I have tried 100 hz with 200hz harmonic at -20 -40 and -60(10,1,0.1).I can barely hear it at 1%,at what i'd say is a decent listening volume,standing about 4 meters away from the speakers.Tommorow i'll try at higher levels and sitting right next to the speakers but I feel at that point the intristic ear distorsion would be much greater that that
This test is much easier to do at high volumes. It also, in my opinion, shows the weakness of back-to-back testing. Consider the application of this test. We can use it to judge, say, how much to spend on a DAC. Should we buy the one that has 0.1% THD or the one that has 0.0001%? If these tests were to be assessed with a bit of distance between the sweeps, I'm not sure many people could actually determine which was better beyond the 1% being the worst. More practically, what are you buying with the extra zeroes? Unless you're testing back to back, is it noticeable as an improvement?
Thanks for sharing.
Thanks for sharing.
0.01% is impossible to hear on my speakers, even though they have THD below 0.1% at 75 dB SPL, which is the level I tested at. 0.01% equals -80 dB, and thus below my hearing threshold. I could increase the playback volume, but then the speaker’s own distortion will increase as well, and probably mask the 7th harmonic on the test track.
The peak level on the test track is too close to 0 dBFS I think. To avoid clipping, peak limiting, distortion in DACs, it might be reduced to something like -3 dB.
You probably shouldn't buy a DAC based on typical distortion measurements. Some of the problems that affect DAC sound simply don't show up well using such 'figure of merit' metrics.
Yes, the harmonic is easier to hear at high volumes. I wanted people to use normal listening levels.
I had to make the test simple so that people would try it. 🙂
Really only power amplifiers have valid reasons for having THD >0.001%.
Interesting. I found that I needed to move away from the panels so that dipole cancellation reduced the fundamental. Even then, I could not perceive a tone at 0.01%.
The DAC should be able to reproduce 0dBFS, right?
I set the level near 0dBFS because 0.001% is below the noise floor of 16-bit audio.
Ed
Yes, DACs should be able to reproduce 0 dBFS without distortion. Truth is that many DACs cant ...
There is always some distortion no matter what. Same for noise. Anyway, are you referring to intersample overs in regard to reproduction of 0dBFS? The distortion that is worst at a sample rate of 44.1kHz, and at the particular audio frequency of 11.25kHz?
Intersample overs do not happen with sinewave.
Take a look at the link below. Very bad DAC performance at 0 dBFS.
https://www.audiosciencereview.com/...iew-and-measurements-of-nad-t758-v3-avr.8912/
Take a look at the link below. Very bad DAC performance at 0 dBFS.
https://www.audiosciencereview.com/...iew-and-measurements-of-nad-t758-v3-avr.8912/
The sample size is now large enough to be statistically valid: one-quarter of DIY'ers can hear 0.01% THD on a distortion spectrum designed to be easily heard. Amplifier designers need to keep distortion below this level.
Ed
Ed
Ummm, this was 7th harmonic right? Maybe the most audible harmonic? So why should it follow that 2nd harmonic should also need to be less than 0.01% THD?
EDIT: A quote from Dr. Earl Geddes: "Lee and Geddes [2] demonstrated that the perception of nonlinear distortion does not correlate with the commonly used metrics of THD and IMD..." http://www.gedlee.com/Papers/AES06Gedlee_ll.pdf
Engineers do worst-case design.Markw4 said:
If non-linearities above 0.01% are audible, some input signal may occur with the right components to produce audible distortion. For example, I can choose input signals that put IMD components at frequencies where they would be quite audible.
Ed
The guys playing around with 2nd and 3rd harmonics over in the Pass forum are doing something wrong, engineering-wise? Or, maybe I misunderstand?
Okay, but I still have concerns. Understood the distortion listening test and its results are well-intended. However, if it were that simple why isn't there already a well-known and widely accepted standard expressed is similar terms? What do the experts think about THD as a metric?
What about the 25% of forum members who could hear the distortion? What should the standard be for those people? How much less than 0.01%?
What about amplifiers that measure higher distortion with a speaker load instead of a resistive load? What's the worst case engineering standard for that? Are resistive load measurements adequate in practice?
The list goes on and on, but I will stop after the above few concerns.
As an aside: Regarding the term 'threshold of hearing,' hopefully by now everyone knows that the term refers to an estimate of average for a population. It means 50% of the population is estimated to be able hear below the limit. What about a standard good enough for those people? Good enough for only an estimated 50% of people doesn't seem like very good engineering, does it?
What about the 25% of forum members who could hear the distortion? What should the standard be for those people? How much less than 0.01%?
What about amplifiers that measure higher distortion with a speaker load instead of a resistive load? What's the worst case engineering standard for that? Are resistive load measurements adequate in practice?
The list goes on and on, but I will stop after the above few concerns.
As an aside: Regarding the term 'threshold of hearing,' hopefully by now everyone knows that the term refers to an estimate of average for a population. It means 50% of the population is estimated to be able hear below the limit. What about a standard good enough for those people? Good enough for only an estimated 50% of people doesn't seem like very good engineering, does it?
Also, as the OP said, "I made the harmonic easy to detect by ... putting it in the frequency range where the ear is sensitive." And (okay, someone else mentioned this) there's clearly no electrical circuit that (without being specifically designed to do so) would only generate the seventh harmonic of an input signal.
I've ranted a few times about people using THD as the ultimate, only known or even a USEFUL metric for comparing anything to anyhing else. I wish I had a dollar for every time I read "But loudspeakers have over 1% THD, how in the world can we hear 0.1% THD in an amp playing through those speakers?