Precision Maximizer - Harmonics

[Dan Duskin]

I just did some frequency analysis, and I found some very interesting things...



Most prominent harmonic = 3rd order harmonic (5th in music)

Second most prominent harmonic = 5th order harmonic (10th in music, a.k.a \"major 3rd\")

Third most prominent harmonic = 7th order harmonic (flat 14th in music, a.k.a. \"minor/dominant 7th\")

Fourth most prominent harmonic = 14th order harmonic (flat 21st in music, a.k.a. \"minor/dominant 7th\")



What's up with the 7th and 14th order harmonics? Are these really that prominent in real tubes? And what happened to the 2nd and 4th order harmonics? This seems really strange to me...

All the spectral analysis of tube harmonics I can find online show the strongest harmonic orders being (in order) 2nd, 3rd, 4th, 5th, etc...
(that equals a major triad in music; octave, 5th above octave, two octaves, 3rd above two octaves, etc...)



PS: shape was set to 50%, single-band, limiter off, nothing clipped 0dB.

PPS: I did this test after a disapointing attempt to try and replicate the sound of a clean bass guitar pushed through some tubes (I have both files, before and after the light tube drive). I matched the amount of distortion, but the sonics were WAAAAY off.

 

[Chi-Squared Mastering]

It depends on the tube type, actual circuit implementation, and the individual tube. Some have a slightly stronger 3rd harmonic, it's just not as prominent as it would be in a transistor or op-amp. Take a look at http://www.milbert.com/tstxt.htm for example, Fig. 6 shows distortion characteristics for a pentode tube amp that produces more 3rd harmonic than 2nd.

Strange for there not to be ANY even harmonics though, and really I don't think the 14th is an intended addition so much as a byproduct similar to aliasing. Even if it is there by design, it's obviously not for the PM to sound like tube saturation.

Still, I don't think it's actually supposed to sound like tube distortion. Most of the PM section of the manual talks about \"soft saturation\", and anytime you have that typical symmetrical saturation you will add only odd harmonics. Only one part mentions tubes, when it says to set Shape to 50%, and it seems like they're comparing the relative amount of high harmonics versus low, not even versus odd.

They probably should reword that paragraph to not mention tubes, considering how much marketing hype capitalizes on telling people about even harmonics. Even so, I don't think the PM is really meant to sound like \"tube\" anything so much as just providing saturation and limiting.

 

[Dan Duskin]

This post is mostly a rant.
The following post (below) contains test based information...

I quite like the PM on bass guitar, & I think it's probably one of the better tube emulators in digital form right now.

I've never been happy with any type of digital emulation/model of analog distortion, & I figured if anyone could do it, it would be UA! Sadly, No.

Precision Maximizer is one of the best harmonic distortion generators avalible in the digital domain.
But 3 questions remain (which include all similar digital processes I have tried).

#1: Is it suppossed to sound like a tube, tape or transistor? Even if it's said to sound like one, it never sounds like any of the above, & realistically sounds more like digital distortion with a low-pass filter. Why is this?

#3: The Sonic (spectral response, etc) is always way off. Why? I wonder if this has to do with simply obtaining the right balance of harmonic distortion...

#3: All digital emulations of analog distortion (mainly attempting to emulate tubes or tape) have linear levels of distortion which are unlike the real thing which tends to have a much more exponentially sinusoidal dynamic response.


Digital emulations of analog distortions are getting better, but it's happening VERY slowly! What the PM does is similar to what other plugins have been doing for years. What technical aspect is being ignored? Obviously something is being ignored, because I can hear it clear as day and night! For some reason the component is either not known to the programmer, maybe considered to be not important to the signal, or possibly considered to be too CPU/DSP intensive. Which is it?

That's my 2 cents...

 

[Dan Duskin]

Comparing the Precision Maximizer to the Inflator...

I N F L A T O R - H A R M O N I C S . . .
Most prominent harmonic = 3rd order harmonic (5th in music)
Second most prominent harmonic = 5th order harmonic (10th in music, a.k.a \"major 3rd\")
Third most prominent harmonic = 9th order harmonic (16th in music, a.k.a. \"2nd/9th\")
Fourth most prominent harmonic = 12th order harmonic (19th in music, a.k.a. \"5th\")

In the case of Inflator, it is creating a major 2nd (or 9th, without the 7th)... and eventually a dominant 9th.
The Precision Maximizer is creating a hard dominant chord.
(I would suspect that the PM should sound better than Inflator in minor keys,
and Inflator should sound better in major keys, but this is only due to the fact that both are very non-linear
and skipping the 2nd & 4th order harmonics, that real tubes have)

Another way to look at it...
The Precision Maximizer is making larger leaps (regarding harmonics)

Yet another way to look at it...
The Inflator obtains the 9th order harmonic (2nd musically) before most tubes do

And yet another way to look at it...
The Precision Maximizer obtains more tube based harmonics \"in order\" than Inflator
(however, both processes seem to be missing the 2nd and 4th order harmonics)

Looking at the spectral graph (regarding the slope):
- Inflator adds harmonic content more linearly
- Precision Maximizer adds harmonic content less linearly


SUMMARY:

Precision Maximizer generates more tube based harmonics
Precision Maximizer generates higher harmonics like tubes do!
Precision Maximizer generates 7th order harmonics when it should not

Inflator generates a more even slope of harmonics like tubes do!
Inflator focases more on even-multiple harmonics (octaves), 2, 4, 8, etc
Inflator generates more 2nd and 3rd order harmonics (a major triad) than PM


ANALYSIS & CONCLUSION:

Both Inflator & PM are similar, but sonically different when pushed hard
Neither Inflator nor PM generate 2nd or 4th order harmonics like real analogue tubes do!

 

[RWIL]

Very interesting report! It will be interesting I hope to read soon the Doctors from the Webzine about how they outline the PM!

 

[LFranco]

I enjoyed reading this. Thanx!

 

[Dan Duskin]

It depends on the tube type as well as the individual tube. Many pentodes have a slightly stronger 3rd than 2nd harmonic, and some triodes do as well, it's just not as prominent as it would be in a transistor or op-amp.

On theory alone... that sounds just fine (i.e... sonically, should sound good).

Strange for there not to be ANY even harmonics though, and really I don't think the 14th is an intended addition so much as a byproduct similar to aliasing. Even if it is there by design, it's obviously not so the PM can sound like tube saturation.

I also wondered if it is an aliasing artifact. Though I thought was still a bit odd for it to be the 3rd most prominent harmonic, even for an aliasing artifact. I quickly checked the degree of flat and sharp cents, and the 14th order harmonic appeared to be about minus 28-34 cents flat (as it should, being a true analog harmonic, and not a digital aliasing artifact).

Still, I don't think it's actually supposed to sound like tube distortion. Most of the PM section of the manual talks about "soft saturation", and anytime you have that typical symmetrical saturation you will add only odd harmonics. Only one part mentions tubes, when it says to set Shape to 50%, and it seems like they're comparing the relative amount of high harmonics versus low, not even versus odd.

Yes, that is why I used 50% shape in my test. I was wondering the same thing... though I was pretty sure (after reading the manual) that a shape setting of 50% is supposed to give a "tube" saturation sound.

They probably should reword that paragraph to not mention tubes, considering how much marketing hype capitalizes on telling people about even harmonics. Even so, I don't think the PM is really meant to sound like "tube" anything so much as just providing saturation and limiting.

I agree, regarding the even order harmonics and tube marketing thing... it is a farce. The only reason that became known is because Tubes contain more 2nd order and less 3rd order compared to solid state designs... but that's only talking about the 2nd and 3rd!
So much sonic detail happens at the harmonics above that! And the really sad thing is that, years ago when those audiophile folks said "even" and "odd" they were talking about just one harmonic (2nd or 3rd), where the mathematic definition of "even" and "odd" defines more than one number. But, even using that old false logic, neither the PM nor the Inflator gives a 2nd order harmonic (i.e., it's lower in level than the 20th order harmonic!). Hmmm.....

 

[Dan Duskin]

Very interesting report! It will be interesting I hope to read soon the Doctors from the Webzine about how they outline the PM!

As long as it's honest.

I only say that because of the PCI-E report they gave, which was blatant marketing hype over fact (unlike most of the Dr. Reports before that). I know they are smart and deserve great praise, but I just hope they will stick to facts and science, and not be swayed by marketing to embellish something where it doesn't deserve it. I would much rather know the facts, to help me decide on mixing choices (yes, of course my ears come first). If they explained the missing harmonics or other components, and explained why, that would make me very happy. After all, the PM sounds great on several of the instruments I've tested it on. I have not said it is poor, not even once! My only question is... does it sound like tube saturation? So far, I think no.

 

[Dan Duskin]

.

 

[lonestar]

Dan (and all who care)

I, too, tried to evaluate the harmonics generated by PM, and found out that with an input around 0dB (100 Hz sine wave), no Limiter, no Split, Shape and Mix at 100%, I get only the 3rd and 5th harmonics, and no (annoying) upper harmonics.

It's the Limiter which makes the big difference, generating harmonics way up to the 14th, as you wrote (actually, I believe it's the 15th...).

Clearly there's some kind of symmetrical waveshaping/distortion going on inside, something which is somehow comparable to the effect of tape saturation. Even harmonics (usually associated with tube devices, namely triodes) are completely missing.

So yesterday, out of curiousity, I created a wavefile with a fixed DC offset at -6dB and added it to the 100Hz sine wave, just by putting the DC file on a track in my DAW and mixing the offset in with the fader, and...
...lo and behold!, all the missing even harmonics are back as expected!

The same thing happens with the Limiter engaged: all the harmonics from the 2nd to the 14th are there.
As everyone can imagine, the colour of this kind of enhancement is quite different from the original one.

The downside of this process is that you lose up to 6dB of dynamics (not a great deal, if you work at 24 bits) and that a HP filter is needed after the PM, to filter out the DC (the Precision Eq's HP filter is perfect for this), not to mention all the hassle of having an extra track to manage, a DC wavefile to create, etc.

In the end: wouldn't it be nice if PM rev.2 incorporated an internal offset control to give us one more option to choose from?

You're all invited to give it a try, of course, and post your opinions.




Ciao,

lonestar

 

[JuergenW]

I think here´s the answer why there are only odd and not even harmonics...

The UA doctor wrote in Webzine May/2007:

\"Harmonic generation is also common for bass processing. Sending the bass through a nonlinearity produces harmonics of the bass signal. Sometimes called \"phantom bass,\" a dense distribution of harmonics results in a perceptual increase in bass, although the energy added is actually in upper bands. Post filtering allows control over the distribution of harmonic energy. By means of synchronous processing, bass subharmonics can also be generated along with higher harmonics. Typically, low frequency energy is sent through a period detector, and a signal is generated at twice the period of the signal. The signal may be a square wave, or, alternatively, can be a wavelet that is repeated periodically. If a square wave is used, filtering is performed to control the spectral distribution of energy. The wavelet technique is convenient for sampled systems because, with proper care, alias-free processing can be carried out without oversampling. Finally, linear processing can be performed to increase perception of bass by advancing the bass-frequency content in time relative to the higher-frequency content.\"

It seems to be that both, UAs new Maximizer and Sonnox Inflator, are
using an applied square wave charakteristc (to enhance the signal). Because
of this there are odd and no even harmonics. I think they use the
wavelet transformation for generating the harmonics, in order to avoid
oversampling and because of their flexible and accurate behaviour.
(AFAIK both have no additional latency due to oversamling).
Wavelet is mostly used in compression algorithms (for example JPG is
based on wavelet) and is good for analyzing structures of source material.

Because of the 3rd and 5th I like it a lot on Bassdrum and Bass (and
I did it right from the start without knowing which harmonics the maximizer
generates).
This fits also with the reports of users of familiar results due to clipping and
transistor (1176) saturation.

Because of this I believe there is NO even harmonic at all. The 14th harmonic
measured by users must be either the 13th or 15th harmonic - which is
frequency-wise very close and not easy to differ on an FFT Analyzer.

I think they made the odd harmonics for one good reason:
Psychoacoustic enhancing of bass without much energy.
It maybe that 2nd and 4th harmoncis (which are octaves) fight to much
with other instruments playing in the upper octaves and/or produce to much
mud in the bass-octave for an \"maximizer\" harmonic generator to be
very effective. I reflect this to EQ tricks where engineers for example
lower 60 Hz content in Bassguitar and lower 120 Hz in Bassdrums to get
them blend better without covering each other. To my ears the harmonics
generated by the UA maximizer softens the Bass, or \"makes it round\"
instead of loud - which is commonly contected with 2nd harmonic octaves.

Interesting would be to compare the Neve33609 / Fairchild THD Spectrum.
Anyone did so far?

BTW: - Transistor and Tape distortion produces only odd harmonics. :wink:

 

[JuergenW]

I also wondered if it is an aliasing artifact. Though I thought was still a bit odd for it to be the 3rd most prominent harmonic, even for an aliasing artifact. I quickly checked the degree of flat and sharp cents, and the 14th order harmonic appeared to be about minus 28-34 cents flat (as it should, being a true analog harmonic, and not a digital aliasing artifact).

[quote="Chi-Squared Mastering":xk75dlho]Still, I don't think it's actually supposed to sound like tube distortion. Most of the PM section of the manual talks about "soft saturation", and anytime you have that typical symmetrical saturation you will add only odd harmonics. Only one part mentions tubes, when it says to set Shape to 50%, and it seems like they're comparing the relative amount of high harmonics versus low, not even versus odd.

Yes, that is why I used 50% shape in my test. I was wondering the same thing... though I was pretty sure (after reading the manual) that a shape setting of 50% is supposed to give a "tube" saturation sound.

They probably should reword that paragraph to not mention tubes, considering how much marketing hype capitalizes on telling people about even harmonics. Even so, I don't think the PM is really meant to sound like "tube" anything so much as just providing saturation and limiting.

I agree, regarding the even order harmonics and tube marketing thing... it is a farce. The only reason that became known is because Tubes contain more 2nd order and less 3rd order compared to solid state designs... but that's only talking about the 2nd and 3rd!
So much sonic detail happens at the harmonics above that! And the really sad thing is that, years ago when those audiophile folks said "even" and "odd" they were talking about just one harmonic (2nd or 3rd), where the mathematic definition of "even" and "odd" defines more than one number. But, even using that old false logic, neither the PM nor the Inflator gives a 2nd order harmonic (i.e., it's lower in level than the 20th order harmonic!). Hmmm.....[/quote:xk75dlho]

IMO we should differ the saturation and harmonic content topic of tubes
and nonlinear devices. The saturation is a dynamic nonlinear process,
the harmonic content is the distortion applied by a device.

When UA writes, that at a shape of 50% it´s most like a tube saturation
we have to take that literally - it´s refering to the dynamic saturation curve.

What about they dropped the 2nd order because they had a reason to do this
for a better application of this plug-in?
I think we have to think different about those maximizers. Today they can
invent improved characteristics, f.e. take the saturation of tubes and add
harmonics of tape.

The upper harmonics are included for that glassy/punchy sound. IMO
it´s not due to aliasing, because in that case they would have used oversampling.
If they used the wavelet transformation oversampling is not necessary.

 

[lonestar]

<bump>

 

[Dan Duskin]

I've been doing a few tests, comparing it to real tubes, both by ear, and by spectral graphs to view the response (and the harmonics), as well as trying other plugins. I'll report back soon!

 

[Macc]

Post screen grabs if you can please Dan

 

[Patrick de Caumette]

Dan, for the record, a major 2nd up two octaves would be a 16th, not a flat 16th (which would be a real bad idea to include in a maximizer, except for trash metal :twisted: )
Same goes for the 5th: it is, if you have to name it that, a 19th, not a flat 19th/b5 (which again may be cool for trash metal or Purple Haze)

Regarding adding a b7 to a major mix, it would actually sound pretty bad if the music's tonal center is major with a natural seven. For blues I guess that may work... b7 with minor can also work but frankly, why add a b7 harmonic?

 

[Dan Duskin]

Dan, for the record, a major 2nd up two octaves would be a 16th, not a flat 16th (which would be a real bad idea to include in a maximizer, except for trash metal :twisted: )
Same goes for the 5th: it is, if you have to name it that, a 19th, not a flat 19th/b5 (which again may be cool for trash metal or Purple Haze)

Whoops. That was a copy & paste error. I was jotting down lots of information about different harmonics (some not listed), and using copy and paste before altering the numbers... but I forgot to remove the word "flat". That would be a cool passing tone for melodies, but chordaly a nightmare!

Regarding adding a b7 to a major mix, it would actually sound pretty bad if the music's tonal center is major with a natural seven. For blues I guess that may work... b7 with minor can also work but frankly, why add a b7 harmonic?

It's not the worst harmonic to add... as you can easily substitute a dominate over a major 7th in some songs and get away with it... and I understand the harmonic divisions make it far more likely than a major 7th.... BUT, all the online data I can find shows virtually no 7th order harmonic at all for real tubes (or, it's so low it isn't even showing up).

 

[1176]

It looks maybe like a case of overzealous marketing. If you look at this month's 'Ask the Doctors' in which the Maximizer is explained, there is no mention of the Precision Maximizer behaving like tubes.

http://www.uaudio.com/webzine/2007/october/index.html

 

[lonestar]

Please take a look at my previous post in this thread (page 1).

I've just edited it and attached two jpeg's to illustrate my findings.


lonestar

 

[XAXAU]

It looks maybe like a case of overzealous marketing. If you look at this month's 'Ask the Doctors' in which the Maximizer is explained, there is no mention of the Precision Maximizer behaving like tubes.

exactly. so if it doesnt sound like tube, then what does it sound (or behave) like?