Vintage mojo... What is it exactly?

[abbey road d enfer]

There is a video somewhere on YouTube where Neve talks about Harmonic distortion and the BS that it is to track THD on anything. He references an old book from the 30s or 40s about building your own radios, where the author lists something like the first 15 order harmonics with maximum acceptable distortions. He also makes a point about even-order harmonic distortion being much easier on the ear than odd order ones. The jist is that a fraction of the amount of say 7th order harmonic distortion is way more audible and gross than a much larger amount of 6th or 8th order.

This is a well-known subject. 2nd, 4th, 8th, 16th and so on harmonics are known to be euphonic.
Third on its own is not bad, as well as 6th and 12th, but as much as the ratio becomes uneven, the ugliest it sounds, 5th, 7th and 11th being most noticeable.
But THD is just a tiny part of the picture. The same non-linearity that would produce mild 2nd and 3rd harmonics results in ugly intermodulation distortion.
For example a very smooth Cmajor chord (C E G or 523Hz, 659 Hz and 783 Hz)) would generate non-harmonic products at 124 Hz (about B) and 136 Hz (about C#). Anybody with even vague musical knowledge knows it's not good. Try to play B, C and C# on the piano (3 adjacent keys).

 

[JohnRoberts]

Not to mention the IMD mud.....

It is perhaps ironic that the design engineers who crafted these "classic gold" old designs were trying as hard as they could to avoid "any" distortion. They would probably smack a junior engineer asking them how to add more distortion.

JR

 

[Bo Deadly]

If distortion alone was responsible for "mojo", that would have been isolated and characterized long ago using spectral analysis and null testing and the usual methods. An op amp, discrete or otherwise, does not contribute "mojo". A tropical fish capacitor does not contribute "mojo". But I will not rule out that a more complex circuit altogether might have some kind of transient effect that could be perceived as pleasing but be overlooked by the usual fft and null testing methods because the stimulus is constant and so the circuit is only being measured in it's steady state.

Humans are extremely sensitive to changes in frequency and find changes in frequency more pleasant than a static frequency. Listening to a pure tone for even a few seconds can be downright annoying. So there could be something that is changing the frequency of fundamentals, harmonics or both. For example, if the attack of note resulted in sharp clipping initially and then, over a few milliseconds, transitioned to a much rounder clipping of each crest in the waveform, that could be perceived as a change in frequency. This could occur if, for example, the input of an amplification stage rectified the signal but after doing so the impedance changed. In a tube the forward voltage for grid current changes from high to low as gas is ionized and the impedance of the grid drops. This might look like a leading spike that drops away over a few cycles. A similar effect might occur in a solid state circuit if the coupling cap charges up and changes the bias of the transistor.

If slight changes in frequency were responsible for contributing "mojo" to a circuit, special time domain analysis would be required to detect and characterize it and so it would be completely overlooked by conventional analysis. My first thought would be to record tone bursts or real notes from an instrument or voice and then do a deep dive into short-time Fourier analysis with Octave looking for subtle changes in pitch and maybe even in the harmonics relative to the fundamental.

 

[JohnRoberts]

If distortion alone was responsible for "mojo", that would have been isolated and characterized long ago using spectral analysis and null testing and the usual methods. An op amp, discrete or otherwise, does not contribute "mojo". A tropical fish capacitor does not contribute "mojo". But I will not rule out that a more complex circuit altogether might have some kind of transient effect that could be perceived as pleasing but be overlooked by the usual fft and null testing methods because the stimulus is constant and so the circuit is only being measured in it's steady state.

Humans are extremely sensitive to changes in frequency and find changes in frequency more pleasant than a static frequency. Listening to a pure tone for even a few seconds can be downright annoying. So there could be something that is changing the frequency of fundamentals, harmonics or both. For example, if the attack of note resulted in sharp clipping initially and then, over a few milliseconds, transitioned to a much rounder clipping of each crest in the waveform, that could be perceived as a change in frequency. This could occur if, for example, the input of an amplification stage rectified the signal but after doing so the impedance changed. In a tube the forward voltage for grid current changes from high to low as gas is ionized and the impedance of the grid drops. This might look like a leading spike that drops away over a few cycles. A similar effect might occur in a solid state circuit if the coupling cap charges up and changes the bias of the transistor.

If slight changes in frequency were responsible for contributing "mojo" to a circuit, special time domain analysis would be required to detect and characterize it and so it would be completely overlooked by conventional analysis. My first thought would be to record tone bursts or real notes from an instrument or voice and then do a deep dive into short-time Fourier analysis with Octave looking for subtle changes in pitch and maybe even in the harmonics relative to the fundamental.

I've probably shared this story before, I do that a lot. Back in the 70s/80s I developed my own tone burst device that was basically a glorified smart mute circuit. It was smart enough to only gate on/off synchronous with zero crossings to avoid adding clicky HF content from the on/off step function. In addition I added a feature to insure that it always presented an even number of positive and negative half cycles to avoid injecting DC content (undesirable in AC coupled stages). A fully on/off gate is too dynamic for designing signal processors with so I made that adjustable with a dry bypass feed so the gate was not fully off when gated off but just attenuated X dB. The final icing on the cake was that this exotic burst generator could accept "any" audio signal input, not just tones. For designing dynamics processors I could punch up the crest factor of normal (cough) music to better stress attack/release circuitry in side chains.

I was not searching for "mojo" whatever that is, but trying to parse out artifacts from handling sudden transients, that were too dynamic for some paths.

Nowadays thanks to digital systems with more bits than I can shake a stick at, there is not much need for dynamics magic, but if/when used they should not burp or fart out of tune.

JR

 

[fragletrollet]

Without trying to express myself technically, I also believe that a part of what makes hardware, and this doesn't go for only "vintage mojo" in expensive units, but analog hardware in itself as opposed to plugins, have tiny variations because of the ever fluctuating environment the hardware unit is operating in. All these factors, like temperature, line stability, component tolerances and lots more I fail to mention influence the processor as to ever so slightly modify the way it behaves, continuously. Sound in nature is usually complex, and we are used to that. So when using plugins that have a set, or coded we should say, response, to the material, we appear it ever so slightly more... not realistic? I for one get much quicker bored listening to a drum sample looped opposed to a drum-machine synthesizing that same bass-drum.

But placebo is both scary deceptive and magically inspiring (for making good music, not science. Or...?)

 

[Bo Deadly]

The final icing on the cake was that this exotic burst generator could accept "any" audio signal input, not just tones. For designing dynamics processors I could punch up the crest factor of normal (cough) music to better stress attack/release circuitry in side chains.

I was not searching for "mojo" whatever that is, but trying to parse out artifacts from handling sudden transients, that were too dynamic for some paths.

Nowadays thanks to digital systems with more bits than I can shake a stick at, there is not much need for dynamics magic

A smart tone burst generator like that might be useful as part of a hardware based transient analyzer. You don't need triple zero THD to study dynamics or transient frequency changes. But in practice it is super easy to use something like Wavepad to digitally create a tone burst that starts and ends at zero for net-zero energy. Then just copy and paste a few times and now you have an absolutely perfect stimulus. Running that through the DUT and mixing with inverted dry might be the basis for some kind of transient analyzer.

 

[Tubetec]

You can piss around trying to quantify tone in scientific terms till the cows come home ,its way less complex in that it either hits or misses, its what happens in the overloaded zone ,which lets face it most solid state/op amp/transformerless men dont dare figure into the equation.

 

[Tubetec]

musical performance has peaks of an undeterminable level .

 

[Tubetec]

thats where correct proportioning of generated harmonics comes into play .

 

[shabtek]

is it possible for mojo to fall in the woods if an observer is unaware of the concept of mojo?

 

[ruffrecords]

This is a well-known subject. 2nd, 4th, 8th, 16th and so on harmonics are known to be euphonic.

Because they are all exact octaves so they are bound to be 'in tune' harmonically

Third on its own is not bad,

Because it is a musical fifth - instant power chord and typical main harmonic of push pull tube output stages

as well as 6th and 12th, but as much as the ratio becomes uneven, the ugliest it sounds, 5th, 7th and 11th being most noticeable.
But THD is just a tiny part of the picture. The same non-linearity that would produce mild 2nd and 3rd harmonics results in ugly intermodulation distortion.

Good point Abbey. This is so often overlooked and rarely if ever quoted these days.

Cheers

Ian

 

[FIX]

This too is a mature topic in the effects discipline. There are many examples of successful effects that effectively add distortion to a clean path. These range from the crude guitar "fuzz" effect created with diode clippers, to more subtle vocal "exciters" selectively reintroducing overtone harmonics. I recall decades ago when I was still writing my "Audio Mythology" column for Record Engineer/producer magazine, I delved into the subject of adding distortion and may have jokingly equated oral exciters with fuzz boxes. I really didn't like the overuse of early exciters and would often try to tune in the FM station better to clean it up. This fuzz box characterization struck a sensitive nerve with the guy who was exploiting that effect commercially, triggering a nasty letter to the editor of RE/P. I guess he didn't know I would get last licks. FWIW I pulled a copy of his patent application and he admits to discovering the phenomenon by accidentally assembling a Heathkit preamp incorrectly. The euphonious early saturation distortion generated by his mistake became the inspiration for his patent and product.

[TMI] Human vocals are rich in harmonic overtones due to the relatively fixed length vocal cords expressing overtones at harmonic multiples of the fundamental. This is true for most stringed musical instruments, and wind instruments. Only drums are the odd ball instrument with non harmonic overtones. [/tmi]

JR

I was an Aphex dealer many moons ago (70s) and we then had the 602b. Which had 2 cards with potted shells and a trim pot. The circuit was a 2 pole high pass filter at 2k, that fed a pot going into an inverter. The wiper had a diode to ground. With +4 in, you would adjust for 22% distortion. That was then blended back with the signal. They later inverted the effect. 2nd harmonic distortion.

they then had the theory that a vca could do the same (a modulator) because 1k in and 1k on the control port would yield 2k. So they replaced the clipper with a vca. The fault was that all the frequencies mixed with all the frequencies and created IM distortion and was harsh.
The early unit sounded quite good on many things, it added clarity without increasing the volume.

 

[FIX]

Most mojo is in the 2nd harmonic area, through it is usually limited to 200 and down with transformers. 3rd is also nice as long as it’s only low end. The API stuff is primarily 2nd and the Neve is more 3rd-ish. Tonelux was also 3rd.

 

[SWAN808]

I thought vintage mojo was simply low level saturation and eq changes caused by running signals though old components, that people found pleasing.
Simple to describe but not create.

 

[abbey road d enfer]

I thought vintage mojo was simply low level saturation

Saturation, low-level or not, results in the creation of harmonics. The distribution of these harmonics is wha makes them pleasing or not.

and eq changes caused by running signals though old components

Old components do not have the magic power of creating pleasing sounds. They may have altered characteristics that may impact distortion, but most often they would be within specs, so would result in nominal performance, or out-of-specs and give relatively poorer performance.

 

[ruffrecords]

Most mojo is in the 2nd harmonic area, through it is usually limited to 200 and down with transformers. 3rd is also nice as long as it’s only low end. The API stuff is primarily 2nd and the Neve is more 3rd-ish. Tonelux was also 3rd.

I may be wrong but I thought transformers produce principally odd order harmonics.

Cheers

Ian

 

[gyraf]

IMO, the "magic" is all about the boundary conditions, how some circuit transitions into overload - that may or may not help/manipulate your auditory system evaluation of subjective loudness in an artistically specified way.

This more or less my rationale for creating the Gyraf stuff the way I do. Quite a few people with known-good ears likes this (although I'm painfully aware this doesn't prove anything)

The whole "it's the n'th harmonic that does it"-argument is inherently flawed - as you discover running a simple convolution of any of your favorite sounding gear..

/Jakob E.

 

[JohnRoberts]

For tube guitar amps IMO the magic was all about the unique overload characteristic of tubes vs solid state power devices. Without getting into nuts and bolts, there are multiple patents covering successful and less successful solid state mimics. There seems to be a larger place for this selective distortion in making music than reproducing music for play back, but as always opinions vary.

JR

 

[abbey road d enfer]

I may be wrong but I thought transformers produce principally odd order harmonics.

Not if there's DC in them.

 

[moamps]

For tube guitar amps IMO the magic was all about the unique overload characteristic of tubes ...

and power and output transformer saturations.
Also with some tube amplifiers, feedback occurs from the speaker to the input and output tubes through the air and the mechanical structure of the cabinet. This relationship is quite difficult to replace with solid state amplifiers.