I’m not disappointed too; you threw in a bait, with no technical description or any other useful information, except the usual FFT “of course” bashing. It fits the profile, too.
Nothing compared to the hearing illusions:
< Can You Trust Your Ears? (Audio Illusions) - YouTube >
More psychoacoustics:
part 1 < 42 Audio Illusions & Phenomena! - Part 1/5 of Psychoacoustics - YouTube >
part 2 < 42 Audio Illusions & Phenomena! - Part 2/5 of Psychoacoustics - YouTube >
part 3 < 42 Audio Illusions & Phenomena! - Part 3/5 of Psychoacoustics - YouTube >
part 4 < https://www.youtube.com/watch?v=WMHyYCk7OqE >
part 5 < https://www.youtube.com/watch?v=YQNsCg4z6L8 >
... and then there are
s who pretend thy can hear the qualityof a solder joint.
ROTFL, Gerhard
Although digital to analog conversion is exclusively time based, just overlap the two pictures I posted.
But maybe you are too busy designing a decent oscillator.
I look forward.
I am ROTFLing since the beginning of this thread, only that the usual peddlers have as much sense of humour as technical knowledge.
Last edited:
Because with two signals I can use two independent soundcard channels to output them, add them with resistors and run them through the DUT, then filter one of them out with known linear components and only keep the second, so the interaction of the two signals, and the distortion it causes only happens within the DUT and not within the measurement setup. With real music, too complicated.
> Do you think it's not possible?
It's like using a notch filter. The fundamental provides no information but it obscures the harmonics which contain the information. So a notch filter is used. Without filtering, the soundcard and the ADC would have to process the whole signal, thus adding their own distortion to the DUT.
I've noted some items on my todolist.
1) the same test as they do for DSL drivers and other RF stuff that has to pass many carriers: filter the music to remove a band of frequencies, run it through the DUT, acquire, check what appeared in the notched out band. But any practical implementation of these filters will have to be digital, which means I can't get rid of the soundcard distortion.
2) lowpass real music and use that as the operating point signal, the test signal being a 20-40 kHz sine. That way the music can be filtered out much easier.
Were the pics from the output of a DAC with a good clock and a terrible one?
But we can count on the "genius at work" board.
Joking aside, the DAC and the ADC have to be better than the DUT, unless we could use a sort of cross correlation.
To do this input and output have to be locked.
Just put the pointer over the pic and magically you read the title.
Basic IT (Mouse over event).
Last edited:
Hi Andrea,
A sine wave has a constantly changing rate of change and is an excellent test for any electronic audio device. That's why we use it. Additionally, it contains one frequency so its easy to see what doesn't belong and even easier to measure it with great accuracy. A two tone test is used for intermodulation distortion tests. Again, very easy to see what signals do not belong and measure with great accuracy.
Music signals are broken down into a series of sine waves (perfect ones). A decent amplifier doesn't know or care what signal is coming, or has just passed, so no magical qualities of memory or foresight. Using music to test an amplifier is valid for a listening test, but isn't of any greater significance to any method of measuring performance.
What you need is a basic understanding of what constitutes a signal so you can sort out the nonsense other uneducated folks are spouting. You can support any viewpoint you want by searching on the internet. Try proving them in a lab, can't be done because there isn't any truth to them. Heck, you can even break it down in math and prove it on paper. No test equipment needed, just knowledge and a brain.
If you want to understand audio, you've got to learn the basics of electronics and build on it. It takes years, some formal instruction and some proof in labs and test benches. If what you say is true, then we wouldn't have guided missiles, manned space craft or satellites. Audio is a very simple implementation of managing signals. Lot's of detail stuff, but there is more going on in industrial electronics and communications. What we see in test instruments and calibration is way beyond that and what we see in space exploration and military electronics is beyond that. The electronics world is really cool and vast. You haven't even got a grip on the basics.
Once you have some understanding, you will see how much you don't know. I am basically an idiot compared to some engineers.
-Chris
No such thing as a time invariant tone. A time invariant signal is some level of perfect DC without any noise. A great voltage reference in other words.
A sine wave has a constantly changing rate of change and is an excellent test for any electronic audio device. That's why we use it. Additionally, it contains one frequency so its easy to see what doesn't belong and even easier to measure it with great accuracy. A two tone test is used for intermodulation distortion tests. Again, very easy to see what signals do not belong and measure with great accuracy.
Music signals are broken down into a series of sine waves (perfect ones). A decent amplifier doesn't know or care what signal is coming, or has just passed, so no magical qualities of memory or foresight. Using music to test an amplifier is valid for a listening test, but isn't of any greater significance to any method of measuring performance.
What you need is a basic understanding of what constitutes a signal so you can sort out the nonsense other uneducated folks are spouting. You can support any viewpoint you want by searching on the internet. Try proving them in a lab, can't be done because there isn't any truth to them. Heck, you can even break it down in math and prove it on paper. No test equipment needed, just knowledge and a brain.
If you want to understand audio, you've got to learn the basics of electronics and build on it. It takes years, some formal instruction and some proof in labs and test benches. If what you say is true, then we wouldn't have guided missiles, manned space craft or satellites. Audio is a very simple implementation of managing signals. Lot's of detail stuff, but there is more going on in industrial electronics and communications. What we see in test instruments and calibration is way beyond that and what we see in space exploration and military electronics is beyond that. The electronics world is really cool and vast. You haven't even got a grip on the basics.
Once you have some understanding, you will see how much you don't know. I am basically an idiot compared to some engineers.
-Chris
I rather suspect what Andrea meant by 'time-invariant' was 'stationary'. A stationary signal is defined as one whose spectral content does not change with time. A sinewave is the canonical example of a stationary signal.
This requires a very deep, detailed and rather boring discussion. Long story short, rigorously, we don’t talk about stationary or non stationary signals. That’s because a signal is nothing more than a recording of something that already happened, or, if you prefer, it’s an observation, or instance, of an output. The output of a process, and if that process generating signals changes its properties in time, then the output is called “non stationary”. Processes that generate stationary outputs are called ergodic, meaning that their properties can be calculated by a single, sufficiently long (but finite), random sample, of an output.
This rather abstract definition leads to a not necessary intuitive classification of signals: white or pink noise are stationary, because any signal value is equally probable to happen, given any other signal value, at any two time instances. An ECG signal is non stationary, since the process that generates the signal (the heart beat) changes over time. It is also interesting that the scale of observation is important. Look at a signal for a short period of time and it may look non stationary. Look at a signal for a very long period of time, at it may look stationary. Look for infinite time, and there are only stationary signals.
Finally, something that is somehow under wraps above: stationarity or non stationarity are concepts that only apply to stochastic signals, not deterministic signals. When statistical tests are applied to determine stationarity or nonstationarity, the deterministic component must be removed. Therefore, any deterministic signal (like a sine) is neither stationary or non stationary. An interesting case is a modulated signal, AM or FM. While these look like non stationary, it can be shown they don’t exactly follow the definition above. We call such signals cyclostationary, since they don’t fit in either.
Boring, isn’t it?
P.S. To bring this a little closer to the current topic, a sine AM or FM modulated by noise (the effect is mostly known as “jitter”) is exactly a stationary signal. Remove the deterministic part (the sine) and what’s left is the noise, which we already know it’s a stationary signal. So random jitter is stationary, surprising? However, deterministic jitter can be non stationary! And here’s another one weird example, is music a stationary signal, even if it is produced by a non stationary process like an instrument? No, if you look locally, within the track. But then if you play a music track in a loop, it is exactly stationary, since after one play you can predict precisely the evolution in the future.
This rather abstract definition leads to a not necessary intuitive classification of signals: white or pink noise are stationary, because any signal value is equally probable to happen, given any other signal value, at any two time instances. An ECG signal is non stationary, since the process that generates the signal (the heart beat) changes over time. It is also interesting that the scale of observation is important. Look at a signal for a short period of time and it may look non stationary. Look at a signal for a very long period of time, at it may look stationary. Look for infinite time, and there are only stationary signals.
Finally, something that is somehow under wraps above: stationarity or non stationarity are concepts that only apply to stochastic signals, not deterministic signals. When statistical tests are applied to determine stationarity or nonstationarity, the deterministic component must be removed. Therefore, any deterministic signal (like a sine) is neither stationary or non stationary. An interesting case is a modulated signal, AM or FM. While these look like non stationary, it can be shown they don’t exactly follow the definition above. We call such signals cyclostationary, since they don’t fit in either.
Boring, isn’t it?
P.S. To bring this a little closer to the current topic, a sine AM or FM modulated by noise (the effect is mostly known as “jitter”) is exactly a stationary signal. Remove the deterministic part (the sine) and what’s left is the noise, which we already know it’s a stationary signal. So random jitter is stationary, surprising? However, deterministic jitter can be non stationary! And here’s another one weird example, is music a stationary signal, even if it is produced by a non stationary process like an instrument? No, if you look locally, within the track. But then if you play a music track in a loop, it is exactly stationary, since after one play you can predict precisely the evolution in the future.
Last edited:
- Status
- Not open for further replies.