The phenomenon you guys are experiencing is the “slew rate” of the circuits in all the electronics involved, the ignition amplifier, the rise time of the coil, the rise time of the voltage in the plug wire and at the spark gap, and last, but not least, the slew rate of the timing light. All electrical devices have a fixed time it takes for them to perform their function. Every part of a circuit has a time constant, in electronics jargon called ‘slew rate’.
http://en.wikipedia.org/wiki/Slew_rate
Every electronic circuit has a fixed amount of time it requires to change state from off to on or from one voltage to another. In the case of ignition amplifiers, magnetic triggered or otherwise, the devices have a fixed time from the trigger pulse to the change of state of the output. It is common to see a ‘rise time’ specification for ignition systems and particularly coils. The difference in rise time between CD and induction systems is often mentioned.
A device that has a lot of transistors in it, like the detector or pre-amp in an ignition system or the micro-chip calculator in an adjustable timing light, can be pretty slow because it has a slew rate for each transistor and the delays are additive. For the device to function each transistor in turn must switch from off to on.
The bulb in the timing light has a fixed time it takes for it to achieve incandescence from its off state. Xenon and neon are the gases in flash tubes. Xenon is most common in modern timing lights and more desirable for two reasons, its brighter white light and faster rise time than Neon, which has an orange light.
The bottom line of all this, and the reason you see the retard in your timing (all else in good shape, no mechanical gremlins like the cam walking) is at high RPM the crankshaft moves more degrees closer to TDC in the same amount of time.
To illustrate (these numbers are not actual, but for example only), if the ignition and timing light use 1 millisecond every firing, and that is one degree at 1000 RPM, it will be 6 degrees at 6000 rpm. The crank will travel 6 times as far in the same amount of time because it is going 6 times faster.
To see the timing be 36 degrees at 1000 RPM the actual trigger event (points open, VR triggers, Hall effect switches, whatever) will have to be 37 degrees. At 6000 RPM the trigger is still 37 degrees but the crank moves 6 more degrees in the same time it took to move one degree at 1000 RPM, so the light shows 30 degrees.
For some perspective do the math: 6000RPM / 60 = 100 revolutions per second X 360 degrees = 36,000 degrees per second. Each degree is 1/36,000th of a second or .00002778 sec. 8000RPM is 48,000 degrees per second.
All systems that don’t calculate the timing from the previous cylinder’s firing event exhibit this retarding phenomenon to some degree (pun intended) and unless someone figures out a way to repeal laws of physics as old as the universe, that’s just the way it is.
Not all ignition amplifiers have the same slew rate. Some are much slower than others. Some are as slow as two degrees (or more) per 1000 RPM. The Ford Dura-Spark and Mopar ‘orange’ boxes are particularly slow. The actual “Chrome NASCAR Box” Mopar sold in the late 70’s is not in the slow category, but your chances of ever seeing one are nil to zip. Genuine GM HEI modules are about the same as MSD and other CD boxes, with about 1° or slightly less retard per 1000 RPM. Aftermarket HEI modules can be as slow as 2° (or more) per 1000 and some of them seem to be voltage sensitive, lower voltage makes them slower. Aftermarket modules of the same brand, part number and appearance can have different behavior. So far, actual GM modules I have handled have been the same (in slew rate) since they were introduced in 1974.
Ironically, because a points system has only one switch - the points, a properly functioning points system will have less retard than electronic systems. 35+ years ago, when a good points system was replaced with electronic it was common to lose performance because of the retard, in spite of more spark energy.
An ‘ah-ha’ moment came in connecting the dots between the advice of Smokey Yunick, Bill Jenkins and FoMoCo, all of whom advised timing curves for specific engines which advanced the timing a degree or two (or more, in some cases) all through the engine’s power range and sometimes beyond the operational ceiling of the valve train. This was proven to work well with points and with a little tweak of softer springs or a different cam shape, was adapted to negate the slew retard of the amplifiers and provide a fixed total or advancing timing as desired. The mechanical advance can be set to advance the same amount as the electronics are retarding.
It is unfortunate you guys who never raced 8000+ RPM engines with points missed the experience. I used several hundred sets of D112P Delco points in the two decades between the early 60’s and the early 80’s. Big Joe knows what I mean, I’ll wager.