Tom Wilson
March 2, 2003

Listen carefully to the buzz about what the '03 Cobra can do--that background noise you hear is the sound of tuners hard at work. Perhaps most animated of these is Jim Bell at Kenne Bell, thanks to the '03 Cobra's Eaton-built Roots-type supercharger. Jim has been eager to strut his stuff--all over the Eaton/Roots competition if at all possible--and what better vehicle for his agenda than the new Cobra? Obviously the car is factory-engineered for blower power, and all Jim has to do is take off the stock Roots-type blower and install his own supercharger along with the necessary supporting pieces. Luckily for everyone, after months of pleading on our parts, we were invited along for what proved to be a screaming dyno thrash--one of the highest-powered bolt-on tests we've ever experienced.

138_0303_01z 2003_ford_mustang_cobra Left_front_view
Horse Sense: History might show the '03 Cobra as the car that finally made the 5.0 Mustang obsolete. Its Four-Valve is the first modular engine to demonstrate the breathing and power beyond what the pushrod small-block is reasonably capable of.

In the following pages is a huge but condensed mass of data on what the stock '03 Cobra is capable of and how the Kenne Bell supercharger was engineered onto it to form a 600hp rocket. Besides highlighting the impressive airflow capacities of the Kenne Bell supercharger, the story illustrates the Cobra's electronic sophistication and hints at the changing state of performance tuning in general.

The Car
To test an '03 Cobra, you have to have an '03 Cobra. Normally Kenne Bell simply buys the car or truck needed, but this time, Randy Kramer provided the all-important subject. He was willing to take his chances on the dyno in exchange for the latest in horsepower. The spanking-new yellow Cobra had all of 904 miles on the odometer when we started, and more than 950 when we ended. Imagine, almost 50 miles of supercharged, all-out acceleration. And if you're wondering how a guy could go without his new Cobra for so long, it's because Randy is a cruise-ship captain, and while he's out to sea, he can't drive his Cobra.

The Testing
All work was done on Kenne Bell's Dynojet. This is a favorite of ours, as Jim's dyno has more instrumentation on it than any other we know. In the cockpit, Kenne Bell technician Brent Morris had a Horiba air/fuel meter, a Fluke digital thermometer, a Toshiba laptop for datalogging, a STAR II tester, the dyno on/off control, a printer for the STAR tester (which was perched in the back seat), a power strip, enough wires and cables to start a phone company, and a clipboard with good, old paper and ink so the dyno and datalogging sheets could be notated and kept together in order. The data sheets were collected into a notebook as they were printed. In the end, more than 250 pages of data were obtained.

Kenne Bell customers will receive their Cobra blowers like this, assembled with the two-hole intake casting, adapter plate, and some small parts such as the pulley. The prototype was polished--the standard finish is a handsome semigloss black.

With the datalogging, horseback analysis, cooldown, and other requirements, each test took about 20 minutes to perform, although some had hours of downtime as a new chip was written or equipment fitted. In all, four long days of dyno time were expended. Though there were some false starts and deadends, there were never any failures. Yes, the test ran long, but given the complexity of instrumenting the test car--there's plenty of drilling and tapping involved in this--once the car was on the dyno, the impetus was to see the test through.

The engine was rigged with numerous probes, many of them for temperature. The stock Eaton blower and intake were removed and replaced so the intercooler/manifold could be drilled and tapped to accept the probes, which were hooked up either to the dyno or the laptop datalogger. This allowed us to see the ambient or blower inlet temperature, the blower outlet temperature, and the intercooler outlet temperature, along with boost and intercooler water temperature. When the Kenne Bell blower went on, it received the same treatment.

Brent had his own electronic Toys-R-Us in the cockpit. The STAR tester is in his left hand, the Toshiba is on the passenger seat, the dyno controller is hanging from the rearview mirror. The Horiba air/fuel meter and computer printer are not visible here.

Datalogging was done with a Toshiba 325CDS computer running proprietary Kenne Bell software. This is non-user-friendly, DOS-based software designed by Ken Christley at Kenne Bell. He used to have his own company, and the software dates from those days.

Nearly all testing was done using 100-octane unleaded, bought by the barrel from a local fuel supplier. At the end, as the power levels soared, 2 gallons of 116-octane, leaded race fuel were added.

Finally, all testing was done with the same eight Denso IT22 spark plugs. Kenne Bell is Denso's biggest cheerleader, as Jim's seen the Iridium-tipped plugs run cleanly long after any other plug has given up. Since we knew these plugs would be used in this test eventually, they were installed from the get-go.

The Eaton
Testing opened with the totally stock Mustang--meaning the standard 31/2-inch blower pulley and nominal 8 pounds of boost, along with a totally stock air path from air filter to tailpipes. As the Cobra warmed up on the dyno, we once again marveled at how smooth the engine was. Clearly Ford has sent its newest engines to finishing school.

The rear-wheel horsepower came out at 375 hp. This is notably higher than expected from Ford's 390 flywheel horsepower rating, but it wasn't surprising given how fast '03 Cobras have been running at the drag strip. Obviously Ford has been underrating the Cobra engine.

Next we tried the smaller 3-inch "Lightning" supercharger pulley, which has already proven a popular '03 Cobra upgrade. This is the pulley used on the 5.4 Lightning pickup, which also uses this same Eaton blower, but mated to somewhat slower engine rpm.

Changing the Eaton pulley is difficult. It can either be done by swapping entire blower drive ("nose") sections, or using a special puller to drag the pulley off the input shaft. The pulley is lightly built, and it uses only a taper fit on the blower input shaft for grip. There is no keyway or other locking device, therefore the force involved in this pressed-on pulley is rather high. Kenne Bell has found most puller attempts resulted in damaged pulleys and broken power steering pump pullers, so they prefer to change entire blower nose sections. However, we had a Performance & Diesel Innovations puller, so that was used along with PDI's 3-inch pulley.

Boost jumped to 13 pounds and power to 417 hp. Remember, this is through the stock air filter and exhaust, so even with a simple pulley change, there is good power to be had. Also, the spark and fuel were well within limits with just the pulley change, so no electronic changes were required at this level. However, as the 3-inch pulley takes the Eaton blower to high speeds (it was at 16,250 rotor rpm at 6,500 engine rpm on the Cobra), and its efficiency was clearly waning, no more boost increases via smaller pulleys were attempted.

The Bazooka
As part of Jim's testing philosophy of reducing variables, he occasionally uses test gear to eliminate restrictions even though that gear is not workable on the street or perhaps even on the track. A perfect example is the "bazooka." Nothing more than a 4-inch aluminum tube with radiused ends to smooth airflow, the bazooka eliminates the air filter, any prefilter silencers, and associated tubing to provide a fail-safe supply of air. On the Cobra it was attached directly to the mass air meter and laid atop the radiator support where it would ingest "outside" air (the hood was open at all times anyway). Also, the shop's cooling fan was diverted slightly to ensure no ram effect--however slight--could take place.

When it's airflow you want, the 4-inch bazooka delivers. It also screams like a fire-truck siren. Kenne Bell runs the bazooka to eliminate concerns about induction restrictions when testing blowers. The KB supercharger is more susceptible to induction restrictions than centrifugal blowers because it draws through both the mass air and throttle body.

Jim's idea with the bazooka is to sidestep issues of whose cold-air kit works best--he just wants to reduce inlet obstruction to the superchargers. This makes more accurate A-to-B compari-sons between blowers, and it shows exactly how much power is available to savvy tuners who can come up with ideal inlet air systems.

It's difficult to say which was more surprising--the shrieking scream from the bazooka when Brent matted the throttle (we were standing just in front of the car at the moment and nearly came out of our skins when this siren-like wail exploded in the shop), or the 35hp gain.

While the 451 peak horsepower reading was OK, nearly the rest of the power curve came with a dangerously lean air/fuel mixture. Up high, at the power peak, the stock electronics in the Cobra had the A/F ratio at a respect-able, power-producing 13.5 to 12.5:1. But in the lower rpm ranges, the A/F ratio was a scary 16:1. Obviously, we needed to richen the bottom of the fuel curve both to produce power and to avoid a melted piston.

On the other hand--wow--451 hp! Again, this is through stock cats and mufflers, stock cast-iron log-type exhaust manifolds, the stock air meter and throttle body, stock engine internals, even without a chip in the computer! Recall, however, that 103-octane unleaded was saving our skins. Had it not been there, the test would likely have been brutally interrupted right here.

Enter the Electronics
Because we could not continue testing with the air/fuel ratio dangerously lean, a new chip was burned by Ken. According to Ken, this was a scary, thrown-together calibration horsebacked from experience, but it came out fine on the dyno.

By increasing the A/F ratio from 15.6 to 12.8:1, power was increased substantially--50 hp--all the way from 2,000 to 5,000 rpm. At that point, the richer calibration ran evenly with the stock, lean calibration. After the power peak, the power dropped off slightly. Ken said this is because in the rush to produce the chip, it ended up with 2 degrees less ignition timing than the stock calibration. While we would have liked to have redone the chip, in the interests of time and adding a bit of a safety margin at the upcoming higher boost levels, we left the timing unchanged for the duration of the tests.

Well worth noting here--besides the big power a chip can provide--is that all you studs with the small pulleys and other breathing mods on an electronically stock '03 Cobra are running dangerously lean in the low- and midranges. The Four-Valve Cobra responds to airflow, sure, but without electronic help it doesn't have the fuel flow to match. Note also that at this power level, it isn't the dual fuel pump's capacity, 42-lb injector size, or other physical limits holding the Cobra back (hardly). It's the computer calibration that requires modification.

There they are, the stock cats. With 20 pounds of boost available from the twin-screw supercharger, getting to some combination of headers, high-flow cats, and mufflers is imperative to experiencing all the supercharger has to offer. At the 600hp level, a set of long-tube headers, a high-flow X-pipe, and an after-cat exhaust is likely worth 40 hp.

The Kenne Bell Blower
Fitting the Kenne Bell blower to a Cobra holds no mechanical surprises, and we're not going to detail its relatively simple installation here. Suffice it to say the stock Eaton blower is removed from atop the engine, and the KB blower assembly is fitted in its place. This isn't to discount the tight working room around the blower, especially in the rear, but overall, the installation is much simpler than a centrifugal thanks to a self-contained oil supply and stock-replacement, bolt-on nature. The excellent stock intercooler remains intact, with the KB supercharger using a short adapter plate to mate with the Ford bolt pattern. The Ford throttle body remains in its stock location, and with more-than-adequate fuel pump and injector capacity on board, those items need no attention--at least not at the 500hp level.

Brent made his first pass with the KB supercharger using the stock computer (no chip) and stock air filter. Wham, the air/fuel ratio hit 16:1 while the power advanced a nearly meaningless 3 hp to 454 hp. The lean A/F ratio obviously wasn't going to work, so the same chip Ken had come up with for the Eaton's 450hp level was installed and another pull made. This gained 100 hp at 3,500 rpm (!), but the top end stayed the same. Ultimately, Ken calculated the stock 90mm mass air meter was still not pegged, but the computer's load calcu-lation tables had run out. That meant writing a new chip, but in a much trickier portion of the code.

Note: Temp In and Temp Out are charge air temperatures into and out of the intercooler. In other words, they are blower discharge air temperature and intercooler discharge air temperature. The Water temperature is the intercooler water temperature, not engine-coolant temperature. All temperatures were taken at 6,000 rpm; cool post-intercooler temps on the final run are attributed to cooler ambient temps/slightly longer cooldown. Note how KB discharge temps at 17 pounds of boost are similar to Eaton at 10.5 pounds. Boost is given at both 4,000 and 6,000 rpm to illustrate any blower drop-off with rpm. Such drops indicate inefficiency, typically from overspeeding. The Pulley diameter is in inches, and the Blower rpm is true blower rpm after all gear and belt ratios are accounted for. Eaton and Autorotor typically quote around 12,500 rpm as a maximum, but that is in continuous operation. Both blowers are commonly turned to 15,000-20,000 rpm in hot-rod applications, but you don't want to go higher due to seal heating from friction.

After working overnight on the problem, Ken had a new chip ready the next day. Bingo! With adequate fuel available, the power shot to 489 hp. As this is the first run with the standard pulley, stock air filter, and where the air/fuel ratio was properly sorted, this 489hp figure serves as the Kenne Bell baseline run.

A major concern of ours was that we wanted to see the Kenne Bell supercharger run at the same boost and configuration as the Eaton, so we could reasonably compare the two from a heat-efficiency and power-production standpoint. Because of the differences in displacements between the blowers, this meant running them at different speeds (different pulleys), as seen in the charts. Note that both tests were run using the stock air filter.

While you have to extrapolate the rising and falling boost curves somewhat, the Kenne Bell made about 70 more horsepower in the same boost range as the Eaton. To be more precise, we had run the Eaton to the end of its useful range, so while it had made 13 pounds of boost at 4,000 rpm, that had fallen to just 10.5 pounds of boost at 6,000 rpm. The KB blower, however, was just getting started, so it happily chuffed out 12-13 pounds of boost right through 6,000 rpm. The extra 2.5 pounds of boost at the peak probably accounts for 25 hp of the 72hp power gain enjoyed by the Kenne Bell. The rest comes from reduced power-drive loss and more efficient airflow through the blower. Impressive.

To more closely compare peak boost power figures, but with an extra variable, you could compare the 451 hp the Eaton made while breathing through the bazooka to the 489 hp the twin screw made using the stock air filter.

All '03 Mustang Cobras are equipped with dual 160-lph fuel pumps and should never need more pump capacity. However, to achieve the necessary 80-psi fuel pressure in the 550hp-and-up range, more voltage to the existing pumps is required. Kenne Bell used its Boost-A-Pump to increase the pump voltage. The BAP mounts in the left rear inner fender where this voltage reading is being taken. A 30-amp fuse in the fuel pump circuit also proved necessary.

Easy Horsepower
The next meaningful event was the smaller 3.25-inch blower pulley, which took the Cobra to nearly 15 pounds of boost and 508 hp while breathing through the stock air filter. Then the bazooka was substituted, which allowed the boost to climb to a tick shy of 17 pounds, and power to 560 hp, but otherwise all this was through the stock throttle body, cats, and so on.

Wanting to see a run using equipment you could close the hood on, we asked for a run using an air filter instead of the bazooka. A 9-inch conical was attached directly to the mass air meter, which knocked the boost off a piddly 0.4 pound and the power 12 hp. Clearly you don't want to be running the stock air filter at this power level (unless you really don't want to hear a muted blower scream through the more open element air filter for some reason). To resume testing, the bazooka was refitted.

It Gets Tough By now we were almost getting numb to the staggering power the Cobra was dishing out, and greedy power-mongers that we are, we were all eyeing the 600hp mark. Getting past 560 hp, however, proved more difficult from a fuel-delivery standpoint. And in a rare, lucid moment, we concluded the mid-500hp level is where most Cobra owners might want to go with their Kenne Bell blowers. Given the IRS and limits of street tires, any more power is just more tire smoke at these elevated power levels anyway.

Pulleys? Kenne Bell has pulleys. Unlike the Eaton design, the KB pulleys easily swap using a single bolt and a simple two-bolt restraining bar. It takes just a minute to change the pulley.

Jim still had plenty of smaller pulleys, though, so we pressed on. But Ken was having increased difficulty navigating the complex Ford software. Working with late-model Ford computers is much more complex than it was just last year, and it took Ken several attempts to arrive at a chip that worked.

Another concern was exhaust backpressure. Obviously 560 hp through stock cats, mufflers, and tailpipes was not representative of typical hot-rodding practice, and it certainly was contributing to excessive backpressure. Was there enough backpressure to block any power increase?

To find out, the exhaust was drilled and tapped before the cats to accept a pressure gauge calibrated in pounds per square inch. Jim said workable backpressure with supercharged engines runs around 5 psi, with the worst he'd seen being 10 psi on a blown 351 (non-Lightning) F-150. He predicted the Cobra would beat that, and he was right by a factor of three. With 30 psi of backpressure, the Cobra might as well have had a potato stuffed up each pipe, and we instantly salivated at the thought of long-tube headers and open collectors. However, in the interest of sticking to a single variable, the stock exhaust remained until Ken solved the fuel-delivery issue.

To shorten a two-day-long story, Ken's challenge was not the pump or really even the injectors at 560 hp--it was still electronic. Add fuel pressure and the EEC pulled injector pulse width. Lengthen the pulse width and the computer pulled fuel pressure! It took some hunting on Ken's part to find the correct tables hidden in the computer code, but ultimately he did, and in the end a combination of his chip and a 17-volt kick in the pumps via the Boost-A-Pump got the necessary fuel flowing.

With the computer code finally worked out, several chips were tried to dial in the fuel curve. All of these were at the 550-560hp level, of course, as no power improvements had been tried.

138_0303_16z 2003_ford_mustang_cobra Engine_view
Here's how the Cobra looked on its final, 604hp test with the 12-inch conical air filter. The box on the right fender is the disconnected Boost-A-Spark. It was tested but found unnecessary as the Cobra and Denso spark-plug combination had plenty of spark. Again, keep in mind that the big 604hp peak was only visible on the dyno graphs, not the charts we're presenting here, so don't e-mail us complaining.
138_0303_17z 2003_ford_mustang_cobra Intake
This is the 12-inch air filter and tubing Kenne Bell will sell in its '03 Cobra kit for inner fender mounting. The corrugated tubing looks restrictive, but KB has used this slightly larger-than-4-inch tubing in many kits and says it flows well and won't pose a restriction.

Our next power-improvement trick was to disconnect the after-cat (we wanted to try much more with the exhaust, but a lack of time, parts, and rounded-off nuts held us to simply dropping the after-cat). This halved the backpressure from 30 to 14.5 pounds, but the power remained right at 560 hp. This looked fishy at first, but then we noticed the boost had also dropped about 1 pound (to 16.6 pounds), which is about a 10hp loss, and the A/F ratio richened back to where it had been in the immediately prior test (chip and after-cat connected), which explained a 23hp loss. We added it up and figured if the boost was raised back to 17 pounds and the mixture leaned with a chip, we'd get to 583 hp.

So, with the fuel situation back in control, we simply went from the 3.25- to the 3.11-inch pulley and reburned the chip. We got back up to 17.3 pounds of boost at the peak and hit 565 hp with 16 psi of backpressure in the exhaust. Not quite what we had hoped for, but a step in the right direction. With 600 hp just in front of us and the day getting late, the 3.11 pulley was swapped for a 2.87-incher, for another 1.6-pound gain to 18.9 pounds of boost (that's discharging from the blower and measured before the intercooler--after the intercooler it was 17.9 pounds of boost). This ran great, but by then Brent was becoming gun-shy and lifted at 5,000 rpm just to take a look at the data lest we burn up something.

Everything looked fine, so all we had to do was get used to the idea of a more-or-less-stock Mustang making 600 hp. Jim said to heck with it--it was getting late by now--and he added a couple of gallons of leaded, 116-octane race fuel. Yes, the cats were still on the car, but Jim was adamant 2 gallons would not hurt them, and he was still big on detonation avoidance. You can't blame him.

Jim also ordered the 2.87-inch pulley off and the 2.66-inch pulley on. Brent climbed back in and held the loud pedal down for a full run--19.8 pounds of boost, 16.5 pounds of backpressure, 2 inches of vacuum on the naturally aspirated side of the blower, and a blazing 616.3 hp and 604.9 lb-ft of torque. That's 725 flywheel horsepower from 281 cubic inches if you use a reasonable 15-percent driveline loss figure. As it was, we had gained 252 hp from the Eaton baseline, and as Jim prognosticated after playing with his calculator, "With 25 pounds it will make 667 hp, guaranteed!"

We had to stop a few minutes and simply soak it in. At least momentarily sated in our horsepower quest, there were smiles all around, punctuated by the most amazing part of all--what hadn't been done. This staggering power was through the stock mass air meter, stock throttle body, stock engine, stock cast-iron exhaust manifolds, and stock cats. Six hundred horsepower at the tires. Unreal.

For a denouement, we asked Jim to remove the bazooka and run an air filter. Saying the 9-inch conical filter we tried earlier would be too small at 600 hp, Jim selected a 12-inch conical. Clamped right to the air meter, the 12-inch cone knocked the boost down to 19.4 pounds and the power to 604 hp.