Richard Holdener
June 1, 2009
Like the smaller headers, the 1 7/8-inch headers were run with a collector extension. The merge collector measured a full 3.5 inches on the 1 7/8-inch headers.

These multiple reflections naturally decrease in amplitude, so it's important to time the first reflected wave since it offers the lowest pressure at the combustion chamber. The ideal situation is to time the first reflected wave to arrive at combustion chamber when the piston has just passed TDC at the end of the exhaust stroke. This means the exhaust pressure wave must travel from the exhaust valve to the end of the primary tube (to the collector) and back during a crank interval of roughly 120 degrees. Timing this is not terribly difficult, but the difficulty is that if the scavenging effect is timed in this manner, it will be timed to be optimized at a given engine speed (much like runner length on an intake manifold). At other engine speeds (high or lower), the scavenging effect will be less pronounced. Therefore, it's necessary to compromise to provide a system that will provide power gains at a variety of engine speeds. Not surprisingly, a short stock exhaust manifold (whether tubular or cast) does not provide sufficient primary length to offer any scavenging effect at normal engine speeds.

With all the talk about changes in primary tubing length, what effect does changing the tubing diameter have? After all, we did test a variety of different diameter (and not length) headers. The discussion on runner length was primarily to distinguish the difference between a long-tube header and the short factory exhaust manifolds. Changing the tube diameter actually increases the surface area exposed to the exhaust gases, so there is an increase in surface friction on larger-diameter headers.

The 2.0-inch headers also featured 3.5-inch merge collectors

This is a nearly insignificant variable. The real effect of larger diameter tubing is to shift the optimized VE point. Basically, the larger diameter header (assuming not change in length) will want to make peak power (and torque) at a higher engine speed than the smaller one. The elevated engine speed will usually come with a drop in power elsewhere (lower in the rev range), but there are a number of other variables that determine what happens to the curve that our space here does not allow us to cover (books have been written on the subject). Much like an intake, the header configuration can be optimized for specific engine speeds, even on a given combination. The header can be tuned to maximize power lower or higher in the rev range.

The tuning effect is not huge, but it is definitely possible to optimize a combination for a given application using the exhaust system. In the case of our test motor, the change in primary diameter had much less effect than the difference between the long tube headers and the stock exhaust manifolds.

Enough theory--let's get on with the test. The normally aspirated GT1000 motor was installed on the engine dyno at Westech and made ready for testing with a FAST management system and a set of Autolite plugs (thanks to Accufab's John Mihovetz). Thanks also go out to Mark Sanchez for once again coming to the rescue with the necessary connectors and E-DIS module, without which there would have been no test.

The first order of business was to establish a baseline with the stock exhaust manifolds. The factory GT500 cast-iron manifolds were run with a set of 2.5-inch pipes to simulate a free-flowing after-cat exhaust. Naturally, no cats were run on the dyno during any of these tests.

Initially we were disappointed by the power output of the 5.4L. No matter what we did, we couldn't make any more than 366 hp and 364 lb-ft of torque. We changed to a different management system, swapped spark plugs, drained the oil, bled the lifters (thanks again Mihovetz), and even ran the motor without the upper plenum. Crazy Tom Habrzyk stood in the dyno room and held down the plenum while we went to full throttle. He then simply lifted the plenum off and allowed the motor to run at WOT as an individual-runner X-ram. This improved the output, but it was clearly not the answer to our missing power.