origamy

The car is 2002 Honda Accord 1.8l, engine code is F18B2. The problem is, that my car bucks/hesitates when I fully depress gas pedal and then floor it. It does not matter if the engine is cold or hot. The car violently bucks a few times, then accelerates normally. It happens at any rpm. It is more felt in low gears, less in high gears. Another thing is, that when i try to keep constant speed in low gears (1-st ant 2-nd at 1000-2500rpm) it hesitates and it's hard to drive in traffic jams, because i have to use the clutch very often. Idle is a bit rough, engine seems to misfire from time to time. When cruising at high speed nothing wrong is felt. Also if changing gears very softly and accelerating softly everything is almost o.k., almost no bucking/hesitation is felt. CE (MIL) light is NOT on. Changed: air filter, fuel filter, plugs NGK, wires NGK, cap/rotor/coil, then tried another distributor from salvage yard. Nothing changed. Cleaned EGR valve and ports. Did put in the new O2 sensor (NTK). Did the valve adjustment. PCV valve is O.K. Tried disconect any sensor that is present in this engine (TPS, MAP, KS, so on), reset the ECU/PCM, then test drive - nothing changes or gets worse. Cleaned IACV. TPS reads 0.5V at closed throttle and 4.5V at WOT. Voltage transition is smooth. Tried another MAP sensor. No luck. Mechanical timing is ok, all marks are where they supposed to be when cyl n.1 is at TDC. Engine idle rpm 's are OK, they do not jump around. Compression in all cylinders is 13.5-14.5bars. No white smoke, coolant level is constant. Eats some oil, but the eating rate is constant and it allways ate some oil. Ground wires to the engine are cleaned and ok. No vacuum leaks I did found. Mileage is 157.000km. Spark plugs look like this: i39.tinypic.com/2vxlqu0.jpg i40.tinypic.com/2evqeex.jpg

Scantool results:

At idle (750rpm):

STFT: from -5.47 to 1.56 (fluctuates all the time)
LTFT: 0 to 0.78 (remains almost constant)
O2 sensor before catalyst: from 0.1 to 0.9 (fluctuates all the time)
O2 sensor past catalyst: 0.18 (remains almost constant)

At 1500rpm:

STFT: from -9.38 to 10.16 (fluctuates all the time)
LTFT: -0.78 to 0 (remains almost constant)
O2 sensor before catalyst: from 0.1 to 0.9 (fluctuates all the time)
O2 sensor past catalyst: 0.2 to 0.5 (remains almost constant)

At 2500rpm:

STFT: from -13.28 to -5.47 (fluctuates all the time)
LTFT: -0.78 to 0 (remains almost constant)
O2 sensor before catalyst: from 0.1 to 0.9 (fluctuates all the time)
O2 sensor past catalyst: 0.1 (remains almost constant)

At 3500rpm:

STFT: from -10.16 to 2.34 (fluctuates all the time)
LTFT: -1.56 to -2.34 (remains almost constant)
O2 sensor before catalyst: from 0.1 to 0.9 (fluctuates all the time)
O2 sensor past catalyst: 0.1 (remains almost constant)

Maybe somebody here can suggest where to look next?

holmesnmanny

My first guess considering you've pretty much done everything is that the knock sensor is bad. They pretty much only throw a code when unplugged or when the wiring is open. It could also be the shielded wire around the wiring is open and not keeping out interference. Furthermore, the greatest stress upon a bad knock sensor system is when you have the throttle opened all the way as the engine is forced to advance timing and it's getting bad input from the knock sensor. This would explain why the hesitation is gone when you gently blip the throttle.

I had issues like that when I did my swap and ran a regular wire to my knock sensor. As soon as I used shielded wire such as factory stock cars already have the hesitation went away.

Last resort if the knock sensor is good is to just run a temporary shielded wire from a used knock sensor connector straight to the ecu after depinning the original ecu pin and temporarily putting the new one in, making sure to ground the shielded metal portion of the wire. I did this on only one side(the ecu side, just run it to ground after cutting back some of the shielded metal and putting a butt connector on it).

This is just a guess though. I would swap out the knock sensor for another before doing that.

origamy

Tried to disconnect knock sensor, reset the ECU, and test drive. Nothing changed, just ECU threw sensor error code, as it is supposed to do

afnan

Few days ago I had same problem and it was a faulty fuel pressure regulator. You have figured out the rest. This may be the last bit

origamy

Injectors, fuel pressure, fuel pressure regulator are checked and ok..

soLame

this could be a long shot, but why not cover all the bases.
Test EGR valve, I know you had disconnected it before, but that only tested the "open" part. on my 98 accord you can send 12 volts down to terminal 4 and ground terminal 6. Doing so on a warm engine (fan is on), is supposed to chock the engine to a stop.
good luck.

origamy

I've tried to block EGR valve completely by installing blanking plate. And that had no effect. Then i've blocked it by installing a blanking gasket directly on the EGR holes into intake manifold - no effect again, just the error code after some driving - that is normal. EGR is not the source of a problem, i'm pretty sure bout that

holmesnmanny

Simply disconnecting the knock sensor will throw the car into limp mode. That's not how to test a knock sensor.

If you want to isolate the egr, even though the egr doesn't work at idle and wouldn't be contributing to your problem you would simply pull the egr valve connector and when it throws a code for egr it will turn off the egr system and if your issue goes away that means it's egr related. It won't go into limp mode for an egr check engine code.

You could try checking for an exhaust leak. If it has an exhaust leak the oxygen sensor is going to think it's running lean and will dump fuel in to try to compensate for a fake lean condition and this could cause your issues. You could try seafoaming and looking for white smoke coming out anywhere from the exhaust manifold to the oxygen sensor.

origamy

EGR was firstly disconnected, tested, then fully cleaned, tested, then blanked, tested. To no effect on bucking/jerking whatsoever. EGR is 100% o.k. and working as it's suposed to. No exaust leaks, and O2 sensor before catalyst is brand new. Catalyst is not clogged too. Tried to disconnect O2 sensor and drive - bucking wasn't gone. In this case even if exaust leak before O2 sensor existed - it wouldn't matter, because car was in open-loop mode. Problem is elsewhere

holmesnmanny

Check your vacuum hoses for leaks ..even very small hairline cracks...especially the map sensor hose. I would then manually test the map sensor. Then I would try wiggling wires while the car is idling. There could be a bad wire somewhere. Then I would take out the ecu and check it for dark spots. It could have a bad spot somewhere.

I'm still thinking maybe knock sensor issue. Knock sensor stops working after 5k does the issue stop after that ?

origamy

All hoses were checked very carefuly, MAP sensor gives 3V when ignition is on, and 1V when car is at idle - so the manual states. When OBD2 scanner is connected, at idle it shows 26kPA - as I understand it is withing specs. If there was a vacuum leak, MAP reading would be off. I even took throttle body off, and cleaned MAP air passage (which was not clogged at all). Then tried another known good MAP sensor - to same results. ECU was taken out and inspected - looks like new, no "cold soldering" spots, no burned components whatsoever. Anyway, I'm going to try another known-good ECU, when I'll have that posibility. The main problem is, that car bucks a few times when gas pedal is fully released, and then pressed again - i guess it will be quite hard to test that at 5k rpm's - even if it's not felt at that high rpm's it doesn't mean that knock sensor has something to do with that - inertia will do it's job. Bucking is more felt at lower rpm's, mostly until 3k and in low gears - 1-st, 2-nd and 3-rd. I'm thinking that maybe ECU somehow sets too advanced ignition timing at the moment of bucking, because it itself is maybe faulty, or some sensor is between life and death, but not bad enough to set a code. I have a posibility to drive and log sensor data, but not experienced enough to interpret it professionally. One more thing I've noticed - when ECU goes into limp mode and sets a code, it does not drop idle rpm's to approx 500rpm (as it did before). Idle stays within normal specs - 750rpms when any code is set.

holmesnmanny

maybe check fuel pressure ... regulated and unregulated

smell the fpr hose for fuel leaking through it

origamy

Fuel pressure is o.k. FPR does not leak..

origamy

Can this be because of clogged intake manifold? EGR passages and holes into IM were blocked with carbon, i cleaned that, but maybe there's much more to clean inside IM runners and this causes problems? Is that common on hondas? It's not so simple to take entire IM off and clean it, but maybe it's worth to give it a try?

holmesnmanny

The egr doesn't work at idle, however it does test the system on startup and if clogged it would briefly do it when you start the car. Then it would level out, especially if you rev it a lil.

Not too familiar with your car(f18b engine) but normally you either gotta pull some plugs to get to the runners or pull a plate.

origamy

Well, I've cleaned EGR system completely, pulled that plate and so on. I'm thinking about intake manifold runners, not the EGR passages, they're already cleaned. I'm thinking maybe carbon buildup is possible not just in EGR system, but deeper in IM runners, ant that maybe restricts fresh air to the cylinders by some degree. I know that it's a common problem on diesels, but maybe this aplies to the gas engines too?

holmesnmanny

Test the engine coolant temperature sensor in the head under the distributor that is two prongs.

Cold engine should be 2000-2500 ohms across the pins.

Warmed up engine should be 180-300 ohms across the pins.

If this checks out my only other thought as I said before is the knock sensor is bad. I would probably test base idle just because.

origamy

ECT sensor is tested and good. Base idle you mean idle speed? It's 750rpm, as it should be. If knock sensor is bad, disconecting it and resetting ECU should cure the problem, but it doesn't. Main thing that I can say, is spark plug color in 2nd cylinder is different from other three cylinders, and that's most probably the core of problem - rough idle (because that cylinder does not put the same amount of power as the other three). http://i39.tinypic.com/2vxlqu0.jpg - that's how it looks like. But there's no vacuum leak (or I cannot find it). I assume there can be some machanical issue with that cylinder - but it's compression is 13.5 bars (195PSI) - so well within specs. Don't know where to look and what to do...

holmesnmanny

Going back to this I gotta think there's a bad sensor somewhere. If it were me, I would go back and look and test everthing. When it goes into limp it uses predetermined values and ignores some sensors. Even looking at the ecu for dark spots.

origamy

I guess it's time for expensive engine diagnostics at Honda dealer

origamy

Did some more driving with data logging. As I can see from that data, everytime I release and then hit the gas, ECU sharply retards timing (from let's say 27 degrees BTDC to 8.5 degrees BTDC), and STFT gets very high - up to 20 and car bucks/jerks at that moment. What could be the cause?

afnan

F18 Does not have Knock Sensor. Timing is pre mapped. What ECU number are you running? STFT getting high means car is getting rich and above 10 its too rich i guess.

Your ECU number should be 37820-PCC-00(1-4) for Manual and 37820-PCC-90(1-4) for Automatic.

origamy

F18B2, not just F18. Yes it does have knock sensor. ECU number is 37820-PDA-E11, manual trans. It has standart OBD2 connector, two O2 sensors (before and after catalyst) which earlier variants did not have (mine is manufactured in 2002, it's the ending year of production, so there were some modifications made since 1998)

holmesnmanny

it should advance timing when u open the throttle ... since its not then something is keeping it from doing that ... like i said i think this is knock sensor related somehow

afnan

Yes check the Knock Sensor wiring, its shield.