E46 M3 Intake Shoot-OutJune 18th, 2005

For most people with a driver’s license and a penchant for sports cars, the E46 M3 is a car without need for improvement. With an impressive 333 crank horsepower from the factory and a cool figure to boot, it is the off-the-shelf Grand Tourer that can fulfill almost any car enthusiast’s every need. That is, unless you are like me and the portion of the M3 purchaser demographic that keeps Excedrin on BMW service managers’ desks and tuner shops buzzing with activity.

These days car makers are doing more and more to extend the boundary of what is considered “regular” in terms of horsepower per liter of displacement or HP/L. Whether they are accomplishing it like BMW with high revving naturally aspirated motors or like Mercedes with forced induction, the HP wars are on, and I couldn’t be happier. Somewhat paradoxically, the price for OEM HP seems to be dropping while the price of aftermarket HP is going up. This paradox is fed by the modern car’s computer(s) - is there such a thing as a car these days with a single computer - which are becoming increasingly complex and apt to circumvent any kind of bolt-on changes made to the car. Couple that with a car that doesn’t have many glaring weaknesses in terms of hardware and you may find that modifying an M3 for power is an emotionally and financially draining experience. In addition, some car manufactures like BMW have added a final fly in the ointment by making it difficult to test aftermarket modifications. Ask anyone who has ever tried to dyno a MY03+ M3 and they will know exactly what I am talking about. (It seems BMW clued into something with the later model M3s since the 01-02 cars will dyno without issue)

Allow me to end my rant and get back onto topic…the E46 M3 and more power. As I have said, for most the S54B32’s broad toque curve and immodest HP levels is more than sufficient, but for others, myself included, it’s just a good start. With a naturally aspirated engine that pushes 104 HP/L out off the showroom floor, gains are not going to come easily or cheaply. Using the analogy that an engine is just a glorified air-pump, we know that for a naturally aspirated engine we can affect output in three major ways without touching the engine’s internals: make it easier for the engine to inhale, make it easier to exhale, and make sure that the ECU takes advantage of these changes. Enter K&N Filters (www.knfilters.com), Advanced Flow Engineering (www.afefilters.com), and Powerchip Inc.(www.powerchipgroup.com).

I conceived of the “Intake Shoot-Out” solely in response to the vexing propensity of self-proclaimed “Internet Experts” to proselytize about which of the popular intakes was better for the M3. These argument, mind you, were raging before one had even come out! Fortunately for me I was in the nearly unique position of being one of two cars in America running the K&N 63 Series intake before it was released to the general public – thanks to my friends in high-horsepower places. Thus, if I could get a hold of the AFE intake then I could put the arguments to rest, or do my best to try. Note: since then the K&N has been released to the general public. I posted this idea on my favorite M3 internet forum (www.m3forum.com) and one of the forum sponsors

(www.mperformance.com) responded with a free intake within minutes. I should point out that that many of the excellent M3Forum community members offered their own money for a pool to buy an AFE intake for the proposed shoot-out. So, having two intakes in hand and an E46 M3, I set out to find a reputable shop with a dynamometer that was familiar with BMWs. Fortunately for me and with the right financial coaxing, the people over at Evosport (www.evosport.com) agreed to open their doors to me on a weekend so we could make this thing happen without the distraction of their many other customers. While planning the event with Evosport, the people at Powerchip found out and inquired about custom tuning software for each intake. When I got wind of Powerchip’s interest, I eagerly surrendered my car for three weeks and about 1000 miles of road tuning to see what we could get. Don’t let my charity overwhelm you, however, as I would be lying if I didn’t hope for the proverbial “scratching of backs” or in this case, I leave you my car, you “leave” some software on it – and don’t scratch anything!

The concept of the shoot-out, as I pitched it to everyone involved, was to test the two (stepping into my ridicule retardant suit) leading choices of intakes for the E46 M3 community and to compare them to the OEM intake. Note that there are many other intakes available for the M3, but with time limited and the level of detail we hoped to get to with the testing, I decided to choose the two most popular if by no other metric than the volume of posts, err, arguments about them. In addition, Powerchip may as well have just kept my car if I threw too many more intakes at them.

As anyone who has chosen one aftermarket product over another knows, the choice cannot and should not be based on any sole aspect…it’s just not that simple. In keeping with that, I set out to measure the results of the shoot-out using a variety of qualitative and quantitative metrics including: cost, build quality, ease of install, aesthetics, fitment, sound, peak horsepower gain, peak torque gain, HP curve comparison, and TQ curve comparison.

It is rather interesting to note that AFE is a company formed from ex-K&N employees. Without diving into the story since I don’t want to take sides, let’s just say that they have different approaches and different prices – to the tune of roughly double the cost for the AFE kit versus the K&N. That being said, let’s explore the different intakes including the OEM.

The Intakes

Pop open the hood of a stock M3 (if you can find one in Southern California) and you will find an OEM airbox designed with a closed box, panel filter, Mass Airflow (MAF) screen, and velocity stack. The tubing leading to the intake plenum includes the MAF sensor and is made of plastic with accordion style bends. Air is fed into the box from a front kidney vent and a duct which comes up via the brake duct.

The K&N intake (as seen in Figure 1) is designed with a large cone filter and a one-piece smooth-bore plastic intake tube that goes directly from the filter to the intake plenum by way of the MAF. K&N left the MAF sensor in approximately the same spot as OEM, but the intake uses no MAF screen. The intake includes an open-top intake box to isolate the filter from hot engine air, but the intake tubing is not mated with the heat box; instead it passes through a hole in the box with about a ½ inch gap around the tubing. The box is sealed on top by rubber trim which meets with the underside of the hood.

Figure 1: K&N 63-115

The AFE intake (as shown in Figure 2) is designed with a large near cylindrical filter that includes a filtering inverted conical section at the top for increased airflow. The filter in the AFE is the larger of the two aftermarket intakes, at least by shadow measurement; however surface area is more than a function of just the filter dimensions. The AFE filter uses rounded pleats versus the folded pleats of the K&N filter. Fans of rounded pleats claim they are superior to folded pleats for several reasons, but I am impartial as long as my engine gets good clean air – and lots of it. The aluminum intake velocity stack is

mated to the heat box so that no engine air can be sucked into the filter area. The intake includes an aluminum intake box top that further seals off the engine compartment from the filter area and a diverter meant to keep the kidney tubing from flowing air directly onto the filter element (to avoid turbulence which has been shown to alter MAF readings in a CEL type of way). The intake velocity stack is mated with the OEM corrugated intake tract for air delivery to the intake plenum utilizing the stock MAF location.

Figure 2: AFE Intake

The Testing Methodology

Let’s get the caveats out of the way first – dynamometers (or dynos) are not perfect. Chief among their weaknesses is their tendency, from one model to the next, to read differently in terms of absolute power i.e. 300 wheel horsepower (WHP) on one dyno may be 275 WHP on another, all things left the same. It should be noted that all numbers in this story are WHP number and not crank HP numbers. Another dyno issue is the inability to recreate actual wind speed sans a wind-tunnel. Because of this, the car cools differently and the computer adjusts power accordingly. As alluded to previously, there is further difficulty with dynos with regard to the E46 M3 and its Siemens MS S54 ECU (pictured in 3b). The passage below is taken from notes I was sent by Powerchip:

Bosch Motronic and Siemens computers are fitted to the vast majority of the BMW range, and Most ‘M’ cars are fitted with Siemens. The E46 M3 is fitted with a MS S54 control unit.

Very detailed BMW training documents on the control unit are at

http://www.unofficialbmw.com/images/m3engineelectronic1.pdf

Some advanced engine management systems feature a separate mode specifically if the vehicle detects it is being used on a dynamometer. On some cars, when the front wheel speed and rear wheel speed vary significantly, the special ‘dyno mode’ setting is activated. On other cars, the mode is known commonly as ‘Cat [catalytic converter] protection’, when the car detects the car is being driven on a dyno, it causes the car to run using mapping that is not used in normal ‘on road’ conditions.

It is important when testing the horsepower of a late model vehicle such as an E46 M3 to ensure that the vehicle is not simply running in a ‘dyno’or emissions mode – as the results of ‘before and after’ testing of modifications may be almost identical, and changes to ignition and fuel maps will not be able to be measured properly. Care also has to be paid to adaptation. Unfortunately, measuring modifications such as performance softwarethat only claim ‘single digit’ horsepower increases (such as Powerchip) on a car such as an E46 M3 needs attention paid to the testing procedure and details to ensure the results are correct and accurate.

Figure 3a: Bosch ECU (for illustrative purposes only, not from an E46 M3)

Figure 4b: Siemens MS S54 from an E46 M3, image Copyright BMW A.G.)

These challenges aside, the dyno is an excellent tool for measuring power changes when using the same car under the same environmental conditions. To address this issue we used a MY01 E46 M3 which was dyno friendly. My own MY03 M3 took to the dyno like oil to water.

Another caveat is the M3’s ECU adaptation process. The M3 ECU adapts to a variety of changing condition which can include gas quality, environmental variables, and it even adapts based on the way you drive so drive hard daily! Because of this, bolt-on modifications can have an instant result that may not last (after adaptation) or vice versa. For this particular test, an example issue is the new intake providing for higher air flows (in terms of oxygen density measured in grams per second or g/s). This higher flow alters the air-fuel ration (AFR) to a leaner mixture which makes more power instantly, but is dialed back via the ECU’s long term fuel trim (LTFT) over time. A caveat to our caveat, a sort of caveat-squared, is that continually going leaner will not

continually increase power; however, continually going leaner will increase the heat of the combustion chamber and eventually turn your piston into a piston puzzle. Again, Powerchip provided commentary on the M3 ECU’s adaptation behavior:

Our research and testing has illustrated the significance of this adaptation. We found that the BMW M3 MUST be driven (on the road) for approximately 50-80 miles (AFTER the adaptation has been reset) over varying load and rpm’s to have adapted properly.

Adaptation data is stored within the ECU’s volatile memory so that if a new engine management software file is loaded into the ECU, or the vehicle is reflashed by a dealer etc then adaptation should be reset, and time allowed to elapse before an assessment is made of the air fuel ratio.

For specific technical information, please refer to an excellent article written by Shawn Fogg from bimmerforums:

http://ackthud.com/shawnfogg/mixture.htm

Unfortunately it was not practical to drive a full 80 miles between setups because of the amount of testing we were looking to get accomplished, but we did take the M3 on a roughly 20 mile fixed course drive that included many different engine load scenarios between each setup (meant to speed up adaptation). As Powerchip’s research has shown, the adaptation process tends to move very quickly for the first few miles and settles down steadily as you stretch out the miles. Thus, we addressed the issue, but in a perfect world there would be more hours in the day and less LA traffic so that we could do our best to reach “nearly complete adaptation.” I guess while we are designing the perfect world, I would request no speed limits either – then we could really speed up adaptation, no pun intended. Note: the M3’s ECU car never completely stops adapting, so there is no true end to the process to strive for.

Before diving into the results, it’s important to explain some of the observations generically so that everyone understands what I am saying. With any modification to the car, the goal is not only power, but safe power. Whenever affecting a cars ability to “inhale” air, we need to be sure that the computer accounts for it the right way and that the air to fuel ratio (AFR) remains safe. Sean Fogg had this to say about AFRs:

For gasoline to combust it needs oxygen. For the most complete combustion gasoline usually needs to be burned at a ratio of 14.7 parts of air to 1 part fuel. This air to fuel ratio (AFR) is called stoichiometric or Lambda = 1. For example all of the following are at the same

7.35 lbs of air to 0.5 lbs of fuel = 14.7:1 AFR = Lambda 14.7 lbs of air to 1 lbs of fuel = 14.7:1 AFR = Lambda 147 lbs of air to 10 lbs of fuel = 14.7:1 AFR = Lambda

Obviously the amount of air and fuel being burned is totally different between each example but the AFR for each is identical. 14.7:1 AFR is normally considered the best trade off between emissions, fuel economy and power production.

Running LEAN is when your AFR has a higher ratio of air e.g. 16:1 AFR is lean. Running lean increases emissions, increases heat, usually increases fuel economy, reduces power and increases the chances of knocking. Running RICH is when your AFR has a higher ratio of fuel e.g. 13:1 AFR is rich. Running rich increases emissions, usually decreases heat, decreases fuel economy, increases power (to a point), and decreases the chance of knocking. Running rich for long periods of time can cause deposits to build up on the plugs and O2 sensors (fouling) and can clog your catalytic converter. Maximum power is usually obtained running around a 12.3:1 AFR. Going richer then that will cost a little power but you loose less power then being leaner then 12.3:1. This can be seen in the graph below.

Figure 5: Fuel Consumption versus Pressure

The above graph is taken from The Sports Car Engine: Its Tuning and Modification by Colin Campbell. The graph shows basically how much pressure is created by burning gasoline at different AFRs and also how much fuel is consumed in doing so. Highest power will be achieved at highest pressure (most force pushing on your pistons).

Now that we have that out of the way, let’s look at the setup. The dyno at Evosport is a Dynojet 248C running Winpep 6.1. The test car was a 2001 BMW e46 M3 w/6-spd transmission (approx. 19,000 miles) and a GruppeM muffler. The environmental conditions were:

Barometric pressure: 29.96’’ Air Temperature: 77F Humidity: 36% SAE Correction Factor: 0.99

The process was as follows:1. Dyno car in stock configuration2. Install AFE intake, make adaptation run, dyno3. Add AFE tuned Powerchip software, make adaptation run, dyno4. Install K&N intake, make adaptation run, dyno5. Add K&N tuned Powerchip software, make adaptation run, dyno

The car was dynoed three times in each configuration in order to stabilize lubricant temperature and create a similar adaptation environment for each configuration. All dyno runs were completed with the hood closed. 5-7 minute cool-down periods were provided between each dyno run and temperatures were actively monitored to ensure that all runs were performed under the same conditions. As mentioned before, the car was driven through a 20 mile loop of city/highway to let the ECU adapt to each configuration change.

The same CA 91 octane gas was used for all runs. Evosport had this to add about the gas:

It should be noted that CA 91 gas is the poorest quality in the entire country due to specific CA blend requirements. The poor gas quality (with an effective octane level of less than 91) has measurable negative impact on power gains and tends to cause detonation at higher RPM’s. Both of these negative situations were encountered during our testing, directly resulting in lower power numbers than the engine and tested modificationss are capable of with higher quality fuel.

That being said let us finally dive into some results.

The Results

Stock Configuration (3 runs)

Figure 6: Stock Dyno Runs

Stock car’s best run was 282HP/233TQ. Air/Fuel Ratio – 12.5 at low end, 13.5 in midrange, and 12.5 at the top end.Engine Coolant Temperature – 190-196FIntake air charge temperature – 95-99FAirflow Maximum Volume – 239 grams per secondShort Term Fuel Trims – (-.8%)Ignition Timing Advance – 19 degrees

Notes: Evosport commented that these power numbers and AFRs are typical for an M3 with an aftermarket muffler.

AFE Intake (3 runs)

Figure 7: AFE Intake Dyno Runs

AFE car’s best run was 291HP/237TQ. Air/Fuel Ratio – 12.8 at low end, 13.0 in midrange, and 11.8 at the top end.Engine Coolant Temperature – 191-196FIntake air charge temperature – 95-99FAirflow Maximum Volume – 260 grams per secondShort Term Fuel Trims – (-.2%)Ignition Timing Advance – 19 degrees

Notes: The AFE added 9HP and 4 FT-LB of TQ to the baseline numbers. There was an observed increase of 21 g/s of oxygen flow past the MAF (8.7% increase in flow). The fuel mixture went too rich at low and high ends for optimal power, but the shape of the AFR curve looked similar to stock. The intake charge temperature was similar to the stock box which is most likely due to the excellent sealing of the AFE intake box. Ignition timing advance remained the same.

AFE with Powerchip software (3 runs)

Figure 8: AFE with PowerChips Software Dyno Runs

AFE car with Powerchip software best run was 291HP/239TQ. Air/Fuel Ratio – 12.7 at low end, 13.7 in midrange, and 13.0 at the top end.Engine Coolant Temperature – 190-194FIntake air charge temperature – 97-99FAirflow Maximum Volume – 260 grams per secondShort Term Fuel Trims – (-.4%)Ignition Timing Advance – 19 degrees

Notes: The addition of Powerchip software showed no increase in peak HP and a marginal 2 FT-LB increase in peak TQ on the dyno; however, the software corrected the sub-optimal AFRs, permitting the engine to operate at the elevated power levels attributed to the AFE intake in a safer and more optimal AFR range. Due to poor 91 CA gas there was noticeable spark retard that limited high end HP. Evosport predicted that, with better fuel, the actual power number would have been closer to 296-298HP.

AFE AFR Comparison

Figure 9: AFE AFR Comparison

Notes: In Figure 9 you can see the comparison of the AFR ratios for the stock car (highlighted blue line), the AFE intake (red line), and the AFE + Software (green line). As you can see, all of the AFE configurations maintain a stock-like AFR curve. The AFE by itself tended to be on the rich side which was not optimal for power or fuel economy. The Powerchip software solved this and tuned the AFR curve to a slightly more aggressive, but very safe level to increase power levels and fuel efficiency.

K&N Intake (3 runs)

K&N Intake best run was 293HP/239TQ. Air/Fuel Ratio – Inconsistent between runsEngine Coolant Temperature – 192-196FIntake air charge temperature – 95-101FAirflow Maximum Volume – 246 grams per secondShort Term Fuel Trims – (-.8%)Ignition Timing Advance – 18 degrees

Notes: The K&N intake added 11 HP and 6 FT-LB of TQ to the baseline run. There was an observed increase of 6 g/s of oxygen flow past the MAF (2.9% increase). K&N had very inconsistent AFRs from run to run, with extended RPM ranges that were too lean for continued safe operation of the engine. It appears that the ECU had a harder time adapting to the K&N intake. The intake air charge temperatures were slightly higher which could possibly be due to more hot air being drawn into the intake air box, but it is also possible that the slightly warmer temperature in the shop as the day progressed affected this as well. Ignition timing advance was retarded by 1 degree.

K&N Intake with Powerchip software.

K&N Intake with Powerchip software best run was 299HP/242TQ. Air/Fuel Ratio – Inconsistent between runsEngine Coolant Temperature – 190-192FIntake air charge temperature – 95-97FAirflow Maximum Volume – 246 grams per secondShort Term Fuel Trims – (-.8%)Ignition Timing Advance – 18 degrees

Notes: The K&N intake with the Powerchip software made 17HP and 9 FT-LB of TQ over the baseline run. Airflow remained at 6 g/s over stock (2.7% increase), but actually decreased to stock levels (240 g/s) on the best run. Again, we observed inconsistent AFRs from run to run It is not known whether this AFR inconsistency would actually be experienced in street driving, but PowerChips’ data logging while road tuning indicated that the K&N with software will run at a consistent AFR of close to 13:1 after more significant adaptation; however, bringing the AFR down a full point will probably bring the power output down in-step Again it seems that the ECU was having a bit more trouble adapting to the K&N intake. Intake air charge temperatures went back down to stock configuration levels which was good to see. Ignition timing advance remained retarded by 1 degree.

K&N AFR Comparison

Figure 10: K&N AFR Comparison

Notes: In Figure 10 you can see the comparison of the AFR ratios for the stock car (highlighted blue line), the K&N intake (red line), and the K&N + Software (green line). Both K&N configurations produce AFR curves which are quite a bit leaner than stock. The K&N by itself has a similar curve to stock, but goes far to lean where the engine is experiencing max torque and thus max cylinder pressure (around 5000 RPMs). The K&N with Powerchip software had a less aggressive AFR (though it was inconsistent between runs), but was still too lean at the top of the RPM band. Further adaptation with the software should further refine the AFR curves (according to Powerchip road tuning data).

Evaluation

Stock vs. AFE

The AFE intake yields 9-10HP/4TQ gains with AFRs getting richer on average by 0.5 point. There are improvements to the torque and horsepower curves across most of the RPM band giving a large improvement of “area under the curve.” The largest improvement came at the top of the RPM band which is expected given the fact that the engine has its highest oxygen needs at maximum RPMs.

Stock vs. AFE intake with Powerchip software.

Adding Powerchip software further increased HP and TQ across the RPMs band, significantly improving the “area under the curve.” The AFR curve in this configuration looks nearly stock which is a major plus. Again, Evosport and PowerChips both believe that better gas quality would yield an additional 5-7 HP on the top end.

Stock vs. K&N intake

The K&N intake adds 11HP and 6TQ. However, the fuel mixture is inconsistent and gets too lean in the midrange. There are improvements to the torque and horsepower curves across most of the RPM band giving a large improvement of “area under the curve.” The improvements across the RPM band are for the most part larger than the AFE bolt-on; however, the AFRs are quite a bit leaner across the same band.

Stock vs. K&N intake with Powerchip software

Adding Powerchip software to K&N improves the lean condition in the midrange but leans out top end. With a slightly richer fuel mixture (needed to make operating A/F range safer), peak HP may be reduced a bit. But, as with the AFE intake, it is believed that top-end timing retardation due to poor quality CA 91 gas reduced peak HP by 5-7 HP.

Final Analysis:

Both intakes are worth approximately 9-11HP (at the rear wheels; 12-15 at the crank) and 4-6TQ. Both intakes can definitely benefit from software tuning as AFE richens the car too much and K&N leans the car out too much. With proper software, the total net gains from either intake should range from 15-20 HP.

Enough with the data, let’s window shop:

Now that I we have analyzed the quantitative performance of both of these intakes, let’s switch gears (yes, I am fully capable of keeping the automotive puns coming) and talk about qualitative measures of comparison. The K&N (as seen in Figure 11) is a nice piece with a very trick molded intake tract that runs from the intake box to the intake plenum. One look inside this piece shows off the smooth air delivery path versus the stock accordion style bends – very nice. The K&N kit could stand to have a little better sealing – read: any at all - where this molded piece passes through the intake box (a modification that people are already handling with a quick trip to their local hardware store). The rubber trim on top of the K&N, while functional, just doesn’t seem to match the caliber of engineering of the car it is going onto. While it may fit in perfect on an

America car, it just doesn’t look right for the kind of German automotive art that it feeding oxygen to. As for installation, the K&N comes with excellent instructions and is a 45 minute job the first time you install it, and a 15 minute one any subsequent time.

Figure 11: K&N Installed

The AFE (as seen in Figure 12) is a great looking kit with a trick aluminum velocity stack that not only seals off the engine bay from the intake box, but really makes the kit seem, for lack of a better word, “complete.” The aluminum lid which further seals the intake box is also a nice touch, but I personally would have molded it in carbon fiber or powder-coated it black – that way it would look nearly OEM. Maybe that’s not what AFE was going for. At any rate, the custom market already has some solutions if that’s your preference. The AFE, just like the K&N, is a breeze to install and since it has a few less pieces it is possible that it could be installed a little quicker, but whose counting when you are talking about something you can get accomplished faster than watching an episode of your favorite sitcom. I should point out that installing either of these intakes is time much better spent!

Figure 12: AFE Installed

To take a further step into the ultra-subjective zone, the “butt-dyno” reports that the AFE intake (with and without Powerchip software) felt crisper up to approx 6,500 RPM, while the K&N intake (with and without Powerchip software) felt stronger at the top. The subjective and objective evidence shows that the K&N flows excellently at the top of the RPM band despite its slightly smaller filter. While driving, the AFE intake was barely audible while the K&N intake much preferred to make its presence known – especially at the higher RPMs where the filter was taking a beating from the torrent of air trying to make its way to one of the six individual throttle bodies. Both kits had excellent fitment.

Now, I saved the price for the metaphorical bottom line. The AFE intake can be found for around $700 at your favorite vendors while the K&N can be found for around half of that, or $350 for the less mathematically inclined. Both intakes are steals in my opinion and depending on your risk profile, wallet, sound tolerance, and aluminum tolerance, making a decision is still going to be difficult.

So, which intake is better? Well, you are going to have to draw your own conclusions. As for which intake I have, well, I have both…

Category WinnerPower

Safety (AFR)

Aesthetics

Ease of Install TIESound

Price

Thanks again to the Contributors!Mperformance.com (www.mperformance.com) – AFE Intake KitK&N Filters (www.knfilters.com) – Series 57/63 Intake KitEvosport (www.evosport.com) – Dyno FacilityEuropean Motor Works (www.emw1.com) – BMW GT1 Tech ToolPowerchip Inc. (www.powerchipgroup.com) – Software designed specifically for each intake