1998 Civic: code P0171: fuel pressure regulator?

I have a P0170 and P0171 trouble codes. The fuel economy is poor (like 9L/100km instead of under 7 where it should be on summer gas), and during warm up, the
engine sometimes sputters once or twice. If this happens while idling, the idle RPM suddenly drops very low and the engine runs rough for a few seconds, like it's about to stall (but never does actually stall). If this happens while driving, there is a loss of power: you can floor the pedal, but the car decelerates. Seconds later, there is a sudden surge of power, which literally feels like a kick. This may happen once or twice, and then driveability is just about normal except for a bit of a low idle.
Running Techron through (a bottle of concentrate into a two thirds full tank) has improved the driveability somewhat, and slightly seems to have mitigated the near-stalling problem. That indicates there maybe had been some buildup which contributes to problems like this.
Also, I tried disconnecting the fuel pressure regulator and pinched off the vaccuum hose. There is almost no difference, though I'd swear there is a bit more power at the lower RPMs. When accelerating normally in second gear, it seems more eager. Also, there is usually a certain little ``power notch'' somewhere above 2000 RPM which doesn't seem to be there when the FP is cut off from vaccuum. The near-stall on warmup still occurs, though.
I don't currently have a fuel pressure gauge to be able to investigate pressures.
But is there enough evidence that it may be purely a mechanical fuel pressure problem, (rather than, say, faulty O2 sensor readings?)
Is there anything I can diagnose with OBDII before I shop for a fuel pressure gauge?
I haven't found any loose vaccuum hoses around the intake manifold.
I don't think I have an exhaust manifold leak, either. I had one previously which was so bad, you could smell it from the ventillation; and there was no such code coming up. The driveability was way worse than now. I recently, about two months ago, changed the top O2 sensor (for a working used one, which came from my old broken manifold), at which time I inspected the exhaust for cracks; didn't see any at that time.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Honda does not use P0170. Are you /certain/ the code you see is P0170?

What's that in civilized English?

Makes me suspect EGR, actually...

Try pinching the /return hose/, not the vacuum hose.
There is almost no difference, though I'd swear

The manual lists a whole bunch of possible causes of P0171. I'm not sure you can do this one yourself without shotgunning lots of uselessly-spent money at the car.
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

With the return hose pinched off using a G-clamp, the driving experience is like this. There is some slight hesitation when accelerating from a stop during warmup. Acceleration and hill climbing is somewhat poor in general. The idle is very low, but not rough.
I would say that with the FPR return hose clamped, the car drives somewhat simlarly to how it drives when the FPR is connected normally and unrestricted. The best driveability occurs with the pinched vaccuum line.
But the car is driveable not matter what we do to the regulator; the fuel pump is clearly working. It may not be a pressure problem at all.
I'm going to do the O2 sensor test described in the service manual. (Hold engine at 3000 RPM until fan comes on, then scan the output of #1 O2 sensor, which should be 0.3 to 0.6V. If it's out of that range, it should be replaced.)
There is also a test described for a stuck or leaking EVAP purge control valve.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I did some driving while monitoring the O2 sensors and fuel trim. Some interesting findings.
The #2 O2 sensor basically has binary behavior it either reads about 0.9xx volts most of the time except during DFC0 (driving in gear with foot off throttle), when it reads a low voltage of 0.0xx at 0% fuel trim.
The #1 fuel trim and O2 reading can be described as having demonstrated two personalities.
During the first fifteen minutes of the run, fuel trim was usually at 46.9%, when accelerating, driving steadily at around 50-60 km/h, and even when idling in neutral at stops. The O2 sensor was reading about 0.485 to 0.505 volts all the time, except when coasting to a stop (DFCO) it would drop down into the 0.2xx range.
Then while I was cruising on a stretch of highway at 90 km/h, the fuel trim value adjusted down to 35%. The #1 O2 sensor value dropped, to about 0.465. Next, after I coasted to stop at a red light, the fuel trim cut to 5.5%, and the O2 reading was only 0.2xx. Big difference from previous stops, when it was 0.4xx and trim at 46%.
After that, there was a bit of street driving at 50-60 km/h before arriving at my destination. Now, the fuel trim did not go above 35%, and the oxygen sensor stayed in the 0.3xx range. At each stop, the trim dropped to 5.5% and the sensor to 0.2xx.
My interpretation is that this may be a bad O2 sensor, since peak value I observed was 0.505V, even at high fuel trim. It never indicates a rich mixture. Come on, the car eats a lot of gas, and its exhaust stinks of unburned fuel. You would think the O2 would indicate rich at some point?
The observed change in behavior could be that the ECU realizes that the O2 lean indication is garbage, and cuts back the rich fuel trim.
Hmm; looks a lot like the repair bucks should be spent in the direction of a new O2 sensor. Will check for that possible exhaust manifold leak, though.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 06/30/2010 12:41 PM, Kaz Kylheku wrote:

those trim percentages seem crazy high. normally, "trim" means deviation in the average from stoichiometric, and that should be just a few percentage points.
i'd two two things at this point:
1. because of the age of the vehicle, i'd go to http://oxygensensors.com/ and get a new oem sensor - ngk or denso, not bosch on a honda.
2. i'd double check on the potential vacuum leak scenario. as scott dorsey says, use a propane source to diagnose.
--
nomina rutrum rutrum

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I put in my vehicle info, and that site didn't come up with any NTK/NGK sensors, just Walker, Denso and Delphi. Hmm.
That site has free shipping in the US; I'm in Canada. They will undoubtedly slap on at least 15 bucks for shipping here, if they even ship here at all. Plus, by the time the item arrives, I will waste at least another $20 in excess fuel consumpion! On a tank, it goes less than 400 km, whereas it should be closer to 600.
Better idea: yesterday afternoon I went to Vancouver Auto Parts and got a made-in-Japan NTK sensor for $78 + tax, and installed it right away. Now it drives better, and the smell of unburned fuel in the exhaust is gone. Haven't seen any rough idling or power dropout.
I will wait for the MIL to go off after a few drives. If not I will reset the codes and see whether they reappear.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 07/01/2010 10:19 AM, Kaz Kylheku wrote:

ok, good.

you should just reset them. if they re-appear, that's when you go to the next stage in diagnosis.
--
nomina rutrum rutrum

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Yay! Went out by itself yesterday while driving: zero effort.
Bring laptop to car, plug in cables, fiddle with software: nonzero effort.
It's more satisfying if the MIL goes out by itself. The ECU diagnosed it, let it confirm the fix.
You never know; the threshold standard for confirming the fix may be stronger than the standard for merely not flagging a recurrence. Hysteresis is an obvious design choice for this kind of thing.
If alarms are programmed trigger and clear on the exactly the same threshold, then they will tend to flip-flop.
Suppose that, in a device, flomposity values greater than 19.7 trigger an alarm, and your fix reduces flomposity values to 19.2 percent. If you reset the alarm, it does not re-trigger. But suppose the management software requires a flomposity level of less than 10 percent in order to clear the alarm spontaneously. If you get it to do that, it's a better fix.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Kaz Kylheku wrote:

I'm not sure that is the case, but your general idea is more or less correct. To set the codes in the first place requires the fault to occur in two consecutive driving cycles. To clear the code also requires 2 or 3 driving cycles without a fault. So having the light go out by itself is a stronger indication that it is working properly consistently even if the threshold for a fault is the same (which it probably is).      -jim
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
"sjedgingN0Sp"@m@mwt,net says...

The big difference in clearing the code yourself vs letting the ECU do it is you immediately know that your fix worked, and you don't have the CEL shining at you for the cycles it takes to shut off. It is not my experience that it requires 2 consecutive driving cycles for the code to set. I have cleared codes and they reappear immediately upon restart. They probably get set according to the polling timing of the ECU. I don't see a reason for error codes to have ranged on/offs. You have an error when you hit a single point of measurement. If it is due to intermittent conditions it may turn off. Every o2 sensor error I've had, the code remained until the sensor was replaced. Just my experience. A home thermostat has a ranged on/off to reduce cycling, a safety valve has a ranged pop/reseat to reduce chattering. But for ODB an error is just an error in my experience, and if the CEL stays on something needs fixing or replacing. It never finds another point of measurement lower than required to set the code. But that's for a single code. Maybe some sensor errors cause other sensor errors to appear for good sensors. That may be classified as an intermittent code for the good sensor until the bad sensor is replaced. I do know that unplugging the TCC connector on some TH125 transmissions can cause the first and second gear sensors to throw their error codes, even though the trans is working fine without the lock-up. Those I usually call permanent intermittent codes.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

250 miles versus 400.

You should reset the ECM /now/. That way you'll know for sure if any subsequent codes are set that are separate from the current P0171.
And if it's fixed, it's fixed; that's a good thing. But if you were smelling raw-gas out the tailpipe, I would have been expecting fuel-trim and/or misfire codes in addition to the lean code.
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Asking for the THIRD time: Are you sure you get a P0170 code in addition to the P0171?
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Yes; I think I can read the screen! Two different OBDII diagnostic programs show the same two codes. Both codes come back eventually after reset.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

Honda does not use P0170.
Does your Check Engine light come on for two seconds when you first turn the key to "II", and then go off and stay off, even after the engine starts?
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 06/30/2010 03:57 PM, Tegger wrote:

it's not a honda-specific code because it's on the universal obdII list!
http://www.obd-codes.com/trouble_codes/

--
nomina rutrum rutrum

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Just to clarify why I've asked this: spurious codes usually mean the ECM is going bad.
I suggest you pull the ECM, remove its cover, and check for moisture or corrosion inside the ECM case. Your ECM should be behind the passenger-side kick panel, under the dashboard, just ahead of the door.
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 06/30/2010 04:09 PM, Tegger wrote:

better yet, just google for this being a real honda code:
<http://www.google.com/url?q=http://www.hondacarforum.com/honda-technical/43700-how-fix-97-accord-cel-p0170-p0171-codes.html&sa=U&ei=ntsrTKnkL8-HnQe21ZXSCQ&ved BEQFjAA&usgQjCNEh3dQLdkwXksw8YlcXEA93gGhgNQ> <http://www.google.com/url?q=http://www.honda-acura.net/forums/showthread.php%3Ft%3D154620&sa=U&ei=ntsrTKnkL8-HnQe21ZXSCQ&ved BkQFjAB&usgQjCNGKF2_kEqCpmKDp3KDlIrxCZOl87w> <http://www.google.com/url?q=http://www.honda-tech.com/showthread.php%3Ft%3D2739671&sa=U&ei=ntsrTKnkL8-HnQe21ZXSCQ&ved CQQFjAC&usgQjCNH6MmgkihAHqrIfbJs38nca6A2-YQ>
accord, civic, prelude.
the two common responses:
1. replace O₂ sensor.
2. adjust valves.
#1 being the most likely imo.
--
nomina rutrum rutrum

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 06/30/2010 01:19 PM, Tegger wrote:

any "P0XXX" code is from the universal list, not a manufacturer specific list like "P1XXX". P0170 is "Fuel Trim Malfunction (Bank 1)"
http://www.obd-codes.com/trouble_codes/
--
nomina rutrum rutrum

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 6/30/10 3:19 PM, in article Xns9DA7A60E6D371tegger@208.90.168.18,

Honda DOES use P0170. It is the "Temporary DTC" for both P0171 and P0172. This according to the FSM for the '00 Acura TL.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

That's odd.
I have TSB 03-020. This is Honda's official OBD-II DTC list as of March 29, 2003.
The TSB specifically says: "This is a list of all DTCs for all OBD II models. No one model has all these DTCs."
P0170 is NOT listed.
Can you give some details on the "Temporary DTC" thing?
--
Tegger

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Motorsforum.com is a website by car enthusiasts for car enthusiasts. It is not affiliated with any of the car or spare part manufacturers or car dealers discussed here. All logos and trade names are the property of their respective owners.