View Single Post
Old 26th December 2007, 09:17 PM   #256
Worthless Aging Hippie
ktesibios's Avatar
Join Date: Feb 2002
Posts: 1,493
Before this thread slips down into backpage Hell, it occurs to me that one bone of contention was the expected response of limiters such as are commonly found in the audio chain of video equipment to suddenly applied signals louder than their limiting threshold and how this would affect other signals present in the background.

This has been discussed only in words; perhaps some measurements would help to clarify the issue. Fortunately, the existence of circuit modeling software makes it possible to do just this sort of testing quickly and conveniently.

So, first I designed a very simple peak limiter in a modeling program:

The gain control element is the industry standard THAT 2180A VCA. The sidechain consists of an absolute-value circuit (an active full-wave rectifier) followed by a threshold circuit, then a peak detector and attack/release timing network and a buffer for the control voltage produced by the sidechain. The circuit is peak-responding, as you would want for a save-me-from-overloading-the-signal-chain limiter, the threshold of limiting is 1V peak, the attack time constant is less than 1 mS, the release time constant is 200 mS and it manages a ratio of approximately 50:1 for signals above threshold.

This screen capture shows how it responds to a simple tone being suddenly incresed in level from 1V peak(below threshold) to 5V peak (14 dB above threshold) (blue is the input signal and red is the output.):

There is a slight overshoot on the first cycle after the level increase, as the control loop takes a finite time to respond to the sudden appearance of an over-threshold signal, but it quickly reduces the gain so that even though the input signal has increased fivefold in amplitude, the output signal remains at a peak level of 1V.

This shows how it responds to the signal level being suddenly turned back down to 1V peak:

The output level initially drops by the same amount as the input and then gradually recovers as the control voltage decays, controlled by the release time constant.

So now what happens if a below-threshold signal is already passing through the limiter and a significantly above-threshold signal is suddenly added? To test this, we start with a 1 kHz 1V peak signal and then add a 100 Hz 5V signal. This is the result:

The 1 kHz "background" signal is reduced in amplitude from nearly 2 divisions peak-to-peak down to about 1/3 division. Since the peak amplitude of the composite "background" and "explosion" signals is 6V, the limiter has to reduce its gain by a factor of 6 to bring the composite signal down to 1V peak and the "background" and "explosion" components are affected the same way.

This shows what happens when the 5V 100 Hz signal is shut off, leaving the 1 kHz background signal:

Again, we see the background signal start out at the amplitude produced by the existing amount of gain reduction and then rise gradually as the limiter recovers from the previous high-amplitude signal.

So, there it is complete with pictures- a comp/limiter can't magically separate mixed signals and affect one without affecting the other. Folks like Jonnyclueless and Maccy, who tried to explain this to LastChild in words, had it right.

Thus it is demonstrated. Thank you and good night.
Ship me somewheres east of Suez, where the best is like the worst, where there ain't no ten commandments and a man can raise a small, bristly mustache.
ktesibios is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top