Online dictionary is available if you don't know the difference.
Playing word game again.
Meaning of the word "proof" is universal whether it's in the judicial case or audio electronics listening test. If one can demonstrate in listening test that he can do what he claimed, then he has proven it to be factual, not sales pitch, BS or the likes.I guess your perception is a sentient being. 🤨
Thanks. Could you also repost the FFT script from https://www.diyaudio.com/community/threads/placement-of-resistors-in-signal-path.384534/post-6983085
BTW there is a small error(?) on your file generation script:
ang=t*2*pi*fcar+2*pi*ymod;
The second 2*pi actually just makes the modulation index 0.05*2*pi
Last edited:
Guessing now?
Better than just slamming in some old audio designer recommended part and claiming on here that it's better surely?
Lars knew I was ribbing him. Also he wasn't claiming any audibility for the 2 files on his blog post. He asked people to listen and decide.
I always get suprised how easily fooled my brain is. But I roll with it. we didn't evolve to listen to 2 speaker stereo after all. It shouldn't work but we can conjure up a vision from the limited information supplied and hormones flow. thank $DEITY we have dopamine receptors.
You made the right choice.
That water meter is way more important and useful for most of human society than a crummy "15 minutes fame" song and similar stuff.
Spoiled Artists think the World revolves around them, or their "Art" ... real life relies on WAY more solid work.
That.
IF the circuit will behave audibly different when fed 1.596 VDC OR 1.595 VDC OR 1.597 VDC THEN it is important to measure with such precision.
I NOT and I am quite certain of that, extra precision might be "interesting" but definitely useless or irrelevant.
Call it "of no practical value" if offended by the other labels.
If so , why argue/worry about that?
Now that Lars provided the Matlab script for generating the files I took a closer look at the issue. BTW I have not used Matlab in more than 30 years
As some members already pointed out the resulting "AM" file is not simply amplitude modulation between 1.38kHz and 43Hz as after inverse FFT of the phase modulated file there are some harmonics. This can be clearly seen if the modulation index of the phase modulation is increased (3x in this case):
(Red is PM and blue is AM).
So the blog claim "One is a pure amplitude modulated signal, the other is its purely phase modulated counterpart. The carrier frequency is 1.38kHz, the modulation frequency is 43Hz." is simply misleading. As a minor nag the blog nonchalantly calls the phase modulated file "Pure FM modulation".
The more interesting bit is the claim that only phase differs. The Matlab script shows that what was actually done is inverse FFT on the magnitude of the phase modulated signal. This is the same as rotating the phase 90 degrees in the frequency domain.
This is what this phase rotation of 90 degrees in the frequency domain does to a sine wave:
Yes, only the phase has changed.
This is what happens to square wave:
B-b-b-but I only changed the phase
My signal processing skills have long faded due to lack of practice so I leave further analysis to others.
Markw4 was eager to point out that operation in time domain has to be equivalent to adjusting some of the frequency domain properties. Perhaps he can quickly enlighten us with what those adjustments to time domain properties are in this case.
As some members already pointed out the resulting "AM" file is not simply amplitude modulation between 1.38kHz and 43Hz as after inverse FFT of the phase modulated file there are some harmonics. This can be clearly seen if the modulation index of the phase modulation is increased (3x in this case):
(Red is PM and blue is AM).
So the blog claim "One is a pure amplitude modulated signal, the other is its purely phase modulated counterpart. The carrier frequency is 1.38kHz, the modulation frequency is 43Hz." is simply misleading. As a minor nag the blog nonchalantly calls the phase modulated file "Pure FM modulation".
The more interesting bit is the claim that only phase differs. The Matlab script shows that what was actually done is inverse FFT on the magnitude of the phase modulated signal. This is the same as rotating the phase 90 degrees in the frequency domain.
This is what this phase rotation of 90 degrees in the frequency domain does to a sine wave:
Yes, only the phase has changed.
This is what happens to square wave:
B-b-b-but I only changed the phase
My signal processing skills have long faded due to lack of practice so I leave further analysis to others.
Markw4 was eager to point out that operation in time domain has to be equivalent to adjusting some of the frequency domain properties. Perhaps he can quickly enlighten us with what those adjustments to time domain properties are in this case.
well done, looks like you are catching up againNow that Lars provided the Matlab script for generating the files I took a closer look at the issue. BTW I have not used Matlab in more than 30 years
As some members already pointed out the resulting "AM" file is not simply amplitude modulation between 1.38kHz and 43Hz as after inverse FFT of the phase modulated file there are some harmonics. This can be clearly seen if the modulation index of the phase modulation is increased (3x in this case):
View attachment 1041472
(Red is PM and blue is AM).
So the blog claim "One is a pure amplitude modulated signal, the other is its purely phase modulated counterpart. The carrier frequency is 1.38kHz, the modulation frequency is 43Hz." is simply misleading. As a minor nag the blog nonchalantly calls the phase modulated file "Pure FM modulation".
The more interesting bit is the claim that only phase differs. The Matlab script shows that what was actually done is inverse FFT on the magnitude of the phase modulated signal. This is the same as rotating the phase 90 degrees in the frequency domain.
This is what this phase rotation of 90 degrees in the frequency domain does to a sine wave:
View attachment 1041473
Yes, only the phase has changed.
This is what happens to square wave:
View attachment 1041474
B-b-b-but I only changed the phase
My signal processing skills have long faded due to lack of practice so I leave further analysis to others.
Markw4 was eager to point out that operation in time domain has to be equivalent to adjusting some of the frequency domain properties. Perhaps he can quickly enlighten us with what those adjustments to time domain properties are in this case.
As the script clearly shows : on the phase differs. This is of course reflected in the tone domain waveform as well. no surprise there. same for your examples.
pure AM means that there is no FM component. Mathematically this means that the instantaneous frequency (look it up) is constant. A pure FM signal has a constant envelope (instantaneous amplitude) but time varying instantaneous frequency. These concepts are exact mathematical definitions not up for debate.
claim of misleading statement: the FM process adds harmonics side bands as easily seen. the AM counterpart is the AM modulation of a 43Hz cosine with a series of harmonics, this is why the envelope is not purely sinusoidal. We also call a 43Hz square wave for a 43Hz signal despite it’s heavy content of harmonics. The harmonics are introduced as consequence of keeping the amplitude spectrum the same for the two signals. All of this has been discussed about 100 posts back. We could add this explanation of the math facts but that is trivial for people skilled in math and signal analysis but confusing for the rest.
At least we think is was interesting that two signals with the same amplitude spectrum seems to sound different (what does masking theory say here?). That the time domain waveforms are very different is trivially true.
Mr. Bohrok, Interesting graph you have there. As you asked me, I will offer a quick thought on it. The most interesting thing about the look of the graph to me is suggestive of a limitation of numerical modeling of an ideal square wave with infinite rise time (if that's what you did?), and therefore an infinite number frequencies. For modeling purposes there should effectively be difference equations rather than continuous. Thinking in that direction I wonder if for some continuous time system being modeled, the blue peaks might going asymptotic to infinity given the infinitely fast edge rise times an ideal square wave, or if maybe they limit at possibly twice the step amplitude. If I saw something like that on a scope I might be inclined to try adjusting the sweep rate to see how the waveform changes, if at all. Also, don't know your exact purpose in creating the graph since we don't really have ideal square waves in the physical world?
Last edited:
It is good that we have skilled math people here who are able to distinguish between phase modulation and frequency modulation.
No. I just used the script Lars provided for generating the "AM" version of a square wave. The purpose of this is just to show that the phrase "only difference is the phase" is quite misleading for anything more complex than a single sine wave.
Too late to edit but in MATLAB this is what I used (based on the script Lars provided):
Code:
x = 0:pi/100:4*pi;
PureFM=square(x); %FM signal
% calculate the AM signal with the same amplitude spectrum
% by taking the abs() of the FFT spectrum and inverse transform
PureAM=ifft(abs(fft(PureFM)));With optical cartridge vinyl or a very low phase noise dac and the big Sound Lab ESL speakers, it possible to get that virtual soundstage illusion pretty reliably here. Do find I have to turn down the HF shelfing a bit to make it most convincing, however. One reason it works with those speakers is they have sort of a asymmetrical figure-8 directivity pattern. Usually the back side is pointed towards a corner of the room a few feet behind the speaker. A little bit if HF damping, not much, is needed on the back wall between the speakers.
The result is that there is sound reflected around in the general area behind the speakers and it is wider than the distance between speakers.
What is more interesting is that the imaging from left to right and from front to back is stable and precise. Virtual performers sound like they are in distinct locations on the the virtual soundstage.
Just saying its probably one of those things one has to experience to believe. Hard to say how I might measure it and post meaningful measurements. Take a shotgun mic and scan the virtual soundstage for virtual performers? Don't know about that. Don't know what measurements would help readers understand it either. That said, no doubt whatever is coming out of the power amp and into the speakers has to be something more well characterized than just THD+N along with 20kHz crosstalk numbers.
Last edited:
good. Luckily its easy: phase is the time integral of frequency. What is the relevance here? Its a pure sine being FM/PM modulated so there is not difference except scaling
Even though for pure sine signal FM/PM modulation does not differ much it is rather lazy to use these terms nonchalantly to describe the same process. Suits marketing purposes (i.e. your blog) but not much else (e.g. this thread).
Finally its becoming more clear what the concern is. Various processes can cause what we call modulation.
So if thinking about it intuitively we might imagine some various possibilities:
We might expect that if a sine wave carrier were modulated by a square wave, it would produce a waveform as in FSK. https://en.wikipedia.org/wiki/Frequency-shift_keying
OTOH, if we have a square wave carrier modulated by a sine wave we would expect to see varying frequency square wave.
Then "intermodulation" would apply to two signals modulating each other, such as occurs when two signals pass through a nonlinear device to produce IMD.
Seems to me lrisbo was talking more about the result of some process, more than the process itself. Some process could produce a signal we could say looks like a frequency modulated sine wave. What should we name the resulting signal if not by referring to it as FM? The name was used as part of an effort to explain potential ambiguity when trying to interpret a spectrum analyzer display. In that context it seems appropriate to me.
So if thinking about it intuitively we might imagine some various possibilities:
We might expect that if a sine wave carrier were modulated by a square wave, it would produce a waveform as in FSK. https://en.wikipedia.org/wiki/Frequency-shift_keying
OTOH, if we have a square wave carrier modulated by a sine wave we would expect to see varying frequency square wave.
Then "intermodulation" would apply to two signals modulating each other, such as occurs when two signals pass through a nonlinear device to produce IMD.
Seems to me lrisbo was talking more about the result of some process, more than the process itself. Some process could produce a signal we could say looks like a frequency modulated sine wave. What should we name the resulting signal if not by referring to it as FM? The name was used as part of an effort to explain potential ambiguity when trying to interpret a spectrum analyzer display. In that context it seems appropriate to me.
Last edited:
I'm sure you will form your own interpretation of this which will be repeated ad nauseam in future threads related to audibility. However you seem to miss that the issue was not about how phase (or frequency) modulation is seen in amplitude spectrum but how 2 seemingly and audibly different signals can have very similar amplitude spectrum. With my graphs I just pointed out that the "amplitude modulated" signal was due to applying quite artificial "signal processing" operation in frequency domain. Once you or Lars (or anybody else) can prove that this "signal processing" can occur in real circuits and can produce audible artefacts then it merits further discussion.
Didn't it come up in the context of loudspeaker distortion, not circuit distortion.
However, even in the case of circuit distortion there are unstated assumptions such as a nonlinearity being time-invariant. That is, that the circuit nonlinearity doesn't have something like a memory effect, maybe due to temperature, or due to some other more hypothetical effect, say, to take a wild guess only for illustrative purposes, maybe having to with the state of stored charge in a transistor at the instant of a fast transient signal, say, as verses what it might be for a continuous sine wave. Another thing with transistors is that state of voltage and current in a transistor when driving a resistive test load my be different when driving an more complex load such as a speaker. Depending on the state of the speaker impedance at some point in time, the transistor instantaneous operating point (voltage and current) and thus nonlinearity may therefore be a little different. I know there are serious people who think about such things, but maybe it can get too complicated for a diy forum? Maybe it doesn't matter that much to some people? But there are small things like that going on in reality.
However, even in the case of circuit distortion there are unstated assumptions such as a nonlinearity being time-invariant. That is, that the circuit nonlinearity doesn't have something like a memory effect, maybe due to temperature, or due to some other more hypothetical effect, say, to take a wild guess only for illustrative purposes, maybe having to with the state of stored charge in a transistor at the instant of a fast transient signal, say, as verses what it might be for a continuous sine wave. Another thing with transistors is that state of voltage and current in a transistor when driving a resistive test load my be different when driving an more complex load such as a speaker. Depending on the state of the speaker impedance at some point in time, the transistor instantaneous operating point (voltage and current) and thus nonlinearity may therefore be a little different. I know there are serious people who think about such things, but maybe it can get too complicated for a diy forum? Maybe it doesn't matter that much to some people? But there are small things like that going on in reality.
Last edited:
well try doppler distortion and force factor modulation in a speaker as that is what Lars was trying to show before Mark started producing word salads.
We have argued for about 12 pages now over a paragraph on speakers.