Erratic idle with codes 22, 34, 35 and 45

This is where we really need a scanner to know what's going on. There's two things that will tell us. The first is to look at where the computer has set the idle air control motor to. Chrysler lists that as a "step" number from "0" to "256". GM uses the same part but they might list its setting as a percentage. It will be in some form that you can identify as how much the valve is letting air in by bypassing the throttle blade. If it's in "steps", step 32 is typical for a properly-running engine. With a single-cylinder misfire on a V-8 engine, around step 50 is what you might find. While watching that number, observe what happens to it in relation to actual idle speed. If idle speed goes up, but the step number goes down, the computer sees idle speed is too high and is trying to bring it down, but without success, That could point to a vacuum leak. If you see idle speed go up at the same time the step number goes up, the computer is requesting that higher idle speed in response to something. Most commonly that has to do with temperature sensors, either intake air or coolant temperature.

The second thing is most scanners have a "record" feature that lets you record a few seconds of data when a problem occurs, then you can play it back slowly, later, to see what changed. Because that data passes through the scanner's memory, the recording actually begins a couple of seconds before you pressed the switch. Things to look for here include a loss of signal from a crankshaft or camshaft position sensor, idle steps that went up as the car slowed down, but engine speed didn't respond, possibly indicating an air passage plugged with carbon, or something else cutting out when the problem occurred. Cam and crank sensors often fail too briefly to be detected and set a fault code.

The MAP sensor is another good sensor reading to look at. That measures intake manifold vacuum which is an indicator of load on the engine. If vacuum drops unexpectedly, suspect vacuum leaks, or less-commonly, a failing sensor.

All car brands except Chrysler use a mass air flow sensor too on most engines. Those measure the weight of the incoming air flow. With this sensor, it's critical that there's no leaks in the fresh air tube between that sensor and the throttle body. Loose hose clamps and cracks in the tube can cause intermittent problems when the engine rocks back and forth. Similar to a vacuum leak, that will show up as the air volume has gone down because some of it is sneaking into the engine without going through the sensor. When that air isn't included in the measurement, fuel to go with that unmeasured air doesn't get included in the fuel metering calculations, resulting in a lean condition, and probably stalling.

Sensor defects become less suspect when you can restart the engine right away, as you described. A different problem that only GM had in their early years of front-wheel-drive cars was a lock-up torque converter that got stuck and failed to unlock. That would be the same as if you had a manual transmission and failed to push the clutch pedal when the car was coming to a stop. This was so common that a lot of mechanics simply unplugged the electrical connector at the transmission so the lock-up feature never occurred, and therefore, it couldn't stick engaged. You'd loose two or three miles per gallon, but the stalling no longer occurred. When that did occur, the lock-up clutch released when the engine stalled, so it could be restarted right away. Whatever caused that problem has been addressed, because it hasn't been a problem for a long time.

Another elusive cause of stalling is a leaking EGR valve. Many engines don't have one, but when they do, they can't be allowed to open at idle and low speeds as it will cause stalling or misfires and a rough-running engine. Often this can be identified by disconnecting and plugging the vacuum hose at the valve assembly so it never opens. That could identify a valve that's sticking open intermittently. With the vacuum hose disconnected, the valve won't open at highway speed, so it can't get stuck.

It's more common for a chip of carbon to get caught in the valve and allow exhaust gas to keep flowing into the engine. You won't notice that at high speeds, but again, at low speeds is when the stalling or rough-running shows up. That chip of carbon is typically going to break free and go into the engine, and therefore it's a one-time thing, or it's going to be permanently stuck and the valve will remain partially-open all the time. Neither of these really fits what you're experiencing.

The last thing I can think of is a lot of people seem to solve intermittent problems on GM vehicles by cleaning or repairing ground wires for the Engine Computer. All Chrysler Engine Computers have four ground wires. Two are "signal" grounds and two are "power" grounds. Power grounds are for things that draw a lot of current in pulses, not steadily. That mainly includes injectors, ignition coils, and relays and solenoids. "Signal" grounds are for sensitive circuits such as most engine sensors. All wires have some resistance, and in the case of those high-current items, when that current pulses on, it results in dropping a small voltage across the resistance in the wire. That might be as much as maybe a tenth of a volt which would have absolutely no effect on those injectors and ignition coils, but to drop that tenth of a volt on a sensor's ground wire would have a major effect on how the computer interprets its readings. To prevent that interaction is why separate ground wires are used for sensors. Chrysler uses a redundant or second ground wire for each one, just in case one develops a little corrosion or other problem.

Here's the diagrams for the plugs for your Engine Computer. I found five ground wires here, but they aren't necessarily two for each circuit. For example, they use a separate wire for the oxygen sensor, (green arrow in the second diagram), probably for the same reason as a slight voltage interference from other circuits would adversely affect the O2 sensor's readings. "Reference" or "reference low" are designations for ground wires.

Do not waste your time trying to use an ohm meter to check these wires. All it takes is one tiny strand still intact to get a good reading that way. Instead, the voltage on these wires must be measured while the engine is running. That's when current flows through them and is when any excessive or undesirable resistance shows up as a "voltage drop". Put the voltmeter on its lowest range. Sometimes the voltage readings will jump around a few hundredths of a volt due to stray magnetic interference, but typically those readings will be obviously too high when there's a problem in one of the wires.

One more thing fairly common on only GM vehicles is magnetic interference from the generator. This problem started in 1987 when they redesigned them and went from the world's second-best generator design to by far the worst. They develop huge voltage spikes that can destroy the internal diodes and voltage regulator, and interfere with computer sensor signals. The battery is the key component in damping and absorbing those spikes, but they lose their ability to do that as they age and the lead flakes off the plates. That battery is the reason for repeat generator failures. Four to six replacements in the life of the vehicle is not uncommon. To reduce that number, always replace the battery at the same time as you have to replace the generator, unless that battery is less than about two years old. It will work fine in an '86 or older model.

This generator problem becomes much worse if one of the six internal diodes has failed. That causes "ripple" voltage to be real high and greatly increases the likelihood of developing voltage spikes. Most professional charging system testers measure ripple voltage automatically on a relative bar chart from "low" to "high". In addition, with one bad diode, the most current the generator will develop under the "full-load" test is exactly one-third of its rated value. Thirty amps from the common 90-amp generator isn't enough to meet the needs of the electrical system under all conditions.

To identify engine running problems caused by the generator, start with a good, fully-charged battery, unplug the small connector on the side / back of the housing, then go on a short test-drive. If the generator is causing interference with sensor signals, the symptoms will clear up when it is disabled.

35 - Idle air control sensor circuit fault
64 - Oxygen sensor right side lean exhaust indicated

Here's what we have listed for those code numbers: Code 35 is to be expected. The computer saw the break in the wiring when the idle speed motor was unplugged. As I mentioned earlier, these two-digit fault codes don't get very specific. They're only supposed to tell you which circuit you need to diagnose. They don't actually include a diagnosis.

When the engine starts, you'll notice engine speed goes real high, as in 1500 rpm, for a few seconds. That's called "idle flare-up". That lasts a lot longer on Fords and on newer vehicles, but on older GM and Chrysler products, it comes back down to normal idle speed within a few seconds. To accomplish that, the computer retracts the valve in the idle speed motor after the engine is stopped. Some models do that the instant the engine is stopped, and others do that when the ignition switch is turned on, just before cranking. It sounds like yours does that when the engine is stopped. Once the valve is retracted to allow additional air to bypass the throttle blade, it can't extend to close that passage when you have it unplugged. Once the computer sees the wiring is disconnected, it will stop trying to adjust idle speed, but it can resume trying when the motor is plugged back in.

It's hard to know where to look without having scanner data to look at. This is exactly the problem cars would have had if it wasn't for the diagnostic fault codes to get us started. It's much worse on '96 and newer models. You could spend weeks checking each circuit if there weren't fault codes to tell you where to look.

The only other thing I can suggest is to remove the idle speed motor, or get a good used one from a salvage yard, and plug it in while it's laying out where you can watch it. Don't run the engine with the valve out as this will create a huge vacuum leak, but when a helper turns the ignition switch on, you should see that pintle valve retract. It should extend fully when the ignition switch is turned off.

There is nothing in the system to tell the computer where the idle speed motor is set, or adjusted to, other than it knows engine speed. Referring back to Chrysler's numbering system that I'm familiar with, lets say the computer sets the motor at step 50 in preparation for the next engine start. Once that is done, if you come along and tug on that valve fairly hard, you will be able to pull it out or squeeze it back in. The physical step position will have changed, but as far as the computer is concerned, it still assumes it is at step 50. Besides you moving the valve manually, the armature could be sluggish, or it might have even vibrated to some other step. Regardless of how it happened, it is out of sync with the computer. To prevent that, on every one of these I've played with, the computer runs that valve all the way out to step "0", then pulses it some more in case it didn't have time to get there. At that point it knows that valve has to be at step "0". From there it retracts it to, in my story, step "50", then sits there and waits until you want to start the engine. What you should see when you turn the ignition switch off is that valve will extend a lot, then retract part way. You can usually hear that if you have your head under the hood when your helper stops the engine. The computer knows now the valve is at step "0", and it knows it has opened it up enough to provide the idle flare-up. If you don't see that happen, that would explain the no flare-up you observed, but we'll have to figure out why.

The idle relearn has been a real common issue for Chrysler products for a long time. It needs to be done any time memory power is lost to the Engine Computer, meaning the battery or the computer was disconnected. Fuel trim data and sensor personalities are relearned right away just like on any other brand, but relearning "minimum throttle" requires a specific set of conditions that let the computer know for sure your foot is off the accelerator pedal. Until that is done, idle speed will be too low and the engine may not start unless you hold the accelerator pedal down 1/4". It will also tend to stall at stop signs, and you won't get that idle flare-up at start-up. All that is required to meet the conditions for the relearn to take place is to drive at highway speed with the engine warmed up, then coast for at least seven seconds without touching the pedals. This goes back to those "idle steps" we've been talking about. When minimum throttle hasn't been relearned yet, the scanner will show "IAC steps" as "0".

When doing any service at the dealership that required disconnecting the battery, we were supposed to take a short test-drive to do that relearn, but it was also acceptable to simply inform the customer of the need to do so, especially if it was one of our regular customers or someone who easily understood the need to do so. There are some other manufacturers that have their own requirements for initiating the relearn of some things but those usually take place without you noticing or even knowing what those conditions are. I'm not aware of any such conditions for GM vehicles. Whenever I've replaced a battery in one, idle speed has always been normal right away.

I did keep seeing the references to the speed-density system on the online diagrams. That's the system Chrysler used where the MAP sensor is the biggest player in fuel metering calculations.

Disconnecting the battery in an attempt to clear something isn't the answer. That thought goes back to the mid '80s where GM had a real lot of trouble with their Engine Computers, and removing memory power had an effect on its performance later. Other than that, the whole idea is the computer learns the personalities of each sensor by comparing them and their operation to each other, then it constantly updates what it has in memory. The goal is to keep that memory alive, and to build on it, not erase it and start all over.

You might also try unplugging the mass air flow sensor to see what happens. The computer will detect that, set a diagnostic fault code, and turn on the Check Engine light, but one of the design considerations is when that happens, most computers will "inject" an approximate value, based on all the other sensor readings and engine performance criteria, and run on that. The idea is even though the engine might not run well, it will run to get you to a repair shop. In many cases the engine runs surprisingly well. The MAP sensor will take over as the main fuel metering calculation, as it is on Chrysler engines. I suspect the software modification you referred to involves telling the computer to ignore the missing mass air flow sensor.

It's impossible to tell. We don't like throwing random parts at a problem because that is the most expensive and least effective way to diagnose something, plus, each one adds another variable to the problem. It would be better to spend your money on a professional diagnosis, or at least on visiting a mechanic who will let you look over his shoulder when he connects a scanner. We don't know what the idle steps are doing in relation to engine speed, and we don't know what sensor signal voltages the computer is seeing and responding to.

Ford had a common problem in the early '90s with their coolant temperature sensors developing erratic readings. Normally temperature sensors have extremely low failure rates because there's just one component inside them. This Ford problem was a notable exception. I saw this on a student's Taurus. The idle speed kept surging up and down. He checked everything we could think of, but it wasn't until we hooked up the scanner that we saw the coolant temperature sensor's signal voltage bouncing all over the place. At first we logically suspected corroded connector terminals, but once we knew what to look at, we found the sensor's resistance bouncing around with an ohm meter. That was a very unusual type of failure, but in this case it was due to a design or manufacturing flaw since it affected almost all of them. The solution was to replace the rather inexpensive sensor, but that would have been almost impossible to find without the scanner.

Good observations, but unplugging the coolant temperature sensor is a common test we do to check if the cooling fan system is working. Doing so will be detected by the computer, and in response on most car models the computer turns on the radiator fan(s) in case the engine is running too hot. The fans will turn off a few seconds after the sensor is plugged back in, but a diagnostic fault code for that will stay in memory. There's different levels of defects as far as operation of the Check Engine light is concerned. Since some defects, such as a thermostat that doesn't let the engine get up to operating temperature, can have an adverse effect on emissions, most coolant sensor codes will turn on that warning light. On most car models the Check Engine light will turn off when the CTS is plugged in, or it will be latched on and will turn off the next time the engine is started.

Since the fan turned on when you unplugged the CTS, I wouldn't worry about it not turning on while driving. Even as far back as the early '80s, radiators are so efficient at dissipating heat, that some fans rarely need to run. Also, GM is pretty well-known to need their coolant to reach an uncomfortably-high temperature before the fan is switched on. That can be higher than 226 degrees.

I read quite often about our other experts recommending cleaning throttle blades to solve running problems, but I've never done that myself. I suspect there's better additives in gas today that makes that less common. That air passage around the throttle blade used to become pugged with carbon on Chrysler's Mitsubishi-built 3.0L engine, but I haven't seen that since the late '90s. That caused low idle speed and stalling at stop signs. I had well over 420,000 miles on my '88 Grand Caravan and only had to clean that passage once when the van was a couple of years old. That was a common service at the dealership up to the mid '90s, but no longer. I have a suspicion those better gas additives are also why cleaning throttle blades seems to be less common.

The good running for five minutes is another dandy observation. All of the engine sensors are in play the instant you start the engine, except for the oxygen sensor. That's called "open loop". It means the Engine Computer looks at all of those sensor readings to calculate the desired amount of fuel to go with the incoming air, then it assumes the fuel / air mixture is correct.

The computer first starts to look at the oxygen sensor readings when the coolant reaches a certain temperature. Oxygen sensors don't work until they reach about 600 degrees. Newer models have electric heaters inside them to get them up to working temperature much quicker, but on yours, it is assumed they'll be hot enough to be accurate when the coolant reaches that magic temperature. That's called "closed loop" when the O2 sensor's readings are added to the fuel metering calculations. The system could switch to closed loop in about five minutes after starting a warm engine.

In closed loop, the other sensors still are used to calculate fuel needs, but then the readings from the oxygen sensor are used to fine-tune fuel delivery. Typically, in extreme situations, the computer can add or subtract only up to ten percent from the fuel volume calculations. If more fuel is needed, you'd see that as a positive number on the scanner for "short-term fuel trim", (STFT). That number is constantly bouncing around and switching as you drive. You'll usually notice running problems if the computer can't adjust fuel beyond that ten percent. By the time it gets that bad, there are usually some other really obvious symptoms, so we don't spend a lot of time looking at fuel trim numbers.

When the computer sees it is constantly making the same fuel corrections over and over under the same driving conditions, it will move those short-term fuel trim numbers to the "long-term fuel trim", (LTFT) number. That is a value that starts out preprogrammed at the factory. By modifying that number and using that as a starting point, less adjustment has to be done in the STFT numbers while driving. The short-term numbers show what's happening right now. The long-term numbers show the amount of correction the computer needed to make over a longer time span. These numbers can only be viewed with a scanner. Normally we only look at them when artificially introducing lean or rich conditions to see how the oxygen sensor responds. A vacuum leak will make the system run lean.
Feeding propane into the intake air tube will create a rich condition.

To add to the confusion, oxygen sensors can't really report a rich condition. They only measure unburned oxygen in the exhaust. They don't measure unburned fuel. Once the system is in closed loop, the computer is constantly switching the fuel / air mixture between too rich and too lean about two times per second. While the exhaust has too much unburned oxygen, it is stored in the catalyst, then when there's too much unburned fuel, that mixes with that stored oxygen and is burned. (In '96 and newer models, a second O2 sensor right after the catalytic converter monitors what's coming out of the converter to be sure that exhaust has been cleaned up. On '95 and older models, it is just assumed that has taken place). The computer watches how long the exhaust gas is lean in relation to how long it isn't lean to figure out when it's rich.

Engine performance specialists look at those switching rates and fuel trim numbers to figure out if a running problem is caused by a fuel problem. It takes some rather extreme conditions before the computer will decide to set a fault code for "running lean too long", or some similar problem. A vacuum leak will do that, but so will a spark-related misfire. This gets confusing too, because with a misfire, the unburned oxygen is what gets detected by the oxygen sensor, but it's the unburned gas we smell at the tail pipe.

Since we're still looking for this surging or idle speed problem, also consider the effects of a stretched timing chain, especially at higher mileages. Many engine models have knock sensors that react to a timing chain slapping against the housing. That causes spark timing to be retarded momentarily and can look like surging. A more elusive problem can be caused by worn bushings around the shaft in a distributor. When the main spark timing sensor is in the distributor, that wobbling shaft will cause spark timing to bounce around. Spark timing isn't really advanced at higher engine speeds like it was on carbureted engines. Those had fly weights that flung out and turned the top half of a two-piece shaft. When the engine uses crankshaft and camshaft position sensors, you can't wait for a signal pulse, then have the computer fire an ignition coil a few milliseconds before that pulse occurred. Instead, they're set to start out with lots of timing advance, then the computer waits a calculated period of time before it fires a coil or injector. To advance the timing at higher engine speeds, it waits less time after the sensor's pulse to fire those circuits.

With a distributor, the computer just fires the ignition coil when it's told to do so, then it relies on the adjustment of that distributor to assume park is occurring at the right time. To see the effects of a wobbling shaft, you'd find that by watching the timing marks with a timing light. When the computer is in charge of adjusting spark timing, you'll see that as a number on the scanner. The slapping timing chain will be detected by the knock sensor and a corresponding change in timing will show up. With a distributor, a sloppy timing chain will cause the camshaft to be out-of-sync with the crankshaft, and that will also bounce around. The distributor is geared off the camshaft, so it will bounce around too.

My service information shows the common HEI distributor, but at some point they switched to three individual ignition coils, each firing two cylinders. I don't know which year it was that they switched systems. If you have a distributor, it will have a really large cap with four screws to push down and turn to remove it. Older engines had the ignition coil built into that cap, but others had a single coil mounted on the engine.

The early distributors had an ignition module inside the under the cap. This was a really nice, totally self-contained system. Service involved switching over the spark plug wires and one 12-volt feed wire when replacing the distributor. By around the mid '80s, they added another plug and more terminals on the ignition module. Its job was still to fire the ignition coil, but now that extra plug went to a computer inside the car. Unlike earlier versions that had fly weights on the upper half of the distributor shaft to advance ignition timing at higher speeds, with the computer, they started out with really advanced timing, then the computer calculated an amount of time delay between receiving a signal pulse from the pickup coil, to the time it commanded the ignition module to fire the coil. Timing was advanced by reducing the amount of delay time.

If you have three ignition coils, they'll be sitting side-by-side on top of the ignition module. To my knowledge, those systems didn't really have many problems, but timing advance is still handled the same way. You can't get a signal pulse from a sensor, then have the ignition module fire an ignition coil a few seconds before that pulse showed up. You have to get that pulse much too early, then build in a calculated wait-time before firing a coil or injector. With this system, since timing is controlled by a computer module, you can see that amount of advance on a scanner.

A sloppy timing chain will allow the camshaft to get out-of-sync with the crankshaft, and move around in relation to it. Engine performance specialists might use a graphing scope or scanner to watch the relationship between the signal pulses from the crankshaft position sensor and the camshaft position sensor. They won't remain steady if the timing chain is worn.

Sloppy timing chains can also slap against the housing and be picked up by the knock sensor. That will cause ignition timing to be retarded momentarily. You'll also see that on a scanner as the listed timing advance bouncing around. This could often be felt as an uncomfortable, or at least noticeable surging while driving.

As for the best course of action, I really want to see what is shown on a scanner. The dealership scanner was called the "Tech2". You can find brand new ones on eBay for close to $300.00, last time I looked. Another member bought one a few months ago, and while he assumed it was made in China, it seemed to work to his satisfaction. These originally sold for a few thousand dollars. I have a Chrysler DRB2 and the newer DRB3 for most of my vehicles. If you added a plug-in card, the DRB3 will do emissions-related stuff on all car brands sold in the U.S, but only on '96 and newer models. Because of that, a lot of independent shops bought them. As a complete kit, they cost just over $6200.00, but I'm on real good terms with the dealership I used to work at, so they got them for me for much less.

What you might want to look into is a Snapon Solus Edge on eBay. My friend has one, and I liked it, so I bought one for my newer truck where the DRB3 is obsolete. One problem with the Solus Edge is it costs around $1000.00 per year to keep it updated to current vehicles, and you can't skip any years. If you find one on eBay updated to say, ... 2012, that will have everything that was available for tests and features for your car. For a shop owner to buy it, he'd have to buy the 2013 update before he could buy the 2014 update, and so on. They only cost around $4000.00 new, not including European and Asian models, so it's a better deal to just buy a new model. That can work in your favor since the more out-of-date it is, the less value it will have to professionals. I've seen these selling for as little as $700.00. That's a real good deal if you can get enough use out of it.

I don't know what the Tech2 had for cables, but with the Solus Edge, you'll have to buy a second cable and the appropriate connector. The main one plugs into any '96 and newer model. They all use the same shaped plug under the dash. '95 and older GMs still had the plug under the steering column, but it's a rectangular design that is different from all other car brands. Those adapter plugs usually cost between five and ten bucks. Once you have that second cable, it's just a matter of buying the different adapters for any other car brand you want to work on.

Buying your own older scanner is by far the best way to go if you'll get enough use out of it, but you're still stuck with having to interpret the results. I haven't had a need to explore more than a fraction of what my scanners can do, but once you use one, you'll wonder why you waited so long to get one.

Autel is another brand that seems to be real popular. You can find various models at Harbor Freight Tools, and they're surprisingly inexpensive. You'd have to inquire as to what is involved and the cost to update them to the latest models every year.

If you can't justify the cost, or you're concerned about being able to make sense out of what is does and shows you, the most effective next step is to get a professional involved. I'd look for a smaller one-man shop like my friend's. He's always busy, but he is free to spend as much time as he chooses to on someone's problem, with no boss to answer to. The problem here is if they come up with a diagnosis, then you do what he suggests, there's no skin off his teeth if he's wrong. When you ask to have the problem repaired, the mechanic is obligated to find the solution, and you'll know it's solved when you pay the bill.

Be aware too a lot of mechanics who were familiar with an '87 model have retired or moved on, and many younger ones don't want to work on older cars. You might consider looking for a nearby community college with an Automotive program. We always had a few dozen people who would sit on a broken car until it fit what we were teaching because they understood the value of having live vehicles to work on. The kids are well-supervised and very responsible, but it can take a few weeks to get your car back. We charged ten dollars per hour for what the job was supposed to take, and we got parts at real good discounts, then marked them up ten percent to form a "breakage" fund in case we damaged something. This could be a very inexpensive alternative and you'll learn a much about the solution as the students do.