code is finally generated popping on the check engine lamp.
The zirconium oxide does wear off the oxygen sensor that slows the response time even further. This increases the time that the engine is running in the lean mode. An engine is very dynamic and it is these peak temperatures that the block and heads can’t pass quick enough that create hot spots accelerating engine damage. These hot spots are not measured, and you cannot depend on the coolant temperature gauge for these readings.
Work is being done to directly measure combustion chamber temperature, but this is a long way off. We still are dealing with a consumer item where the color of the car is more important than any fail-safe system. If antifreeze hits the oxygen sensor from a leaky head gasket, this sensor must be replaced.
Only one or two oxygen sensors are used that assume all of the spark plugs and injectors are working properly. Many a good oxygen sensor was replaced due to a fouled spark plug or erratic injector. If a spark plug fails, a rich mixture is sent to the oxygen sensor that in turn, leans out the rest of the engine. If an injector fails, a lean mixture is sent to the oxygen sensor that enriches the rest of the engine. A leaky injector acts just like a failed plug, so if an engine is running rich, it’s a guess as to which one is the culprit.
The OBD II system found on most 1996 and later vehicles uses a 36-tooth crank angle sensor where the MC tries to determine variations in angular velocity. By noting dips in velocity during the firing of each cylinder that it already knows, and generally over a two minute time interval, as other engine loads cause instantaneous velocity dips, it is supposed to record a misfiring cylinder and store this in the error code RAM. Short term misfires due to hitting a water puddle are eventually erased.
This system also adds a pressure sensor to the fuel tank; the draw of fuel creates a vacuum in the tank to assure the EPA that your fuel system does not have a leak – actually a sizable leak that can reduce tank vacuum. The most common problem is not tightening the gas cap, or having a bad gas cap seal that lowers the tank vacuum beyond the threshold point.
Extra oxygen sensors are added to the exhaust system that are used for monitoring the quality of the exhaust and do nothing for engine control. The bottom line is that the EPA is more interested in what the vehicle is kicking out into the atmosphere with the OBD II system then adding safeguards to prevent engine failure.
At least the OBD II system finally attempts to detect a cylinder misfire. However, this system has many shortcomings in that if two more cylinders are misfiring, say one with a failed injector the other with a misfiring plug, this is almost too much for a short MC program to detect. In my humble opinion, more emphasis should be placed on detecting correct engine operation that would result in cleaner emissions.
Two simple things could accomplish this. One would be adding a tube prior to the catalytic converter providing access to a calibrated air/fuel ratio meter. How can any system self-diagnose if it does not have calibrated references to work from? You can still have many undetectable errors in the OBD II system as the MC still has to go by the word of all the sensors.
The second item would be monitoring the exhaust gas temperature. This has been used in aircraft engines for years as a reliable means to tell if the engine is running too rich or too lean. This would be great back-up for the oxygen sensors. If the EGT drops below a predetermined level, the mixture is too rich; if it rises, it is too lean.
OBD II systems have a three-way check engine lamp: off, everything should be okay; on all the time, you have a minor problem; and if it is flashing, you have a major problem. OBD II scanners are more expensive. The very cheapest is about $200 ranging up to $3,000. However, if you could afford $40,000, you could get a computer interface that is suppose to tell you more about the system. Many now have flash ROM, so new programs can be downloaded. This has its good and bad points. The bad points are, whenever it is easy to change a program, programmers tend to get sloppy and the storage time of flash ROMs is limited, some say 10 years. A 10 year old car may simply loose all of its programs; maybe by then, there will be a cheap way to download them.
The final program the MC runs is the limp mode. If more than one sensor fails, or if the information available is not sufficient to control injector dwell or spark timing, the system goes into a kind of fixed timing mode barely sufficient to get the vehicle to a shop.
Electronic engine control is great; we are getting fuel economy and pure raw power from our engines undreamed of a few years ago. You can’t be persuaded by marketing schemes that these systems will tell you exactly what is wrong with them. On the contrary, very little accurate information is provided by these systems. Of far more importance to you, the engine rebuilder, is that certain types of failure in these systems are not detected. These can lead to premature engine failure that will inevitably be blamed on your workmanship.
Engine rebuilders have a major strike against them in that the engine they rebuild may be installed in a vehicle that ruined its engine in the first place by having an undetectable or undetermined fault in the electronic control system. Therefore, rebuilders have to practically demand that their engine installation also include a complete check up of both the cooling and electronic control system, including the ignition, fuel and EGR systems.
Rebuilders, for example, don’t want a reused, untested fuel injector causing a return. Also rebuilders should be aware that both custom hot-rod chips and complete aftermarket controllers are available where the user can write his own programs that may be detrimental to your rebuilt engine.