Here are some dyno test results...
Here are some dyno test results of four different primary tube lengths on an 8.5:1 compression 350 Chevy. The shortest length at 18 inches (black curves) show the engine did not like really short primaries. But the 29-inch (green), the 32-inch (red) and the 38-inch (blue) headers all performed well with only a moderate bias toward better low end from the longer pipes.
With primary length demoted from the most important dimension, the primary pipe diameter becomes the number-one dimension to get as near optimal as possible. The target, for a street motor, is to get the sizing such as to deliver the most area under the torque and power curve along with good fuel efficiency at part throttle. For a street/strip application the bias is toward making more top end while sacrificing a little low down. For a race engine, it’s all about power output in the top 1,500 to 2,000 rpm. There are a lot of formulas that address the sizing of the primaries, but a method that I’ve used to great effect over the last 20 years employs a chart that shows the correlation between exhaust port flow at the peak valve lift and primary pipe diameter. Use the chart in this story (shown above) and you’ll find sizing in this department to be really close.
With the diameter selection taken care of, it’s time to consider primary lengths. Here we are fortunate to have a wide operational window due to the two-plane crank configuration of a Detroit V-8. In broad terms we can say that shorter primaries favor top end output while longer ones are better for low and midrange. That said, dyno tests show that the variance in output with a primary length change is, over quite a wide range of lengths, small. In my testing of a small-block 350 Chevy, I found that if the primary lengths fall within a 29- to 38-inch range, there is little difference in output. In reality, this test did not show the full range of primary lengths that would work. For street use, the primaries can be as long as 46 inches, and yet will still show good (though not optimal) top end results. We can conclude that fixating on precisely (or even closely) equal primary length is not a necessary requirement for a good header.
This is the conventional collector...
This is the conventional collector form, and it works well.
Just so we’re all on the same page, we’re going to define the collector as the part of the header that brings the four primaries together to form one pipe. After they have become one, the pipe itself will be referred to as the secondary pipe. In essence, we have three styles of collectors. First, we have the four-into-two-into-one collector that has been popular with NASCAR teams for the past decade or so. This type of collector rarely finds its way into the general hot rodding community, probably because of its cost and that it’s less easily installed. That said, it works, but the return on investment is generally small. This leaves us to consider the two most popular collector styles. The original simple parallel collector where the four pipes merge into a pipe the size of the intended secondary, and the merge collector where the four primaries come into a venturi-like form before expanding into the secondary pipe diameter.
A merge collector like this,...
A merge collector like this, however, often has the edge over a standard collector.
Fortunately, there is not much to cover here. First, the original parallel collector works well, but my dyno tests indicate that the merge collector has a slight edge in terms of power bandwidth, torque, and horsepower. In another area, if the design is sufficiently well developed, the merge collector has the ability to pull more vacuum via the evac-u-pan system, if one is being used. Notwithstanding, the merge collector is a little more costly to make, so ultimately you will have to go with whatever suits your budget.