Car, Car, C-A-R,
Stick Your Head in a Jelly Jar

Regionalism. We used "mustard jar" in my neighborhood in the 60s.

We were all about alliteration in New York in the 60s.

We were so poor in Maine in the 70s that we only had the first line. Didn't even know there could be a second one.

Only said it when it was time to get out of the street to let a car go by.

Well, duh.

j/k :D

A whole line? Luxury! We only had the scream and the thump to alert us to get out of the way. That, and a mouthful of hot gravel.

BigV @ 2:11

We were so poor we couldn't even pay attention.

i was so poor i had to fart to have a (s)cent in my pocket!

Once the Car Chip arrives, it must be setup. After downloading the software, the Car Chip must be told what parameters to monitor.

The first is vehicle speed - a default. Next is Engine Speed. O2 Sensor voltage. And then Short term fuel trim. The first three parameters should be set to a 5 second interval.

Attach to the car (when still cold). Take a maybe 5 minute drive. Best to take a road where long steady conditions occur. So that multiple 5 second data from the O2 Sensor can be compared to engine speed and short term fuel trim.

First notice what an O2 Sensor voltage is when the vehicle is cold. Ideal voltages will probably be between 0.6 and 0.7 volts once the sensor is hot. Voltage will increase to something like 1 volt when the engine is dumping lots of fuel into the engine. When more fuel is being burned and more fuel is exhausted unburned (wasted energy and a lower gas mileage).

Fuel trim says how long each injector is on; how much fuel is being dumped into the engine.

That is what the O2 sensor does - adjust the amount of fuel to no longer do gross waste; what a carburetor did. Carburetors would dump massive amounts of fuel hoping the engine would burn as much as possible. Leaving a catalytic converter to burn what remains.

If too much fuel is detected in an exhaust, the computer assumes a defective catalytic converter. Not necessarily. The error code only says where to start looking. Some mechanics will not think through a problem. Automatically replace what they first suspect (ie blame the catalytic converter). Do not assume what is wrong before replacing anything.

I'm sorry tw (I really am, not being sarcastic,) but you misunderstood. I didn't get the Car Chip product after all, only a regular OBDII sensor. My goal was to be able to turn off my own check engine light, both now and in the future. I don't think the item I got (this one) can read those values for you.

Is that a hand held OBD2 display? Not the Car Chip Pro that reads and stores data from a car? And is then read via a computer's USB port? That is the only CarChip Pro I knew of. Did not know a handle display existed.

Yes, mine is not a CarChip Pro product at all. It's a different brand of OBD2 sensor, of which Amazon had at least a dozen. It doesn't connect to the computer, and I don't think it can be used while the car is running.

But at this point I'm really sure it's the catalytic converters. (I know you're not sure, but you take an inordinate amount of evidence to be sure, and I am sure enough to meet my threshold. :)) The really good news is that I successfully ordered compatible parts online myself, and plthijinx is completely. awesome. and is going to install them for me, with the help of a friend and a welding torch.

The awesomeness of plthijinx comes through x 2!

Thats awesome Clod! Way to step up plthijinx. :thumbsup:

BWAHAHAHAHAHHAAA! now THAT was funny!!!

it's nothing really. gonna be a PITA but certainly easier than when my friend and I changed out the transmission on my old truck.

<------- FAIL!

hate to admit it but i couldn't fix it. the cat's were the wrong part to start off with so we thought, ok well, lets just change the sensors. way in the fk easier said than done. one sensor was right in front and easy to get to but i'll be damned if i could bust it loose. i even put a torch on it twice and still couldn't get it loose. i gave in. said hell with it. i don't want to make it any worse than it already is. so we rode go karts instead.

ETA: pics
first pic, albeit a little fuzzy you can see the spark plug looking front sensor

second pic: torch

bummer
:(

But the go karts were awesome! And no trip wasted, I was gonna be here to pick up my stepkids today anyway, and I can return the unusable parts. It was worth a try.

I think those have a left-hand thread.




(Trouble maker)

The one thing you need is a left handed monkey wrench.
.
(accomplice)
.

Normally the first sensor (completely different from another oxygen sensors) is located where the exhaust manifold connects to the engine block. Is that sensor, instead, relocated to be one in your picture?

no, that was the first one. this engine, er exhaust system has the manifold and cat's as one piece so to change the cats you in turn have to remove the manifold. there was that front one there, another down stream and i couldn't find the fourth. (1 on each cat then the downstream one + the anonymous one)

i tried going both clockwise and counter. those suckers are on there. 9 years of cooking on hot ass exhaust. i was afraid i was gonna cause more damage than good so i quit

But is that pictured oxygen sensor the one normally located (on most cars) only six inches from the engine block? Meaning of the word 'first' is not clear.

Are there two (the picture only shows one) sensors - each one for three cylinders? Of course another two sensors are after catalytic converters. Those would be reporting a failure. A report that implies those 'after' sensors are 100% OK.

A torch should be sufficient to remove most 'decade rusted' parts. Especially any oxygen sensor mounted on that stainless steel exhaust pipe. But if overheating due to excessive gasoline, then removing a part (ie oxygen sensor) would be difficult.

Implied: catalytic converters damaged and now seized by an engine always dumping excessive gasoline into the converters - as discussed earlier. Diminished (not yet failed) converters may be a symptom of a long existing problem that also caused lower gas mileage.

Federal law says an exhaust manifold must not fail for first 100,000 miles. Which is why most pipes are good for less than 200,000 miles. That pipe and cat converters typically do not fail. Says why those parts are so expensive. Parts that don't fail often have highest profit margins. We used to charge a 100% markup.

So, if an oxygen sensor is the first (my original question), then doing anything to replace it (soak it repeatedly in WD-40 for 24 hours before using the torch to heat it while using a socket wrench to remove it) is a best shotgun solution.

BTW, never saw a reverse threaded oxygen sensor. If reverse threaded, then a shop manual was blunt about that unusual threading.

Spending another $25 for the Car Chip Pro would have said much more. But if any part is to be replaced, a most suspect part is probably that 'first' oxygen sensor. Not any sensors after the cat converters. Replacing only that sensor is by far the least expensive of all 'shotgun' options.

I thought WD-40 was flammable - apparently not?
.
.
.
.

It doesn't matter, after you spray and let soak, the flammable component is pretty much gone.

He was pretty sure of the outcome.

I wouldn't have tried that right next my car :eek:

And that, folks, is a really good mechanic/engineer.

lol, hardly. i hate not being able to do/complete a job. this one won. it beat me. what can i say? still irritates me though.

That's exactly what I mean. I also hate starting a job and not finishing it, I'm sure we all do. But I hate it more when I over-reach my abilities and break something and turn a maintenance need into an emergency fix. Recognising when you are at the limits of your abilities and the tools at your disposal is a prime skill. Well done.

Not necessarily. It may have set a trap. What were consequences of, for example, a partially removed pipe? You would have to replace an entire exhaust manifold and catalytic converter that cannot be removed.

You may have outsmarted a car that had laid a trap.

However, just as fun is getting into that mess. And then discovering how to solve it. Either you outsmarted a car. Or you just missed out on having fun. Either way, you did not lose. Only and maybe did not win.

A simplest ‘shotgun’ solution is replacing an oxygen sensor. Soaking in WD-40 for at least 24 hours may be helpful.

I was thinking I might get a small above-ground pool, fill it with WD-40, and soak the whole car in it. Sure, it would smell funny, but the kids would love sliding off the seats.

I love the smell of WD-40 - anytime. It also cures bed bug problems.

omg don't soak the whole car in wd-40! It just occurred to me that every bolt will loosen simultaneously and the car will fall apart while driving down the highway, like a cartoon!

Like the giant lego ball!

For the last two months we've been putting nothing in it but Shell. No improvement. I put a bottle of fuel injector cleaner into it. No change. Still a rough idle after it warms up.

So I bought a CarChip Pro so I can monitor RPMs and other stuff.

I just started it and let it idle for 8 minutes or so. I didn't touch the accelerator. The first 4 charts I got are below. It started off fine, and the idle got progressively rougher, until at 8 minutes in, the roof liner started vibrating it was so bad. Then it was dinner time, so I turned it off.

Chart 1 is RPM. You can see it gradually slows down. I didn't touch the accelerator ever.
Attachment 36998

Chart 2 is the throttle position. I had no idea what this would show, but here you go. It shows little when you don't step on te gas.
Attachment 36999

Chart 3 is the first of several oxygen sensor readings it can take. I ahve no idea what this means, but this is O2 B1, S1
Attachment 37000

Chart 4 is also an oxygen sensor reading. It's called O2 B1,S2, whatever that is.
Attachment 37001

I'll have to double check the shop manual, but if I recall correctly, it should read between .1 and .9 volts. It goes a little below .1 volts.

There are a whole bunch of other readings I can have this make, but you can only do 4 at a time, and you have to take it out of the car and hook it up to a PC to change what you are monitoring.

It's a neat little gadget.

A car at idle runs completely different from one at normal engine speed. At idle, a car must dump more gas in the engine while making the car work so inefficiently as to run at only 800 RPMs. Most pollution occurs during idle. Also why cars sitting in bumper to bumper traffic consume so much gas (another reason why hybrids are so necessary).

IOW a better test of your O2 Sensor and other factors is to be moving at constant speeds as indicated by little or no throttle movement.

Now, some facts. In the first minute, an O2 sensor must get warm before it reports anything. So newer technology sensors have an internal heater to make them hot (work) sooner. In your case, sensor one got to temperature quickly. Sensor 2 seemed to be doing nothing until engine exhaust finally heated it. Sensor 2 may have a defective heater or for some reason is taking longer to get the temperature. (BTW, I am assuming this is a 6 or eight cylinder since fours only need one O2 sensor. If this is a four, then the second sensor may be after the cat converter making this analysis bogus.)

Sensor 2 appears to be reporting inaccurate voltages. An O2 sensor is typically between .6 and .7 volts when the engine is running optimally. When dumping gas excessively, then the voltage is typically about 1 volt. Your sensor two appears to be reporting numbers that are low. Or one side of the engine is using more gas than the other.

However, how electronics measure those voltages might explain different numbers. Your particular car may measure or report those voltages differently. Normal voltages for an O2 sensor must be confirmed from the shop manual. But something is different on both sides of the engine.

Throttle clearly remained closed the entire time. But after 8 minutes, your idle was not yet at 800 RPM? The car should have been hot and at idle after 8 minutes. You might repeat the test while monitoring engine temperature. Your thermostat might not be fully closed (partially sticking) when the engine is cold. Therefore the engine is taking too long to warm up; wasting gas in the first ten minutes.

What you are looking at? Fuel injection is a switch (much like Cloud's idle adjustment valve) that is turned on and off quickly by a microprocessor. Time of each opening (called fuel trim) determines how much fuel is dumped into each cylinder. An O2 sensor measures how much got burned. Microprocessor reads the O2 sensor to determine whether to lengthen or shorten that opening time (adjusts short term fuel trim).

Graph is each O2 sensor number (maybe) sampled only at 5 second intervals. The actual numbers vary significantly in milliseconds. Actual O2 voltage is varying many times more often and more gently. Your chart simply shows some of the highest and lowest voltages. Therefore the steep changes.

Those variations may be normal for idle. But if you see variations that large for normal RPMs, then that may also be symptoms of a problem. An engine at normal speeds should see O2 sensor voltages varying little between 0.6 and 0.7.

Notice that at seven minutes, the variations decrease to what I am more used to seeing.

Voltage should increase when, for example, when a throttle is pressed. An open throttle means intake manifold has no vacuum. So the computer dumps more fuel into cylinders (short term fuel trim increases). More fuel comes out unburned. The O2 Sensor (that I am used to seeing) reports 'more unburned fuel' at about 1.0 volts.

Again, large variation may be normal for your engine. If it was any of my Hondas, I would be concerned. Approaching eight minutes, the variations decreased. Whereas your seven minute voltages to me are too low, still, the O2 sensor varies more like I am used to seeing it.

Meanwhile, the 'Shell' test implies gasoline was irrelevant to your original problem.

You can view a relationship between short term fuel trim, throttle angle, O2 Sensor voltages, and intake manifold vacuum. To see what the microprocessor sees and responds accordingly.

Another parameter is long term fuel trim. I have never confirmed this by doing it. But a long term number would increase if the gasoline is inferior. If gas (ie WaWa) does not burn as efficiently, then engine software may respond by increasing (over a long time) the amount of fuel it dumps into the engine. Long term trim is how a microprocessor learns and adjusts from engine overall performance. Short term fuel trim is how it responds due to immediate road, throttle, load, and other current variations.

O2 Sensor is not the only input. But it varies according to how much gas remains unburned. Other inputs include a sensor that measures atmospheric pressure (are you one mile up in CO or at the shore?), and one that measures manifold vacuum. When does it know to dump more fuel into the engine? When vacuum decreases (due to an open throttle or a sudden air leak). Ignition timing also changes (advances) when more fuel is dumped. You can also view that relationship.

I don't remember what your car's technology is. However older cars adjusted engine timing mechanically (centrifugal force and manifold vacuum). So a sticky or binding ignition timing could be seen on that chart. I have seen where distributor grease got sticky after ten years causing engine timing to intermittently stop changing. That was also indicated by a slight knock. However some auto companies (ie GM) decided knocking is normal rather than fix an inferior engine designs. Knocking alone might not report a problem. But sticky mechanical ignition timing might be seen on the chart. Unlikely. But an example of what the microprocessor can see. But the failure is not serious enough to report. An example of why many detected problems can remain unreported by engine check lights.

Hopefully this successfully introduces simple operations inside a computerized engine. Computerized engines are simpler than carbureted. But involve fancy nouns (ie short term fuel trim) that at first sound confusing.

But until then, every squeak and rattle will have been eliminated.

Thanks for your input, tw.

FYI, the car is a 4 cylinder.

I was reading the shop manual as I was drifting off to sleep last night. Boy, what an annoying book. You find one list that talks about rough idle possible causes, and each cause sends you to a separate section of the book, so you go to those separate sections, and they each send you to other separate sections, so you spend all your time just flipping around from section to section. I need to just sit down for a couple hours with a notebook and take my own notes from this manual, since trying to find where I read something is pretty much impossible. I've got almost every page in this one section folded over.

Anyway, for the O2 sensor, the voltage ranges are supposed to remain within .1 to .9 volts during operation. It doesn't send you a code until the voltage drops below .05 volts, and in my idle test on my car, it never did.

I also have a few stop and go trips around town recorded, but they are confusing because there is so much variation in road and engine speed. I'll try to fit in a short trip on the highway this weekend.

One amusing thing is that this CarChip Pro is also marketed as a way to snoop on your teenage driver and tell you if they accelerate too fast or stop too fast. My wife never triggered the thing, but almost every time I stop when driving, I set the damn thing off. So I come to a stop at a stop sign and the damn thing beeps at me. Then I stop at a red light, and it beeps again. I'm not jamming on the brakes. Honest. But my wife thinks it's funny. There's some way to turn it off, but I haven't figured it out, and maybe it will make me into a "better" driver.

There was a review on the Amazon page by a dad who had purchased it to monitor his son's driving habits... and when his son got a ticket for going 90 in a 55, the son swore he wasn't going that fast, and the dad's all, "Yeah yeah, sure son. Let's check the record, shall we?" And it turned out, no, the CarChipPro had recorded a speed of only 62 at that time. Still speeding, of course, but obviously not 90. They presented the data in court and got the ticket thrown out.

Nice! I like!

A year and 150 posts ago, I complained in this thread about a rough idle my car has. If you are a glutton for punishment, you can go back and re-read it all, starting at post 55. The problem is still there.

The executive summary is that after driving the car for a while and the engine gets warm, I stop at a light and put the car in neutral or step on the clutch, and it begins to idle. And vibrate. A lot. It's annoying. It's loud. It didn't use to be like that. It used to be smooth and quiet. Now it's rough and loud and vibrating. However, the fuel economy is as good as ever. I got 39.5 mpg on the highway on a trip back from Maine over the summer. Very pleased with that. Normally it's city driving and around 30.

We recently bought a second car, which means I can take a day here and there to work on this one, and the other one is available for use. No pressure to get it fixed in a half hour window before the next soccer game. So I am returning my attention to this problem.

So far, I've tested the engine mounts by putting on the parking brake and trying to drive forward and reverse while watching the engine. The engine barely moves. I think the mounts are good. I ran a few bottles of fuel injection cleaner through the gas tank according to instructions. No change. I paid the corner mechanic to clean the throttle body when they were doing an oil change. No change. But I recently took the air filter off to check the throttle body, and noted that it was clean, so they did the work. I tried to clean the mass airflow sensor, but discovered that this car (1996 Geo Prizm) does not have one. It has a pressure sensor instead and I understand you don't clean those.

Here's the throttle body.
Attachment 41652
Looks clean to me. The little square-ish hole at the bottom of the throttle leads down to the Idle Air Control valve. It looks like it might be a little dirty down there, so I may try cleaning that next. It's kind of complicated though. I'd need to remove a few hoses, including coolant lines, some data cables, and a throttle cable before I can get the throttle body off, and then take the IAC off of the throttle body and clean it out. Maybe this weekend. Seems like if I'm having idle trouble, the problem might be the idle control valve.

How do you take off a couple of coolant hoses without getting coolant everywhere or bubbles in the coolant system? Do I need to flush or bleed the coolant afterwards? Sounds like a lot to go through to clean something in the hopes it will fix the problem.

I would not waste time cleaning things. Cleaning rarely fixes such obvious problems. And might even create more problems if disconnecting the coolant is necessary.

If I remember correctly, we went through a list of suspects from the manual and other suggestions. Eliminating many including fuel and O2 Sensor. Mileage and operation when not in idel mode eliminates most suspects (including motor mounts). Listen to the exhaust pipe both on idle cold and idle hot. Any missing that corresponds to the roughness?

When cold, the idle is probably faster. Might not feel rough. A problem not apparent in cold (faster) idle might still exist and be heard in the exhaust pipe. Missing in the exhaust pipe would eliminate motor mounts and other static reasons for roughness.

Does the roughness appear suddenly or gently become obvious as idle RPMs slowly decrease?

Does this fuel injection system do a fast idle by holding the throttle plate slightly open or by opening a fast idle air valve (not to be confused with the Idle Air Control valve)? IOW is there a relay or vacuum line that partially holds the throttle open for a fast idle?

As noted previously, an idle control system operates different from high speed operation. Parts that are common to both operations (ie spark system) can be eliminated since they work fine in high speed operation. Apparently performance and gas mileage say all high speed subsystems work OK.

Some previous questions I don't recall seeing answered. How does this car perform timing? For example, does the distributor have vacuum and centrifical advance and retard to mechanically adjust when spark plugs fire? Or is engine timing done electronically? Carchip may report timing as the engine gets warmer.

One part that operates only in idle mode (not in high speed mode) is the Idle control valve. A binding valve might explain rough idle. However, if difficult to access, then better is to eliminate other possible suspects first.

EGR valve can be in that suspect list. EGR valve is open during high speed operation. But must fully seal always during idle. That valve leaking exhuast gas during idle means it too would be a part that could only affect low RPM idle. And is often easily tested. Two reasons for that leaking would be vacuum on that EGR valve at idle (when no vacuum should be opening that valve) or the valve does not fully seat (close) at hot idle. The later can only be confirmed by temporary removal and inspection. Former becomes obvious by disconnecting an EGR controlling vacuum line and covering that hose with a finger. No vacuum should exist during idle.

I don't remember the long list of suspects. Don't remember any error codes. And don't remember why we spent so much time eliminating an O2 Sensor and gasoline from the suspect list. But gas mileage and high speed performance also eliminate both from the suspect list.

Thanks for the post! I'm not excited about removing and cleaning the IAC valve because I don't want to mess with the coolant unless I have to. I just read the section of the manual that talks about this, and the second step after relieving fuel pressure is to drain the coolant system. What a job!

A little more information:

1. I've only been using name brand fuel (Shell) for the last year.

2. Car has an O2 sensor next to the exhaust manifold by the engine, and one after the cat under the car. I've looked at the charts from the carchip pro, but don't really know how to interpret them. One potentially interesting thing that I noticed this weekend is that the outer protecting sleeve of the wires going the the first O2 sensor by the manifold is all chewed up and torn. There had been a cable tie holding that sleeve to a coolant hose away from the fan, and the cable tie was broken and slipped down the coolant tube. The O2 wire harness was hanging loosely between the fan and the exhaust manifold. But when I inspected the wires within that damaged sleeve, the insulation on the wires was undamaged, and the sensor sends readings to the carchip pro. So I assume it is fine, even though it looks like hell.

3. No error codes.

4. I think the timing is controlled by the computer. And I just confirmed. Manual says "All spark timing changes in the distributor are performed electronically by the Engine Control Module Powertrain Control Module." The OBD2 system will send data to the Carchip for "timing advance." But I don't know what those numbers mean.

5. Roughness is not present at cold start up, but I wouldn't say the engine is smooth either. The problem may exist when cold. Hard to tell. It gradually gets worse as the car warms up and the RPMs drop a little, but then suddenly gets a lot worse. I assume this abrupt change happens at the point when the car decides it's warm and changes from rich to lean mode.

6. There is a throttle opener that increases idle speed during start up conditions. A vacuum hose goes from the top of the throttle body to the throttle opener and appears to be what activates the throttle opener. I haven't studied this during a warm idle, so I don't know if it does anything after the engine has warmed up. The manual discusses ways to test this throttle opener, but it requires a real time scan tool to monitor RPM while you fiddle with the hoses at a specific RPM and then confirm that the engine drops to another RPM when you let go of the throttle with the hose plugged. I don't own a real time scan tool that will display RPM, so I can't test this component right now.

I am intrigued by the EGR. It looks easier to test than the IAC. I think I'll focus on that this weekend instead.

Also, listening at the exhaust pipe sounds easy. Could you describe in more detail what I should be listening for?

A good car will have a constant 'hum' (as it were). Misses will sound like a 'poofff'. Listen for this on any car. Best designed (high performance) engines have a smoothest and least noisy exhaust. (Listen to Harley, Camaro, Mustang and pickup exhausts to hear inferior engines.) But a cylinder miss (the pooofff) is quite obvious in an exhaust. Especially listening within feet of that exhaust pipe during idle.

Another sign of a defective engine is backfire as the car decelerates. Only scumbags think that crap sound is cool. An engine without defects outputs least and smoothest sound when a foot is removed from the accelerator.

EGR valve is typically so easy to access and test as to be a first thing to remove and inspect. (Some may require a new gasket to reinstall. Better cars typicaly have no gasket damage when an EGR valve is removed.)

EGR valve opens with vacuum that you can even create with your mouth. Flow from exhaust manifold to intake manifold can also be tested (for a good seal) with your mouth (and with something to keep your lips looking fashionable such as black lip balm).

If an EGR valve has a wire connection, then a computer is monitoring its operation. If no wire, then the computer might not know an EGR valve is failing. So disconnect a vacuum hose to that valve. Rev the engine. A vacuum will be felt only when a warm engine is at higher revs (above 1500). And never when the engine is at cold or idle. A leaky EGR valve can create rough idle.

Only EGR failure a computer cannot see is leakage from exhaust manifold to intake manifold. Your 'lip' test would expose a leak.

Disconnecting coolant hose is easy (once a few tricks are learned). Often best done by partially emptying the radiator into a pan via a cock located conveniently at the bottom of the radiator. Refilling is also easy on better designed cars. An air bleed port at the highest point of engine coolant is opened to let any trapped air bleed out when refilling. But disconnecting coolant hoses and removing the IAC valve is a pain that usually results in a little of your blood left in an engine compartment. (Or get construction gloves from a hardware store.) Requires care to not harm wires or control cables. And mark each vacuum hose before disconnecting.

BTW, one remote problem is an inspection mechanic who inadvertantly swapped two hoses. Better cars actually number each hose. Sometimes a vacuum hose diagram is on the bottom of the hood (bonnet).

EGR valve is typically 'inspected' with your mouth. IAC valve typically requires better equipment and knowledge. Just another reason why inspecting the EGR valve is better done first.

Noted previously - the idle system and normal operation operate completely differently. To idle, a car must change engine parameters to become very inefficient. And timing is different.

Car Chip can view engine ignition timing verses RPM and engine loading (if your computer reports loading to Car Chip ). During idle, timing is at zero degrees top dead center. Or a few degrees off center. As engine speed increases, timing (when the spark plug fires) advances. Generally, fuel is ignited before fully compressed. And continues to burn (pushing the piston) after the exhaust valve opens. When an engine becomes more loaded, then timing may retard closer back to zero. Causing more power but more unburned fuel.

To make an engine inefficient at idle (so that it can run at a pathetically slow 800 RPMs), timing is moved to zero. If timing does not move back to that point, then it might run rough at low RPMs. Shop manual may report what that timing number at idle should be.

Computer is apparently monitoring your timing. Therefore viewing these timing numbers would only be informative and interesting. The computer was not seeing timing 'out of spec'. So idle engine timing also should be OK. View those timing numbers for various RPMs to appreciate what is happening. And to confirm timing at idle is at the right spot. Obviously timing is OK at high speed (no knocking, no hesitation, good mileage). Timing was only relevant to your timing roughness when it should be at or near zero degrees.

BTW, TDC or Top Dead Center is when piston #1 is at its highest point. That is when a spark plug fires for idle. At higher speed, the spark plug fires much sooner (before cylinder gas is fully compressed). IOW timing advances ten or as much as 30 degrees Before Top Dead Center (BTDC). If not, then an engine can be less efficient and 'knock' due to unburned fuel. But again, your mileage number suggest timing during normal operation is correct.

Timing at idle sometimes changes with engine temperature. A cold idle might have a different timing from a hot engine at idle.

Engine timing was why so many spent maybe $100+ every three months for a tuneup (new points and condenser). EPA requirements (1975) made that tuneup unnecessary by replacing mechanical parts (1900 technology) with electrical parts (1962 techology). And later (1980s) with something also more reliable - a computer.

Four items to consider - EGR valve, IAC valve, engine timing, and swapped vacuum hoses. Also useful might be other items on their 'suspect' list. Since something unique to your car design (something else that works differently at idle than at normal speed) may be overlooked by me.

Glad I can test the EGR with just my mouth (I'll use some clean tubing so I don't get gunk in my mouth.) I was thinking I'd have to buy one of those hand operated vacuum pumps to test it. I like excuses to buy tools, but I've been spending too much lately.

There are so many vacuum hoses! I was looking for possible disconnected hoses and kept finding new ones to inspect. They are everywhere.

Better designed cars have a number on each hose. A vacuum hose diagram confirms each connection. I once found two hoses reversed. Car still started.

EGR valve will not open fully with 'mouth vacuum'. But just enough to confirm its works. More important is no leak (ie stray piece of carbon) from exhaust to intake manifold.

I listened at the exhaust, but didn't hear anything other than a steady hum. After a while some water started consistently sputtering out, and I noticed some pinhole leaks in the muffler where water was dripping out.

I turned my attention to the solenoid that opens and closes the EGR. I checked for the proper resistance between the prongs, and then check that there was no continuity with the outer frame.
Attachment 41686

I blew in the lower nozzle, and saw that the air came out the filter vent, and then I applied battery current to the prongs, heard the solenoid activate, and when I blew in the lower nozzle, the air came out the upper nozzle. According to the manual, this confirms the solenoid is OK

Then I went to the EGR transducer and checked to make sure the filter was clean. It was very clean. I blew in one nozzle, and covered the two opposite nozzles, and confirmed the air came out the sides. Attachment 41687
According to the manual, this confirms the transducer is OK.

Finally I took off the EGR valve. This was a little trickier because it was underneath and there wasn't much room. But I got it off and saw that it wasn't terribly dirty.
Attachment 41688

A little dirty, but not bad. I tried sucking on the hose that leads to the diaphragm, and could easily open and close the valve by sucking on it.

The air intake was a little dirty too, but I couldn't get at it to clean it. It was hard enough to take this picture.
Attachment 41689

As long as I had the valve off, I might as well clean it. So I did. Sprayed lots of carb cleaner in the different holes and sucked on the hose that leads to the diaphragm so the cleaner would get in valve area as I sucked the valve open and closed.

Attachment 41690

Then I put everything back together and started the car. It still vibrates a little, but it's not as bad as it was before. I can't get it to make the really annoying vibrating and sound. I'm optimistic!

There were a couple other tests I could have done if I had a vacuum pump, a vacuum gauge, and a scanner tool that showed RPMs in real time.

Maybe this fixed it. I'll pay attention for a while and see if it's fixed. Hardly drive this car at all though, so I don't know when that will be.

Did not realize you had a vacuum gauge. Smooth sound in an exhaust suggests the vacuum gauge should show a constant vacuum in the manifold at idle. However confirm this. Intake manifold vacuum at a constant above 17 inches should be observed. If the vacuum gauge needle vibrates, then the problematic cylinder is indicated when the gauge point dips to a slightly lower vacuum.

Assuming the vacuum gauge is contant at idle. Then this is puzzling. Vibration at idle implies a missing cylinder. And yet the exhaust pipe sound suggests no missing. Vacuum gauge will probably report the same 'no problem'. If so, move on to some other suspect. This engine appears to be operating smoothly on all cylinders during idle mode. And gas mileage suggests it is also working properly during normal operation.

Every part of a drive train and engine must be isolated from the car's body. Even a radiator does not make a direct connection; connects to the body via rubber isolators. Is it possible that something (ie exhaust pipe) is touching the body as it gets warm? For example, an exhaust pipe was hung slightly twisted? Or a heat shield place moves slightly with heat?

Well, I don't have a vacuum gauge. Maybe I should get one. Seems like a waste of money for a tool I'll only use once. But I wish I had one.

I listened at the exhaust, but don't trust my skill at identifying the sounds I hear. It sounded like an exhaust. Didn't seem abnormal at all.

I'm confident that I eliminated the EGR system as the cause of the problem, and even cleaned the EGR valve, but last night I drove the car to a party in DC (because it's small and easy to park) and it's as bad as it ever was. The problem is still there.

When I watch the engine, one of the radiator hoses is vibrating like crazy. The engine doesn't seem to vibrate much. At times the battery is vibrating a lot too, so you can't read the writing on it, but yesterday it wasn't so bad.

I guess it's back to the drawing board.

Under the section for vibration diagnosis, the shop manual talks a lot about road vibrations, but those don't apply because the problem is only noticed when it's stopped. The only part of the vibration diagnosis that applies is the engine speed sensitive section. They list a bunch of scary sounding possibilities:
-Belt or engine driven accessories like generator, A/C compressor, power steering
-drive belts and their tension
-bad bearings
-bent shafts
-faulty or loose flywheel

For rough idle, which seems to me like it might cause cabin vibrations and loudness, they list the following:
Sensor checks
Check O2 sensors
Check Throttle position Sensor
Check Engine Coolant Sensor
Check MAP sensor (but they don't say how)
Fuel System Checks
Check if condition occurs under Rich or Lean conditions
Check EVAP control system
Check fuel injectors for leakage
Check fuel pressure
Ignition System Checks
Check ignition voltage with spark tester
Check spark plugs for wet, cracks, wear, gap, burned, blistered, dirty
Check spark plug cables for resistence
Check ignition timing
Additional checks
Check for vacuum leaks (I did this)
Check ECM/PCM grounds for clean, tight, routing
Check ECM/PCM with scan tool to look for A/C signal
Check EGR (I did this in the post above)
Check battery cables and ground straps (battery cables look fine)
Check Crankcase Ventilation Valve
Engine Mechanical Check (It's like they are just listing engine parts now)
Motor mounts
valve timing
low compression
bent push rods
worn rocker arms
valve springs
camshaft lobes

I thought I would check the PCV valve since that was suggested. I pulled off the vacuum hose to the PCV valve, and the engine almost stalled and then started revving and then cycling between to two. When I put my finger over the end of the vacuum hose the engine settled down into its normal rough idle. It had been sucking air in like crazy.

The manual says that if you put your finger over the PCV valve inlet, it should snap back into place. I tried it and the valve did nothing. So I pulled the valve. It was perfectly clean, and I could shake it and hear the plug inside rattling back and forth. I guess the PCV valve is fine.

So then I started disconnecting the spark plugs one at a time and listening to the engine. Now I know what missing sounds like. It's very distinctive. The engine wasn't missing before. As I disconnected each spark plug in turn, the engine would sound worse each time. So I think it's not missing.

But now my check engine light was on! Disconnection a spark plug will do that. Fortunately, the CarChip Pro did exactly what it's supposed to do and I was able to turn that light back off.

So I haven't solved anything, but it's been fun playing with the pencils on the bench.

Maybe cars just sound like shit as they get older.

Understand what you did. Fuel injection monitors intake manifold vacuum. If the vacuum decreases, then the injector system assumes you have opened the throttle plate (pressed on the accelerator). So it pumps more fuel into the engine to accelerate. But then it sees the throttle not open (throttle sensor) and faster engine RPMs. So it cuts back on fuel (there is an RPM number above which injectors cut off all fuel when the throttle is fully closed). Confused because the amount of fuel to maintain intake vacuum is not desireable if you have taken your foot off the accelerator. So it stops injecting fuel. But then intake vacuum falls too much. So it again thinks you have pressed the accelerator.

Exhuast - is the sound constant? Then no "poofff" exists. Its simple. Either the sound is constant - good engine. Or is has a change (or short loss of sound) - cylinder miss. If in doubt, go around listening to other cars. Then notice, for example, how much smoother Acura, Lexus, Audis, etc are because they have higher performance engines.

Mouth the word "poofff" using lungs and not using any vocal cords. That is exactly what you are listening for. A not properly working car should make that sound.

Sensors will be seen by, for example, Car Chip as the O2 Sensor was also reviewed. MAP changes with throttle changes. The shop manual may provide some graphs that relate voltage to air pressure. Which could probably be confirmed by viewing those voltages via CarChip.

Don't know how you checked vacuum lines for leakage since that usually requires a vacuum gauge. Some lines can be tested by simple squeezing the line to learn what happens. Then comparing that line to its function defined in the shop manual.

A small vacuum leak could be apparent by temporarily plugging the offending line and its manifold connection. You saw what happened with a major leak. What would a minor leak do? The engine would constantly be searching to inject enough fuel to maintain vacuum but then lower the resulting high idle.

I do not know what they recommend for a throttle sensor. In at least one case, that throttle sensor was only factory adjustable. Any inadvertant change meant buying a new throttle sensor assembly. Don't know if that applies to your design. A precaution about doing any throttle sensor adjustments without consulting the shop manual.

EVAP system would create problems similar to a vacuum system leak. And located by similar procedures.

Low fuel pressure would result in poor acceleration. You would need a pressure gauge to measure a maybe 45 or 60 pounds pressure on gasoline. I built one once by using a water pressure gauge from Lowes and some fitting from a company selling pneumatic hose fittings. Then discovered low fuel pressure on a pump the mechanic said was good.

He could not bother to read spec numbers. He saw pressure. Then assumed everything was OK. And yes, I was pissed I had to do his job. Low pressure did not affect idle. But did restrict acceleration and eventually (but rarely) triggered the check engine light if I did full petal accleration with multiple, consecutive engine restarts. I had to create the same problem with every restart by doing a full pedal acceleration - that I never do.

Why did I find it? I would be damned if a check engine light was reported by a mechanic as an 'unknown problem'. He failed to do what all solutions require - numbers.

Low fuel pressure would be apparent when accelerating in normal mode. Pressure would be higher (and sufficient) in idle due to less fuel consumption.

Any ignition system failures would be indicated by a missing cylinder (as reported by the vacuum gauge or "poofff"). Eventually by poorer mileage. And by poor operation (ie inconsistent acceleration) during normal operation mode.

Valve timing, bent valves, low compression, worn cam lobes, valve springs, etc would result in poor normal mode operation or in noisy 'ticking' in the overhead cam. I am surprised they also did not list valve lash which is adjustable on overhead cams for better engines. I cannot think of any condition here that would cause rough idle, not create the 'missing cylinder' symptoms, but permit good normal mode operation.

Those visible vibrating parts may simply be frequency reasonant to the vibration. May stop vibrating as engine speed (and therefore vibration frequency) increases slightly.

However, is some other part (ie heat shield on the exhaust pipe) vibrating against the body only when it also resonants at that frequency? Might be heard inside but not outside the car.

I saw in some youtube video that you can spray carb cleaner onto the exterior of the vacuum hoses. If there is a leak in the hoses or at the connections, the carb cleaner gets sucked into the vacuum hose and the engine either stalls or revs. This guy also said you can use an unlit propane torch to do the same thing. I was out of cleaner, so I used a propane torch. But it was a little bulky and hard to get into some locations. So I should try it again now that I have a pretty full can of carb cleaner.

Make sure it's not idling rough because the rpm is dropping too low when it's warmed up.

It's dropping down to about 700. I can't find the number anywhere for what the idle should be for a 1996 Geo Prizm Lsi, but I think 700 is in the right ballpark. It's certainly getting rougher as it drops. Maybe it should be 800? You'd think the shop manual would give me that number, but I haven't come across it in any of my reading. Maybe there is a chart of specs in the front or back or something. I should look harder to make sure that number isn't hiding somewhere.

That's not uncommon these days since the mechanic isn't required to set the idle speed. It's all the computer saying stand back, I'm in charge here. But that makes it hard to find out if the computer is fucking up because of bad feedback.

Back feedback means a computer also fires spark plugs at the wrong time.