THD

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
THD is the sum of all the harmonic products that a circuit adds to the fundamental, the original frequency. It doesn't differentiate which products it contains, so THD numbers between different circuit types are not directly comparable. Different harmonic products relate to the fundamental differently: 2H is double the frequence, 3H is 1.5 times the frequence, 4H is four times the frequency and so on. These different harmonics are not heard the same way by listeners, so simply stating a device has a certain THD does not necessarily tell you how good or bad it sounds.

It does give a ballpark figure; a high THD is probably going to be worse than a low.

THD is influenced by circuit design.

What is acceptable? There's no answer for that. Depends also a lot on what harmonics the THD happens to contain.
 
Let us look at it from a musical standpoint:)

2H is two times the fundamental and 3H is three times the fundamental.

Let us use low E on a guitar as an example. Low E is 82Hz. 82 times 2 is 164. 164 is an octave above E, and is itself another E note. So it reinforces the fundamental frequency. Now let us look at 3H. Three times 82 is 246. 246 is a musical twelfth and is a B note. 4H is another E note, but 5H is close to G# and is a major third.

I guess I am trying to say that TOTAL Harmonic distortion doesn't tell you enough, to know exactly what dominates the total will give more of a clue to how the DUT will sound like.

Anybody who studies and plays music should be able to tell you what these notes sound like. If you look at a Major chord, it is built from a triad. A triad has three notes, the root or one, the third, and the fifth. Aint life strange.
 
Last edited:
but 5H is close to G# and is a major third.

Close, but not close enough, especially if any of the on frequency major third is present. Most trained musicians will find 5H unpleasant, especially if it is the dominant harmonic. It can be masked by a strong 3H.

THD is DEFINED as the sum of all the harmonics 2H + 3H + 4H + 5H +......

Most if not all old style THD analyzers actually MEASURE THD by simply notching out the fundamental and measuring what's left. This residual is simply called THD, but also contains a hum and noise component, and anything else that is not the fundamental.

With the advances in computing power in the last 20 years, it is now possible to perform Fourier Analysis in real time on an audio signal, and identify the level, frequency, and phase of each individual component in the audio signal. You can do this in a PC with a good sound card.

It is also possible to do the reverse. We can generate a single tone, and each harmonic in a PC, and adjust the level of each harmonic individually such that we can listen to the result. This way you can decide for yourself whether you like or dislike THD, and which flavors sound ugly to you.

There is software used by musicians to write, record, perform and produce music. This software is called a DAW, or Digital Audio Workstation. These programs vary in cost from free, to over $10K (Avid Pro Tools). Most have built in (or plug in) software synthesizers that allow playing with harmonics.

There is a freeware or low cost (depending on platform) DAW that has a Hammond B3 Organ simulator. It is called Caustic, and is free for the PC, and low cost for Android or iPad. The Hammond B3 was a mechanical harmonic synthesizer based organ. Each "drawbar" controls a particular harmonic, and allows manipulation of it's level. This was done with magnetic tone wheels spinning on a common shaft. The drawbars, like most organ controls are labeled in feet, corresponding to pipe length in an original pipe organ.

The 8 foot pipe, or drawbar is the fundamental, and all others are numerically related. The 4 foot drawbar corresponds to a 4 foot pipe length. The pipe is half the length of the fundamental pipe, and therefore operates at twice the frequency, corresponding to the second harmonic. Likewise the 2 2/3 foot drawbar corresponds to the 3rd harmonic. The 2' is the 4th harmonic. The 1 3/5' is the 5th, the 1 1/3' is the 6th, and the 1' is the eighth. The 16 foot drawbar is half the frequency of the fundamental, and is a sub harmonic.

You can download the PC version and play with the drawbars to see and hear the effect of some of the harmonics. Note that this simulator does not have the 7th or 9th harmonics. They are not considered musically useful for most music.

There is a synthesizer included with Ableton Live (I think) that allows manipulation of all harmonics up to the 16th. Both Ableton Live and FL Studio have the ability for the user to "build your own synth" allowing all sorts of musical combinations.
 
Back to the OP question, since it has been posted in the Tubes section: THD is caused by nonlinearity of components, mostly active components and the output transformer. As said before, different active components (tubes) have different distortion spectra. Triodes have 2nd harmonics dominating that is more pleasant than pentodes having strong 3rd harmonics and higher. Global negative feedback reduces THD at the price of the resulting distortion spectrum spreading to higher order harmonics so that the end result might be more unpleasant for the critical ear.
 
Close, but not close enough


I was waiting to see if somebody would pick up on that.....


Yes 5H is unpleasant indeed. It is between a minor third and a major third, but closer to the major third. Either way to stay neutral in musical terms one usually opts to omit the third, hence the modern day "power chord" which is just the one and the five (from our example the E and the B notes). 5H is a dissonant sound.....but in my opinion even if 5H was a perfect major third or minor third you wouldn't want it in the distortion spectrum........it would have an extremely biased effect on the fundamental. Push Pull amps might sound more powerful because of the reduction in 2H and the added 3H, it's like taking the fundamental frequencies going in and making power chords with them:eguitar:
 
This guy I've been building amps for asked me why 1) high order 2) odd harmonics sound bad.

My explanation centered on the fact that most natural sounds contain low order, and mostly even harmonics.

High order harmonics are very rare in nature, so the human ear must be very sensitive to them - "something very strange is happening, be alert now!"

Think of a sharp object touching things, maybe scraping something. Most sharp things in nature (especially before humans learned to make them themselves) were very dangerous and it was probably good to have them "steal our attention".

All just speculation. But what do you think, why evolution made our ears so sensitive to these things?
 
This guy I've been building amps for asked me why 1) high order 2) odd harmonics sound bad.

My explanation centered on the fact that most natural sounds contain low order, and mostly even harmonics.

High order harmonics are very rare in nature, so the human ear must be very sensitive to them - "something very strange is happening, be alert now!"

Think of a sharp object touching things, maybe scraping something. Most sharp things in nature (especially before humans learned to make them themselves) were very dangerous and it was probably good to have them "steal our attention".

All just speculation. But what do you think, why evolution made our ears so sensitive to these things?
I agree.
even harmonics are natural, odd ones are odd!
so we are sensitive to them.
 
Hello

Questions beginners

What is defined as total harmonic distortion?

Root mean squared of the harmonic components:

THD= sqrt(h22 + h32 + h42 + h52 + ... + hn2)

from what influenced design circuit? quality of Materials? something else?

It's the natural consequence of their being no such thing as a perfectly linear active device. Some passive components can contribute some incidental harmonic distortion by virtue of nonlinear materials, including some resistance materials, capacitor dielectrics, and, of course, the nonlinear H v. B characteristic of ferromagnetics.

what is the acceptable harmonic distortion for hi end tube amplifier?

No way to say, other than lower is preferable, but the whole story is more complex than that. If you look up the specs for the 807, you'll see a suggested Q-point and load that gives THD= 1.8%. For the 6V6, THD= 3.0% so do 807s sound better? Open loop, they don't. The 807 likes to make a lot of higher order harmonics, and you can see this with the Twin-T test. After nulling the fundamental, what's left looks like a near perfect sawtooth at three times the frequency.

The residual from a 6V6 (or the 6BQ6 -- calculated THD for this type was 3.0% and 2.8% measured) looks like a slightly distorted sine at three times the frequency. Open loop, 807s sound very irritating, like fingernails on a black board irritating. 6BQ6s just sound overly "aggressive" or "edgy" due to the mainly h3 nature of the harmonic distortion. This is what gNFB is for: to correct for those errors. Proper application of NFB can get excellent sonic performance from 807s.

There are other sources of distortion that are excluded, but possibly more harmful to sonic performance, such as IMD. This is especially bad since it makes frequencies that aren't musically correlated to the original frequencies. That's one reason why you'd like to design for low THD, as both are caused by the same thing: nonlinearity.
 
That's one reason why you'd like to design for low THD, as both are caused by the same thing: nonlinearity.

With all being right what you say, we have to admit that low THD does not necessarily equate with pleasant sound.

One well known case is crossover distortion, which produces rather low THD values, but sounds horrible even in the slightest amount.
 
The 7th harmonic is "even worse" than the 5th musically

Yes it falls between a flat seven and six musically, once again it just sounds dissonant against the fundamental. These distortions that don't harmonize nicely with the fundamental get worse by the polyphonic nature of music.........intermodulation distortion arises and is the real offender of a good rich harmonic structure.
 
Root mean squared of the harmonic components:
THD= sqrt(h22 + h32 + h42 + h52 + ... + hn2)

actually
THD= [sqrt(h22 + h32 + h42 + h52 + ... + hn2)]/h1*100 if expressed in percentage of the fundamental
or
THD= [sqrt(h22 + h32 + h42 + h52 + ... + hn2)]/[sqrt(h12 + h22 + h32 + h42 + h52 + ... + hn2)]*100 if expressed as percentage of the total harmonic content

using the 1st formula:
sawtooth 80%
square 50%
triangle 12%
now what would be the THD of a major triad chord ... ?
 
With all being right what you say, we have to admit that low THD does not necessarily equate with pleasant sound.

One well known case is crossover distortion, which produces rather low THD values, but sounds horrible even in the slightest amount.
That might be true for full power reference, but the distortion percentage increases when you go to lower level. Most pleasant is the distortion that decreases with decreasing signal level (apart from the harmonic content).
 
Another way which can help assess an amplifier is to investigate its voltage transfer function over the entire frequency range. This chart should be a 3D chart with function Vo = f(Vi, Fr). Ideally, it should have a regular wedge shape with constant slope (dVo/dVi).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.