There are two dynamics to consider when adjusting valve lash. First, there are valvetrain dynamics, and second, the gas dynamics of the engine's intake and exhaust system. In addition, there are two distinctly different types of lash adjustments to consider-those for a solid lifter and those for a hydraulic. On top of that, and surprising as it may seem, there are issues that also separate flat from roller lifters when it comes to lash adjustment. Lastly, there is thermal expansion to take into account, which can be an aggravating issue to deal with. In all, setting the lash is not quite the minor job it might first seem. Let's start with solid tappet cams and work from there.
The principal reason that lash is in the valvetrain system is to offset the effects of thermal expansion and extraneous manufacturing tolerances. A part of the cam lobe is designed specifically to deal with this issue and is known as the tappet clearance ramp. There are many forms, but the constant velocity and constant acceleration types are among the simplest and are commonly used. Their whole purpose in life is to pick up the valvetrain without unduly "shocking" it. The goal is a smooth transition from gently taking up the lash and transitioning into a rapid opening of the valve.
All this would be relatively easy to do, but the fact that parts expand when they heat up means what is set cold may not exist when the engine is at operating temp. For an all-iron engine this is not so big an issue. But for an engine with aluminum heads or one with aluminum heads and block, things can be different. Most cams call for the tappet clearance to be set when the engine is hot. This is not so easy to do at realistic temperatures (warm won't cut it), but there is a way to go without getting first-degree burns. First, set the lash cold and, on one or two cylinders, be picky about getting it just right. Next, run the engine until it is at operating temperature, pull the valve cover, and remeasure the lash. This gives you the amount of change. When the engine cools off, you can take your time and set the lash to a figure that allows for the expansion.
Valve lash (arrow) is so size...
Valve lash (arrow) is so size sensitive that just a few thousandths can make the difference of 20-30 hp.
Just how tight or loose the lash should be is largely a question of valvetrain dynamics. First, let's dispel a common misconception about valve lash and opening duration. I have heard it said more often than I care to recall that tightening the lash of a solid lifter cam (flat or roller) is good for top-end output because the duration has been increased. Not so.
Here's how it falls into place. Yes, the duration is increased but the rate of valve opening during the tappet ramp phase of the cam's profile is so slow compared with what is needed that it represents something more like a slow leak. One thing that a lot of cam testing has shown is rapid rates of opening and closing are what makes torque and horsepower. If setting the lash wider produced less output (less duration but faster opening) than tighter, it was probably because the closing rate of the valve was such that it momentarily bounced off the seat. Usually the best power is made when the lash is such that the tappet is just short of the end of the tappet ramp. Remember, we are setting the lash through a rocker ratio. What we are really trying to do is set the working gap between the lifter and cam. If the tappet ramp on the cam profile is, say 0.010 (ten thousandths), then through a 1.6/1 rocker, the lash to position the lifter exactly at the start of the opening phase at the end of the ramp, as measured at the rocker, will be 0.016 (sixteen thousandths). It's worth mentioning that when higher-ratio rockers are installed, the lash setting needs to get wider by a proportionate amount. As for output, best results are usually seen when the lash is as wide as it can be with the lifter just short of the end of the ramp.
Lifter Side Load Flat vs....
Lifter Side Load Flat vs. Roller Both cams and lifter combos deliver an identical lift curve, but because of the rollers' form the cam lobe needs to be a significantly different shape to do so. The shape of the cam in conjunction with the roller lifter is such as to create a considerable side load. This distorts the lifter body, allowing it to leak much faster and eventually collapse.
Cam companies typically quote the lash setting on the conservative side. That is, the follower takes up the lash reasonably well before the end of the ramp. What this means is that more often than not best output occurs when the hot lash is just a few thousandths wider than the cam manufacturer recommends.
Many race classes call for engines that have hydraulic follower valvetrains. Crate motor classes are a good example. Although it at first may seem like a hydraulic flat tappet will react the same as a roller (and vice versa), the reality is this is not so. The drawing labeled Fig. 2 shows where the key difference lies, and this difference can completely reverse the techniques for making the most power from either a flat or roller hydraulic.
It would appear the side loading on a hydraulic roller causes the body to distort sufficiently to cause the lifter leakage to increase to the point the lifter can hydraulically collapse under load. The way to fix this is to limit the plunger travel by some means. The usual technique is to make up a sleeve/shim/washer stack to limit the plunger travel-if the rules allow. The other alternative is to first disassemble the lifters and drain them of oil. Reassemble them and lash the lifters to within about 0.010-0.015 (ten to fifteen thousandths) from the bottom of their travel. Doing so limits the amount of lifter collapse by allowing the plunger to bottom out.

This scope trace from a Spintron...

This scope trace from a Spintron test shows what happens when the valve lash is too wide. The valve hits the seat with excessive closing velocity and proceeds to bounce back off (arrow). This costs a lot of power.

Polylocs are about the most...

Polylocs are about the most common form of setting and locking the valve lash once set. The problem is that tightening the Allen locking screw can upset the adjustment. This makes the job a little finicky, but it's time well spent to get it right.

It may not seem as if the...

It may not seem as if the hydraulics for a roller lifter can react much different to that of a flat tappet lifter. In reality they do, and understanding why can mean a sizable power edge over the competition.