Nissan Forum banner

21 - 40 of 47 Posts

·
Registered
Joined
·
35 Posts
Discussion Starter #21
Well, I’m still at a loss

VACCUM TEST:

So here is a better vacuum test. I connected the vacuum to the brake booster, but before the one way valve.
In the video you can see the rough idle:
@ 0:05 Near stall
@0:38 Near stall
@0:53 & 1:30 I revved up to 4000RPM
@1:55 Briefly opened throttle ½ way
@2:42 Near stall
This is the first time I do a vacuum test. Does the little vibration in the needle qualify as unstable vacuum?
The tinging/sputtering vibration noise you hear in the video is not the exhaust but noise from inside the vacuum gauge, perhaps the needle fluctuating?

My Vacuum test: https://youtu.be/g7GcgtUDa9o


This next video shows idling only. As I noted down the times when a near stall occurs, I see they occur almost every 30 seconds regularly. Perhaps that has some significance?

My vacuum at idle only: https://youtu.be/oBqMvjeotjI


COMPRESSION TEST:

Seeing whatever needle vibration I saw above, and seeing the paper sucked into exhaust in the “paper by exhaust” (like this guy shows on youtube: https://www.youtube.com/watch?v=PI5bBV0ZFek), and reading that it may indicate sticky valves, I decided to do a compression test.

Test was done with all spark plugs removed, throttle full open, about 7 turns of the engine (cranking to 15 turns seems to yield 3 PSI more):
Cyl 1: 162 PSI
Cyl 2: 162 PSI
Cyl 3: 158 PSI
Cyl 4: 158 PSI

I tried adding about 1tsp of oil to Cyl1 and got 170 PSI.

Nissan specs are: Standard: 178 PSI. Min: 149 PSI, Max acceptable difference between cylinders: 14 PSI.

So looks ok to me. No sticky valves? No bunt valves, no major ring and seal issues?

LOOK FOR BLOWN GASKET:

Oil looks fine, not milky
I opened the radiator and started engine cold. No coolant came out.

FUEL PRESSURE TEST:

I imagined that fuel supply issues would occur at high power, not idle. But since I have now pretty much checked everything, I went ahead and did a fuel pressure test.
I Td a fuel pressure gauge after the filter and got:
36 PSI with vacuum hose attached to fuel pressure regulator (Spec: 34) OK
45 PSI with vacuum hose NOT attached to fuel pressure regulator (Spec: 43) OK

So, all systems are checking out ok

I now do get a pending P0171 (lean) code once in a while after letting it idle for a while. I also got a P0300 but I’m not sure about the conditions since I was doing all these tests with vacuum, compression and fuel pressures.

In any case, when I took out the spark plugs to do the compression test, here is how they looked (engine had been rough idling for about 10 minutes before I took them out).

I cannot see any unevenness, implying something might be wrong with one or more cylinders.


So what is going on?

As a reminder I have also:
-Changed the MAF (ok with cheap non OEM part, since my original was noisy)—disconnecting MAF does not fix the rough idle anyway.
-Cleaned the Idle Control Valve and checked operation (spring by itself closes valve only ¾ of the way, but I assume that is normal?)
-measured the TPS resistance for linearity and the TPS switch
-Checked the fast idle settings (thermos-mechanical opening of throttle, throttle completely closed once engine warm)
-Checked the manual portion of the idle air (screw on throttle body) per shop manual procedure (If I repeat the procedure per shop manual I end up in the same ½ turn of the screw open-same as it was)
-Checked ignition timing per shop manual procedure (look for 20 deg BTDC with TPS disconnected) -- in case timing chain had skipped.
-Hand pumped the EGR valve at idle to verify that it did indeed cause engine to stall—in case EGR stuck open at idle.
-Changed front O2 sensor—in case it was about to fail and giving erroneous lean
-Put Techron in gas tank—to clean injectors
-Opened injectors to look for dirt—very little found
-Checked for vacuum leaks (not very thoroughly but vacuum measurements look good, don’t they?)

So what’s left:

The only thing I have not checked is ignition. But obviously I do have spark, so any static or simple continuity tests on ignition and coils are unlikely to reveal anything. Besides what sort of ignition issue is it that shows up only at idle? I yanked on cables, banged on distributors, disassembled distributor cap, nothing, no difference. Any ideas on the ignition?

A mystery remains, that “Paper by the exhaust test” which at my exhaust does indeed show the paper being occasionally sucked towards the exhaust pipe when engine gets into its funky idle spurt (by is this cause or effect of the fact that engine is in near stall?)

Driving me nuts !!

As an added irritation, I have to pass a smog test by the end of the month…
 

·
Sup Mod keeping the peace
Joined
·
6,375 Posts
So here is a better vacuum test. I connected the vacuum to the brake booster, but before the one way valve.
In the video you can see the rough idle:
@ 0:05 Near stall
@0:38 Near stall
@0:53 & 1:30 I revved up to 4000RPM
@1:55 Briefly opened throttle ½ way
@2:42 Near stall
This is the first time I do a vacuum test. Does the little vibration in the needle qualify as unstable vacuum?
The tinging/sputtering vibration noise you hear in the video is not the exhaust but noise from inside the vacuum gauge, perhaps the needle fluctuating?

My Vacuum test: https://youtu.be/g7GcgtUDa9o

I now do get a pending P0171 (lean) code once in a while after letting it idle for a while. I also got a P0300 but I’m not sure about the conditions since I was doing all these tests with vacuum, compression and fuel pressures.

In any case, when I took out the spark plugs to do the compression test, here is how they looked (engine had been rough idling for about 10 minutes before I took them out).

I cannot see any unevenness, implying something might be wrong with one or more cylinders.


A mystery remains, that “Paper by the exhaust test” which at my exhaust does indeed show the paper being occasionally sucked towards the exhaust pipe when engine gets into its funky idle spurt (by is this cause or effect of the fact that engine is in near stall?)
The small amount of rapid vacuum gauge needle vibration at idle shouldn't be there which might indicate sticking valves but more so dirty fuel injectors. Removing the injectors and looking at them doesn't prove that they are dirty or not. The dirt would be inside that you can't see.

You may have leaking fuel injectors or dirty injectors causing an uneven spray pattern. To test them, first disconnect the ignition wire from the coil, then unbolt the fuel rail from the intake manifold and pull off the entire assembly. Keep the fuel hoses still connected and don't remove any individual injectors. Also keep the electrical connectors on the injectors. With the assembly now away from the intake, turn the ignition key to the run position WITHOUT STARTING THE ENGINE. Now observe each injector to look for leaks. There should be no drips.

You can follow this up by having someone trying to start the engine while you observe each injector spray pattern; the spray pattern for each injector, of course, should be identical and uniform.

Make sure you have at least 4 spare insulator rings for the injectors in case some are damaged during rail removal
 

·
Registered
Joined
·
35 Posts
Discussion Starter #24
Thanks. I put the injector test in the queue… as far as pinpointing a particular cylinder at fault, no the only codes I have gotten occasionally are P0171(lean) and P0300 (random misfire) and neither compression tests or spark plug appearance show any difference between cylinders

... meanwhile,

SMOG TEST PASSED:
Well, it passed the smog test. So now at least I have time to fix this.
Here is a snapshot of the smog test results:


OCCASIONAL P0171 and P0300:

Engine still works well, including reasonable idle while cold and well after it has reached operating temperature.
But once it has been running for about half hour, the rough idle starts.
I’ve been monitoring for codes and when it starts rough idling, within about 5-10’ I often get a pending P0171 (lean), but not always. Typically I don’t get the code again in subsequent trips so Engine light typically never comes on and the pending P0171 quickly clears (one or two trips, I’m not sure). Sometimes, together with the P0171 I’ve gotten also an occasional P0300 (random misfire).

drive with MAF DIDSCONNECTED test:

Another interesting fact: Rough idle is there even if I turn on the AC. However, if I disconnect the MAF then the rough idle is still there, but less than with the MAF connected. More interestingly, with the MAF disconnected when the AC compressor is on the idle is now steady at about 1000RPM.
 

·
Registered
Joined
·
43 Posts
"...as far as pinpointing a particular cylinder at fault, no the only codes I have gotten occasionally are..."

Yet according to the video the truck seemingly misfires at idle. If it were me I'd yank the spark plug wires off one at a time, and ground it, start the truck, and see if I could isolate the misfire to a particular cylinder. That's pretty much what the self diagnostics does: a cylinder balance check.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #26
Well, I did capitulate and have taken the truck to two more garages now. But no success. Nobody seems to be able to find anything.

I guess the dealer is next in line to prove their worth. Hopefully I will not have to just pay for overpriced attitude.

To their credit, the two garages did not want to charge me anything since they were not successful, though I insisted on paying them $40 for the one hour or so they spent making an honest effort.

If I heard correctly, one of the mechanics told me that my Long Term Fuel Trim is at about -10% while my Short Term Fuel trim is about +24% (almost maxed out, I guess?). That seems puzzling to me. At a certain RPM after a while the LTFT should start adjusting so that STFT goes back down closer to 0. Not?

I'm thinking I should buy one of those ODB2 bluetooth or wireless tools and then download one of the many available programs (I heard Torque was good?) to verify these things myself. Anyone has any experience with those?


The garages also double checked many of the items I had originally investigated, including smoking the whole system in search of vacuum leaks but did not find any. In one mechanic’s opinion, the lean condition could cause the vacuum needle to vibrate a bit, there is not necessarily something wrong with the cylinders, valves or gasket, he said, things that were more or less excluded by the other tests anyway.
 

·
Registered
Joined
·
3,193 Posts
This is one of the bluetooth OBD adapters I've got (have 3, all in the same style case, different labels, etc.). Can find them on Amazon as well.

http://www.ebay.com/itm/-/310678002251?ssPageName=ADME:L:OU:US:3160

"Torque" is good enough. I use the free "Lite" version. Works for me.

Not sure about the Nissan's, as I haven't had the opportunity to try it, but... If I clear the codes on my Vibe, the LTFT and STFT values get zero'd out and the ECU basically starts from scratch. Even the idle gets stupid and I have to get it out on the highway to let the 'throttle by wire' reset itself.

You're thinking right about your LTFT and STFT trims. But, if you haven't driven the vehicle long enough, the ECU won't have a chance to do what you're wanting. Takes awhile to get the LTFT trims to adjust themselves. The STFT gets updated almost immediately based on O2 sensor feedback, while the LTFT gets adjusted only after a complete drive cycle has been completed and even then only after X amount of time, miles, etc. has gone by.
In the absence of of really good hard data and all other things being equal, your STFT does suggest a lean condition, while the LTFT could just indicate an older engine.
Bouncing needle on the vacuum gauge--It's a 4 cylinder engine. If you've got an 'undamped' vacuum gauge, the needle is going to bounce at idle. Just the way it is. You end up kinda smoothing the needle out in your mind.

Next step might be to pull the injectors and do an injector leak down check. Pull the injectors out, put a cup under them, turn the key on/off a few times (pressurize the fuel system), let it sit, see if the injectors drip into the cup. That'll rule in/out a leaky injector.

After that, you get a fuel injector pulse tester. Basically, you pull the injectors one at a time, hook up a fuel pressure gauge to the rail, hook up this pulse tester, turn the key on/off a few times (again, pressurize the fuel system), pulse the fuel injectors with the tester, which pulses the injectors X amount of times for a known time, and you read the fuel pressure gauge to see how far the fuel pressure dropped. Each injector should drop the fuel pressure by X amount of psi, within a certain tolerance. Any injector that doesn't drop the fuel pressure enough isn't shooting enough gas. Any injector that drops the pressure too much is shooting too much gas.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #28
Thanks jdg. I ordered something similar waiting on it…

Sorry for the long silence. Had to get some business travel out of the way.

Took the truck to the dealer first time Friday. But their diagnosis seems vague to me and leaves a lot unexplained.
Here is how it went:

I talked to the master tech, gave him a two page summary of symptoms and what I had already done and spent 10’ explaining verbally.
As I was talking he told me I should have done an “idle learn” after changing the MAFS. But I don’t think this ’98 Frontier has an idle learn procedure. I did not find one mentioned in the Service Manual. In any case, I cannot expect techs to know by heart all the models and years that have an idle learn procedure. He did not mention doing idle learn himself afterwards.

He later called me and thought it may be a short or something of that nature. He said the MAFS signal was really noisy “I have never seen it that bad, and wonder how does this truck even run?” were his words. He was then nice and sent me the diagnostic page through email. Here is what I got:



BTW, I’m not sure I understand the two middle signal traces (especially the Air/Fuel Alpha, isn’t that value too low? ), so if someone has something to add I’d appreciate it.

Sure, looking at that first trace, seems like there’s virtually *NO* MAFS signal. That trace looks to me more like the signal from Eddie Van Halen’s guitar amplifier.

So first mystery is how can the truck even run without major drivability issues (other than a rough idle) with a MAFS signal like that? (No MAFS codes from ECM)

Second, this does not corroborate at all with my measured MAFS signal. Below are the videos showing my voltmeter measurements. Of course it could still be that the noise is so high frequency (looks to me in the order of 10HZ ~100msec period) that I cannot pick it up with my analog or digital voltmeters. However, as you can see in the video, the MAFS seems to respond to changes in engine load as expected and is also within voltage specs. I also blew into the MAFS with ignition on, and it does indeed seem to respond as expected.
I’m thinking that the noisy signal trace I got from the dealer shows a MAFS that should be totally unresponsive, not what I observe.

In any case, here are my videos. To rule out problems in the wiring harness, I took out the ECM and measured the MAFS signal directly at the ECM input, as shown here:


First is video of measurement using a small analog voltmeter:
https://youtu.be/vdCYtiSyeIc

As you can see my video shows:
@idle my MAFS signal = 1.3V (spec is 0.9-1.8V)
@2500RPM no load my MAFS signal = 1.9V (spec is 1.9-2.3V)
@3500RPM no load my MAFS signal = 2.3V
The signal seems to increase linearly with increasing RPM – as expected.

In this second video I repeated the measurement with a digital voltmeter:
https://youtu.be/-jb_MRQKADM
@idle my MAFS signal = 1.40-1.45V (spec is 0.9-1.8V)
@2500RPM no load my MAFS signal = 1.97V (spec is 1.9-2.3V)
@3500RPM no load my MAFS signal = 2.35V
Snapping the gas (higher engine load) brings my MAFS signal up to 3.5V


As an extra test I just turned on the ignition (engine not running) and just blew into the MAFS hole in the throttle intake. As you can see the airflow registers as expected.
https://youtu.be/5ylCFTawZtk

The dealer’s suggestion was to replace the MAFS with a genuine Nissan at over $500 – just to continue testing. I already paid him $150 for this diagnosis.

I’m not expecting this cheap MAFS I put to be that good. I’m sort of expecting skewed long term fuel trims as the MAFS is likely to have a different airflow/signal characteristic. But I do expect the truck to run and idle with the fuel trims taking up the slack. My hope was to confirm that the original MAFS was the culprit and THEN go ahead and try to find an OEM part. But the evidence makes me doubtful that the MAFS is indeed the cause of my problems.

The biggest discrepancy is that I see nothing like what the dealer sees in the MAFS signal. Of course he’s (presumably) measuring the MAFS as the ECM sees it, while I’m measuring the signal at the ECM input. If there is a disparity indeed, that would point to a bad ECU. An ECU whose only malfunction is to misinterpret the MAFS signal. What is the probability of that? Even taking into account the fact that I should start looking at oddball things at this point.

I guess next thing would be to borrow a scope again and see if I can record the MAFS signal the dealer sees.

P.S. Will I be able to see traces like the one the dealer sent me with torque? Or at least quickly updating real time values?
 

·
Registered
Joined
·
3,193 Posts
Took the truck to the dealer first time Friday. But their diagnosis seems vague to me and leaves a lot unexplained.
Here is how it went:
Overall, in this post, looks like the guy is actually trying, vs. trying to pull the wool over your eyes.

Second, this does not corroborate at all with my measured MAFS signal. Below are the videos showing my voltmeter measurements. Of course it could still be that the noise is so high frequency (looks to me in the order of 10HZ ~100msec period) that I cannot pick it up with my analog or digital voltmeters. However, as you can see in the video, the MAFS seems to respond to changes in engine load as expected and is also within voltage specs. I also blew into the MAFS with ignition on, and it does indeed seem to respond as expected.
Gotta throw this out there.
A 10HZ (10 times per second) signal would be a signal that runs about 600 times per minute, much like a single cylinder engine running at 1200 rpm would be ingesting a gulp of air 600 times per minute.
See where I'm going with that?
If the MAFS was really sensitive, then that trace looks 'normal' as it is actually registering each and every 'gulp of air' when an intake valve opens up.
Now do I think that's realistic? No, probably not. A chunk of wire in a hot air style MAFS can't possibly change temperature that quickly. Certainly is plausible though.

P.S. Will I be able to see traces like the one the dealer sent me with torque? Or at least quickly updating real time values?
No. Your ECM probably runs on the ISO9141-2 protocol (most '98 Nissans do).
Assuming you turned off every other parameter and only leave the MAFS input on, at the very most you'll get around 10x/second updates, on a good day.

I think the MAFS is a red herring and leading you down the wrong path. Sure, could be the root of the problem. I don't think so this time. Especially since you've gotten the same results with 2 different ones.

Am I to assume the fuel injectors themselves haven't been checked yet? Leakdown, flow rate checks, etc.?


Also, on a side note, if you take those meters, put them on DC, check the voltage, then put the meter on AC and check the voltage again, you'll be able to tell approximately how much your voltage is bouncing.
For instance, if you read 1.4V DC at idle, then switched over to AC and got .2V, you can be relatively confident that you signal is actually a jumpy signal bouncing between about 1.2V-1.6V.
It won't be exactly those values as most meters don't work over a wide range of frequencies in the AC mode (unless you've got a good Fluke or other brand name), but it'll give you a ballpark. Similarly, if the reading on the AC scale is close to zero, you can also be confident that the signal is a steady one that's not varying by a whole lot.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #30
I JUST CANNOT SEE ANY NOISE ON THE MAFS SIGNAL WITH A SCOPE:

I went ahead and measured the MAFS signal entering the ECU (pin 54) with a scope.
I cannot see anywhere the noise that the dealer sent me (see my previous post).
Here is my scope trace for MAFS signal entering the ECU:



The scale is 2seconds per square the voltage 0.5V per square. It shows me revving up the engine from idle.
I cannot tell if the voltage levels are correct (they are within spec), but the behavior is as expected. You can see how the signal surges by depressing the accelerator. Certainly I do not see the noise. The signal looks quite smooth. Definitely nothing like what the dealer sent me ? ! ?

Here is one of the videos I took showing the scope trace:

https://youtu.be/dUm6E1512tU

And here is a video of the MAFS signal I made during rough idling. You can actually see how as the idle drops below the pre-programmed 850RPM, the ECM opens the IACV (I can hear this when the air filter is off) in an attempt to bring the idle back up (presumably injects more fuel too) and the increased airflow is seen in the increased MAFS signal.

http://youtu.be/r4VBFGXGxok


INDIVIDUAL CYLINDER AIR INTAKE VISIBLE?

Jdg, your intuition on individual cylinder air intake being picked up by the MAFS might be correct. Interestingly, the only high frequency I see is exactly when I snap the gas -- when the butterfly opens and the individual cylinder air gulps become large -- then I can see oscillation in the signal. I think these might indeed be the individual “gupls” of air intake from the cylinders, since the frequency of the oscillation is 2x the RPM. So I guess this hot wire MAFS seems sensitive enough to pick up some of these high frequency airflow variations !?! If anything, being a cheap one, I thought it would be slow in responding…

Here is the MAFS signal trace when snapping the gas:



And here is a similar video but with a fast trace, where you can see the signal oscillations when snapping the gas:

http://youtu.be/8saghlBiZYg

But other than those brief periods where the gas is snapped, I see no noise on the signal. Certainly nothing like what the dealer shop sent me.

So bottom line…

If the dealer indeed sees noise in the MAFS signal when he reads it from inside the ECM while I don’t by reading it as it enters the ECM, there are three main possibilities:

a) I am measuring wrong, but serendipitously read something that looks exactly like a reasonable MAFS signal
b) The dealer is reading the MAFS wrong from the ECM
c) The signal is ok entering the ECM but there is noise that scrambles it in the internal ECM signal conditioning, or the analog to digital converter, or something like that

a) seems quite unlikely. So does b). As I understand it, there is nothing to connect that the dealer tech might have gotten wrong. You just connect to the NISSAN consult port and the Nissan diagnostic just reads the MAFS digital value from some predetermined location in memory. Does not seem like there’s much opportunity for error.

c) would imply a bad ECM. A bad ECM where the only malfunction is scrambling of the MAFS input signal. Perhaps I’m biased but I’m also very skeptical of diagnoses that blame the ECM.

In any case,
To test this last scenario c) I wondered if I could just feed an artificial constant MAFS signal to the ECM. If the ECM is scrambling the signal then it should scramble my constant voltage too, and continue to run rough, or worse. So I just used a potentiometer as a voltage divider to create a constant voltage of about 1.7V (what the MAFS typically puts out at idle). I disconnected the MAFS, and fed the 1.7V signal directly into the ECM MAFS signal input (pin 54). THE ENGINE IDLED BEUAUTIFULLY. If the signal were getting scrambled, I should still have the rough idle, shouldn’t I? Of course with this artificial signal you cannot do much other than idle. As soon as you press the accelerator the engine stalls. Poor ECM sees no change in the airflow, probably does not increase the fuel and thus with throttle open the engine stalls. I’m pretty sure the ECM gets thoroughly confused by that.

Another thing I feel this proves is perhaps the absence of anything mechanically wrong with the engine, since I got it to idle beautifully by simply feeding it an artificial constant MAFS signal.

WIRELESS OBD2 READER:
Anyway. I got the wireless OBD2 reader today. This one:


I have not connected it yet. I was hoping to get it to work with my iPhone but none of the 4 free apps I tried seem to work that well (dropping data etc). I’ll keep looking or may have to resort to using laptop computer. I don’t mind buying an app for this, but I first want to see the limited functionality demo version work, which has not been the case yet.

Yes, nobody has looked at the injectors yet.

This is becoming quite the intriguing problem, I must say…
 

·
Registered
Joined
·
3,193 Posts
You won't see any noise at such a slow sample rate. The earlier 'scope shots were sampled at 100Hz. You're sampling at .5Hz. Won't see anything there except an overall snapshot. Noise won't show.

I cannot tell if the voltage levels are correct (they are within spec), but the behavior is as expected. You can see how the signal surges by depressing the accelerator. Certainly I do not see the noise. The signal looks quite smooth. Definitely nothing like what the dealer sent me ? ! ?
See note above.
As far as voltage levels being correct, they can be off quite a bit and the computer will compensate up to a point (fuel trims).

Jdg, your intuition on individual cylinder air intake being picked up by the MAFS might be correct. Interestingly, the only high frequency I see is exactly when I snap the gas -- when the butterfly opens and the individual cylinder air gulps become large -- then I can see oscillation in the signal. I think these might indeed be the individual “gupls” of air intake from the cylinders, since the frequency of the oscillation is 2x the RPM. So I guess this hot wire MAFS seems sensitive enough to pick up some of these high frequency airflow variations !?! If anything, being a cheap one, I thought it would be slow in responding…
I was guessing. Could be that MAFS's are fast enough to catch each intake stroke. I've never 'scoped one myself, just watched the values coming out of the OBD port.

But other than those brief periods where the gas is snapped, I see no noise on the signal. Certainly nothing like what the dealer shop sent me.
...because you aren't sampling fast enough. If your 'scope is capable of it, bump up the sample rate, do a capture, eventually you'll see the same 'noise'.

a) I am measuring wrong, but serendipitously read something that looks exactly like a reasonable MAFS signal
b) The dealer is reading the MAFS wrong from the ECM
c) The signal is ok entering the ECM but there is noise that scrambles it in the internal ECM signal conditioning, or the analog to digital converter, or something like that
a - looks like a good MAFS to me
b - mechanic sounds like he's got good intentions with bad execution.
c - doubt it. If that was the case, it would run like serious crap at high revs.

a) seems quite unlikely. So does b). As I understand it, there is nothing to connect that the dealer tech might have gotten wrong. You just connect to the NISSAN consult port and the Nissan diagnostic just reads the MAFS digital value from some predetermined location in memory. Does not seem like there’s much opportunity for error.
Except for the mechanics interpretation of the data.

In any case,
To test this last scenario c) I wondered if I could just feed an artificial constant MAFS signal to the ECM. If the ECM is scrambling the signal then it should scramble my constant voltage too, and continue to run rough, or worse. So I just used a potentiometer as a voltage divider to create a constant voltage of about 1.7V (what the MAFS typically puts out at idle). I disconnected the MAFS, and fed the 1.7V signal directly into the ECM MAFS signal input (pin 54). THE ENGINE IDLED BEUAUTIFULLY. If the signal were getting scrambled, I should still have the rough idle, shouldn’t I? Of course with this artificial signal you cannot do much other than idle. As soon as you press the accelerator the engine stalls. Poor ECM sees no change in the airflow, probably does not increase the fuel and thus with throttle open the engine stalls. I’m pretty sure the ECM gets thoroughly confused by that.
That's a helluva bit of troubleshooting :D
If you're thinking that a rotten input signal is the problem, then put a low pass filter inline with the MAFS signal...
MAFS signal -> 100 ohm resistor -> ( 10uf capacitor to ground ) -> ECM (kinda looks like a T, resistor goes before the capacitor, - on the cap goes to ground).
100 ohm resistor and 10uf cap gives you a rolloff a bit below 100Hz dropping to practically nothing above 1Khz.

Or...try wiring in a hard ground wire from the MAFs connector to ground. The ground wire might meter out good (eg. close to zero ohms, etc.), but might be hanging on by one single conductor in a piece of stranded wire somewhere and won't carry any load. For that matter, might want to try hard wiring in the signal and power wire as well.

While I'm thinking about it, ever build a wire load tester using a tail light bulb?
If you end up thinking it's a rotten wiring problem, get a small 12v battery, disconnect the wire at both ends, and try to run enough juice thru the suspect wire to make the bulb light up. Like I said, a wire might continuity test good, but wouldn't be able to flow any current. Sorta like a garden hose that's almost completely plugged. It ain't leaking, but you can't get any water thru it.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #32
jdg, thanks for hanging on with this long thread. More information is coming.

Regarding the scope trace, the time shown (2seconds to 500msec) is the scale, I.e. the time per square on the display, not the sampling rate. The sampling rate is much higher. So you would see noise even on a slow trace. It would just all smudge together.

I did the ground current carrying capacity test (tail light bulb) you mention for the MAF signal shielding (see post #13). I have not done it for the MAF ground itself.

GOT SOME NEW DATA WITH OBD2 WIRELESS DONGLE:

I gathered quite a bit of sensor data on a few trips using the new OBD2 dongle in combination with the OBDCarDoctor free app on my iPhone. It is very useful.

I’ll have to process the data and then I’ll post it. I may have finally isolated the issue, but there are also a lot of things I cannot explain… could use some help… let me process the data and I’ll post again…
 

·
Registered
Joined
·
3,193 Posts
So you would see noise even on a slow trace. It would just all smudge together.
That's what I was getting at. Can't see the trees in the forest.
We're on the same page here.

I gathered quite a bit of sensor data on a few trips using the new OBD2 dongle in combination with the OBDCarDoctor free app on my iPhone.
Again, since the '98 is running the ISO9141-2 protocol, even at the highest sampling rates, assuming you're only pulling one single parameter, on a good day, you won't get much more than 10/second. Not that that's a bad sample rate, but it sure ain't 1000/second.

It's gonna turn out to be something stupid like a tail-light bulb loose :D
 

·
Registered
Joined
·
3,193 Posts
That there is some troubleshooting data right there...yep...that's what I call it.

More later on your data. Long night at work.

Yep, not a lot makes sense up there in the captures, except when it does make sense.
I think you're looking at rotten wiring. The spiking on the MAFS signal is acting a bit like a what I used to get in fuel quantity systems on aircraft with broken signal wires or rotten shields. All kinds of goofy stuff. And if you put a meter on it or a 'scope, the signal itself would magically (capacitive coupling) go up and down by itself with zero inputs (either mechanical or electrical) otherwise.
I think my next step would be to let it sit and get overnight cold. Go out the next morning, hook up the OBD/datalogger, turn on the key and let it sit for an hour. See if the MAFS magically starts spiking by itself. My thinking is that there's a single rotten wire in the harness that's getting warmed up, possibly expanding, shifting, moving somewhere somehow.
After that, next step might be to cut the MAFS wires, one at a time, and splice in your own wire. Start with the grounds/shields, then powers, and do the signal last.
Maybe easier than that would be to add a 100K resistor across the signal wire to ground. If you're getting anything capacitive shooting thru to the signal, that should be enough to dampen it out. If not 100K, maybe a 10K.
Analog-Digital converters (such as the types inside ECMs) are high impedance inputs, which means they hardly draw any current and it's very to induce a bit of 'extra' current into a wire and make the voltage spike upwards (or downwards).
 

·
Registered
Joined
·
3,193 Posts
Notes on the graphs...

Startup/warmup...
LTFT not 0 - just reset codes ... Resetting codes doesn't necessarily reset the fuel trims. I used to think it did too. Apparently not the case.

Warmed up/idling...
No idea why the STFT is ramping up and down. The only thing that comes to mind is when it hits a limit (eg. 25%), it triggers a change in the LTFT, and the STFT resets to 0. And O2, yes, very lean, but may not be warmed up fully yet.

Pulling out/idling at stoplight...
The only other halfway reasonable thing I can think of with the ramping is the above explanation, or possibly that the ECU is trying to complete a 'drive cycle' (checking whether or not the O2 sensor actively responds to programmed changes in fuel delivery).

Continued driving & on highway & still on highway...
Looks normal there.

Still on highway #2...
Yes, deceleration = fuel cutoff as long as RPM (and I'm sure a few other conditions) is above X amount of RPM.

32 minutes in...
Maybe the ECM (and you) has completed a 'drive cycle' and the ECM thinks everything is 'ready' for anything (eg. everything has been tested).

After that...MAFS spiking...
I'm outta ideas other than the rotten wires thing.
Wouldn't it suck if you got 2 identically bad MAFS? Not totally out of the realm of possibilities.

Get an average reading of your MAFS output from OBD (in grams/second) after you fire it up, before it starts to go nuts. Let's just say it's around 10 g/s (just throwing numbers here, my example is NOT the correct number).
When the MAFS goes nuts, is it mainly dropping towards 0 or spiking up towards some ungodly number, as based on your initial observations of an idling MAFS value?
I can't tell what your actual MAFS numbers are from the pictures, and photobucket isn't any help either.

One other thing that makes me think you've got bad wiring (or another bad MAFS) is the fact that when you blow into it, it quiets down for a bit. The fact that you had to handle something...air filter, the MAFS itself, wiring, tubing, whatever, tells me that something got moved somewhere...thereby leading me in the direction of the rotten wires.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #37
Thanks jdg, those suggestions were quite helpful.

ELECTRICAL CABLES:
Now that I have the scan tool with graphing capability I did try to yank, shake and disturb any cable I could, bang on the dashboard, shake the truck, the ECM etc. … while observing the MAFS signal through the OBD interface. No noise or disturbance was ever noticed. I repeated the test by placing electrical tape over the MAFS to make sure I had a steady signal, so that I could more easily spot any perturbation in the signal. If you don’t put tape over the MAFS opening you just see variations in natural air movement (eg. a breeze).

THERMALLY SENSITIVE MAFS?
So, unable to blame the cabling I pursued the theory of a heat sensitive MAFS. So I took the MAFS out of the throttle intake, left it hanging by its wires, turned on the ignition and used a hairdryer to start heating up the MAFS, while monitoring the signal with the OBD2 app. I was hoping to see some noise develop at some point. The MAFS opening was taped, to keep the hairdryer airflow out of the signal.

So, in about 15’, I heated up the MAFS to probably 160-170F (the temperature I thought it develops during driving) and waited 15 minutes, but I saw no noise. I decided to heat it more (did not care about destroying a $30 MAFS at this point). So I put the MAFS in a glass jar (to keep the hairdryer airflow from dispersing) together with a BBQ thermometer. Temperature inside the jar climbed to 230F. I really don’t think it heats that much normally, but what that heck, I thought, enough trouble already. I must have waited a good 30 minutes warming up the MAFS, …and then finally voila the spiking signal started!!
I verified for about 10’ that what I was seeing was indeed spikes and noise. Yes it was!. Airflow as seen by ECM was spiking between 0 and 12gr/sec even though the sensor was blocked with electrical tape!! Then I let the MAFS cool. But it kept spiking even as it completely cooled!! I must have taken no more than 10’ for the MAFS to cool, but I waited and it kept spiking for a good 30’ – spikes with MAFS cold now!!

I thought, “Could it be something else that is warming up and causing the spikes, perhaps a wire (jdg, your intuition about just turning on the ignition and waiting was quite insightful!) or the ECM?”The temperature I had to reach to get MAFS to spike seemed unreasonable, and the fact that spiking remained well after MAFS had cooled made me suspicious that perhaps some other element is heating up and causing the spikes instead of the MAFS itself.

So, without turning off the ignition and while this new MAFS was still spiking, I swapped it with the old Hitachi MAFS. No heat. I waited about 5’, got no signal spikes.

So, I put back the new MAFS which by now had not only cooled down completely but had also been disconnected for 5’. No heat. Within 1’ the new MAFS started spiking again!!!. I watched it spike for another 3’ and then,

I swapped it again for the old Hitachi MAFS. I waited about 8’ but did not see any noise/spikes. So,

I put back the new MAFS, which had spent even more time cold and disconnected. Within a few minutes it started spiking again. Stone cold and spiking!! (to be exact, not completely cold, it was about 10F above ambient temperature – I think the hot wire and circuitry inside the MAFS generate enough of their own heat to keep the device a little above ambient temperature). I observed this (new) MAFS spiking for another full hour.

As a last test, I put back the old Hitachi MAFS and waited one hour and 45 minutes. I saw no noise/spikes.

I did though verify the defect of the old Hitachi MAFS, whereby pressing on the housing generates noise on the signal. But it’s a completely different type of noise than the spiking I get with the new MAFS. So my original Hitachi MAFS does seem bad, but in a different way.

So it looks like a thermally sensitive bad (new) MAFS right?

Two things though are not convincing. A) that it took so much time and so high a temperature to cause the misbehavior. But more importantly, B) The fact that even when the MAFS signal is misbehaving a voltmeter (analog or digital) shows the same voltage as when the MAFS is healthy!! It’s as if in spite of the spikes and noise, the “average” signal is still the same, hence the voltmeters show the same. That does not seem very convincing.

So to do one more test, the next morning I just turned on the ignition with the new MAFS and let it sit there for about 3 hours (thanks for the suggestion jdg!!), without heating anything. I figured that if it were the ECM (perhaps the DAC or other input signal circuitry) or some thermally sensitive wiring that was causing the problem, I should see spikes/noise after a while. But I did not see any noise/spikes.

I’d be happy to get a new MAFS and get this over with at this point -- and stop torturing you all guys here on this very helpful forum. But I’m reluctant to shell out the $550 the dealership wants just to get another Hitachi part that likely has a design or systemic manufacturing weakness. And if I order from another party, I’m afraid I’ll get a third MAFS lemon… which will then further drive me nuts for another month or two (though I think, if anything, I’ve now learned how to debug them – so thanks for the support, regardless of what happens I have learned a lot about how to debug contemporary engines).

I think I’ll get a more definitive conclusion when I get the MAFS to misbehave and can actually see the noisy signal entering the ECM with the scope. Problem is, it’s not that easy to borrow the scope. Apparently, when I looked at the signal with the scope the last time, I must have not waited long enough and the MAFS was not misbehaving yet.

So what about that funny STFT behavior when the MAFS is working? That STFT ramping up and down rhythmically over a 30 second cycle? It could be that the noisy MAFS, when it does eventually happen, throws the ECM for a loop and teaches it bad habits which then it carries on even when the MAFS IS working well. Well, one thing at a time. The MAFS signal is definitively going crazy. Let’s fix that and we’ll see if there is anything else.

Thanks and sorry this has dragged on for so long. Hopefully it’s not a total waste.
 

·
Registered
Joined
·
3,193 Posts
Thanks jdg, those suggestions were quite helpful.
Dumb luck on my part :D
20+ years of troubleshooting dumb avionics problems on old military aircraft gives a guy a number of chances to try oddball troubleshooting techniques (eg. heating/cooling wiring, letting it sit, etc.). Then again, sometimes you play the game of shotgunning parts (random swaps) until something changes.
And again, nice to see somebody else with a clue for a change vs. the usual pinhead asking about cold air intakes, fart can mufflers, or the next bigger size rims he/she can put on a vehicle.

turned on the ignition and used a hairdryer to start heating up the MAFS
That's the way to do it! Beat it into submission!

Temperature inside the jar climbed to 230F. I really don’t think it heats that much normally, but what that heck
I don't think 230 is out of the norm. Think sitting at a long stop light on a hot day with no wind after a hard highway run. Sure, you've got the radiator fan going, but that's just circulating the exhaust manifold heat around a bit.

I must have waited a good 30 minutes warming up the MAFS, …and then finally voila the spiking signal started!!
Nice...finally a repeatable(ish) result.

Two things though are not convincing. A) that it took so much time and so high a temperature to cause the misbehavior. But more importantly, B) The fact that even when the MAFS signal is misbehaving a voltmeter (analog or digital) shows the same voltage as when the MAFS is healthy!! It’s as if in spite of the spikes and noise, the “average” signal is still the same, hence the voltmeters show the same. That does not seem very convincing.
The pieces/parts inside the MAFS are probably filled with some sort of potting material, an epoxy of sorts. Takes a long time for the heat to penetrate to the guts...and also takes a long time for the heat to soak out of the guts.

I think I’ll get a more definitive conclusion when I get the MAFS to misbehave and can actually see the noisy signal entering the ECM with the scope. Problem is, it’s not that easy to borrow the scope.
I'd be willing to bet if you got in touch with a local Ham radio operator in your area, one of those guys would have an old 'scope they'd let you borrow for a few days.

So what about that funny STFT behavior when the MAFS is working? That STFT ramping up and down rhythmically over a 30 second cycle? It could be that the noisy MAFS, when it does eventually happen, throws the ECM for a loop and teaches it bad habits which then it carries on even when the MAFS IS working well. Well, one thing at a time. The MAFS signal is definitively going crazy. Let’s fix that and we’ll see if there is anything else.
Ya, I don't have a halfway decent answer for that. Like I said before, I was kinda guessing it had something to do with completing the 'drive cycle' where the ECM basically jockeys things around a bit looking for something else to happen downstream in the process of basically self-testing various items...but. Sure, might be that noisy MAFS.

Don't know how you'd go about it, but, search around for other vehicles that use the same MAFS, hit the junkyard for a 3rd MAFS. 2 MAFS having a similar defect, unlikely but possible. 3 MAFS having a similar defect? Now that's getting wayyy out there...even in my book.
 

·
Registered
Joined
·
35 Posts
Discussion Starter #39
Well, I ordered a new MAFS. That is why I kept quiet. This time I went with the original Hitachi (or so I think, I’ll have to see what arrives).

In the meantime to offer some entertainment, I’m posting the OBD2 graphs I was able to plot.

As a reminder, I ordered a cheap OBD2 dongle from Amazon (see previous post) and paired it up with the “OBDCarDoctor” app on my iPhone. I paid the $4 to get the paid version of the app, which allows you to monitor up to SIX parameters simultaneously and also allows you to SAVE the values to a comma separated text file.

Then with modest spreadsheet manipulations I imported the text file into Microsoft Excel and created a graph. I was monitoring 6 parameters (the maximum the tool lets you simultaneously monitor) so the update rate on each parameter had dropped to around 1.3 seconds, but the information is still useful, I think.

The six parameters I monitored are: Throttle Position (TP), Engine Load (EL), Short Term Fuel Trim (STFT), Long Term Fuel Trim (LTFT), engine RPM, and vehicle Velocity.
It was a long graph which displays nicely on excel, here I’m posting it broken into 9 snapshots.

The abnormal behavior I can see is:
a) How the STFT does this rhythmic ramping up and down between 0 and +25% at idle. It’s as if it’s seeking something but I don’t understand what. The mixture is very lean during idle (I verified by reading the O2 sensor) but LTFT does not seem to make an attempt to adjust to the fuel deficit. Not sure what the ECM is doing in this mode.
b) How the mixture turns lean and STFT trends towards +25% on deceleration. But perhaps that is normal?
c) How at the end of my drive, arriving home, I waited about 9 minutes addling in front of my house, at which point, a little over 42’ since originally starting the engine, the MAFS signal goes nuts, the idle gets quite rough, combustion turns rich etc. You cannot see the MAFS in the graphs (again 6 parameters is all it lets me log simultaneously and I had already looked at the MAFS so I was more interested in what was happening with the other parameters). As a reminder, on another drive it took about 32 minutes for the MAFS to go nuts. This last run was in the evening, when its cooler so seems like it took longer, consistent with the heat sensitive MAFS theory.

In any case, these graphs seem very useful. And all for just $24 total !! (~$20 for the OBD2 dongle and $4 for the app). Seems like anyone can do it now.








 

·
Registered
Joined
·
3,193 Posts
A) Still, the only thing I can think of as far as the STFT ramping goes would be the ECM looking for a change in a downstream O2 sensor...part of the "drive cycle". The ECM trims for a rich mix, loads the cat up with fuel. Eventually the cat can't "store" the fuel and the downstream O2 sensor pegs rich. Then the ECM trims for a lean mix, emptying out the cat. The downstream O2 sensor pegs lean. That's how upstream (pre-cat) and downstream (post-cat) O2 sensors do their job. The upstream is supposed to wiggle (a wiggle you probably can't see because you can't scan fast enough to see it...it's a relatively fast wiggle) a bunch and the downstream O2 sensor is supposed to stay steady...if the cat is doing it's job of storing and releasing and "burning" HC's. If the upstream and downstream were wiggling at roughly the same rate, it would signal to the ECM that the cat isn't doing anything and would flag a code for catalytic converter efficiency. At least it is ramping at regular intervals, suggesting it is a programmed thing in the ECM. If it varied a bunch, I'd begin to think surely wasn't right...but this almost seems like it's normal.

B) Yes, on decel, that's normal. It's called 'deceleration fuel cutoff'. Sort of like built in engine braking. Totally normal. If you watch the flag that says open loop / closed loop, you'll see the ECM switch to open loop on long decelerations, like letting off the gas going down a long hill or on a highway off-ramp.

C) I like the jittery MAFS theory, even though it seems like it's been a serious pain in the ass so far :).

Entirely possible that the whole thing has been overthought a bit and the whole problem has been the MAFS all along and there is nothing wrong anywhere else.
 
21 - 40 of 47 Posts
Top