Can anyone advise what my problem is?

The car keeps stalling itself when i let the revs drop, if it doesn't stall then it is idling at really low revs like about 2-6K, from what I've read it sounds most likely to be the MAF sensor but i don't want to shell out for a new one incase its not!

I have unplugged the sensor while the engine is running and it does not stall but revs higher, sometimes when i plug it back in then the engine stalls??

Could it be that it just needs cleaning or is it best to get a new one?

Advise on this would be much appreciated as it is making me look a right fool at junctions and traffic lights!!

 

Take your car to a specialist, and he can diagnose it. All we can do here is offer pointers and educated guesses. Those guesses would be better educated if you tell us what type and model year of car you're talking about.

Quite, there are other causes for this sort of problem.

Why did you do that? If it was as a result of reading something on here about a "test" in which you unplug the sensor with the engine running, and if the engine stalls, the sensor's good, and if it doesn't stall, the sensor's junk, unfortunately you've just followed some complete bullsh*t, this "test" proves nothing, one way or the other. It's a complete waste of time, and also has a small amount of potential to damage both the ECU and an (otherwise good) airflow sensor.

Unfortunately there isn't a 100% certain way to define a good or bad sensor without some diagnostic instrumentation. However, one thing you might try is getting the car fully warm (and at a point where your idle problems are prevalent), switching off, unplugging the airflow meter and then restarting the engine. You will see the Check Engine light. Do not drive the car - but blip the throttle gently, and try edging around at parking speeds. If your idle issues have completely (or largely) gone away, then it is very likely that the airflow meter is the cause.

Reconnect it and drive slowly/off-boost until you have the replacement fitted (and reset the ECU after you've done so).


Your engine should be equipped with a good, effective air filter capable of removing foreign particles very effectively, right down to microscopic sizes.

As such, the value of cleaning is likely to be minimal. In addition, when airflow meters fail, they often do so via methods other than physical contamination of the sensing element - and with this type of failure, cleaning will make absolutely no difference.

Short answer is that if you want to be sure, don't muck about - fit a new one. And, if you are running a non-standard air filter, or induction kit, you might want to read the forum carefully on the issues that can present.

If you do have a failing airflow meter, the car's behaviour at junctions will be the least of your worries. Keeping off boost til you have the cause of this problem identified and fixed would be a good idea in these circumstances.

 

Thanks for the advice. The car is a classic shape WRX STi Year 2000, and yes i did read on a forum about unplugging the sensor!

 

I have swapped the MAF sensor with my other scooby which is currently off the road and it seems to have now resolved the issue. Just need to buy another one before i get the other car back on the road, does anyone know how much to expect to pay for one I have been quoted £85+VAT by HDC Subaru in Hebden Bridge?

 

Just over a ton is about the right price, if they have one in stock that is

Tony

 

Thanks Tony is this the price you'd pay for a brand new one? Could i get it cheaper if i went to a breakers yard and took one off a write off?

 

Although gettin one off a breaker is tempting and 'sounds' like a good idea trust me its a cack one.

For one you have no guarantee's as to how old the sensor is, how long it will last or indeed even if it works at all.

But most of all if a scooby is in a breakers yard its usually cos its had a sizeable whack of some kind and MAF sensors are very sensitive to vibration... so chances are even if you can find one in a scrapper the right age etc it'll be knackered anyway!!!

I'm afraid the only sure fire way of getting one that you KNOW will work properly is your local dealer IMHO....

 

Scooby715 has already ably covered this but to reiterate the point, the above is a downright silly idea. Aside from the fact that secondhand sensors are best avoided in general because you never know their providence, buying one off a write-off is asking for trouble, as the biggest killer of the 99-00 type sensors is vibration damage. And you're probably going to buy one off a car that may well have just hit something quite hard, or, worse, the car may be in the scrapper's yard because the MAF died and killed the engine...

I've seen one, scratch that, two of these sensors killed by front end shunts that didn't actually damage the airbox area. One was quite a serious accident, the second occurred at little more than parking speeds.

Yet, it was the second that was the potential engine breaker as the sensor was fine at idle (so there were no obvious/immediate symptoms), but under-read like a tw*t under boost. The only "symptom" was that the car felt noticeably quicker. Luckily it also had an AFR gauge which caught the misread immediately - which was then confirmed by a quick bit of logging.

Bottom line is to buy a new sensor, it's the only way you can be guaranteed one you can trust (and new they come with a warranty), and it's a damn sight cheaper than the consequences of driving around with a damaged one. If you don't like the price above, £95 inc VAT (but + delivery) from Import Car Parts is one alternative, and may work out a couple of quid cheaper.

 

Hope the OP forgives me joining in...I have very similar idle issues that I am trying to track down. I will try Splitpin's suggestion for unplugging the MAF, and am also trying to borrow a working MAF to try.

I also have RomRaider so can log things easily - what would be useful parameters to log in order to conclusively diagnose a faulty MAF?

In case relevant this is a somewhat non-standard engine with an 80mm induction kit (all rescaled and mapped properly by JGM).

 

If a modified newage simply won't run when doing this, does it mean anything?

 

you dont remove the maf sensor period

 

The word "conclusively" is a tricky one, as it often isn't possible to diagnose, on a black/white, yes/no basis, a sensor with a relatively small amount of degredation on it. As your car's (quite heavily) modified it's also more difficult to give you specific instructions to follow. However, hopefully the following will help.

If you have a "known good" sensor and a "possible bad" one, log MAF voltage, engine speed, throttle position, boost pressure, O2 sensor voltage, ignition advance multipler.

Conduct a quick acceleration run with the known good sensor and the above data parameters. Repeat with the test sensor under as close to the same conditions as you can manage (i.e. same gear, same start/finish engine speeds, same location). As the engine accelerates you should see a steady increase in airflow meter voltage without any significant troughs or peaks away from the curve.

Then try and cross-reference the run with the known good sensor against the possibly dodgy one. You should see a close correlation between the two. For example, once you get the car onto full throttle and consistent boost, if you're getting a MAF voltage of 4.00 at 4000rpm and 1.2 bar boost on one sensor, you should be at pretty much the same - within a tenth or so of a volt, on the run with the "other" sensor.

Incidentally, one note of caution. If you have a suspect sensor, you may not want to run flat out to test it, as if it is dodgy, you don't want to risk running hard on it. You should still be able to spot a difference between a badly degraded sensor and a good one at idle or low throttle speeds, but it's more difficult to do so as the differences won't be so stark.

It probably means something, but unfortunately I don't have the detailed knowledge of the newage ECUs necessary to give you an authoritative answer to this question, especially given you're running quite a bespoke setup. However, based on knowledge gained elsewhere I can hazard a guess what might be happening:

On a 99-00, if the MAF input voltage disappears, whether because the sensor is unplugged, or because it has failed/is failing, the ECU, after a short timeout value has elapsed, activates MAF error mode and then uses a number looked up from a table (based on RPM and throttle position) as an "approximate" airflow value. This is fed into the engine load and fuel calc sequence. It's not particularly accurate but sufficient for limp mode/get you home.

It's because of that function that I suggested the "unplug MAF and then restart engine" procedure you followed earlier - although of course I suggested it to Greg and his classic. The theory is that if you have an airflow meter screwed enough to cause inconsistent idle, if you start the engine without it, the ECU will have no choice but to use that emergency lookup table value I mentioned above as its airflow input.

If the engine immediately idles more consistently on the emergency value than on the "real" MAF input, and there's no other causal issue (i.e. a post-MAF air leak or similar), then the likeliest cause of the idle problem is the sensor itself.

It would be reasonable to assume that the newage ECUs implement a similar contingency in the event of a MAF failure. If they do, and if you've had your main MAF sensor lookup table rescaled, you would also theoretically need to rescale the emergency lookup one too, otherwise the "emergency" values will bear absolutely no relevance to the amount of air going into the engine, and hence, it can't set a mixture viable enough to get it to run.

None of that is a criticism of Simon, btw, as above this is a theory based on what the classic ECUs do, rather than knowledge of the way the newages work.

 

Thanks Splitpin - hugely helpful as always

Was out grabbing some logs before I saw this post, and have some logs of 2 or 3 stall events and am comparing them to non-stall events.

At the moment I'm suspicious of the O2 sensor. Most of the time driving around on light throttle (i.e. closed loop) the voltage is varying 0.2-0.8V or thereabouts but every now and again I get several seconds of 0V or just a few 10's of mV...or several seconds with a fixed voltage not varying (e.g. 0.86V for 4.5 seconds)

Stall events seem to be preceded by such an event.

So - with continued apologies to the OP - what is the best way of investigating further to see if it is, indeed, the O2 that's at fault?

I am, in the meantime, trying to locate a MAF to do just what you have suggested there

 

PS...O2 sensor has 75,000 miles on it and has been removed/replaced several times through various exhaust and turbo upgrades and repairs.

 

Sounds good, exactly as one would expect.

The behaviour you're describing could, potentially, be legitimate, depending on whether you're coasting (i.e. trailing the throttle slightly), accelerating ever so gently, and so-on. If you're expecting the 0.2 to 0.8 volt bounce to be constant and unwavering, that's not how they should be.

Easiest next step is just to unplug it and see what happens. Get the car nicely warmed, and then, first time you feel it acting up, pull over at the next convenient moment, disconnect the primary O2 sensor, start back up and drive away. If the problem goes away, plug the lambda back in and see if it starts recurring again. If it does, you have a likely culprit. Given the mileage, it wouldn't be too surprising to find this is at least a contributor to your problem.

 

Thanks...and just to be sure, the primary O2 is the one in the downpipe? Romraider only showed a "rear O2 sensor" that could be logged and, as far as I know, there isn't another on my JDM?

 

Not the O2...unplugging it did not fix the problem

 

Pish, shame it's not that easy - although, also while it might not feel like it you've still taken a step forward as it's one causal factor ruled out.

Are there any other trends in your log that accompany the "incidents"? For example, what sort of MAF sensor voltage does it typically read under hot idle, and does the sensor dip significantly beneath (or is it significantly higher than) that number when it's acting up?

As a matter of course it might also be worth a word with Simon, as he knows far more about your particular setup and may be able to make a more educated guess.

 

Coming to (hot) idle without a stall, by closing the throttle to 0% from the MAF voltage drops from maybe 1.5V-ish steadily to somewhere around 1.3V, but could be anywhere in the range 1.14-1.3V at idle, and does fluctuate slightly.

When it's playing up, I would say it's a little lower before hitting idle rpm - 1.0-1.1V at 1200rpm or so. Here's MAF voltages as it stalls:

rpm MAF V
1264 1.16 (throttle 0.39%)
1237 1.12 (throttle 0.39%)
1206 1.08 (throttle 0% from here onwards)
1182 1.1V
1008 1.1V
648 1.2V
344 1.4V
184 1.28V
196 0.98V
196 0.98V
196 0.8V
0 0.74V

Not sure if it all means anything though as I don't know what the usual spread of MAF voltages are at idle.

Only othe thought I've had is the PCV valve. This is blanked off due to the catch can, but I know the blanking tube is not clamped. I did notice once that it was a little loose, but in theory it shouldn't matter since if there's a vacuum it will pull it back on and if on boost the PCV valve will block it.

But if the valve is sticking then on boost it could blow the blanking pipe off and cause an airleak. Half-baked theory might suggest that lack of oil passing through the valve as it's no longer venting crankcase fumes could cause it to stick maybe.

I can't check whether the pipe's off until tomorrow, but I certainly will do so - and at least it's easier to see as there's no TMIC any more!!

 

That's normal. You get a bit of natural fluctuation and the numbers you're quoting there are in the ballpark. Although if you're running a larger MAF tube and have had the ECU rescaled, a lower number would probably also be legitimate.

That's interesting. I'm not an expert on the newage MAF either, but on the 99-00 one, any voltage lower than 1.015 or so is regarded by the sensor and ECU as reverse flow (i.e. travelling from the engine out of the filter). If this applies to the newer ones then you are clearly below that value.

What sort of MAF voltage do you get when you've just turned the ignition on (i.e. engine not running)? I'd also ping Simon a quick message to ask what a typical "engine off" voltage on the newage MAF should be, and see how yours compares. And, speaking of which, has this problem cropped up only recently or has it been an occasional issue since the remap?

If it's a recent thing have you eliminated all the other possible causes - sticking idle speed solenoid, gunk in the ISCV pipework, post-MAF air leak somewhere etc. etc?

Yeah sounds unlikely but always possible that a small amount of air could leach through when the engine's idling and there's a strong vacuum inside the inlet manifold. Stick a jubilee clip or whatever on to make sure there's a really good seal and then it's ruled out, innit.

 

0.7V, consistently. Will, indeed, ping Simon.
It has been a very occasional issue - but is now happening pretty much all the time.

There have been a whole heap of niggling problems with this build so it is really difficult to recall exactly when any one issue occured!

Have removed and cleaned ISCV as well as throttle body; cleaned out BCS; cleaned and/or replaced all boost-related vacuum pipes - there isn't a pipe to the ISCV on a newage as it's integral to the throttle body AFAIK.

I have checked all the FMIC clamps countless times (they had a habit of blowing off until I replaced some of the shoddy clamps the APS came with). I have used carb cleaner on joints to try and find leaks.

I did have a small 3mm split in the Samco turbo intake pipe, where the metal insert had worked loose and damaged it. That's fixed now (by removing a little of the end of the pipe so the insert covered the split, as well as using silicone sealant in the split).

I also had to remove samco intake pipe to get access to sort leaking fuel rails and, to replace it, had to lift inlet manifold slightly. It is possible that there's another tear in the Samco, or that I have damaged inlet manifold gaskets.

I know that I will need to eliminate those at some point - but that means manifold off and new gaskets so I want to eliminate the easier stuff first.

Thanks, as always Splitpin, for you continued interest and helpful suggestions

 

Not the MAF - swapped for a known good today and behaves the same.

Inlet air leak is, apparantly, ruled out since the vacuum at idle is, I'm told, normal at -0.8 bar.

Running out of ideas now

 

Tim, inlet air leak isn't ruled out simply on the basis of idle vac being "normal".

Don't forget that the idle vacuum only exists across the throttle plate. You can still have a leak somewhere between the throttle and MAF sensor - and it won't affect your idle pressure at all.

Incidentally, when I made that reference to the ISCV pipework earlier, I'm aware the newage one's bolted to the TB - but there's still a section of "pipe" within the manifold casting Have you tried taking it off and checking for oily carbon-type buildup in the airway beneath? Still worth checking with a fine tooth comb for a leak somewhere too.

Incidentally, that car you borrowed the MAF sensor from. Could you swap the idle control solenoid over for a little while too?

 

I will take the TB etc off this weekend and check everything over and give it a good clean.

Is there a procedure for aligning a newage TPS? Although the logs show 0% throttle plate angle, the voltage is 0.7V and - if it's like a classic - I thought it should be 0.5V? I did mistakenly remove the TPS a month or two ago, when I was going about cleaning the ICSV...unfamiliarity caused me to remove the wrong bit.

My hunch is that it is something in the ICSV/TPS/TB area. There have been a number of times when I've "fiddled" in that area (e.g. cleaning the ICSV) and the problem has improved for a few days - but then got worse.

 

OK - spent a happy hour or so on Sunday removing throttle body from inlet manifold and cleaning up all the sensors, ICSV etc etc. ICSV spindle spins freely, no obvious signs of wear anywhere...and there was some oily gunk in a few places so must have done some good.

And...it is HUGELY better now...and although there is, I'm sure, some kind of psychological effect, the engine seems to run smoother and quieter as well

It generally drops into idle mode most of the time now (90% perhaps) but occasionally the revs drop and it nearly stalls.

Anyway, I can now characterise down the symptoms of occasions when it nearly stalls.

When you disengage the clutch as you come to a standstill, and it's behaving, the revs seem to drop to about 1500rpm, hover there and then drop to idle at 800-900rpm.

When it's misbehaving, the revs plummet without that hovering at 1500rpm and, not surprisingly, it nearly stalls (bear in mind lightened flywheel and pulleys etc).

It is worse if the revs when you push the clutch pedal are lower (2000rpm say).

So, given that it's intermittent, I no longer think it's an airleak. Since it has vastly improved since working on the TB - and, as I've said before, tinkering in that area has frequently improved things but just for a few days, etc I'm pretty sure I'm homing in on something.

I think the question now revolves around the ECU's idle strategy...what inputs does it take to "decide" that it will hover at 1500rpm and then control the descent to idle? If it's something specific, then I can attend to that sensor (or whatever) and maybe fix it for good

 

An update after having driven for a week since cleaning and rebuilding TB/ICSV/Sensors and also adding a jubilee to the blanking pipe on the disused PCV valve "just in case"

It has very evidently self-learned how to manage idle properly. Now, when dropping from almost any rpm to idle, I can see the descent being well controlled with no stalling

There is just one case where it still struggles and that's from about 1400-1500rpm when the revs still plummet to almost nothing but, at least, it doesn't stall.

I suspect this may be as good as it gets given the lightened flywheel and pulleys - the revs probably drop too fast for the ECU to "catch" it. And I can drive around it by disengaging clutch a fraction earlier (my daily driving style is not to go down through the gears, so I generally disengage the clutch at low rpm in a high gear).

Another interesting fact from logs - previously (ever since the build and map) the A/F learned correction for minimum load and low rpm has been nudging towards +14% correction which is near its limit. JGM was always happy with that, but I thought it a little odd from everything I'd read.

Since last weekend and 400 miles, the learned corrections in all ranges are within 2%

So, fingers crossed, the problem is near as damn it sorted