Fine tuning the S38

[Technician117]

Hi All, 

Last year I picked up my first E28, a US spec '88 M5. I have only ever driven this car, so I wasn't 100% sure what to expect in terms of drivability, smoothness, etc. The car has had a ton of things done to it prior to my ownership and I've done some things as well. When I bought it, the engine started well and ran "ok" with it seemingly wheezing at higher RPM's. I now have it running better, but not sure it's running "best". I'm looking to see what hard data I can use to know it's running as good as it should. Here's more detail:

Under prior ownership the ECU was resoldered and supposedly the AFM rebuilt. Various sensors were replaced as well, like the TPS and CPS (about 10 years, 20k miles ago). Fuel pumps were also replaced around the same time. 

Under my ownership, I had an intermittent crank/no start that I diagnosed as a bad battery cable (the skinny one to the glovebox area). That wire feeds the fuel injection system/pump, so fixing that high resistance definitely helped some. I also did more basic stuff: spark plugs, wires, coil, fuel pressure reg (before I diagnosed the power wire). I also tested the TPS, CPS, IACV, etc. Next I determined there was a wiring issue with the O2 sensor. I had to rewire the plug and replaced the sensor. At this point it was running better, but still some stumbling or "unclean" running. I knew it was time to look at the AFM (condition and screw setting), gross idle air screw, TB balance, etc. 

TB balance seemed fine when I checked. 

AFM wasn't dirty and seemed to move fine through the flap sweep. 

I decided to "start from scratch" with the AFM and idle screw settings. I set the idle screw around 2 or 3 turns out and the AFM mixture screw at 3 turns out. I then bridged the TPS and warmed up the engine. The idle was low and when the bridge was removed, the throttle response was sluggish. So I reinstalled the bridge and set about making adjustments to get it to idle at around 850 RPM. I had to turn the screw out to about 7 turns! Next, I connected my DVM to the diag port and ground for the O2 sensor voltage. I then slowly tweaked the AFM screw to get an avg value around 5v give or take. 

It runs pretty good now but still unsure about higher RPM power. It's certainly not bad and way better than before. When still fairly "cold" it doesn't run as clean. Seems like a tight ambient temp is when it runs best. Colder or hotter isn't as good. On/off throttle transition seems really good. Occasionally it will be abrupt, but mostly smooth.

I'm wondering if I need to go through the process again, or maybe several times, to dial things in? Now that I'm at this point, I'm looking for recommendations on what steps to go through to fine tune.

Also, is 6-ish turns out on the big idle air screw too far out? What would cause that? AFM spring too loose? I'd guessing extra turns out means it's otherwise running rich at idle. Speaking of the AFM, I believe the circuit board is newer/was replaced which would verify the "rebuilt" claim by the PO (I don't have a record, just his word). If I make a spring tension adjustment, then I need to go through the tuning process again, correct? 

Sorry for multiple questions at once. I now know enough about the systems to be dangerous. I just want to collate the knowledge into a more precise process to tune in the drivability. 

Thanks in advance!

[turbodan]

What I would recommend is to find someone that has a wideband oxygen sensor.  Go put it on a dyno and you'll likely get a high fidelity AFR plot included for free.  This will indicate if the AFM, DME, and everything else is working properly and getting the job done.

If the AFR is good, compression is good, and timing is good, it will make power.  The factory rating of 256hp is at the crank, so you would hope to see a dyno run of 210-220 rear wheel horsepower out of a healthy, stock US spec S38.

The e28 M5 is a cool car, very unique.  Relative to modern cars though, it's not fast by any means.  Car and Driver tested it at 14.6 seconds in the 1/4 mile at 96 mph.  0-60 in 6.3 seconds.  In simpler times, that was blazing.  It won't feel as fast in 2025, where everything has a turbo and an 8 speed gearbox with paddle shifters.

https://www.caranddriver.com/reviews/a1 ... hive-test/

[Technician117]

I agree it's not going to be fast. I'm not looking for that. More so drivability fine tuning. As part of that, I think/thought that higher rpm acceleration was not quite there, possibly indicating it running a little lean or a lot fat. I was already thinking about installing a wideband sensor in the exhaust and temporarily installing a gauge. Now that it's running better I should pull some plugs and check their color as well. 

I guess I'm really looking for an order of operations for adjustments. For example (not sure if correct) start with the AFM spring tension, then reset gross idle monitor O2 sensor voltage and readjust the AFM screw?

[turbodan]

Its easier if the AFM bypass screw hasn't been tampered with. Then the spring tension is the only variable.

Either way, I would set the spring tension for AFR at full throttle. I don't know what the original calibration would have produced but I think 13.5:1 to 14:1 should be alright. You will melt down the cat if it's too rich.

With the WOT AFR set, you could check the idle AFR. If the bypass has not been adjusted and the spring tension is correct it should be in the 14.2-14.7 range. Before you turn this screw it would be a good idea to measure it's exact depth. Molested or not, it would be good to know exactly where it was before any changes were made.

The bypass has a fairly limited effective adjustment range. If it is fully open and the AFR at idle is still rich, your spring tension is insufficient. If it is fully closed and the AFR at idle is still lean, your spring tension is excessive. You will have a range of a couple teeth to find what works best in terms of WOT AFR within the ability of the bypass screw to set the correct mixture at idle.

You can dial it in pretty quick with a wideband. I have an old school LM1 that works great for this sort of thing.

[Technician117]

Thanks for the additional response. If I understood correctly, the AFM spring tension really shouldn't need to be adjusted provided I can reach a solid closed loop operation at idle with the screw in a reasonable setting, AKA somewhere in the middle of it's travel? The caveat being what AFR the engine is running at WOT. I have read through a bunch of the various threads on the AFM, setting idle, etc.

A little further detail on how it's running. It starts right up, goes into the "cold start" idle which comes down as it warms. Acceleration/throttle response when the engine is still pretty cool is not great. It should be running in closed loop by then, but it might simply be that - it's not in closed loop yet. Once fully warm, dipping deep into the throttle more often than not results in some surging while the RPM are rising and over 5k? rpm it just feels wheezy instead of pulling harder. If I'm not mistaken, these engines make their power higher in the rev range? sometimes it will pull "cleaner" without surging which is why I haven't mentioned potentially adjusting the sweep arm on the AFM. Especially if they board/plate is supposedly nearly new. I can see a line in the carbon but it's all the same color in the "groove". 

I hate running "blind" and am pretty convinced to just weld in a bung to the exhaust and run an AFR gauge into the cockpit. I suppose I could drop down the ECU, back probe the O2 sensor voltage pin and look at that while driving, but I'm not sure that's as effective even with the decent DVM I have (Fluke 115). 

Side note - I had previously smoke tested the engine for air leaks, etc. and only found a small one at the IACV. I applied silicone around the connector where the leak was and everything looks good. The brass idle screw being turned out 6 turns tells me there isn't an air leak anyway. 

[jimoreno]

I don't own an M5 or this engine however some of your symptoms describe what my M30 did. Sometimes (not always) when accelerating, it felt like it was doing it in steps, increase RPM's some, kind of pause for a second, then keep rising, then another kind of pause. Then my in-tank pump died as I was returning from the Vintage about 180 miles from home while having a quarter of a tank of gas.
Not sure if you have checked your fuel pressures but it might be something simpler!

[turbodan]

Thanks for the additional response. If I understood correctly, the AFM spring tension really shouldn't need to be adjusted provided I can reach a solid closed loop operation at idle with the screw in a reasonable setting, AKA somewhere in the middle of it's travel? The caveat being what AFR the engine is running at WOT. I have read through a bunch of the various threads on the AFM, setting idle, etc.

A little further detail on how it's running. It starts right up, goes into the "cold start" idle which comes down as it warms. Acceleration/throttle response when the engine is still pretty cool is not great. It should be running in closed loop by then, but it might simply be that - it's not in closed loop yet. Once fully warm, dipping deep into the throttle more often than not results in some surging while the RPM are rising and over 5k? rpm it just feels wheezy instead of pulling harder. If I'm not mistaken, these engines make their power higher in the rev range? sometimes it will pull "cleaner" without surging which is why I haven't mentioned potentially adjusting the sweep arm on the AFM. Especially if they board/plate is supposedly nearly new. I can see a line in the carbon but it's all the same color in the "groove". 

I hate running "blind" and am pretty convinced to just weld in a bung to the exhaust and run an AFR gauge into the cockpit. I suppose I could drop down the ECU, back probe the O2 sensor voltage pin and look at that while driving, but I'm not sure that's as effective even with the decent DVM I have (Fluke 115). 

Side note - I had previously smoke tested the engine for air leaks, etc. and only found a small one at the IACV. I applied silicone around the connector where the leak was and everything looks good. The brass idle screw being turned out 6 turns tells me there isn't an air leak anyway. 

I would not adjust the spring tension unless there was reason to believe it may have been tampered with in the past. I was under the impression that you suspected this.

The big question with Motronic is exactly how it handles fuel trims. If it works like the MS2 narrowband EGO control, it's a continuous, real-time correction trying to maintain the desired afr. This is where precise calibration of the AFM is essential. If its out of whack Motronic will always be behind the curve. It'll be pulling fuel out at a cruise, then as you tip in it'll go lean, or vice versa. Every throttle movement will throw it out of whack and send the mixture into oscillation.

What you can do for calibration is completely remove the factory o2 sensor and use that bung for a wideband. Get it dialed in while open loop, then you can reconnect the stock narrowband sensor.

Another possibility for restricted power output is an obstruction in the exhaust system or catalyst. This is a possibility worth investigating. Before you go too far though I really would suggest putting it on a dyno. That will confirm definitively if it's actually down on power or not. A dynojet readout should have the AFR too. This would be very useful data.

[Technician117]

Thanks for the additional response. If I understood correctly, the AFM spring tension really shouldn't need to be adjusted provided I can reach a solid closed loop operation at idle with the screw in a reasonable setting, AKA somewhere in the middle of it's travel? The caveat being what AFR the engine is running at WOT. I have read through a bunch of the various threads on the AFM, setting idle, etc.

A little further detail on how it's running. It starts right up, goes into the "cold start" idle which comes down as it warms. Acceleration/throttle response when the engine is still pretty cool is not great. It should be running in closed loop by then, but it might simply be that - it's not in closed loop yet. Once fully warm, dipping deep into the throttle more often than not results in some surging while the RPM are rising and over 5k? rpm it just feels wheezy instead of pulling harder. If I'm not mistaken, these engines make their power higher in the rev range? sometimes it will pull "cleaner" without surging which is why I haven't mentioned potentially adjusting the sweep arm on the AFM. Especially if they board/plate is supposedly nearly new. I can see a line in the carbon but it's all the same color in the "groove". 

I hate running "blind" and am pretty convinced to just weld in a bung to the exhaust and run an AFR gauge into the cockpit. I suppose I could drop down the ECU, back probe the O2 sensor voltage pin and look at that while driving, but I'm not sure that's as effective even with the decent DVM I have (Fluke 115). 

Side note - I had previously smoke tested the engine for air leaks, etc. and only found a small one at the IACV. I applied silicone around the connector where the leak was and everything looks good. The brass idle screw being turned out 6 turns tells me there isn't an air leak anyway. 

I would not adjust the spring tension unless there was reason to believe it may have been tampered with in the past. I was under the impression that you suspected this.

The big question with Motronic is exactly how it handles fuel trims. If it works like the MS2 narrowband EGO control, it's a continuous, real-time correction trying to maintain the desired afr. This is where precise calibration of the AFM is essential. If its out of whack Motronic will always be behind the curve. It'll be pulling fuel out at a cruise, then as you tip in it'll go lean, or vice versa. Every throttle movement will throw it out of whack and send the mixture into oscillation.

What you can do for calibration is completely remove the factory o2 sensor and use that bung for a wideband. Get it dialed in while open loop, then you can reconnect the stock narrowband sensor.

Another possibility for restricted power output is an obstruction in the exhaust system or catalyst. This is a possibility worth investigating. Before you go too far though I really would suggest putting it on a dyno. That will confirm definitively if it's actually down on power or not. A dynojet readout should have the AFR too. This would be very useful data.

I got side tracked with many other small jobs on the e28. Radiator, trans fluid change, valvecover and rear cam housing/cover gaskets, new trunk seal, etc. 

Getting back to this topic - first - DUH!. Excellent idea on swapping out the factory O2 sensor for a temporary wide band. I'm not used to working on this era car where something like that is possible to do. LOL. I will do this when I get a chance to take my wideband setup off my track car. I had it on there for tuning as well but it's set now so not necessary to leave installed. 

Second - I removed the cap and wires while doing the VC gasket. I noticed carbon deposits on the leads. I lightly sanded them off. Not sure it made any difference. The cap and rotor are less than a year old and 5k miles. I was surprised to see that and wonder if it's the quality of the part. I had to go with something off the shelf at a parts store when I bought the car and was road tripping it home. I have a Bosch cap and rotor now as spares. Wondering if maybe I should install them and save the parts store brand as backups in the trunk?  

[Technician117]

Small update. I know I need to get the wideband hooked up, but I had the urge to tinker last week and this past weekend. I kept having a nagging feeling that there was a fueling issue as much as anything. I had already checked and knew that I had the correct fuel pressure. Last year I ended up swapping the regulator anyway. This doesn't check or help me verify fuel volume though. The PO had replaced the siphon pump in the tank but I couldn't remember the history on the external pump, so I ordered a new Bosch unit. I swapped that in as well as replaced the rubber hoses down there. I won't go into detail, but the pump and filter weren't mounted correctly and the feed hose from the tank was too short once I had everything installed correctly, so I replaced it. The pump I removed was Uro brand, for what that's worth. I'm not sure how many miles were on it. 

Anyway, I'm happy to report that the car is running better than ever before. It still needs fine tuning via the wideband, but now I can actually give it hard acceleration and have it feel pretty good with power delivery. 

I also have the gross idle screw set around 2.5 turns out. Per my Digisync, the idle is right around 850 RPM despite the tach showing roughly 700-750 rpm. TPS is set perfect. Also the Digisync shows my ITB's are balanced within 2 kPa of each other, which is a good sign. IACV seems good too as I can load up the engine letting the clutch out without having to use the throttle (if I'm careful, and only for testing purposes). 

[Technician117]

Another update, I think mostly for any future searchers. 

I temporarily installed a wideband using the factory O2 sensor location. She was indeed still running lean even after the new fuel pump made it actually run "right". So far I have the AFM enriched two "clicks" from where it was and the idle bypass screw is at 3.5 turns out. Probably good enough to call it set, but I might try one more click just to see. I feel like the numbers I see while steady state cruising and lighter throttle input acceleration is still a bit lean. I did see 12.5 at WOT near redline (maybe 5,800 rpm or so) today, but it was also near 90F. I'm wondering how much the AFR's are influenced by the brass idle screw and the AFM idle bypass while operating in low/light throttle scenarios? Going to mess with those while watching the wideband as well. Might as well learn everything I can while the wideband is still installed. 

[Philo]

Hey, cool progress with your car.  Flying blind when it comes to tuning these older engines is normal, and takes some old school knowledge and gut feel to get things right.  Checking plug burn, throttle balance, valve lash, setting the idle screw with the 02 connector under the plenum disconnected, will get your engine running pretty good.   Provided you have good compression and new tune up parts.     I didn't see mention of checking the engine temp sender (blue plug from memory, in the coolant pipe).  Bentlys discusses the sender test.  

When I had the S38 in my car I got bored and decided to install an aftermarket engine management system.  I ran a MAP setup and boy was it difficult to get a clean vacuum signal.  But eventually nailed it with a few tricks.  The engine dynoed at around 280hp at the crank.., with B36 cams.  

If you want to tinker.., these days there are so many aftermarket ECUs to choose from and will give you lots of clarity when tuning, but don't kid yourself there's a ton of work to get the engine drivable.   

If the car is clean, original, I'd leave it alone.  

Enjoy !

[tschultz]

Another update, I think mostly for any future searchers. 

I temporarily installed a wideband using the factory O2 sensor location. She was indeed still running lean even after the new fuel pump made it actually run "right". So far I have the AFM enriched two "clicks" from where it was and the idle bypass screw is at 3.5 turns out. Probably good enough to call it set, but I might try one more click just to see. I feel like the numbers I see while steady state cruising and lighter throttle input acceleration is still a bit lean. I did see 12.5 at WOT near redline (maybe 5,800 rpm or so) today, but it was also near 90F. I'm wondering how much the AFR's are influenced by the brass idle screw and the AFM idle bypass while operating in low/light throttle scenarios? Going to mess with those while watching the wideband as well. Might as well learn everything I can while the wideband is still installed.

 

I found this recently with my early E24 which doesn't have o2 sensor or closed loop.

You can tune the AFM if it is indeed lean by the wheel position or the central wiper position. Once those are set, you then finalize the idle mixture if it seems like the o2 sensor is still having some issues near idle:

[Randomg]

If you want to tinker.., these days there are so many aftermarket ECUs to choose from and will give you lots of clarity when tuning, but don't kid yourself there's a ton of work to get the engine drivable.   

I wish someone would have emphasized this before I went down the mod route. Car is way faster, but getting it to stock smoothness that drives in all conditions is tough. It's really easy to mess up an engine with a bad tune.

I'd be nervous about touching the AFM, especially if I hadn't made sure that everything else is perfect first, but those pictures look legit as far as understanding how it works. Diagram doesn't match the M5 with it's 5 pin AFM connector.

The AFR graph at the end is wrong for an s38 in my experience (even though you aren't going to be controlling that unless you get a tunable chip). I'd say 11.5 is way too rich to ever run it at, and might wash the cylinders if it happens at lower loads. Everyone I've talked to has said to target 13.0 for peak power on these engines.

[Technician117]

Hey, cool progress with your car.  Flying blind when it comes to tuning these older engines is normal, and takes some old school knowledge and gut feel to get things right.  Checking plug burn, throttle balance, valve lash, setting the idle screw with the 02 connector under the plenum disconnected, will get your engine running pretty good.   Provided you have good compression and new tune up parts.     I didn't see mention of checking the engine temp sender (blue plug from memory, in the coolant pipe).  Bentlys discusses the sender test.  

When I had the S38 in my car I got bored and decided to install an aftermarket engine management system.  I ran a MAP setup and boy was it difficult to get a clean vacuum signal.  But eventually nailed it with a few tricks.  The engine dynoed at around 280hp at the crank.., with B36 cams.  

If you want to tinker.., these days there are so many aftermarket ECUs to choose from and will give you lots of clarity when tuning, but don't kid yourself there's a ton of work to get the engine drivable.   

If the car is clean, original, I'd leave it alone.  

Enjoy !

Thanks

I'm not planning to put that much time/effort into the car as it's just meant to be a fun cruiser. I have other cars with much more HP when I need that fix. 

The car is indeed clean - not a spec of rust, but it's not exactly a show car. Definitely a driver, which is what I wanted because it's what I like to do. I've already put 6k miles on it since I bought it exactly a year ago (goes into winter storage as well). Up to 326,000 miles on it now. :-) I did indeed test the coolant temp sensors as well. The cold start loop works fine. 

[Technician117]

If you want to tinker.., these days there are so many aftermarket ECUs to choose from and will give you lots of clarity when tuning, but don't kid yourself there's a ton of work to get the engine drivable.   

I wish someone would have emphasized this before I went down the mod route. Car is way faster, but getting it to stock smoothness that drives in all conditions is tough. It's really easy to mess up an engine with a bad tune.

I'd be nervous about touching the AFM, especially if I hadn't made sure that everything else is perfect first, but those pictures look legit as far as understanding how it works. Diagram doesn't match the M5 with it's 5 pin AFM connector.

The AFR graph at the end is wrong for an s38 in my experience (even though you aren't going to be controlling that unless you get a tunable chip). I'd say 11.5 is way too rich to ever run it at, and might wash the cylinders if it happens at lower loads. Everyone I've talked to has said to target 13.0 for peak power on these engines.

I have indeed gone through the rest of the engine management system. Adjusting the AFM via the wheel is by far the preferred method IMO. It easily allows you to put it back exactly where it was before you started. Adjusting the AFM changes the AFR's through the entire rev range except arguably idle. That includes WOT, which I noted during my testing to look lean, so that's what lead to my initial "2 clicks" on the wheel. Subsequent testing showed 12.5 near redline, so a touch fat according to what you posted - which I don't doubt as I didn't have any S38 experience prior to this. I already removed the wideband sensor and reinstalled the O2 sensor, but I may try going back 1 click to see how it runs. At least now with all the data from the wideband I know what the car is doing for a baseline. 

[Philo]

I ran an external wide band 02 sensor (pre cat) on my S38 when I had the AFM installed and at WOT in the RPM sweet spot I'd see 12.5.    Just for reference.  Have fun.