New Project - Tuning my 3.4L for POWER
 

i have goodies coming this week.

the problem
i am getting pulled by cayman S's on the back straight at road atlanta. i am lighter, ergo i am down on power!

the hypothesis
all my mods (3.4L swap, custom intake, cat bypass, headers, race exhaust, pulleys, AC delete, etc) are too much for the stock 996 ECU tune. AF ratios likely incorrect.

the solution
generate current fuel & spark maps
set car up to dyno itself
use AFC to modify AF ratio
determine how fuel is affecting spark
calculate ideal AF map
dial it in


this will be fun! i have an Innovate LC1 wideband O2 sensor & controller. it will hook up to an Innovate OT-2. basically, the wideband sensor AND my OBD II data will be integrated & wirelessly logged to my iPad. the software will take inputs for weight, wheel diameter, frontal area, Cd & gear ratio. i will be able to log spark, AF ratio, RPM, TPS, MAP, calculated load, OPEN/CLOSED MAF state, etc. the log files can be emailed from my iPad to my PC for detailed analysis.

the logs will actually calculate HP & torque. accurate or not, it should at least be precise since it is using the mechanical relationship between RPM & vehicle speed to determine velocity. should be great for analyzing whether a particular change is helping or hurting.

i will have my current spark & AF maps w/ initial dyno in a week or so. after that, the games will commence.

questions / comments in the meantime are appreciated; i know some you have done this before (JAAY!). cheers.

Subscribed. There's valuable information in this to be had. Wish you luck.

Guessing: Injectors &/or fuel pump not putting out like they should. Air induction mods are allowing too much extra heat into engine &/or engine compartment.

good guesses; we'll see. my first thought was similar: inadequate fuel. JAAY's experience, and those of others who have done this swap, is an extreme rich condition (he was 10 to 1!!). in my case, i do see a lot of soot across the back of my bumper that gathers over time.

i wonder if the flow charicteristics of the 996 air box don't translate well to a straight-through MAF housing. on the 996, the MAF is VERY close to the airbox inlet & not entirely round.

variocam+ would account for some of the hp on the table vs the cayman.

but if we are not comparing to caymans but rather looking at 10:1 afrs using stock tune, then why? if the maf was consistently over-metering the air (resulting in more fuel than required) then the o2 sensors would pick up on it and adjust the fuel trims. similarily if you had leaky injectors or an over-performing fuel pump producing more fuel than expected then the 02 sensors would notice and adjust trims.

so, perhaps new o2 sensors? a quick trip to auto atlanta shows that the 986 02 sensor has a different part number that the 996; most likely due to lead length, but perhaps they are different beasts in other ways?

otherwise, the ecu will add more fuel to keep the heads cool if it is nervous about something - timing, octane, etc. - what gas are you running and is it a row or na 996 tune?

nope; i have a 3.4L from a '99. i have varioRAM (which is hooked up & functional), but only standard varioCAM.

I recently installed 987 airbox & MAF tube with 997 plenum "T" & T/B, then reflashed with '00 LEV 996 map for my 3.6 boxster engine & it seems to work very well. Pulls to 7400 very smoothly.

A good project to investigate and a common problem on 996/Boxster conversions.

Some observations while running regularly with 3.4L Boxster conversions and 987 Cay S 3.4L:

The 987 has a nice wide, fat torque curve which allows them to put down power early on corner exit. A 996 3.4L tends to have a more narrow torque curve requiring more shifting to stay in the sweet spot. I suspect a well tuned and displacement-matched intake and exhaust along with software remapping will fatten up the torque curve and make the most out of a 996 3.4L in the Boxster.

One problem that has been reported is sonic pulses from the valve train traveling back up the intake and actually fouling the MAF sensor readings. Does the intake require a helmholtz chamber to neutralize these pulses?

i was reading over some chatter about a similar issue on another forum. one guy said his AF issue was solved by raising his fuel rail. he was using the boxster fuel rails w/ the 996 injectors. when he raised the rails to the right height, his AF problem went away.

this doesn't make sense to me. anyone see what i'm missing?

(FWIW, i know that the 986 & 996 injectors are identical. my personal solution was to modify the 996 rails to work on the 986, so this issue is not a concern to me).

interesting observation; i wonder if that is why there was a resonator on the original intake? anyway, i should be able to log MAF voltage vs. RPM at WOT. if there are any anomalies, i can probably program them out.

Fuel pressure and injector setting differences in conjunction with intake airflow differences....just a thought

ltft should adapt to all that fuel stuff unless it is out of adaptable range, at which point it should chuck a code. freaky maf readings due to turbulence or pressure beat waves should also trigger a code. i think either the 02 sensors are off or the computer wants it to be at 10:1 if it is running that way without codes.

long term fuel trims don't really do much except use the 02 sensors to set the AFR to 14.7 near idle. at higher load & RPMs, the (narrow band) O2 sensors can't really tell what the AFR is. the ECU uses the MAF to calculate, using a table, how much fuel to send based on a trimmed out 14.7 idle AFR + some offset values.

once you are over about 3k RPM and 30% throttle, the fuel amount is basically an educated guess based on MAF readings and offset tables.

since my engine will perform so much differently at wide open throttle (WOT), there is a good chance that the ECU tables won't get me to the right AFR's. there really is no way to know without measurement.

on another note, my logger came today. pretty slick piece of equipment! i fired it up to look at my fuel trims and MAF readings. fuel trims are small (<3%). one interesting note: the MAF at idle should be around 15 kg/hr. mine is over 20! so, at least at idle, i am flowing a LOT more air than stock. my guess is that flow advantage decays as RPM's go up, so i am likely getting way too much fuel at high RPM/Load values.

i will get the wideband hooked up & start logging later in the week.

How many of you actually know your afrs???? I'd love to see a stock curve. I'd bet there all rich as hell on Boxsters. - detuned to preserve the carrera.

Here is one guys 3.4L tuning solution that was featured in Excellence magazine. He is a regular street driver and not a shop owner or track rat.

986 Boxster story

There used to be tons of Dyno data for 2.5 Spec Boxsters regarding different exhausts, underdrive pulleys, intakes etc. but all of it was lost when Brad's site went down. As I recall there was concern for the cars running too lean with straight pipes and actually ran better with a muffler of some type. Too bad it's gone.

These cars are all running lean with a stock 2.5 showing the best tune and a steady AFR.
http://boxcar-racing.com/forum/index.php?topic=688.0

I LOVE this.

from what i understand there are three base maps - timing, idle, and fueling.

the fueling map is three dimensional and uses rpm and airflow to determine how much fuel to deliver.

typically, the lower-rpm end of the map delivers a stoichiometric air fuel ratio (14.7:1) which gets richer (more gas) as rpms increase (to cool engine, avoid knock, etc.).

the o2 sensors are the only true indicator of afr that the computer gets, but the narrow band o2 sensors are only really creating long term fuel trims for the low-rpm area of the map as they do not work outside of a small range centred around 14.7:1.

at low rpm the stock porsche engine is designed to have a pulse-tuned intake and backpressure from the exhaust to improve cylinder scavenging and fill. on an engine with intake and exhaust modified to increase airflow at high rpm, some of these low rpm benefits will be reduced. this is why most folks see a reduction in low to mid-rpm torque but with an overall increase in power in the upper rpm ranges. to the computer this means the engine is drawing less air at low rpm. the o2 sensors should detect this and reduce fuel as a response. when the rpms increase and the engine is breathing more freely the computer doesn’t know it and you should actually go into a lean condition. the leaner condition at wot is supported by the comments from topless.

another possibility is that the 02 sensors are no longer working properly. in this scenario the computer doesn’t detect the reduced airflow at lower rpm and continues to fuel as planned, resulting in a rich condition. this should transition to a less rich condition as rpms increase.

oddly, you are getting more airflow than expected, but that is at idle which is a different map.

otherwise, the computer is looking at throttle position as well as maf output, so one presumes that any wild fluctuations in maf output would get detected and ignored. a maf that is consistently reading off due to steady-state turbulence might be a culprit. my understanding is that there is also a maf calibration map, so any consistent maf inaccuracies should be addressed by the computer as determined by low-rpm o2 sensor feedback.

if the maf is metering more air than is actually flowing (ie, located in a high pressure area of the intake such as the outside of a bend) and, as rpms increase, the inaccuracies of the maf readings increase in a greater than linear manner such that more air is metered than is actually flowing beyond what has been accounted for in the maf calibration map, the mixture will get rich and the computer would not know it. in this case the best solution would be to simulate the 996 intake geometry and locate the maf as close as possible to the air box.

regardless, we are talking a 996 engine with 996 tune. presumably Porsche isn’t leaving any power on the table with this combo. i find it hard to believe that intake and exhaust modifications can have such a drastic effect on afr. i think the real test would be to compare your wideband o2 results to a stock 996, or even get your hands on the base map for a 996 to see what the afrs are supposed to be from Porsche. my thinking is that you’ll find that porsche wants a 10:1 at wot to deal with vagaries in fuel octane, etc. it’s all really moot because, regardless of the cause of bad afrs, if you can improve them …

insite, I am watching this and will be next in line; you have proven results from jaay and the guy topless posted about, and now, hopefully you!

Insite is your MAF housing tube I.D. 3.5" ?

certainly there are lots of different variables. one thing is for certain: the stock ECU map for a 996 contains no fuel ratios near 10:1. it's a losing proposition: less power, more emissions, worse mileage. the only cars that need to run in the 10's are augmented (turbo, supercharger). the extra intake heat & cylinder pressures raise the probability of knock. the extra fuel is used to cool the charge (latent heat required to vaporize fuel is around 300kJ/kg) & stabilize the flame front.

the accuracy of MAF readings in a given system is largely irrelevant so long as it is PRECISE. the ECU is tuned to the specifics of that vehicle. if the MAF reads 20% low compared to reality because of flow idiosyncrosies, it doesn't matter, because they program the ECU with maps that take that into account. that means that with a completely different intake, the stock 996 maps on my car may be irrelevant even though the engine is from a 996.

most NA cars will make good power between 12.3 and 13.5 AFR. i will be generating maps at WOT that model lambda from around .83 to .92 in 1% increments across the RPM band. i will plot HP vs. RPM vs. AFR, chosing the power path & programming the computer accordingly. i will also monitor knock to make sure things are safe. one thing i need to learn more about is ethanol in fuel. does RON+MON/2 account for the anti-knock properties of ethanol? i.e. does ethylated 93 octane fuel have a higher EFFECTIVE octane rating?

i digress. at any rate, as i do all of this, i will have many pretty charts, graphs and tables to spell it all out. it will be cool. my guess: 50HP (crank). quote me on that.

yup. i use a BMC airbox w/ silicone plumbing to a 3.5" MAF housing.

most NA cars will make good power between 12.3 and 13.5 AFR. i will be generating maps at WOT that model lambda from around .83 to .92 in 1% increments across the RPM band. i will plot HP vs. RPM vs. AFR, chosing the power path & programming the computer accordingly. i will also monitor knock to make sure things are safe. one thing i need to learn more about is ethanol in fuel. does RON+MON/2 account for the anti-knock properties of ethanol? i.e. does ethylated 93 octane fuel have a higher EFFECTIVE octane rating?

I have wondered this too. I'm guessing that it is unregulated other than to insure it has the posted octane as a minimum. Consumer reports a 15% reduction in MPG when using "Gasohol"

996 is 3.25" ID, no? and, from what i recall about your install, your maf housing is located right after a bend?

i just read up on octane ratings. aparently, the RON & MON measurement methods actually use a running engine! shocking. at any rate, the anti-knock index (octane rating on the pump) is related to the performance of the fuel & not necessariliy to the CONTENTS of the fuel.

also, turns out E85 is only like 96 octane. looks like 93 octane is probably 93 octane, regardless of E content.

pretty sure the 996 is 3.5".

MAF sensor should be at least 1 and ideally 1.5 pipe diameters from a bend. in this case, that would be 3.50" - 5.25". i am at the upper end of that range.

EDIT: if you look at the 996 MAF housing, the inlet is a funky shape and VERY close to the airbox itself. probably not a comparable measurement environment to the one below.

here are a few notes from others as i went through the intake design process:

tholyoak

blue2000s

i think the wideband 02 and afc initiative is fantastic and am really excited for the results, but in the interim perhaps look at revising your intake to see if any power gains result (with your wideband 02 you would be able to definitively determine what, if any, changes to maf readings this would effect). if the 996 maf housing is indeed +/- 3.25" ID then it is a direct match to a bmc 85; you could couple the maf housing direct to the end of the filter then elbow towards the throttle body. room is tight in there, however, and such a config may not fit (didn't for me) but i've since found that some room can be gained by pulling the insulation off the inside of the engine bay.

i'm in that process right now; my intake design is similar to yours (copied in fact) and i am looking to use the 987 airbox instead (3.25" ID) as it gets my maf housing closer to the filter as per oem. i also worry about vibration affecting readings from a maf that is hung out there instead of tied to the air box.

Sounds like a very methodical way to accomplish it. Are cylinder temperatures at WOT a concern to you?

How are you going to create an AFR altered map from all your data and then code it, test it, upload it to the ECU? I am talking hardware and software ....coding language etc. Are we simply talking about manipulating the MAF signal as a means of correction?

I ask as it seems difficult to get any custom tune work done and I would like to do this for my 3.2

Sounds great let us know how it works for you

not messing w/ the intake anymore. i think it's reasonable. the bigger issue IMO is calibration of the MAF w/ respect to the ECU's AFR maps. for me, i think getting rid of the giant airbox all together was a smart move. the BMC is slick & flows well; it also allows me to elbow over to cold air easily.

Yes so I have been wondering what will come about with the 987 air box in a 986 with a 986ish tune. The 987 MAF sensor is right around a bend and convergent section where as the 986 is not

jaykay -

cylinder temps are a concern, but only insofar as they induce knock. i will be pulling a knock signal & also checking spark to see if the car starts to pull timing. if that happens, i will richen up a little.

to make the actual fuel corrections, i'm using a piggyback computer that will modify the MAF signal. i'm then checking the actual AFR's with wideband to verify / tweak the result. i will be able to measure rough spark & fuel maps by monitoring the OBD II port and a wideband sensor. i will drive several test points to gather data at 10% TPS intervals from 30 to 100.

when i change AFR, the car will likely change spark as well. i will try to monitor that. over a certain RPM, the spark advance should be fairly constant. we shall see.

BTW, there is a product called MAF Translator that i almost bought. i couldn't verify at the time whether it would work w/ my car. i have since come to the conclusion that it would. it has the capability to run closed loop wideband setup. basically, it hooks up to your wideband & adjusts the MAF signal to achieve the desired AFR's in the map you load to it. pretty cool.

had a few minutes to log some data today. very interesting to say the least. this was quick and dirty; i will perform a lot more testing later. first, let's look at the fuel map:

http://986forum.com/forums/uploads01...1366415154.jpg

what do we see? first off, at idle and low revs / TPS, the car is doing a great job maintaining roughly 14.7.

next, at WOT, it looks like we average AFR around 12.0. that's a little rich, but not bad. we have to run it against power numbers later at different AFR's to see empirically what works best.

now here is the funny thing: WOT is only 83% throttle on the TPS!! the car is using this to pull fuel and spark maps! also, it's possible the throttle isn't opening all the way. i need to look into this.

next, spark. here is the quick and dirty map:

http://986forum.com/forums/uploads01...1366415343.jpg

what do we see? at partially open throttle, it advances nicely. once i hit 40%+, it looks to me like the car is pulling timing. i have to read up on spark maps, but i would think at WOT and heavy throttle, the timing would be in the upper 30's / lower 40's.

it's possible the knock sensors are sensing one of many things. my car is stripped out and LOUD as hell. race exhaust, no cats.

Fertile ground for tuning there if your data is accurate. I see another 40-50 horses.

ok, had a chance to log more data this weekend. still dealing with some choppiness to the ECU output. for whatever reason, the OBDII data is only pulling roughly one data point per 0.8s. this isn't too big an issue if one pulls 3rd or 4th gear runs. i can't do that on public roads; i'm stuck with second gear. in second gear, i can pull 1500 RPM to redline in about six seconds. that's not too many data points.

i talked with a guy in australia who used the same hardware to log data on his early 996. his pulled a lot quicker for some reason. FYI, he has OEM spark and fuel maps for the 3.4L 996 on his website (oz951.com). he does lots of slick tests & checks. good site.

i am a data guy. i decided to take what i have & make it workable. i ran the 'choppy' data to get some maximum points & used smoothing and scaling to develop workable curves. i then performed several runs on the identical section of road in the same conditions. the goal was PRECISION, not accuracy. i am most concerned with the ability to compare one change to another.

here are the parameters i used. the HP / Torque calculations use SAE correction factors. drag is considered.

Weight: 2890lb (includes the car, me, 1/2 tank of gas + the spare tire)
Tire Diameter: 25.5" (NT-01 275/35/18)
Gear Ratio: 2.117 (2nd gear)
Final Drive: 3.889
Frontal Area: 19 sq. ft. (WAG)
Cd: 0.40 (WAG based on the fact that i have my top down)
atm press: 29.92 inHG
Temp: 68F
Driveline loss: 15%
Logworks smoothing factor to RPM channel: 0.90

so here are three runs, back to back, plotted against each other. all in all, the precision seems to be good. this seems especially true over 3500RPM, which is great. also of note: the torque curve is flat and i appear to be very close to the OEM 296HP at the crank for this motor. that is a great starting point.

http://986forum.com/forums/uploads01...1366656311.jpg

next, i chose a single run to represent my baseline state. i chose the strongest pull, which was run 2. here it is, plotted against AFR.

note the AFR doesn't seem to drop below stoic until over 4k RPM!! Gets down to low 12's / high 11's way up in the power band. that's too low; too much fuel.


http://986forum.com/forums/uploads01...1366656786.jpg

next up, fixing my throttle. recall last week, my TPS was showing 83% open at full throttle. now look in the baseline plot above & see how my AFR is 14.7 up to 4500RPM.

the car uses throttle position & RPM to determine whether to run in open loop (fuel maps) or closed loop (O2 sensors / 14.7AFR / 'Econo Mode'). in the plot above, i have the throttle pegged, but i'm still running 14.7. that's WAY lean for WOT & possibly dangerous to the motor. i have been driving it on the track like that!

i popped the engine lid to have a look. sure enough, lots of slack in the cable. i repaired it with a cable stop from ace hardware. now at WOT, the TPS signals 100%. what does that bring us?

http://986forum.com/forums/uploads01...1366657959.jpg

so i gained a handfull of HP and torques, but the torque curve is flatter & the area under the power curve is a lot higher. more importantly, look at the AFR! looks like the threshold for open loop w/ our ECU is somewhere north of 83% TPS. i will log later to find out. so, AFR now much safer at low revs. also note that i seem to be open loop even at 2K revs w/ WOT. that's great, because i can ONLY tune the open loop zones with my fuel controller.

what's next? first, determine EXACTLY where open loop starts & stops. next, install the fuel controller & log more data. HP vs. RPM & AFR at both WOT and 50% TPS. this is going to be fun.

quick update. i got the AFC computer installed & initialized. my wideband is acting up; i called innovate & they gave me a reset procedure which i will try this evening. i had to swap out my cat bypass pipes for the cats; my emissions are due this weekend! once i get the bypass pipes back on the car, i will do some tuning.

earlier, i mentioned data rate issues. the data coming in from the ECU was slow & i had to apply some smoothing to get good results. i brought this up to the innovate guy when i called them about the wideband. by default, the OT-2 computer (the one that montors the ECU & outputs to the iPad) is set to 'Auto' for OBD II protocol recognition. his suggestion was to change the setting to ISO 9141. i did this...and it worked! nice, smooth data.

once i get the test pipes back on & the wideband reset, i will do a baseline dyno again. i will also start re-logging baseline fuel maps & begin logging power vs. AFR. good times.

3.4 tunning
 

Anymore info on your project? I'm planning to do the same on my 3.2

I have some dyno time this week. I'm testing some exhaust and intake parts

all the equipment is working right now. i had to pull off my bypass pipes to get emissions test done. i will do a LOT of tuning this weekend, as next weekend i will be at the track.

i will post some updates next week.