Spurred on by Joel's post on pushrod engines and some of the comments that it generated, I think it's time to take a closer look at the differences between OHC and OHV engines. We all "know" that OHC engines rev higher, produce less torque, increase specific output (power per unit displacement), operate more smoothly, and so on and so forth. Those are stereotypes, but ones that are indeed based on some honest real-world experiences. So were do these characteristics come from?
To figure this out, we need to first understand that engines are air pumps, and the volume of that air is what affects both power and torque characteristics. Peak power is basically determined by the limit of airflow per unit time. Peak torque is determined by airflow per combustion cycle; essentially, how much air is crammed into the cylinder on every intake stroke. This should hint towards the intertwined relationship between horsepower and torque, but we'll cover that at some other time.
Peak power is relatively easy to achieve, airflow on a naturally-aspirated engine mostly being a function of the cross-section of the intake tract (the exhaust tract is important as well, but less so because there's more pressure available to expel the exhaust gases). Specifically, the smallest cross-section of the intake port is going to present the largest restriction to flow if an even velocity is achieved throughout the port (and that's a big "if")*.
Usually, the area of the intake valve(s) is the ultimate determining factor, but this assumes that the intake valve is fully opened. By this, I mean that it's lifted off the seat by at least 0.25 times the valve's diameter, which is where the valve "curtain area" (the circumference of the valve times lift) equals the area of the valve itself. This should allow the valve to flow at its maximum - we'll come back to this in a moment.
To achieve a large valve area, we need to maximize the diameter of the valves. In a two-valve head, this places each valve alongside the edge of the cylinder, a condition called "shrouding". Due to the proximity of the cylinder wall along a part of the valve diameter (shown to the right as a red line superimposed on the combustion chamber), a significant portion of that curtain area becomes unusable. This is why it's often necessary to lift the valve well beyond the 0.25 x diameter figure of merit in order to achieve peak flow. This is also why, all things being equal in terms of displacement, larger bores usually make more peak power - it frees up room for more valve area. Four-valve setups take advantage of more of the cylinder area, but almost as importantly, they suffer less from shrouding since the smaller diameter of the valves doesn't as closely follow the cylinder wall (the valves can shroud each other, but that's less of an issue).
All of the above points to an obvious airflow advantage for four-valve heads, given a fixed cylinder bore size. But there's yet another advantage, this one related to the curtain area. Since multiple valves necessarily result in smaller valve diameters, this means that less valve lift is required to maximize flow. Less lift and smaller (read: lighter) valves makes the job of the valve springs much easier. Indeed the difficulty of closing the valves is often a limiting factor to how high an engine can safely rev, and it's an extremely difficult problem to work around in a production engine where a valve spring life of a few thousand miles just isn't acceptable.
This clearly points towards a multi-valve design - and almost by default, overhead cams** - as being superior for peak power. No big surprise, eh? But peak power is rarely what we're after in a production engine.
More important is maintaining a healthy amount of torque over the usable rev range. No, this isn't some sort of claim that Torque Is King, since proponents in that camp are usually interested only in peak numbers, and preferably at a low RPM. To obtain this, we need to fill the cylinder as much as possible across the rev range. Simply maximizing the valve area is not the way to accomplish this task, and choking down the intake tract to maintain velocity at low revs isn't the way to go, either.
Assuming that sufficient bore diameter is available, and that there is enough displacement - such a nasty word! - to keep the maximum operating speed under, say, 6000 RPM (for those engines not employing exotic valvetrain components such as titanium or hollow-stem valves), it is quite possible to achieve great results with a 2V pushrod engine. Spend enough money on the valvetrain, and that arbitrary rev limit goes away, too. Additionally, the low-RPM airflow characteristics of a 2V wedge-type head are usually superior to those of a pent-roof narrow-angle 4V design, with more swirl (airflow rotation parallel to the cylinder axis) and tumble (perpendicular to the cylinder axis). Additionally, the area of the chamber that's not occupied by valve allows the addition of addition quench area, which adds further turbulence to the mixture during the compression stroke. All of this can add up to excellent low-speed and part-throttle performance, which is why an engine like Chevrolet's LS7 can offer nearly 75% of its peak torque anywhere between idle and redline, offer up 10% additional usable revs after hitting peak power, and manages to pull down some extremely respectable economy numbers. Hey, there's a reason that Honda's VTEC system on its V6 Accords barely cracks open one of the two intake valves below 3000 RPM.
For those that evaluate an engine based on mass, packaging volume, and fuel efficiency, OHV designs are very attractive, for stuffing a pair of cams into the cylinder heads adds volume and mass at just about the worst possible place on a V-configuration engine. Add in some roller followers and tall valve springs, and all of a sudden we've got V6s that are larger than V8s, and "small" V8s that are larger than the big-blocks of the 60s. During an SAE presentation that I attended, Chevy's Dave Hill stated that the Nissan VQ35 DOHC V6 was benchmarked during the development of the C5 Corvette, and was clearly found to provide significantly less power per unit mass and unit volume than the GM's GenIII V8. Peak-power-per-unit-displacement is strictly an amateurish way to compare two engines.
What about smoothness, NVH, power delivery, the touchy-feely stuff - do OHC engines really offer an advantage? To some extent, yes. The OHV valvetrain tends to create a rather long string of mechanical interfaces, each bringing with it the potential for noise and vibration. And rocker arms can make a heck of a racket as well (as anyone who as installed a set of aftermarket roller rockers knows). But OHC engines necessarily place the cams far away from the crankshaft, which means that the cam drive system often has an opportunity to emit noise. As well, any noise generated by the cam-follower-valve interface is now generated at the top of the engine, instead of being buried in the block. Sure, we can throw a set of thick and heavy cast valve covers on the head, but now we're adding even more bulk to the outer corners of the engine. From a practical standpoint, I call this one a draw.
So why do OHV engines have such a poor reputation for refinement? Well, where have most people experienced an OHV engine in the past decade? It's almost assuredly been in a pickup truck, full-size SUV, or a GM car. Truck and SUV engines are not exactly the place to seek out refinement, and until GM's introduction of the new 3500 and 3900 this year, that company was using a pair of architectures that are older than many readers of this site (let's not forget that the 3800 was indeed a 90-degree V6, a configuration not known for smoothness regardless of camshaft placement). Hop into a vehicle equipped with GM's GenIII or GenIV V8 or Chrysler's Hemi, and I assure you that there will be no complaints with regards to the engine's behavior. Even GM's older GenII LT1 pushrod V8 behaved itself extremely well when compared to other engines of its time.
So what's my final conclusion on this topic? For most inline engines (either four-bangers or straight-sixes), OHC makes perfect sense. Smaller engines that make use of extremely high revs are also obvious applications. But if we look at passenger-car applications where the displacement is over 3 liters or so and revs rarely top 6K, then there is absolutely no reason to believe that a properly-designed OHV engine cannot deliver power, refinement, and compactness comparable to an OHC engine.
* The extremely knowledgeable Judsen Massengill (of S.A.M. fame) considers the 0.5" before and after the intake valve to be the single most critical area of the engine when it comes to making power, since the geometry is so complex in this area.
** GM was rumored to be exploring a three-valve head for its GenIV V8, using an in-block cam to actuate two intake valves via a single lifter and pushrod, and a rather complex rocker arm assembly. Many modern diesels also actuate four valves per cylinder with pushrods and block-mounted cams.