Yesterday the wife's 1998 4 Cyl Accord EX lit the CEL and it read as P0108. It was running OK although she reported it was stumbling a little when accelerating. It was around 50F and had been raining for days before, in case that might be relevant.

So, today got around to working on it. First pulled the connector off the sensor and verified no corrosion on either side. Put the car to "ON (II)" and measured the voltage relative to ground on all 3 pins on the connector, and none of them were significant (all like 20 mV, but wandering). Not sure how to interpret that. Plugged the connector back into the sensor, started the car, let it warm up. Meanwhile fired up Torque Pro. It didn't have a "MAP sensor" field, but it did have a "Vacuum" display. Hopefully that is the MAP sensor, if not any suggestions as to what other sensor it might be?

At idle "Vacuum" read "-21.96 in/Hg", step on the gas to jump to 3000 RPM and it would got to around "-7 in/Hg" but then rapidly return to -22 (ish) like before. This was with the car in Park, not driving around. The readings seemed reasonable and the motor sounded normal. Cleared the CEL and turned the car back on, rev'd it up and down, and the CEL did not reappear over 10 minutes and a couple more restarts. Have to drive it some tomorrow to see if it reappears.

The car does have one other known problem. If it drops from higher RPM to idle too quickly it will stall. It has been like that for years. My working theory was that that was from the IAC being a little slow to respond. I suppose if the MAF was somehow slow it might be a factor.

Looked at a few P0108 videos and some suggested back probing the sensor connector and measuring the voltages there. That won't be so easy in this case as the wires seem to be embedded in rubber seals of some type. It doesn't seem necessary though if "Vacuum" is from the MAP sensor, since it reads pressure more or less OK.

Thoughts?

Edit: Page 11-48 of the service manual says that the MAP should read "30 in/Hg or higher". "Standard" atmospheric pressure is 30 in/Hg, goes a bit higher for some weather conditions. Torque Pro gives vacuum in negative units which would I guess mean roughly 8 in/Hg of pressure (for -22 in/Hg of vacuum). This suggests the MAP is bad - but only if Torque Pro's conversion from volts to in/Hg is correct. For all I know it is calibrated for a different sensor. I also have an Autel AP200 with a subscription for Toyota. If I pony up for a year's subscription for Honda I can check the MAP values with that. Or maybe I will stick a pin through the insulation in the signal wire (can't back probe it because of the rubber seal) and just measure the voltage directly. That section of the manual gives voltage tests too, and then there will be no possibility of a conversion error.

 

Plugged in the OBD2 dongle, fired up Torque Pro, and watched the values for Vacuum, RPM, Speed, and Coolant Temperature while my wife drove it around. As before, the vacuum pressure was around -22 in/Hg when the car was stopped and idling, and went more positive (more pressure, less vacuum) when the throttle was opened. When the car was "On(II)" but the motor wasn't running the pressure was -.4 in/Hg, which is more or less standard atmospheric pressure. At one point when the car was made to accelerate hard pressure fell briefly to -3 in/Hg. The most negative value observed was -25 in/Hg when the car was moving at 40 mph and my wife took her foot off the gas quickly. In short, pretty much what one would expect. The CEL did not come back on in 20 minutes of driving.

Since the problem initially arose after a hard rain the carpet over the computer was pulled back about 6 inches and the padding underneath was checked for moisture. Dry all the way around. The carpet in the passenger area was dry on top too. I suppose there is a possibility a very small amount of water might have run down a cable to reach where it plugged into the computer. When I have more time I will pull the carpet all the way back and check that connector.

Since Torque Pro was still hooked up, the IAC was "sort of" checked. With the car stopped in Drive and idling, foot on the brake, gas pedal untouched, "Vacuum" read -21 in/Hg. Then the AC was turned on, and that rose to -18 in/Hg. Toggling the AC on/off would move the pressure back and forth between those values (with a little bit of variation in the measurement, like plus/minus around 0.2 in/Hg). So the IAC does open/close, but I don't know if that is the appropriate amount or if the speed at which it does it is fast enough. Perhaps not, because the RPM were a little lower when the AC was on then when it was off. The car still has its one known PITA issue, which is that it will stall if allowed to drop too rapidly to idle. For instance, put it in Neutral, rev to 3K RPM, then quickly release the gas pedal, and the RPM will drop through the ~700RPM idle "floor" and the car stalls. However, in normal driving this never happens.

 

This code reappeared. The car has a bottle of Techron in this tank and since it has not been getting much use, it was taken out for a 1 hour highway drive, mostly at 70-75 mph to let that chemical work and maybe burn off some carbon. No problem, it drove just fine. On returning the car was backed into the driveway, and while slowly creeping backwards (slightly uphill) a CEL appeared and the car immediately began running poorly. Turned the car off, fetched an OBD2 reader from the other car and the cell phone with Torque Pro from inside the house and it once again read P0108. No other codes. Cleared the code but it came right back. Opened the hood and with the car running (poorly) wiggled the connector to the MAP sensor. No difference. Turned the car off. Waited 5 minutes and turned it back on - the CEL was still lit but the car was running normally. Used Torque Pro to clear it again, and it stayed off. The coolant temperature was reading 205F at this point (and falling), which was around 40% on the Temperature gauge, which is typical for the car once it is warmed up. Stupidly I did not try to read the manifold pressure with Torque while the symptoms were still present. Once they were gone and I did look it read normally.

It had rained about 1/2" 4 days ago, and the car had not been driven since then. Last time this code showed up was also after a heavy rain. Guess I need to stop procrastinating and fully excavate the wiring to the computer to see if there are signs of water intrusion or damage there.

Is it possible for a MAP sensor itself to have intermittent issues? This issue seems to miraculously heal itself when the car is turned off for a short time and it doesn't come back for months.

 

Based on your postings, this is what I would do if it was my car........I would unplug the connector to the MAP sensor and see if the car runs OK. -- To my knowledge, even though the CEL will come on, the computer might "compensate" for no MAP input data and interpret a correct fuel mixture using other sensor data. This works on some rides as I have done this for testing on both MAP sensors and MAF sensors.. I haven't tried this on a Honda. If this is not how the Honda works.....I would unplug the MAP sensor when the car is running badly, to see if it makes a difference. If you get a marked improvement in the way it runs, the MAP sensor is the issue....Intermittent failures are hard to diagnose with certainty.

You could also go ahead and replace the sensor and hope for the best. It is not prohibitively expensive in my opinion. 37830-PAA-S00 Shop around for the best price. Aftermarket ones are even cheaper, but I have had some bad luck with aftermarket sensor parts for imports. And yes, the answer to your question is that I have seen various MAP and MAF sensors work intermittently.

 

Thanks for the input. I suspect that the car running without the MAP will be pretty much the same as it runs when P0108 is being generated. As in, just enough to limp to the side of the road, probably.

I think maybe I will buy a new O-ring for it, which is only going to be $5 or so. Apparently the O-rings can crack and cause mysterious problems. Then pull the MAP sensor, clean it with MAF cleaner, and put it back in with the new O-ring. Then wait for the problem to appear again, possibly another month, or hopefully never.

The car has two other known problems:

1. An evap leak which sets the "gas cap loose" code occasionally. Also the car sometimes smells like gasoline near the trunk but that smell isn't very well correlated with the code.
2. If the car is rev'd to 3000 RPM or so, and then the gas pedal abruptly released, the RPM will fall through the idle point and it will stall. No code. It is a peculiar symptom and In normal driving it never stalls. It has stalled on a panic stop from high speed though, just as the car stopped moving.

The first might be involved if it somehow causes an intermittent vacuum leak. I suppose that is possible if there is a crack in a hose or plastic that opens spontaneously on the pathway to the manifold. Seems pretty unlikely that a little leak that far away would have that large an effect.

The RPM crash to stall symptom could actually be MAP related, for instance if the MAP sensor responds too slowly. And it might respond slowly if it was dirty. And it might be dirty because the car ran for nobody knows how long with a split on the bottom of the air intake hose between the air filter and the throttle. The air here tends to be dusty must of the year. That hose was replaced a couple of years ago - which is also when we noticed the RPM crash to stall behavior. Never thought to clean the MAP at that time.

Also it is pouring out right now, so if it is water or humidty related perhaps it will show up when next tested.

 

How does the O-ring fit on this sensor?

I have searched and searched for a cross sectional diagram of one installed, but came up empty. Does the end of the sensor extend below the metal of throttle body and into the air stream, or is it recessed up in the cylindrical hole it passes through? Does the O-ring go between the sensor extension and the bore (down in the bore), between the top of the throttle body and the bottom of the sensor, or in a seating area cut into the top of the throttle body?

The size is 9.8 x 1.9 (mm) which is JIS P10. The two metric O-ring sets I have only have JIS P10A, which is 2.4 mm CS instead. Not knowing if that difference would matter I bit the bullet and ordered a 16075-P07-000 from Honda, around $6 with pick up at the dealer. The usual price for a P10 in bulk is about $0.04. Grrr.

Anyway, from the pictures I did find it doesn't look like there is groove around the tube that sticks down into the throttle body. That suggests rather than sealing the extension in the bore (like a piston in a sleeve) the O-ring just seals the bottom of the sensor to the top of the throttle body. Assuming there is no seating groove in the top of the throttle body, then a P10A will work just fine, in fact, a lot of different sizes would work well. If there is a groove then the exact size will be needed. If there is no seating groove, and the slightly larger CS is used, it seems unlikely that shifting the bottom of a pressure sensor up by .5mm is going to be a problem, even if that brings it up out of the air stream slightly.

 

Finally found a picture showing how the O-ring seats. This is for a CRV of the same era, but probably the same in the Accord. Look at 1:18 here:


The O-ring sits in a circular groove around the hole in the top of the throttle body. It looks like the groove is very close in size to the O-ring, so the P10a is probably too large.

 

(Odd, I posted this but it seems to have disappeared. Second time's the charm?)

Progress, but not in a good way.

Installed the new O-ring, cleaned the MAP sensor with MAF cleaner, reassembled. Drove car until it was warmed up and a few minutes past. CEL came on pulling into driveway. Pulled connector and tried to measure VCC and continuity for the ground on it, with no success, but it was getting dark and I couldn't see if the multimeter probes were making contact. On the plus side I guess it is better that it now fails in minutes semi-reliably than in hours to months randomly.

This morning tested VCC again. Wasn't able to measure VCC in IG(II). Cleaned contacts with CRC cleaner, let it dry. Then was able to see 5V on VCC and continuity from the ground to the battery. Plugged connector back in and started the car. Wiggled the 3 wires and tapped on the connector and it didn't change anything. Drove the car as before. Once it was warmed up, pulled into drive way and held engine at 3K RPM for a couple of minutes, then did it again. Let the car go to idle and suddenly it became rough and the CEL lit. Torque Pro had been running this whole time and the rough idle corresponded to the "Vacuum" display becoming "Boost" and then back again. Repeatedly. Then the car stalled. Turned it off and back to IG(II) (without going to start). Unplugged the connector. Measured VCC against the ground in the connector and it was going from 5V to, I'm not sure what, open I think, the display wouldn't stabilize when it wasn't 5V, and back. Just like the Torque Pro display. So the problem is not the sensor. (Or not only.)

Carefully inspected the cable back for the short distance I could follow it and didn't see any damage.

So, when the car is heated up thoroughly either the ECM goes nuts or there is an intermittent wiring problem. Not good either way.

When the car belonged to the in-laws a tree branch fell on the roof. (About 3" in diameter.) The in-laws told us years later that the driver's side carpet was getting wet every time it rained, but they never did anything about it. When we inherited the car I found that the body shop that "fixed" it didn't reconnect the sun roof hoses. They also didn't get the roof quite flat around the sun roof, and there is a gap in the weather strip at the top of the rear driver's side door because the roof isn't in the right place there. Anyway, reattached the hoses, then put globs of silicone under its outlet to stop water from following the hood release cable into the cabin. It still leaks a couple of table spoons of water near the release handle in a big storm, but the fuse box over there is no longer getting wet. I have never seen water on the floor near the ECM, and as noted in an earlier post, the carpet and padding below it were not wet recently when this MAP sensor issue arose, and after a rain. I'm afraid though that all that water just being in the cabin may have over the years caused corrosion at some of the connectors, or even in the ECM itself. Also the ECM might have been swimming in some instances before I fixed the hoses.

Once the gardeners are done today, and all the dust has settled, I will at least pull off the C connector at the ECM and see how that looks.

 

Removed negative cable from battery. Exposed the ECM and pulled all the cables off. Couldn't see the metal inside the cable connectors but all the pins on the ECU were nice and shiny. No signs of moisture or water damage. With the ECU unplugged also unplugged the MAP sensor. Used a Klein Tools "Tone and Probe" to trace each of the three wires from the Map sensor connector. Tone was detected on cable C at the ECM end (as expected) for all 3 wires, and not detected on any of the other cables. With a multimeter checked that no two of the three wires were shorted to each other, and that none of the three had continuity to ground.

Simplest explanation at this point is that the ECM is failing. It is also possible that some other 5V sensor is drawing too much current, and that is taking out the MAP sensor because they somehow share the 5V line inside the ECM. The MAP sensor itself might do that, but since VCC stays, well, not sure what, but not at 5V, when the MAP sensor is unplugged, it cannot be actively causing that odd voltage at that time. So either the MAP sensor damaged the ECM or some other component is screwing with the 5V. Not getting any other codes though. I have seen plenty of circuits that were damaged when they had to supply too high a current, and once that happens they tend to stay damaged, not work for several minutes and then go bad. The repair manual doesn't have details about the inner workings of the ECM, I need to figure out which other sensors use 5V and unplug them to see if the 5V for the MAP recovers.

 

It just occurred to me that Torque Pro switching from "vacuum" to "boost" for that display probably does not indicate a problem with the ECM reading the value of the MAP sensor signal line, but rather a loss of communication to the ECM as a whole. Because "boost" is what Torque Pro shows when the OBD2 is not communicating, like before the car starts. Once it can talk to the Accord it somehow realizes that it needs to use "vacuum" instead.

 

Triggered this failure again and then pulled off the EGR connector. There was 5V on that connector. Detaching that connector made no difference in how the engine was running, at least once it was already running like garbage and the CEL was lit. Also discovered that "boost" vs. "vacuum" label in the Torque Pro display shifts with the reading from the MAP sensor. When it is reading near zero it can be like +.01 then -.01 (made up example) and that causes the display to flip back and forth. It was not taking the OBD2 dongle off line. Pretty much rules out a general 5V failure in the ECM.

Pulled out the ECM, took off the top plate, and inspected the circuit board. No signs of burning, bulging caps, or the like. I don't have a microscope and so could not inspect the solder joints.

Tried to read the 5V line on the MAP connector with an old analog multimeter and it could not do it. The geometry of the probe was such that it simply couldn't reach the conductor. Now I'm wondering if the faltering 5V reading with the DVM was also because the probe was only marginally making contact. Guess I will go by the nearby junkyard and pick up a MAP sensor. Maybe somehow that sensor really is temperature sensitive? Seems really unlikely. This is beginning to feel more like a temperature dependent vacuum leak.

 

TLDR anything.

Check the engine vacuum with a gauge. Vacuum reading is dependent on your altitude. Is the reading good for your altitude?

if the reading is good,does the MAP sensor agree with your gauge? (MAP sensor readings are often times given in absolute pressure on scan tools, so youll have to convert to gauge pressure)

If not the MAP is probably bad(could be a wiring issue, but that generally will set a different code)

 

Yes, a gauge reading would be good. Tried to hook up the one my Son had and none of the adapters fit the skinny vacuum ports on the intake manifold. Ran around to the local auto stores and found a roll of Dorman 5mm tube. It was loose on that port too. Also the one T he had was too big for the existing vacuum line. So we ordered one of those Amazon kits with a bunch of different size metric tubing and adapters. Should arrive tomorrow and then we can plug it in.

In the meantime picked up a second MAP sensor and a pigtail for it from the junkyard. Today I'm going to solder some gold plated pins onto the ends of those wires, and then shrink wrap a little so they can't bump each other and short out. Already established that those pins make a solid connection when inserted into the connector. Will first use those to test the junkyard sensor (computer 5V supply and a syringe for vacuum). With this modified pigtail plugged in between the sensor and the car's wiring it will finally be possible to get reliable readings for the sensor voltages. That will nail down whether or not the computer supplied 5V is really dropping out.

Another possibility might be a gross leak on the far side of the evap purge valve. The problem only shows up when the car is up to temperature and (probably) in closed circuit mode. That is when most cars will open the purge valve, and is roughly when the P0108 shows up (at least now). The car has some sort of evap problem already, it throws a P1456 occasionally (although not lately), if that was a leaky hose or fitting perhaps it has progressed to a larger hole? Odd though it would trigger P0108 and no evap code.

 

99% sure it was the MAP sensor. I have not yet taken it for a long drive yet, but it was running fine in the driveway, even though warmed up and run at intervals at 3K RPM for 30s at a time.

Today's work:

1. Hooked up the pigtail with the gold plated pins soldered on, between the MAP sensor and the car's MAP connector.
2. Attached one side of a DC voltmeter with a sprung hook probe to the VCC1 gold plated pin (pulled slightly out of the connector) and the other side of the voltmeter to body ground.
3. Attached the Vacuum gauge through some little bits of tubing and a T at the fuel pressure sensor vacuum port.
4. Ran the car until it was warmed up, rev'd to 3K RPM a bunch of times, let it idle, eventually it went to very low RPM and ran rough (CEL was already set, it probably would have lit then otherwise).
5. Observed at all times: Vacuum = typical, VCC1 = 5.0V.

This ruled out both a vacuum leak and the ECM 5V failing. Apparently the previous instances of "no 5V" were actually "multimeter probe not contacting reliably". The geometry of that connector are such that the tips of these probes can only barely make contact with the conductor inside (and another multimeter's probes couldn't contact them at all). The gold plated pin resolved that.

Swapped in the Denso MAP sensor from the junkyard, off of a 1999 Accord. Reused the new O-ring which was put on a few days ago.

The car started up and ran normally.

The replacement MAP sensor also fixed the "rev to 3K then release gas pedal and it stalls" issue:

https://honda-tech.com/forums/honda-...talls-3366772/

Interestingly the first few times I tried that test it still overshot and went down below 600 RPM but recovered instead of stalling. It seems to have quickly learned from that and afterwards it dropped smoothly to 600 RPM.