BMW X3 - 5.3L 6 spd with big cam, erratic idle

[Fulltilt]

Hi everyone. Long time listener, first time caller. I just jumped into LS tuning. I built and tuned a Megasquirt 15 years ago, and just dusted off the DIY WB I built at the same time. It still works great. I ran it through the EGR and I am ready to play.

The car is a 2007 BMW X3 with a 5.3L, 6 spd. I milled the heads to gain 1 point of compression, and installed the Summit stage 3 Pro LS cam with 232 degrees duration.

Fuel injectors are from 2001 Corvette, Bosch 12561462 (28lb/hr)

I have a short commute and I am driving the car to work. It drives decent, but surges a lot. It will idle sitting still, but hunts and when I throw in the clutch at a light I have to pedal it to keep it from dying.

I did not originally intend to tune it myself, so a while back I sent the computer off to Wait4me and that is where my tune started.

I have been researching a ton and still can't figure out if I should focus on the idle first or the fuel. I tried the idle setup tips in the sticky and it made it surge worse so I undid them and decided to work on the MAF, but my fuel is clearly jacked up.

Just tonight I disabled closed loop and dynamic airflow, and as I went to turn the high RPM disable down to 100 RPM, I found that wait4me already turned it down to 400. The stock tune is at 4000. I found that interesting.

Anyway, I just went and logged and not only is my spark erratic, but my AFR also jumps around a lot. I'd love it if someone could take a look and see if some obvious things jump out at your trained eyes? I am excited to fine tune, but if I knew I was going to tune it, I wouldn't have installed the cam until I learned more. I think I attached the necessary files. I don't have a lot of PID channels chosen. Let me know if I am missing critical channels. Thanks for reading!

[ryan_axberg]

If your having idle issues you should log different idle channels. tps, iac if it is drive by cable. stit, ltit, throttle cracker, follower. You should also have fueling pretty dialed in also

[kingtal0n]

Ditch the maf unless you are a maf expert and have tuned lots of maf cars

Tune idle last because adjustments you make to fuel and spark will change it in the end anyways

Start by increasing airflow base numbers (many values/tables) to keep it from dying or stalling and use intuition to derive spark idle compensation (using 1959-1983 distributor/mechanical advance theory) to stabilize the idle for now

First tune the injector tables (voltage/pressure compensation and injector size) and VE table so the air fuel ratio on the wideband makes 'sense'
in other words if those tables are incorrect you will see strange VE table behavior, like huge numbers at idle or right off idle or something
The VE table should be smooth and generally increasing from left to right until peak VE of the engine (with a huge cam it will peak very late and likely not come down much)

Start with cruise 30mph, 40mph, 50mph, etc... safe easy to hit regions of RPM where the wideband is functioning properly. Exhaust at idle can be confusing to the WBO2 with large camshafts so avoid idle (again avoid idle tuning for now).

Always use OPEN LOOP (disable narrowbands) for tuning. Unplug them if you have to. Make sure there is no compensation in fuel from any memory. I never use closed loop on my cars for 20 years its not for performance, its so you don't have to 'retune' the engine when conditions change (like with age or weather) much.

The ideal A/F ratio for cruise is 14.8 to 15.2 most engines. Idle is usually similar. It's ok to run up to 15.5 or even 15.8 with light load but depends on compression and timing. This will keep the plugs hot and clean so the engine can achieve high mileage without building up carbon on the pistons/rings/valves as much.
Anything under say 55KPA for example should be near those a/f values. 15kpa-55kpa its fine. You can richen up slightly as you approach 55-60KPA to 14.5 to 14.7 a/f for safety.
Once you pass 60KPA start bring down the a/f ratio in a somewhat linear fashion from 14.5 to 13.5 eventually by 75-80KPA you may decide to go 'all in' at 12.5 or 12.0 at some point. Sooner with high compression than later. The lower the octane fuel, the more compression, the more spark, the sooner you need to richen up and more you need to richen up.

I assume you are N/A. With a modern Combustion chamber and high compression on gasoline typical max timing values are around 22-24* BTDC so keep this in mind, it won't take much timing to extract max power. From idle to say 3000rpm you need to start much lower than that, perhaps even close to 0* around 1200rpm for 100KPA region since you are MANUAL transmission and there is no converter to "slip" out of the way when you nail it at such a low RPM.

Once the fuel and timing is tuned the idle will be much easier, and you can use your CRUISE values to extrapolate (guess out) the remaining idle values in many maps.

[Fulltilt]

Thank you kingtal0n. I feel good about my game plan for working on VE tables now.

[kingtal0n]

the most important part is really nailing the injector delay and injector size tables. Once those are really dialed the rest of the VE tuning will 'make sense'. But if you started tuning the VE table and later on realize you fudged the injector delay or sizing... it means having to re-do the entire VE map usually.

[Fulltilt]

This was a big miss that I am ashamed to admit. My BMW fuel pump puts out 70 psi, and I was using injector flow rate based on 58.5. That would explain why it was so rich. I corrected that, but I am still not clear on delay, and can't find specs for the stock Corvette injectors related to that. Is there a specific parameter or table that I am looking for related to delay?

[kingtal0n]

Injector manufacturer provides delay sheet. For example goto fuel injector clinic and look at their spec sheets for HPtuners. Very specific values. You cannot guess.

If you dont have it, you need to buy new injectors with the correct sheet for HPtuners.

Now, I have tuned a hundred cars without the delay sheet, it can be done. But ALL of them were Stand-alones (haltech/aem/pfc/etc..) which allowed real-time tuning so I could adjust the delay and VE map simultaneously with the engine running to dial the injector close enough that it will work fine. Also the stand-alones tend to use a very scaled down approach, like one to five styles of graph with a scale factor (something you can easily drag with the mouse to fit a curve "curve fitting") whereas hptuners just has this enormous table of values to deal with afaik. So... what I am saying is... its possible in theory. But I wouldn't want to do it without real time tuning at least.

[kingtal0n]

I just re-read your post and realized you have stock injectors from a corvette? If so you simply need to find the exact tune file that the corvette used and copy the injector delay table over. In theory.

The specific table you are looking for shows a Pressure (MAP in KPA usually but sometimes its in bar or kg/cm^2 or PSI) vs voltage. It is used to tell the computer how long the injectors take to open at each voltage and pressure condition, usually fuel pressure vs voltage.

More info If you want:
The pressure below (intake manifold pressure or MAP KPA/PSI) and the pressure above (fuel pressure in PSI or KPA etc...) Affect the injector opening time delay. Furthermore, applied voltage also has an affect. For example very low voltage will open the injector slowly. So will very high Intake manifold pressure, or high fuel pressure, I believe. Since the fuel pressure is usually several times higher (and thus more 'effective' at altering injector delay) they usually only go by fuel pressure and system voltage and just neglect the manifold pressure. Or there may be another static calculation for the manifold pressure idk. The point is to understand that the pressure on both ends of an injector orifice combined with applied voltage all affect the rate of injector opening (from fully closed to fully open, like a transistor, I guess).

The (HPTuners ECU) computer uses airmass and desired air fuel ratio to determine fuel mass that needs to be injected. Then, it will use injector size to figure out how long to spray to get that desired fuel mass into the cylinder. It takes that value, and then ADDS the injector delay look-up table to that, to compensate for the delay to open the injector and get a spray going.

So lets say your idle fuel pulse needs to be 1mS = 1 milliseconds (.001 seconds)
But it takes 600uS (.6 mS) to open the injector.
If the delay table lookup value is, say, 800uS,
you would get 1.2mS of actual fuel spray when you really only wanted 1mS of fuel spray. So, 20% more fuel from just a slight miscalculation.
Yeah it will totally throw the VE table out of wack at low fuel spray values (cruise, Idle).
It doesn't matter nearly as much at High output situations because the fuel injector pulse is so long (you might see 12mS or 15mS of spray at upper RPM WOT) that the tiny injector delay is negligible. So the engine can be tuned well at high RPM if you don't have delay values but getting the VE map to 'make sense' will be very difficult in idle and possibly cruise regions ( the larger the injector, the more difficult) using the wrong delay values.

[Fulltilt]

I just re-read your post and realized you have stock injectors from a corvette? If so you simply need to find the exact tune file that the corvette used and copy the injector delay table over. In theory.

Boom, there it is. I checked tunes from a Vette and Camaro with those injectors and they matched. Offset is the table you are describing, which makes sense now, it just took me a while. There 5 other parameters or tables that were different as well.

It is running better already. Nearly ready to work on the VE table.

THANK YOU SO MUCH!

[Fulltilt]

Lots of learning this week. Got the VE table safe, as it was WAY off and way lean in certain areas. Made the switch to Lambda from AFR. Learned that IAT heat soak issues are worse without MAF and can have you chasing your tail. And I learned that too little advance can be a very bad thing.

If my VE table is close, I want to re-enable the MAF, as I intend to run the MAF and I feel like it will help me dial in my idle. Does that logic make sense?

[kingtal0n]

MAF is a fine thing if you know how it works professionally. Understand completely. For example, how air is a fluid and it can flow through the tube towards/from a maf and small perturbations in the fluid flow cause erroneous maf readings. turbochargers create all kinds of disruptions under certain circumstances but I guess you dont have that issue.
The throttle valve slamming shut creates an "water hammer" effect in the fluid flow of air, which can disrupt the maf sensor.
The maf hotwire needs a constant, smooth airflow over it, to be effective. So there should be plenty of straight plumbing before and after the maf, and no sharp corners in the flow of the tube. It needs to be positioned away from the throttle valve and turbo so their perturbations do not influence maf sensor.

All in all, more trouble than its worth in most cases. I dont know why people keep trying to use the maf sensor when there is no issue using the map sensor. A/F ratio can be tuned either way, its the same end result. I suggest you learn to deal with the MAP.

The issue you face is typical a/f swing due to IAT heat soak... This can be managed.
1. Relocate the IAT sensor so it doesnt heat soak AS MUCH. I put it approx 20 to 40" before the throttle body, after intercoolers.
2. You can adjust the a/f bias towards CTS or IAT as necessary. I just disabled the CTS side of things in my application and *deal* with the IAT alone.

3. Tune the engine in open loop so that the a/f values around "heat soak regions" (idle, near 0mph, parked, is when u heat soak the most) Can run 14.7-14.9 when the car is NOT heat soaked. This way when the vehicle heat soaks they lean out to 15.5-15.8 range, which is fine. The engine can run lean like that safe when near idle speed. It will help keep the plugs clean and improve fuel efficiency a tiny bit.

4. If necessary (if it makes you feel better) Just let closed loop add fuel to compensate for heat soak. You can even use a wideband analog output (0-1v analog output from some widebands) to simulate the narrowband sensor so you can target any a/f ratio you want. Once you understand what is happening (IAT rising and leaning out the engine) You can live with the fuel trims adding fuel for those situations. Just make the swing "small" so it cannot add or subtract too much fuel, let it "max out" easily, so you always stay within some decent range.

5. And last the best for last. Get yourself a Variable resistor 0-1000Ohms is fine. Maybe even 0-1500Ohms. Put this inline with your IAT sensor so you can control the temperature of IAT manually. That way if it really heat soaks bad for some reason you can simply dial the temp back down by adding resistance from the driver seat.

[Fulltilt]

Interesting. I am convinced. I ordered an IAT sensor and pigtail and will ditch the MAF.

[Fulltilt]

The car is now starting and idling amazing! VE table is close in the cells I am driving through on my short tuning runs around town. I need to get out somewhere where I can try to hit a lot more cells. I am being very cautious and have not taken it above 4000 yet or done any WOT.

Regarding timing, I started with a truck tune and even backed timing off by about 10 degrees, which I learned was a mistake, as I was already much less timing than my engine needed. As I stepped deeper into the throttle and introduced more load, I was getting popping or misfires. AFR looked fine, so I started bringing timing back in. This seemed to help a lot. I brought about 13 degrees back in for much of the table, and the car started feeling much better. Then I looked at a Corvette tune and realized my whole table was still shifted down a ton, so I brought in even more. I finally started seeing 1 degree of knock in a couple cells. I see a lot of discussion around real vs false knock, and I have no idea which I have. Here is my dilemma: I'm still getting this popping and major loss of power as I induce more load, and it is right next door to the cells that are 'starting' to see knock, which makes me think it is not an underspark situation. I haven't messed with PE yet, but that's because I don't think I am even 50% throttle position yet. I'd love your thoughts and a look at my log, but would you be able to see much without my math channels for my EGR/wideband? I also had to create custom math to give my spark and knock tables the cylinder airmass it needed. Maybe just looking at my spark tables would give you an indication if that is al jacked up?

[kingtal0n]

I cant look at logs or tunes on this computer. And my tuning laptop is in another city atm.

However if you post them log/tune somebody can look at least.

Also we can discuss what timing SHOULD be and what can lead to misfires and pops.
First just making sure you are N/A (no forced induction)
Typical NA max timing is near 22-24* btdc for peak VE (volumetric efficiency). To find peak VE you look for peak torque. The peak should be close to 99% VE when engine is NA.

You will reduce timing near peak VE / peak torque but lets come back to that.

The timing for low rpm is reduced to give the engine a chance to move out of it's own way. So while you may have 24* of timing at 3800rpm, you may only use 12* of timing for 2200rpm. Or less. The timing can even be negative some low rpm places.

Do not depend on knock sensors to tell you when to remove timing. They are *ONLY* Last resort/ditch efforts. This is a rule stemming from forced induction because if you make a very high power/displacement with a cast piston and just tickle the knock sensor it may fracture a piston. So we avoid triggering the knock sensors at all costs.

That means using very low timing and depending on EGT to determine when to add timing AS NECESSARY. Again this is a forced induction safety tactic. You may not need be so cautious but it still worth teaching you how to advance timing safely without a dynometer.
As timing is input the EGT will drop. If you add 1 degree and see a LARGE drop in EGT then that 1* was very important. However if you add 1-2* and see little to no change in the EGT Then you should immediately remove that timing, it was unwelcome in terms of HEAT input to the exhaust system which means the fuel is burning in the CYLINDER instead of the exhaust. This should make sense logically / common sense, you'd prefer if most of it burns in the cylinder and does not overheat the exhaust.

The ultimate ratio of EGT to cylinder pressure is an infinite number of potential outcomes which depend on engine rate of change (which varies with load and vehicle weight and gearing), heat input/output (to exhaust, coolant, air, into metal or from metal contact, etc...), displacement/power/compression ratio, cylinder head design/piston design, and fuel quality/type (alcohol can reduce inlet temps whereas C16 leaded racing fuel does not, so even though the C16 racing fuel is more TOLERANT of high temperature it does not help the engine parts themselves to RESIST that high temperature, they can still become damaged by high temp so alcohol is a "safer" fuel).

For low rpm timing, think like a mechanical distributor. From say idle to 2800RPM you will begin near 50KPA, 15* at idle, and ramp in towards 22 to 24* total by 2800rpm for example at 100KPA. LS engines do not seem to enjoy alot of timing before 2800rpm.
The timing at low RPM near 100KPA is bit a less, for example 1000rpm and 100KPA (not all cars can hit this spot) this might be near 0* of timing if the engine is high compression and the vehicle is heavy. It could even be -2 or -4* btdc of timing. Take a look at some truck graphs and compare to car graphs, maybe that will help.

Without a dyno, and without forced induction, you can easily use low timing values and go WOT at low rpm and then see how the engine behaves at proper A/F ratios.
For example say you input -3 to 15* of timing and WOT the engine pops and misfires, you know you can add timing.
I believe you have done this already. Good.
By watching a/f ratio you can make sure the engine will not over-heat (melt parts or warp parts or damage parts). The only thing is to consider the EGT (which won't get too "out of control" since the engine is not Forced induction, and the WOT pulls are SHORT use 1st or 2nd gear).
Keep adding a bit of timing till the a/f stays constant, and the engine pulls smooth. Use the least amount of timing necessary to achieve this. As the gear ratios numerically decrease (3rd, 4th, 5th etc) this "LOW" timing you set back in 1st and 2nd gear will become "nearly perfect" because longer gears means more heat buildup in the chamber which will call for reduced timing and create faster combustion reactions, so it will be very *GOOD* that you set the timing low to begin with. In other words, the first time you make a pull in 1st or 2nd on a recently started engine, the timing will be "too low" to feel very strong. But once it runs for a bit, heats up, makes a pass in 3rd or 4th gear "longer pulls" the timing will start to make sense and the engine will be tuned 'perfectly'. This can be confusing to people because they start their engine, immediately make a pull, feel lack luster then add a few degrees and the car pulls much harder and they think it was necessary to add that timing when in reality they just fucked their engine once it fully warms up, it will eat rod bearings and possibly damage a ring-land (forced induction is unforgiving). Perhaps without ever triggering a knock sensor.

Cruise timing I like to use 38-40* in most engines, the LS seems to be OK with anywhere from 38-45 sometimes even 47* I've seen in factory maps. I hope you know that cruise timing is far more advanced than WOT timing (Im sure you do)

So lets say you have the a/f ratio at 12.2:1 and the timing 24* by 3000rpm and still have misfires.
You should check the plugs, compare them. make sure the gap is correct for application. I use tighter forced induction gap. You should be around .035" or .044" range. Wider gap is better spark delivery but wider can blow out from boost pressure. I suggest using stock gap for NA application, nice and wide.

Check the plug wires obviously. Look in the dark for arcs.
Make sure the injectors are all working... somehow. Do continuity tests (test from the ECU-> through the injector plug-> out the other side of the injector) and wiggle the injector clips alot to make sure they dont lose connection.
The injectors can also be clogged. This will create misfires as some cylinders fail to receive enough fuel. Then you can't trust the number on your wideband because its an average. This could be a death sentence so always use high quality injectors and CLEAN old injectors before using.

theres other stuff but start with that.

[Fulltilt]

Wow, that?s a lot to think about. The car is NA. I would need a EGT sensor to read EGT, right? Is that common for a NA car?

[kingtal0n]

no i wouldnt bother with an EGT on an NA car lol. I was just trying to show how important EGT is to the tuner, he must always consider EGT and EGP (temp and pressure) of the exhaust, its just that NA cars can set their own exhaust resonance using headers (there is no forced induction manifold collector), and since there is no way to adjust pressure beyond exhaust design diameter (no boost controller on an NA car) flow rate can not exceed whatever 100-105% Volumetric efficiency allows for at some RPM range (for many stock cams, 3000-5000rpm for example).

without forced induction the factory engine can work well within its designed expectations using typical gasoline fuels without worrying about overheating the EGT usually. If tuned properly, of course. If you run an engine at 15:1 WOT for a while it could easily overheat the EGT I bet. But we are targeting and verifying the low a/f ratio using wideband so that is not usually an issue.

[kingtal0n]

Wow, that?s a lot to think about. The car is NA. I would need a EGT sensor to read EGT, right? Is that common for a NA car?

It just seems like a lot because its typed out completely. If, instead of typing, I can just show you a timing map example, then it would look alot simpler I think.
here is my base timing for 87 octane and 93 octane, I use this table for both on a 2002 5.3L 9.5:1 compression LM7. If you have an engine with more compression you would likely need to be even more conservative.
mytiming.png

notice soft spot 2800-3200rpm from 0.60 airmass. I am using a very tight 9.5" 2800 stall, the tightest lowest aftermarket 9.5" you can get so I think the engine loads into that range just before boost. I also use AFRadvance table to adjust timing in boost, so Warning to anybody reading my timing map values in boost get adjusted significantly by my AFR advance settings, see the table in the next picture


My table should answer alot of the questions you have about timing, and then ask whats left

[Fulltilt]

New information: I was running 89 in the car before the swap. When it was at 1/8 tank, I filled the tank full with 93. At the time of my recent post about popping, I was down to the bottom of the tank. I just went and filled it again with 93, and my latest log shows no knock at all. I am going to post my tune to a new post asking for a look at my spark tables only, just to see if I am way off in areas. I'm starting to think I still need more spark.

[Fulltilt]

spark.PNG

Take a look at this. Based on your table, I'm thinking I am way high.

[kingtal0n]

If your engine is high compression and has a manual transmission then yes that table seems aggressive in timing

also it doesnt make sense, there is a timing peak near where peak torque would be. You want the opposite, low timing when cylinder pressure is highest. Actually if the cam is really over-sized (8,000rpm cam lobe) it won't peak VE at 3200-3600 anyways. More like 4,000rpm start to peak maybe even 4,400 for very large camshaft duration. We would need to see a dyno graph to be sure where peak torque wants to wind up.

High timing is 26-28* at WOT. The only time an LS can use those numbers is where cylinder pressure is falling off, torque is being reduced. In other words, a stock engine can carry 26 27 28* from 5,500+ because the stock camshaft is reducing torque and therefore cylinder pressure, some 20 to 30% usually by redline. So when the engine torque drops and cylinder pressure drops, the engine can withstand more timing, so we might see values ranging from 26-28* range. A well/large cam'd engine on the other hand will have a pretty flat torque as it approaches redline, so timing would be kept low 23 24 25* ranges usually.

So in theory that bump 3200-3600 could be due to poor torque from having a too-large camshaft. In other words if an engine has a cam that big when racing the driver would down shift or if in 1st use a large enough converter to avoid it entirely (drag racing). But we would need a dyno graph to figure out when its really safe to increase timing.

Another thing is you don't want the same timing number from cell to cell. Never use two identical numbers next to each other. The electronics isn't perfect, if you command 35.0000 you really get a variability like 35.0154, 35.0284, 35.0488, 34.998, 34.915*, etc....
Always use a timing number difference from cell to cell to keep the variability going in the same direction, e.g. so the numbers always vary in the same direction. The computer might be imperfect, but the physics of a rotating engine is keeping very careful track of all forms of energy and mass, and the slight perturbation timing error has an influence on the frequencies and smoothness of a running engine. To put this one more way, as an engine accelerates its rate can be increasing or decreasing, it can be speeding up or slowing down all the while still accelerating. In other words you can accelerate an engine while simultaneously slowing down the rate of it's acceleration.
As the engine is accelerating and gaining or losing rate of acceleration (2nd and third derivatives) the timing profile needs to keep pace.
For example an engine which is accelerating while also gaining rate of acceleration needs to be pulling timing out to compensate for the new arrival time of the piston each time it comes back around on the 4th stroke (4 stroke engine). The rate of combustion is influenced by not just the fuel and temp of the fuel, but also the rotating speed of an engine and its derivatives (acceleration and so forth). Our job is to time the combustion event beginning so it can culminate so some useful reasonable predictable pressure integral over a useful portion of the power stroke rodxcrank angle without having the pressure go too high that it blows a hole in something or so low that it raises EGT and melts exhaust parts and potentially melting the head or warping it at least. Also be aware there are some fuels which have such LITTLE octane that they simply cant be used- they always explode suddenly, violently, with a pressure spike. For example n-heptane, is Zero octane rating. It contains enough carbons to be classified as a type of gasoline alkane and yet completely useless to run an engine because of its terrible octane rating. This example shows us that it doesn't matter what value we choose for timing if the fuel quality is reduced for whatever reason (high temp, contamination, age, etc...) the engine can still be blown apart with low timing conservatism so be aware of fuel quality all the time.
For another example if the engine is accelerating but losing rate of acceleration (such as near the top, 5600-8000 high RPM while torque is falling off due to a mild camshaft or stock camshaft) then it is usually safe to start increasing the ignition timing to compensate for the reduced torque and reduced rate of acceleration because the piston will be arriving later and later with respect to time and things may also be 'cooling off' a little with the reduced heat and pressure of a falling torque.

so here is a typical high performance mild 5.3 (lets say 218* @ .050 duration) camshaft curve, 100KPA WOT timing with lets say 9:1 compression
1200rpm -3*
1800rpm 3*
2200rpm 7*
2600rpm 10 or 12*
3200rpm 13 to 15*
3800rpm 15 to 20
4200rpm 17 to 23, approaching peak torque
4800rpm 16 to 21 just starting peak torque
5200rpm 16 to 20 through peak torque
5800rpm 17 to 22 leaving peak torque
6200rpm 18 to 24 torque is falling off, power somewhat flattening out
6800rpm 20 to 26 torque is gone, power dropping

Notice we were ramping in timing up until peak torque, then take a bit out to nurse it through peak torque, then add it back in once torque drops out
MILD cam engine share this distribution, it looks like an upside down U on the dyno curve. I can post my graph if you want...
Large camshaft engines tend to move the "U" shape so far to the right that it looks like one straight line from left to right and we never see the "falling off" part of the U because its over in the 8000+ region which is why those cams are for racing