Hunting a vacuum leak (s)

[decoupe]

Title says it all.

Symptoms are fluctuating vacuum reading at idle and deceleration (all the foot off throttle conditions) through a MAP sensor that provides info to the ECU. Result is a crappy rough idle at 1250 that stalls if I try to go lower and frequent stalling when foot of gas decelerating to a stop. No effect on accelerating and wide open throttle.

Background. Running an after market standalone engine management system (SDS, not MS) that ran last years L Jet based 3.5l flawlessly (other than a persistent ear piercing throttle plate whistle associated with that series of throttle body). Great setup other than the whistle.

Current setup. No changes to the 3.5l engine. I refurbished a '89 e32 B35 intake, throttle body, fuel rail, new 30lb bosch injectors, Tii brake booster plus the existing SDS engine management with no changes - just plugged it all in and it fired right up. Just idles poorly.

To date I have confirmed the spark, timing, fuel pressure, MAP sensor (swapped in a new one to check) checked all of the electrical and hose connections and sprayed liberal amounts of engine starter at likely areas for vacuum leaks. No luck. I've been told that propane is a better method of finding vacuum leaks and will try that tomorrow.

The symptoms just scream vacuum leak to me. My suspicions are that the brake booster hose or one way check valve are at fault but no amount of wiggling the hose shows anything wrong. Other possible culprits? Any of the vacuum ports on the b35 intake I guess but the vacuum line to the fuel regulator and to the MAP sensor are new, the valve cover vent hose to the intake is hose clamped and new rubber (had to home build this).

The whole thing is a bit if a hybrid - no start valve assembly, there is a fast idle bypass relay that closes at 73c - engine warm up does not seem to be seriously effected while the relay is open. That fact and that when using lots of throttle shows no issues leads me to think the leak is small but persistent at idle.

As I said earlier this started, idles and ran flawlessly last year with an L Jet setup. Only the brake booster and b35 intake (fuel rail, injectors etc etc etc) were changed.

Anything come to mind - I'd appreciate hearing.

Doug

 

[DaveG]

I'm running MS3 on my B35, and I had the same issue with stalling on decel. Tried for quite a while to make it go away and failed (before I activated the CL idle..)....MS3's closed loop idle seems to be the only way I can keep it from stalling. Even then the idle will momentarily drop to ~600 after a long decel before the CL idle catches it.
I'd sure like to know what really causes it....
As far as the rough idle....We found that the injection timing really had a big effect on idle quality....basically ...you don't want to inject onto an open valve.
Not sure if the timing is adjustable in SDS though.
Also a quick question... What is your idle Kpa?
edit: 30 lb injectors may be a bit too big. That will definitely effect the idle quality..... I'm running 19 lb injectors and my idle PW is 2.7ms. Your idle PW is probably around 1.7ms which is getting a bit short for good resolution...May be another contributor to the lousy idle...
DaveG

 

[CSLCOUPE]

Fluctuating vacuum at idle is usually a weak cyl,,, valve, piston, ETC ?????
Compression test????

 

[61porsche]

big daddy

smoke test- go buy a cheap cigar.8)

 

[Sven]

Doug - I agree a larger injector could affect idle quality. When you get less than 2 ms on the pulse width then the injector "dead" time needs to be figured in. I believe with the MAP set- up a vacuum leak (if consistent) will be accounted for in the MAP reading. Does the new throttle butterfly close far enough? Is there an extra hole in the plate or body that is letting in too much air (that the ECU can't deal with)?

Also a quick question... What is your idle Kpa?

Dave - With idle advance at 12/13 I get a MAP reading of ~50 kpa. With idle advance 18/20 (or higher) I get 40/42 kpa. I noticed with my set up that when the idle advance was up in the 18's 20's, then even with the idle valve all the way closed I could not slow down the idle below 1,050/1,100 rpm. With idle around 12/13° I could get idle down to 850 (a bit rough still but steady).

In MS you can use the "Closed loop idle initial values (duty)" table to set your idle valve a bit more open than steady idle, so that at off throttle-return-to-idle state the idle will begin a bit higher. The PID will then bring it down to the target rpm. I never liked the dashpot adder feature much. Also, you can add a couple degrees timing to the lower rpm column (left of the idle rpm column) that will help pick up a drop in rpm as you come off throttle - by sloping the idle advance curve and bumping up the spark table numbers in the left column.

 

[decoupe]

Rough Idle

Thanks to all for the comments - I'll respond in order of postings.

DaveG
Closed loop is adjustable for rpm - currently it's disabled below 2000 as CL tries to average the air/fuel to 14.7 and idle in my case has always liked a bit of richness - 13 or so and CL was part of the problem getting the idle stabilized.

Yes the idle timing is adjustable -currently 13 advance at 1000 and have tried as much as 15 with no improvement.

My MAP readings are in Absolute Pressure where WOT is 0 (1 BAR at sea level) and not in Kpa. At idle the rpm the reading fluctuates between 1000 and 1500 and MAP between -13.8 and -16.1 (a range of six reading points). The best I have been able to do is get it to stabilize to 2 or 3 points and more or less stable at 1250. Below that it falls apart.

CSL
All that is good numbers (175 - 180psi average) but I should pull the plugs to see if there is any indication of a problem cylinder.

61porsche
I've considered something more potent we grow in BC but not yet. The propane revealed nothing.

Sven/DaveG
Last years smooth running setup used 24lb injectors so going to 30lb didn't seem unreasonable and since I'm getting a ported and flowed head as my next upgrade I went to the larger size. I have a set of 19lb injectors that are my last resort to try but I'm not there yet. My controller allows considerable adjustment of pulse width and duration and the SDS guys have tried that for me with some improvement but the fluctuation persists.

Sven
The butterfly has no hole and closes to and nearly complete seal - effectively zero air passage. This creates a problem setting the idle as all I have is the throttle stop screw.

I continue to fiddle and will report any changes.

Thanks to all once more,

Doug

 

[DaveG]

Doug,
Sorry for the mis-understanding... The Closed Loop Idle control I am referring to is the MS3 controller controlling the idle SPEED using the idle air control valve. Not the closed loop mixture control.
The timing I refer to, is the time the fuel injection occurs during the cycle. We found that when the fuel is injected into an open intake valve that the idle quality is not as good as you would like. Moving the injection event to occur onto a closed valve turned out to be much smoother, as well as more efficient.
If the MAP numbers that you gave are in inches of mercury then they are the same as I have. The B35 idles at a much higher MAP (~50 kpa) then I am used to on my '02(~30kpa). (~100Kpa= ~30 inHg=~1Bar)
This must be because the cam has more overlap then the older cars.....
DaveG

 

[DaveG]

Sven,
Your idle issues are interesting....I have the idle stop set all the way open!
And my IACV duty is about 50% at idle....I've got it tuned pretty well now, with only the occasional drop to ~700 rpm on decel.
BTW
The latest MS3 firmware update (V1.1.1) has a MUCH better CL idle code... Lots of other subtle improvements as well. It was well worth the effort to do the upgrade.
I idle at 1000 rpm, ~50 Kpa,and 20 deg BTDC. I make use of the idle advance, idle timing compensation, and the idle VE features. They really stabilize the idle speed a lot.

 

[Sven]

Dave- my 3-wire valve goes from 39 % (closed) to 64 %(open). At closed a bit of air still gets through. I have some overlap because I have the valves set 10/10, which may be part of the problem. What is your injector table set at for idle? Mine is 200 deg NTSC. I think the intake valve is closed by then. Also running latest 1.1.1. and using most of the features. Generally quite Pleased with the results.

 

[DaveG]

Sven,
My injector timing goes from -300 at 700 rpm to -220 at 5000rpm.
PM me and I can send you my msq...
DaveG

 

[decoupe]

I mentioned pulling the plugs as the next step and found that cylinders 1+6 plugs showed black and wet (looks very rich) and 2, 3, 4, and 5 were dry and tan/grey (just about perfect). Since my coil pack is wasted spark and fires paired cylinders and coil A handles cylinders 1+6 it is highly likely that a defective coil and not vacuum is the cause and the poor idle and stumble/stall on deceleration.

http://s54.photobucket.com/albums/g120/decoupe/EFI/?action=view&current=P1020317.jpg

If I can find a timing light, I can check for a consistent spark by comparing 1+6 and either of the other two coils using the other plugs.

Doug

 

[sfdon]

That timing light light won't show you a weak spark.

 

[decoupe]

The spark is fine as far as strength (voltage?). Hopefully the timing light will show an eratic spark. In any case both the coil pack and ECU will go in for testing. The company is about 45 minutes from home.

Doug

Edit:

As suspected, the timing light confirmed a regular miss in the spark on one coil that does not show on either of the other two. Not likely to be a wiring issue because of it's regular miss as opposed to random. Bench test to follow.

 

[decoupe]

Vacuum Leak!

After SDS bench tested the coil pack and ECU and gave me two loaners to swap over, we determined there was nothing wrong with the components or anything related to the wiring. A comment on the mixed messages we were getting as symptoms (surging, wet plugs/dry plugs, mileage decrease etc) made vacuum the only logical source of the problem.

Turns out that the fast idle relay which, pipes in extra air at warm up (the ECU adds the appropriate richer fuel to match) is ported into the b35 intake manifold between the ports for MAP (vac sensor) and the vac port to the fuel regulator. The hose at the intake manifold was hose clamped but the other end at the air valve relay was not and was lose enough to leak air. So both the sensor feeding vacuum readings to the ECU and the vacuum line to the fuel regulator were compromised. Fixed by a hose clamp and the whole thing works as it damn well should.

Duh

 

[rsporsche]

Doug,

gives new meaning to verify the simple things first - especially on cars the age of ours. glad its all back to the way it should be - enjoy the ride!

scott