I have a Question about Peak Room in Limiter

HI Kali:

What do you mean by the term "Peak Room".

Please comment.

miguel

Dear mmmike

mean of Peak Room

Thank you~!

Hello,

The ADAU1452's 8.24 format number system allows for an internal signal reaching peaks from -128 to +127.999 -- a level of about 42 dB over the peak external level of -1 to +0.999. The Limiter works over this entire range, as shown by the simple experiment below. Notice that no audio modulation scheme is needed for a test like this, because the limiter operates down to DC.

We see that the output stays at 1.0 (zero dB) while the input approaches +126 (42 dB above 1.0).

Thus you have plenty of room, however it's defined.

Best regards,

Bob

Dear KJBob

Thank you so much about your answer

KJBob I have a Question.

Please Can you answer me?

Please help me..

[My Question list]

What is Relationship about "RMSTC" and "DECAY" and "RATIO"?

Why is "RMSTC" and "DECAY" and "RATIO" together change?

I don't understand....

Hello kali,

Not having designed the algorithm, I can only guess: The decay values you're using (5 to 7) are all very fast -- 5 corresponds to a time constant of only 0.6 mS. To see a table of time constant to decay numbers, scroll down to the bottom of this reference:

https://ez.analog.com/message/164321#comment-164321

With a fast attack time such as 10000 dB/s as well, the limiter is now acting upon the waveform itself, not merely its level. This is actually an undesirable condition as it distorts the incoming waveform. To prevent distortion, the limiter's gain should remain constant over an entire audio cycle. For good voice and music reproduction, this means that the limiter must always remain slower than the lowest audio frequency input. Thus for most applications, use a decay number of 12 (the default value) or higher. In any event, the well-researched measurements you have made show a level ranging from 0.51 -- 0.57. In dB terms this works out to 1 dB -- not very much.

Best regards,

Bob

Dear KJBob

Thank you so much.

your answer is best of best!!!

so decrement of output level and ratio cause is I uesd big RMStc value or small Decay value?

What is Relationship between decrement of output level and Limiter sensing time?

can you tell me about that?

I need help of you!!!

Hello kali,

For most applications, you would use a slow value for decay (such as the default value of 12). With a slow decay, the limiter's gain won't change much during one audio cycle. The RMSTC value regulates how fast the limiter responds to rapidly increasing audio level -- make it only as fast as you need to get the desired limiting. Set up this way, the limiter acts like the "automatic level control" found in portable cassette recorders (remember those?). When someone is speaking the record gain remains relatively constant, but when they stop talking, you hear the background noise slowly increase as the ALC rides the gain upward.

If you're running the limiter so fast that its speed settings affect the output level, it's also distorting the waveform. To illustrate this, I built the simple circuit shown below. It runs a 100 Hz sine wave into the limiter, then measures both the limiter gain and output waveforms. A simple "equivalent time" sampling arrangement slows the 100 Hz waveforms down to 1 Hz, allowing us to see them on the Real-Time Displays:

With the stock settings of RMSTC = 50 and Decay = 12, the Limiter's gain remains constant through each audio cycle. Thus, the audio waveform comes through clean:

To dramatize what happens with a very fast limiter, I set it for RMSTC = 10000 and Decay = 5. Now we see the gain shift during each audio cycle, and the resulting audio is a distorted mess:

So we may conclude -- don't run the limiter too fast, and its output level won't change.

Best regards,

Bob

Dear KJBob

thank you for help of you in every time.

you are very smart and kind men

real thank you so much .

You're welcome! By the way, if your application needs a limiter that acts instantly and without distortion, you can make one yourself. See an example at:

Limiter with "Zero" attach time?

Bob