Peak Limiting

[K9ACT]

I am beginning to think that I have a fundamental misunderstanding about peak limiting, especially negative peak limiting.

I set up my Volumax so that whenever the needle enters the red area, the bottom of a sine wave or a recording of my voice starts to clip.

If I feed this into the speech amp and put out RF, I can see this limiting at the base line.  It appears as a bulge, more precisely, like a breast with the nipple facing the previous wave.  As I increase the gain toward 100% the nipple end separates from the previous wave and looks just like overmodulation just as the top reaches the 100% line.

The net result is no different from bypassing the Volumax.

I am usuing the 811 rig for testing and there is plenty of audio for over 100% modulation.

It seems to me that, as long as negative peak limiting is done in front of the modulator, no matter what the audio input looks like going into the speech amp, one can still overmodulate with the audio gain control on the speech amp.

Is this correct?  If so, what good is peak limiting if one still has to ride the gain on the speech amp?
Perhaps not ride it but set it to a point where it can not overmodulate but I don't need the Volumax to do that.

js

[WBear2GCR]

Jack,

Yes, you can clip the negative peaks and then send a level that is in excess of 100% negative (baseline the transmitter) anyhow.

However, if you put in the 3 diode limiter circuit into the modulator it self, that will never happen... it will still clip the negative peak, but just before you can baseline, and splatter. That's why it works - no splatter.

In theory you should never have to ride the ,odulator gain control - it should be set, fixed and locked down! Then the Volumax should ride the gain preventing overmodulation, in whatever fashion you set it up. There shoudl be a gain control in front of the Volumax, that will set up the "density" of your sound. So, below a certain mic level in, the Volumax is not doing much and your sound will be fairly unprocessed (depending of course on how you set up the Volumax...) then as you crank up the mic gain (before the Volumax) it will begin to compress/limit/ride whatever it does the audio signal. In theory, if you really blast the input to it, you'll get a highly limited/compressed but very "dense" audio out of it...

The other thing, a flattened baseline look that is NOT actually at the baseline due to the negative clipping/limiting in a processor might look a lot like "baselining" the rig! You may have to look closely at the flat part to see if it really reaches the baseline or stops just short! Or put another way, as you hit it harder, the flattened area on the negative side will get wider due to the clipper/limiter working longer on that wave because it is flattening a "higher mountain" making that width greater...

Shooting from the hip...

        _-_-Want Both Two Gooey Chicken Ranch

[Ian VK3KRI]

Is this correct?  If so, what good is peak limiting if one still has to ride the gain on the speech amp?
Perhaps not ride it but set it to a point where it can not overmodulate but I don't need the Volumax to do that.

As Mr Bear said, once you have set the correct level at modulator it should be left alone.  You should be setting it up so that the carrier doesn't quite dissappear on negative modulation peaks. Being a peak limiter with a relatively fast release time (140ms) , you wont be able to see the effects directly  on a sine wave signal , and probably not on voice either.

The Volumax isnt an automatic level control, it alters the dynamics of the signal, effectively reducing the level of the peaks so that the average level can be higher before you hit 100% modulation.
                                                                                   Ian VK3KRI

                                                                       

[W2INR]

Here is an article written by W2ZE on Processing in the Audio section of our handbook. This may help.

http://amfone.net/ECSound/processing101-1.htm

G

[steve_qix]

Negative peak limiting (clipping) ahead of the modulator only "really" works if the modulator is DC coupled, and doesn't have any overshoot or bumps in the frequency response.  Otherwise, the modulator will modify the clipped waveform and you will not achieve the desired result.

Someone in a previous response pointed to the 3 diode negative peak limiter (clipper).  For a transformer-coupled modulator, this is the only way to achieve true negative peak limiting.  The circuit is simple and very effective !

With DC coupled modulators such as PWM, negative peak limiting is performed ahead of the modulator and audio filtering, and the modulator does not alter the waveform. (if everything is designed properly).

Here is the 3-diode circuit:


Regards,

Steve

[k4kyv]

But doesn't that diode clipper circuit still produce distortion and spurious sideband products as it hard-clips the negative peaks, just the same as overmodulation would?

So-called "ultramodulation" and "negative cycle loading", where the negative half of the audio cycle is attenuated but the positive half is unaffected, result in a signal that is identical to controlled-carrier modulation combined with even harmonic distortion of the modulating signal.

I once tried the original 3-diode ultramodulation circuit as described in the 1956 QST arcicle.  It worked perfectly as observed on the envelope scope pattern, but over the air, the signal reports I received said that my audio wasn't any louder than without it, but it was audibly more distorted.

The only advantage I  can see with any of those circuits is that they can be  made to protect the modulation transformer from operating without a load whenever 100% negative peak modulation is exceeded.  A simple circuit to remedy that problem without affecting the waveform of the modulated signal is to use one diode and a loading resistor the same resistance as the modulating impedance.  Ground the anode of the diode, connect the cathode to the resistor, and connect the other end of the resistor to the modulated B+ line.  A wirewound resistor rated at about 1/10 the audio output power of the modulator will be sufficient.  Replace the diode with an 866A, and it makes a good overmodulation indicator if you place the tube where it is readily visible - it flashes blue upon overmodulation.  A "keep-alive" voltage can be applied to make it flash at any desired percentage less than 100% negative.

Any time you modify a sine wave, that is by definition, distortion, and additional frequency products will inevitably be generated.

[Ed/KB1HYS]

You are right Don, There will be some distortion with negative clipping. But I think this addresses why it is not objectionable. The article was about a zener diode peak regulator, but I think any diode regulator would operate along similar principles.

Steve WB3HUZ on the AM window Tech Pages (http://www.amwindow.org/tech/htm/zenlim.htm)
...clip the negative audio peak near 100% negative peak modulation, the resulting distortion is minimal. Remember, at the instant the clipping occurs, the voltage on the final tube(s) is only about 5% the normal B+ voltage. This also means the instantaneous final plate current is also only about 5% of normal (or what you read on the transmitters meter which reads the average value). Knowing power is the product of voltage and current (P = EI), then the input power to the final at the instant the negative peak clipping occurs is only 0.25% normal. Thus, the distortion products created by the clipping will be no worse the 0.25% of the input power or unnoticeable. For example, if you are running 100 watts average input, the power input at the instant the negative peak clipping occurs is just 0.25 watts!

[AB2EZ]

As another view on this:

My audio chain has a steep low-frequency cutoff (more than 6 dB per octave) that is ~3dB down at 50 Hz, and I usually set the brick-wall high frequency cutoff at either 4.5 kHz or 7 kHz... depending upon band conditions.

All of my high-level-modulated AM transmitters (Ranger, KW-1, Class E, externally modulated AC-1) have a pretty flat frequency response, and a pretty linear phase response between 50Hz and 7 kHz (i.e., no significant roll off or non-linear phase characteristic until the frequency drops below 50 Hz or rises above 7 kHz).

As a consequence, the waveform that is produced by my broadband off-air monitor nearly perfectly tracks the audio waveform at the output of my audio chain (which is, of course, the input to the transmitter).

I can observe this on a modern, dual trace, real time, digital oscilloscope.... which can capture and display the waveforms... with the triggering set to capture negative peaks. After scaling, and offsetting for the carrier level, the traces superimpose nearly perfectly. If they don't superimpose nearly perfectly, then I know that something is wrong (usually softening of positive peaks due to improper loading of the RF output stage, or residual distortion in the Ranger's Class AB1 modulator when I run the Ranger at full power)

Thus, in my case, if the audio chain is doing negative peak limiting (which it is), and if the audio gain of the transmitter is properly adjusted ... the negative peaks of the modulated rf output of the transmitter will also be tightly limited.

However, I do agree with Steve (who, I believe, generally prefers a much lower low frequency cutoff than I use; and which is certainly a matter of personal preferences in audio) that IF the low end frequency content of the audio signal that you have negative-peak-limited is not passed with high fidelity by your transmitter/modulator, e.g., if your transmitter (for some reason) introduces a non-linear phase vs. frequency characteristic within the band of frequencies contained in your input audio signal... then your r.f. output envelope waveform will not track the input audio waveform. In that case, for illustrative example, an audio signal with desirable 125% positive peaks (vs. 95% negative peaks) might easily be converted by your transmitter/modulator to one with undesirably large negative peaks. Anything that disturbs the phase relationships of the various frequency components (whether intentionally using an "all pass" filter, or unintentionally due to non-linear phase vs frequency characteristics within the band of interest) will change the shape of the waveform vs. time, and likely mess up the control of the negative peaks.

Other than the limitations imposed by the modulation transformer (both linear and non-linear), I don't think there is any reason for a high-level-modulated transmitter to have an audio response with a non-linear phase vs. frequency characteristic within the band of interest.

If your modulation transformer is introducing funny effects due to saturation... then using a negative peak limiter on the output of the modulation transformer will probably not remove the distortion products (including out of band "splatter") that the transformer has already inserted.


Stu