It probably isn't useless but it is not a defining metric by any means. It can be an indicator that both even AND odd order artifacts are low. I guess it is better than nothing but we should expect one metric to be sufficient to describe the performance of a part.
And it depends on the application. If we had 1% THD in a transducer at full output we would be jumping for joy. Some perspective in the overall playback chain is called for.
I'll disagree with you... single number THD is useless... once the underlying spectra has been squashed out there is NO context to give the number any meaning at all.
dave
Ok.... sometimes it is useful when your looking at gross levels and looking for a solution. For example, if you design a speaker and you need a given SPL at a given frequency, you may take a quick THD measurement and can see when it is trending extremely high. You know at that point, you are reaching the limits of the device and you either design filters to take care of it, or make plans for more devices.
It is useful in that context.
Given that the quick THD measurementhas to do the spectra 1st, why bother looking at the total without context when there is much richer information right at hand?
If you can see the spectrum you also get an idea of what limit it is reaching which gives a much better idea of the direction to go...
... as well the topic is amplifiers not speakers.
dave
If you can see the spectrum you also get an idea of what limit it is reaching which gives a much better idea of the direction to go...
... as well the topic is amplifiers not speakers.
dave
Well.. in the context of a transducer, when it spikes at the limit of the device all forms of distortion spike. It doesn't really give you any more meaningful information to look at the spectrum unless your trying to improve the design a of transducer and that is a different topic.
I'm talking HIGH distortion though not what most people argue about when they are comparing amps. If people are looking at 0.05% vs. 0.005% then the number is meaningless without looking at the spectrum. When you compare 25% THD you can pretty much say that it is undesirable no matter the content.
I think we are pretty much agree for the most part though.
THD simply gives a number for a certain error the amp produces; by definition a high-fidelity amp must have also low THD. What low means, is up to you.
Looking at distortion spectra certainly gives more information, but so far I haven't seen anybody succeeding in predicting accurately an amp's sound by looking at it.
That reminds me of the urban myth that an amp sounds like a tube if it's distortion components are of 2nd order. Any single ended amp will do that. Even Ovidiu's high NFB amp does that.
As funny sidenote, more people prefer in listening tests amps with dominating 3rd harmonic. Push-pull amps fall into this category as the 2nd order simply cancels.
It's so easy to shape THD the way you want - does that guarantee a great sounding amp? C'mon
Have fun, Hannes
Hannes,
I`m now noticing an unfortunate deep decay in your contribution quality. I cannot naturally accept any of the above phrases.
I`m now noticing an unfortunate deep decay in your contribution quality. I cannot naturally accept any of the above phrases.
By what definition ? The venerable industry standard DIN 45500 ?
Which information does predict an amp's sound accurately?
Not necessarily.
No way.
That´s a moderately funny side note. Even order harmonics being brushed aside, dominating odd, high order would be more far from the urban myth, Not a very delightful category.
OK, but it`s of little value. Sadly, it's not at all easy to shape the harmonic spectrum in any desirable direction, I`d say impossible. Again, it´s fundamentally about spectrum not level.
Lumba,
When Hannes says:
"THD simply gives a number for a certain error the amp produces; by definition a high-fidelity amp must have also low THD. What low means, is up to you"
- he means that a high-fidelity amplifier should not add anything or subtract anything from the signal it is processing. That implies that THD must be low, otherwise it would add something. In that sense, by definition, a high-fidelity amp must have low THD.
jd
This is strictly true in an ideal world. But in the real world all amplifiers lose information*. The problem with single number THD is that in throwing away the context it does not produce a metric with usable meaning. Is a 0.01% amplifier better than a 0.1% amplifier? Or a 0.001% amplifier better than a 1% amplifier? There is no way to tell -- too much has been thrown away, and even a comprehensive set of uncollapsed data only partially (a small part) characterises an amplifier. That single number by itself is useless. Less than useless, actually misleading.
*(this term is more inclusive since many think distortions we typically measure when someone says distortions, and that is only the tip of the iceberg. The very act of thinking in terms of distortions in this manner limits the designer, and certainly limits the purchaser)
Yes, I see your point. But high-fidelity is what it is, reproducing with high fidelity. That means 'low' THD. An amp with 'high' THD is not 'high-fidelity', by definition.
It isn't enough, granted, you also need low hum/noise, for instance, and reasonable damping and low IMD and and.... But THD must be 'low'.
jd
I agree that information loss should be low. What is low? With single number THD there is no answer since (just to fabricate an example), 1% 2nd order is likely "lower" than 0.01% 7th. If these were the only harmonics (very unlikely situation), then you'd have an amp with 1% THD that has lower distortion than the 0.01% THD. And the kinds of things that often cause the presence of higher order harmonics are specifically done to reduce lower orders, so it is really easy to generate real examples where amps with lower numeric THD have higher distortion.
One cannot remove 2 ears + a brain from the analysis. THD (single number) ignores this.
Also, what is reaonable damping? It is speaker (+ speaker wire dependent). Another spec that out of context is useless. We have a variable transconductance amplifier here, and each speaker likes a different setting.
dave
Exactly.
And for the crossover distortion which is no longer there - Sayonara!
And for the "mysterious so far" harmonic distortions(read poisons) that amps of this class create - a perfect amp won't exist as soon as it sees that it has to please the death-metal-lovers as well as smooth-jazz-lovers, as humans come in so many variety of tastes.
The good thing to think is that there is one amp for everyone(this sentence sound so similar to something I heard elsewhere).
Each chooses his/her own amp, the one that matches his/her style(of enjoying the music he/she likes).
And I think there are many to choose from. Just that one was fooled by his/her amp and/or has an amp which he/she thinks is "exceptional", don't mean he/she should be in despair seeing someone else's amp running "hot" or be too proud that the "someone else's amp" is just a cold thing.
I used to listen to a DIY class-ab amp, my pc's altec lansing 2.1 speakers, my FM radio's headphones and my pc's headphones, amongst which I liked the pc headphones most even though the others were MUCH louder.
The last thing I made is a SE-Class-A, which initially rendered the headphone and the other amps taking a nice rest of quite some couples of months since it ran first.
I have chosen to overlook the distortions, etc. and objection, etc. and sure am enjoying the music at its best. To say that I haven't heard an amp better than the SE. Ya I reject anyone offering me 1000 bucks for considering the "spec". But I don't push my pals to have the same type of amp I own even if it gave me so much. One of my friends have a pair of USB powered Perx mini speakers which, according to him, were just made for his ears. This is just cool, as long as he like to listen to it. Hope he never read the specs also, coz the sheet does come in between the speaker and the listener and add a lot of "numeric" distortions to the music!
Wavebourn,
my mother-in-law says the class A - open-loop mode (for minimal interaction between stages) is best, the rest is artificial trickery. Furthermore, Shaan´s amp does almost certainly not qualify as "high fidelity" according to DIN 45 500, but... what the heck. An approval would surely entail a massive loss of subjective quality, so rather the measurements standards should adapt to the ear`s way of perceiving sound...Maybe she`s right, I don´t know...
my mother-in-law says the class A - open-loop mode (for minimal interaction between stages) is best, the rest is artificial trickery. Furthermore, Shaan´s amp does almost certainly not qualify as "high fidelity" according to DIN 45 500, but... what the heck. An approval would surely entail a massive loss of subjective quality, so rather the measurements standards should adapt to the ear`s way of perceiving sound...Maybe she`s right, I don´t know...
A first, instead of quality of sound distortions were legalized, accepted and normalized, so instead of competition for higher quality "Q" manufacturers started to compete for lower "D". Then, a sneaky "H" was added, in order to measure Harmonics only. Then, one more sneaky "T" was added, to hide from customers which harmonics they are getting. Finally, letter "N" was added, the sneakest one: go away, nothing to measure, THD is under the N floor!
The result is THD+N, a Magical Formula that sells a garbage.
Does your mother-in-law have a Magical Formula? If she does not, her criticism is not constructive.
The result is THD+N, a Magical Formula that sells a garbage.
Does your mother-in-law have a Magical Formula? If she does not, her criticism is not constructive.
She is Right.
You are right.
Are you supporting me or are you not? I'm confused.
This way the entire thing would become as boring as music with 0.1% 3rd harmonic, 0.001% 5th harmonic, 0.0001% 7th harmonic... And to tell the truth, objective quality measurement often saves life, as the user ultimately believes he's got the best in the world amp. Don't you think that's right? I Do.
You will never need to know, that knowledge will only bring despair. I don't want to know too. Good news, your mother in law will neither have a chance to suggest me listen to her amp and decide, and nor do I think she would even if given a chance.
Anyway, I'm a fan of SE-Triode. But don't own one, have never listened to one and maybe I will dislike the sound of those amps if ever listen to one. That definitely does not mean the one who likes these amps is stupid because he/she ignores the spec-sheet and goes with his/her ear. This is cool as I said before.
The harmonic spectra could tell a lot about the sound of an amplifier with 10% THD, but next to nothing about an amplifier with 0.001% THD as the harmonic components are below any demonstrated threshold of audibility.
Please provide just one good, peer-reviewed scientific paper demonstrating the audibility of, say the (favourite around here) 7th harmonic at 0.001% (-100dB)
Incidentally (just for fun and to put some kind of perspective on the absolute levels):
Jet at 100 m:
110 – 140 dB
Jack hammer at 1 m:
100 dB
Calm breathing
10 dB
Auditory threshold at 1 kHz
0 dB
I've been near a few jet engines and have used a jack hammer a number of times. In such circumstances I have difficulty hearing someone near me breathing.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.