Stumbles/shuts off on cold starts, can't figure it out

[91 318is]

First off, I've searched on a few boards to see if this is a common problem, everyone seems to say it's the vacuum hoses under the intake.  I took care of that last year and replace every gasket that I took off, still have this problem.  When you first start it up, for some reason, it idles at about 600, and almost everytime, it will stumble, sometimes it will catch itself and come back up, sometimes it stalls.  I checked the plugs, change the coolant sensor for the ECU, and changed all the gaskets on the intake and bypassed all the vacuum hoses.

Anyone have any ideas?  I'm thinking it might be the AFM, anyway I can confirm this?  I would hate to drop $250+ for a new AFM and it not be the problem.

Thanks

[DesktopDave]

You've really done a lot of work so far, I'd check all the same stuff in the same order.  Have you tested the fuel pressure regulator or pump?

I'd try to adjust the AFM before replacing it.  I don't think they can be easily tested.
There's an idle control screw in there, but I've never messed with it.
There is a temp sensor too, never messed with that either.

There is also a gear wheel inside that puts tension on the flap door spring.  It's easily adjustable.  I HAVE messed with that.  It helps a little with some idle and throttle response issues.
Get a knife, carefully cut the silicone all the way around the little black cover on top of the AFM, and paint a location mark on the gear and housing.  Then you can return to stock if necessary.
Then change the gear setting by a tooth or two.  Test and repeat.  If I recall correctly, turning the toothed wheel clockwise tensions the spring (leaner), CCW loosens it (richer).

[AcSchnitzer318is]

Only does this on cold starts?  Not warm starts?

[quinn11m20]

I have done that myself.

So 91 318IS, you can adjust the spring tension to accomodate the lack of idle. I would suggest leaning the motor out, loosen the tension on the AFM spring. This kinda guarentee's that the motor wont start. Then you work your way back up. But too much tension and of course it wont run right.

When I first got my 91 318is (Natasha) back in 97 she had the same issue. The first thing I checked was all the hoses.
 Then for S n G's I changed the O2 sensor. Good luck.

[91 318is]

You've really done a lot of work so far, I'd check all the same stuff in the same order.  Have you tested the fuel pressure regulator or pump?

I'd try to adjust the AFM before replacing it.  I don't think they can be easily tested.
There's an idle control screw in there, but I've never messed with it.
There is a temp sensor too, never messed with that either.

There is also a gear wheel inside that puts tension on the flap door spring.  It's easily adjustable.  I HAVE messed with that.  It helps a little with some idle and throttle response issues.
Get a knife, carefully cut the silicone all the way around the little black cover on top of the AFM, and paint a location mark on the gear and housing.  Then you can return to stock if necessary.
Then change the gear setting by a tooth or two.  Test and repeat.  If I recall correctly, turning the toothed wheel clockwise tensions the spring (leaner), CCW loosens it (richer).

Haven't checked the pressure yet, I will though.  I actually did take the top off the AFM and wanted to look inside at the potentiometer, I cleaned it with contact cleaner, but that didn't help.  I will take the top back off and try to adjust it.  

Here's another question, is it normal for the idle to rise as the engine get's warmer?  For every other car I've owned, it was always backwards.

Only does this on cold starts?  Not warm starts?

You know, now that I think about it, it does it on warm start up sometimes too.

Sorry, it's been 9 months since I've driven it (got back from paint) so I forgot all the little quarks of her.

[DesktopDave]

Here's another question, is it normal for the idle to rise as the engine get's warmer?  For every other car I've owned, it was always backwards.

It's normal for the computer (DME) to raise idle on cold starts, but the car will idle very badly if the DME doesn't 'see' that the engine is cold.  The DME uses the coolant and intake air temp sensors to determine temperature.

[91 318is]

It's normal for the computer (DME) to raise idle on cold starts, but the car will idle very badly if the DME doesn't 'see' that the engine is cold.  The DME uses the coolant and intake air temp sensors to determine temperature.

Right, so on a cold start, it should idle higher then on a warm start?  Mine is backwards, it idles at about 600 on a cold start and when it's fully warm, it's about at 1,000.  

Also,  where is the fuel pressure test port?  I figured it would be somewhere on the rail but I didn't see it, am I not looking hard enough?

Thanks again for all the help guys, really want to get this bug out.

[DesktopDave]

Right, so on a cold start, it should idle higher then on a warm start?  Mine is backwards, it idles at about 600 on a cold start and when it's fully warm, it's about at 1,000.  

Also,  where is the fuel pressure test port?  I figured it would be somewhere on the rail but I didn't see it, am I not looking hard enough?

Thanks again for all the help guys, really want to get this bug out.

I'd guess that the car isn't getting the right signal to know it's cold, so it's not rich enough and it never gets to high idle...  I think the fuel pressure valve is only on M44 fuel rails, it's not on our M42s.  Not sure...anyone else know?

Did you do the stomp test to see what the computer thinks is wrong?  That was where I started.  I hate to tell you this, but I had to eliminate almost every major system before I found out what was wrong:
1. Stomp test told me code 1215, AFM problem.
2. Fuel pump test gave me acceptable flow, pressure was a little low.
3. Spark plugs were oily, but gave a good spark anyway.
4. After working my way around the sensors, I found that the AFM plug was waterlogged and the CPS was well oiled from my weeping main seal.
5. Car started after I cleared up those two problems.  DME still tells me code 1215 - AFM is bad.  I'll be looking at that again and posting my findings...

The lesson I learned (other than that so many systems are close to total failure), was that I had to single out each component for testing.  I wasted hours guessing before I started eliminating the good sensors one at a time.  I can't afford to replace every sensor...for that price I could be probably be leasing a new BMW.

[91 318is]

I'd guess that the car isn't getting the right signal to know it's cold, so it's not rich enough and it never gets to high idle...  I think the fuel pressure valve is only on M44 fuel rails, it's not on our M42s.  Not sure...anyone else know?

Did you do the stomp test to see what the computer thinks is wrong?  That was where I started.  I hate to tell you this, but I had to eliminate almost every major system before I found out what was wrong:
1. Stomp test told me code 1215, AFM problem.
2. Fuel pump test gave me acceptable flow, pressure was a little low.
3. Spark plugs were oily, but gave a good spark anyway.
4. After working my way around the sensors, I found that the AFM plug was waterlogged and the CPS was well oiled from my weeping main seal.
5. Car started after I cleared up those two problems.  DME still tells me code 1215 - AFM is bad.  I'll be looking at that again and posting my findings...

The lesson I learned (other than that so many systems are close to total failure), was that I had to single out each component for testing.  I wasted hours guessing before I started eliminating the good sensors one at a time.  I can't afford to replace every sensor...for that price I could be probably be leasing a new BMW.

Thanks for all the good tips man, I will definitely try some of this out.  I've actually never heard of the stomp test, but I'll look into it.  I've worked on american cars my whole life, so german cars are a new thing for me.

[91 318is]

bump on the pressure test port, still haven't found it.

Also, been looking and haven't found any info yet but what exactly is the "stomp" test?

[AcSchnitzer318is]

Get in the car and turn key to position 2... stomp (literally) the pedal 5 times in rapid succession.  Then the car should start blinking the check light at you.  It will do a code so count how many short blinks and long blinks there are.  Then return with answer.  

Also try the "Did we ever figure out that warm start issue" thread.  Simple to check one wire.  Did the trick for a bunch of us.  If the car has always done this then it might be that wire.  If the car developed this problem then it's not likely.

[91 318is]

Get in the car and turn key to position 2... stomp (literally) the pedal 5 times in rapid succession.  Then the car should start blinking the check light at you.  It will do a code so count how many short blinks and long blinks there are.  Then return with answer.  

Also try the "Did we ever figure out that warm start issue" thread.  Simple to check one wire.  Did the trick for a bunch of us.  If the car has always done this then it might be that wire.  If the car developed this problem then it's not likely.

Thanks!  Ever since I've driven the car, it's done this.  I'll go out and try check it and let you know what happens.

[91 318is]

Check the wire, unfortunely the wire wasn't the problem.

AND

I was trying to count service engine light flashes and I confused myself so I didn't get it.  How many numbers are in the code so I know how many I should look for?  

I guess if you don't catch the code the first time, you can't get it to repeat?  I tried to get it to repeat a few times and was unsuccessful.

Thanks

EDIT: Nevermind about the codes, realized it's 4.

[AcSchnitzer318is]

Was the wire plugged in or unplugged?  

Did you manage to get the code?  I should keep repeating the series of 4 numbers over and over separated by one long period (4 or 5 seconds) of no light flashing.

[91 318is]

Was the wire plugged in or unplugged?  

Did you manage to get the code?  I should keep repeating the series of 4 numbers over and over separated by one long period (4 or 5 seconds) of no light flashing.

The wire was unplugged.

Hold on, now I'm confused, am I counting the flashing of the light?  Or the pauses between the flashes?  I've read codes by the flashing SES light, and they were counted by the number of times the SES light flashes.

Hopefully I'm not confusing you.