OK. you're dealing with a post-96 car that's loaded with the OBD2 emmissions system. fun fun shit there. i HATE it.
anya, you are somewhat in luck here, because i just had a conversation with a friend on OBD2 equipped cars not too long ago. he also just took his emmissions tests again and passed with flying colors. when there's a question in his "region," techs in other shops are told to call him up. he hates it, but that's what happens when you're good.
you're going to have to run the tests specified in a repair manual on the EGR system. you could have a faulty EGR valve, solenoid, no voltage running to it, bad vaccuum lines....there's a number of things that could be causing the problem. you'll need a vacuum gauge to do this as well. appears that's the only tool you'll need besides a multimeter or a regular 12V test light. indepth testing of the EGR control system requires the proper scan tools (above and beyond what you can do in your home garage). you might have to bite it on this one and let the shop deal with it. i'd suggest checking all the vaccuum lines and you can probably try a few of the vaccuum tests w/out a gauge...just pull the proper line and feel for suction/blow. won't be 100% accurate, but always worth a shot.
the o2 sensor, on the other hand, if it was installed properly the code should be gone. HOWEVER, the OBD2 systems are set up kinda funky. on older cars, you can clear the codes, fire the car up, drive it a little, and check your codes again. problem solved.
on OBD2 equipped cars, there is a NASTY procedure involved to fully test out all the systems once you've cleared the codes. some of this involves getting the car up to 55mph and holding it at partial throttle for a few minutes...some wild stuff. when you clear the codes on an OBD2 equipped car, you've pretty much got to put it through all forms of regular driving to get the system back up to snuff again. if you have an trouble code popping up and you simply clear the system and take the car in for an emmissions test, you're SOL because as soon as they hook the computer up to it, they're gonna know that it was recently cleared. all of this was mandated by the government in order to make sure people weren't cheating the emmissions system. isn't it wonderful?
the O2 sensor, however, should've been a quick fix. i replaced an O2 sensor on our 96 civic (after it got bent in a little doozie i had...i'll have to post pics of that some day), cleared the codes, and was fine.
did you connect all the wires of the O2 sensor correctly? if i remember correctly, the downstream O2 sensor doesn't do much to begin with, but i could be thinking of regular OBD equipped cars, not OBD2 equipped ones. since the O2 sensor should've been a quick fix but apparently wasn't, you need to check your wiring. if your wiring is correct, you need to trace the wires through the car to check for any damaged wires. if your wiring is correct (and it's properly insulated - not shorting itself out) and the wiring leading to it is ok, you've got a faulty O2 sensor. most come with a limited warranty. pull it back out, take it back, tell them it's bad.
where do you live and what type of emmissions tests are we talking about here? if you would happen to live in SE PA, i can tell you what shop to go to and who to have work on your car (specifically request the tech).
anything else i can help with? i'm flying by memory here on quite a lot of it (and also whipped out my 96 civic book for some references). i can reference out of this book for most things on emmissions because when the government mandated the OBD2 systems, it's all the same across the board. honda used to call an O2 sensor (just an example) one thing, chevy called it another, ford called it something different....it was near impossible to work on one car to the next and have a clue as to what was what because things were named so differently. the standardizing of parts and systems has, in effect, made a technician's life a bit easier considering the complexity of the system.
now if my final on monday was diagnosing cars and not on experimental methods...i'd get an A
