My research and findings have shown low frequency phase shift is meaningless and not heard by listeners... Phase shift in the upper audio spectrum will affect the transient response....According to my many customers over 40 years of listening test.. cymbals and piano losses realism due to Phase non-linearity.. I have over 40 years of designing and building Audio equipment as an EE.. I always use science not emotion or opinion..... Designing and winding transformers just as long... Sounds like you would be best as a lawyer...
Just think for a moment. How could a simple time delay possibly alter the relative timing
of a waveform's frequency components, and hence the waveform's shape? It cannot.
In an amplifier the issue is phase shift relative to the mid-band frequencies.
Last edited:
You're right (and so was DF96). Looked at as a function of time, clearly this is true.
I looked at the signal as a function of frequency and space, which should be equally valid. But apparently I came to the wrong conclusion. Now I'm wondering why transforming f(t) to f(w,x) and letting that propagate through space leads to that wrong conclusion. I haven't figured out the mistake yet.
-Gnobuddy
Your research has been published in reputable, peer-reviewed journals in the field? No?
Then why should we trust it over the data collected by many different scientists and researchers over several decades, which was published in reputable journals, reviewed by peers, discussed at conferences, reproduced by other researchers independently, and finally collected together in the summary paper by Dr Toole that I linked to earlier?
Then you know that claims of unpublished data are essentially worthless. Anyone can claim anything on the anonymous Internet. Without proof, it can't be taken seriously.
My daddy's bigger than your daddy. So there!
-Gnobuddy
The bandwidth is a very advanced design specification for feedback use.
In single ended you don't need much bandwidth if you plan not to use global feedback. 35khz is enough but could have some ringing and some attenuation of the signal at 20khz, NOT a big deal, nothing you cannot fix with an oscilloscope.
No compelling reasons for choosing the 100khz vs the 35khz.
The main criteria for your transformer should be the highest possible inductance in all conditions. This will be what you will hear. The less inductance and the faster the inductance drops as the current increase Will be the determining factor for the amplifiers quality.
Secondly, the transformer input loss is to be considered because it can easily transform a 40 watt tube into a 20 watt tube!!!
Lastly, you will never know until you build it, is how bad is the IMD of the transformer, how well it was winded and interleaved. Ever heard of airtight? Their amplifiers THD is almost bad. but if you look at the level of feedback and the IMD, they are very special amplifiers.
The recommendation to go solid state is probably very wise because tube amps cannot sustain any bass note like a solid state can with a high current regulated supply, transformer quickly saturates in tube, feedback cant fix it.
If you can build an Accuphase quality SS, it will cost less and sound probably better.
According to my many customers over 40 years of listening test.. cymbals and piano losses realism due to Phase non-linearity..
How can they lose reality when they have none?
1. first thing you listen to is your room. Sounds like your room.
2. second thing you hear is a cone vibrating with so many reflections, modes, and time problems that anything broadband is impossible to realistically reproduce, like a piano or a cymbal.
3. The amplifier has to match the optimum resonance, SS with feedback will screw up most of audio in the low frequency but have good high frequency.
4. Tube amps will have better relaxed behavior but higher distortion in bass and problems in the midrange 1k and higher, lack of details for example and more dynamism.
5. cables, transformers, capacitors are only enhancers, they are the frame not the painting.
I can build you a mediocre amp with a super high quality perfect OT, and a killer amp with a normal EI core.
How can they lose reality when they have none?
1. first thing you listen to is your room. Sounds like your room.
2. second thing you hear is a cone vibrating with so many reflections, modes, and time problems that anything broadband is impossible to realistically reproduce, like a piano or a cymbal.
3. The amplifier has to match the optimum resonance, SS with feedback will screw up most of audio in the low frequency but have good high frequency.
4. Tube amps will have better relaxed behavior but higher distortion in bass and problems in the midrange 1k and higher, lack of details for example and more dynamism.
5. cables, transformers, capacitors are only enhancers, they are the frame not the painting.
I can build you a mediocre amp with a super high quality perfect OT, and a killer amp with a normal EI core.
I was not disputing your arithmetic, but the false conclusion you draw from your arithmetic. The fact that we cannot hear thousands of degrees of phase shift on a particular signal tells us nothing because phase can only be relevant up to 360 degrees, and a pure time delay preserves relative phase between different elements of the signal.Gnobuddy said:
Is it that you were looking at it as a 'standing wave', while it is really a 'travelling wave'?
from post 16 :
Quote: Originally Posted by waltube Also the value of H. It easy to find a good declaration on response but not on inductance; generally a trafo with a great BW hasn't a high value of H. The LF behaviour depends on inductance, so for good LF bandwidth you need high H.
My definition is not correct, my english is not so fine. What I am sayng is that normally on trafo with good hi freq. response the H is reasonable low. Of course to get a good LF response we need an high value of inductance. And the right balance of both goals is a problem for the majority of OT trafo.
Walter
Quote: Originally Posted by waltube Also the value of H. It easy to find a good declaration on response but not on inductance; generally a trafo with a great BW hasn't a high value of H. The LF behaviour depends on inductance, so for good LF bandwidth you need high H.
My definition is not correct, my english is not so fine. What I am sayng is that normally on trafo with good hi freq. response the H is reasonable low. Of course to get a good LF response we need an high value of inductance. And the right balance of both goals is a problem for the majority of OT trafo.
Walter
1. The impact is no need to bypass the high frequency of feedback to compensate for the transformer loss. The feedback must work faster than the audible frequency it is correcting. (aka multiple types of feedbacks, multiple filters used in output and feedback loop).
2. Transformers cannot differ only in bandwidth.
wisdom:a) you need to know how stable the inductance of the transformer is for the current you will be using, b) for the class of amplification A1, AB, AB2, B(never used much in audio) the bias imbalance tolerance when going into those modes.
Reputable companies: C core audio note, Lundhal, James, you cannot miss.
Then you have more ordinary: Sowters, Hammond, Edcor ( a new welcomed player)
Exotics Like Plintron, Toroidi,
High end, boutique, Electra Print, J&K audio (I recommend!), silk
Budget: electron one, maybe dynaco (tube4hifi) can count as budget, but they are good.
Look into the step function. Its used to shift functions, including fourier transforms.
Narrower bandwidth due to low primary inductance and high shunt capacitance increases distortion at both frequency extremities because the tube is loaded more due to lower impedances. Simple.
The phase shift is an inevitable "bonus".
Although some people prefer the sound of more distortion in the low frequencies, I suspect HF distortion is more offending to the ear. It is also mentioned in RDH4 IIRC.
The phase shift is an inevitable "bonus".
Although some people prefer the sound of more distortion in the low frequencies, I suspect HF distortion is more offending to the ear. It is also mentioned in RDH4 IIRC.
Last edited:
What do you hear, phase in different situations.
1. Can you hear relative phase of a note and its harmonic in monophonic playback?:
Look at the time domain waveform of a french horn playing a 1080 Hz note. The french horn has a dominant 2nd harmonic (well established in the literature, search for yourself if you feel the need to). You will easily see the second harmonic in the time domain (I did so with a french horn player, a shure microphone, and an oscilloscope). At the sound intensity that was used, neither the shure microphone nor the oscilloscope have any significant 2nd harmonic distortion them selves; so all of the 2nd harmonic in the waveform was from the french horn.
The velocity of sound is about 1080 feet per second. A 1080 Hz note will have one cycle per foot. A 2160 Hz tone will have 2 cycles per foot (2nd harmonic of a 1080 Hz note).
Start with just the Left channel of a stereo system. CD, CD player, power amp, loudspeaker. Use a CD that has a Left channel recording of a french horn.
Position the loudspeaker exactly 10 feet from the listener’s ear. Pretend the 1080 Hz note, and the 2160 Hz 2nd harmonic are in the same relative phase versus the original french horn 1080 Hz note and 2nd harmonic phase relationship. Call that 0 degrees relative phase.
Now, re-position the loudspeaker 10.5 feet away from the listener’s ear. The 1080 Hz note will be -180 degrees relative to the original 10 foot position. But the 2160 Hz 2nd harmonic will be - 360 degrees relative to the original 10 foot position.
So, the 1080 note is -180 degrees, and the 2160 Hz 2nd harmonic is -360 degrees. At the 10.5 foot position, they are 180 degrees relative now, versus the 0 degrees they were at the 10 foot position. Can you hear the difference of the 0 degrees relative versus the -180 degrees relative?
2. Can you hear relative phase of Left and Right channels in stereo playback?:
If you can not hear relative phase between the left and right channels, that means one of two things: A. You only hear the position of the stereo image based entirely on amplitude differences of the two signals as they come out of the two loudspeakers. B. You only have one working ear, and so can not hear the stereo image at all.
1. Can you hear relative phase of a note and its harmonic in monophonic playback?:
Look at the time domain waveform of a french horn playing a 1080 Hz note. The french horn has a dominant 2nd harmonic (well established in the literature, search for yourself if you feel the need to). You will easily see the second harmonic in the time domain (I did so with a french horn player, a shure microphone, and an oscilloscope). At the sound intensity that was used, neither the shure microphone nor the oscilloscope have any significant 2nd harmonic distortion them selves; so all of the 2nd harmonic in the waveform was from the french horn.
The velocity of sound is about 1080 feet per second. A 1080 Hz note will have one cycle per foot. A 2160 Hz tone will have 2 cycles per foot (2nd harmonic of a 1080 Hz note).
Start with just the Left channel of a stereo system. CD, CD player, power amp, loudspeaker. Use a CD that has a Left channel recording of a french horn.
Position the loudspeaker exactly 10 feet from the listener’s ear. Pretend the 1080 Hz note, and the 2160 Hz 2nd harmonic are in the same relative phase versus the original french horn 1080 Hz note and 2nd harmonic phase relationship. Call that 0 degrees relative phase.
Now, re-position the loudspeaker 10.5 feet away from the listener’s ear. The 1080 Hz note will be -180 degrees relative to the original 10 foot position. But the 2160 Hz 2nd harmonic will be - 360 degrees relative to the original 10 foot position.
So, the 1080 note is -180 degrees, and the 2160 Hz 2nd harmonic is -360 degrees. At the 10.5 foot position, they are 180 degrees relative now, versus the 0 degrees they were at the 10 foot position. Can you hear the difference of the 0 degrees relative versus the -180 degrees relative?
2. Can you hear relative phase of Left and Right channels in stereo playback?:
If you can not hear relative phase between the left and right channels, that means one of two things: A. You only hear the position of the stereo image based entirely on amplitude differences of the two signals as they come out of the two loudspeakers. B. You only have one working ear, and so can not hear the stereo image at all.
Last edited:
Here's an analogy that I find useful:
Imagine a ticker tape machine with three motorised pens writing transversely on the tape (a bit like an old fashioned recording barometer). Lets say the tape is coming out at 1 metre per minute. Pen 1 is writing a sine wave at 10 cycles per minute, pen 2 is writing a sine wave at 20 cycles per minute, pen 3 is writing the sum of the current positions of pen 1 and pen 2.
Look at what is coming out of the machine for a minute, now quickly walk 3 metres away from the machine and have a look at the tape at your new position. After you wait 3 minutes, you will see exactly the same sine waves and composite wave that you saw coming out of the ticker machine.
The speed of the tape being produced is analogous to the speed of sound. The traces on the tape are the 'travelling waves' which are analogous to sound waves travelling through the air. The ticker machine is analogous to the loudspeaker, and you looking at the tape is analogous to you hearing the sound.
Imagine a ticker tape machine with three motorised pens writing transversely on the tape (a bit like an old fashioned recording barometer). Lets say the tape is coming out at 1 metre per minute. Pen 1 is writing a sine wave at 10 cycles per minute, pen 2 is writing a sine wave at 20 cycles per minute, pen 3 is writing the sum of the current positions of pen 1 and pen 2.
Look at what is coming out of the machine for a minute, now quickly walk 3 metres away from the machine and have a look at the tape at your new position. After you wait 3 minutes, you will see exactly the same sine waves and composite wave that you saw coming out of the ticker machine.
The speed of the tape being produced is analogous to the speed of sound. The traces on the tape are the 'travelling waves' which are analogous to sound waves travelling through the air. The ticker machine is analogous to the loudspeaker, and you looking at the tape is analogous to you hearing the sound.
Last edited:
No relative phase changes of the tape as you scroll it out, and no relative phase changes of phase on the CD player as you playback the recording. OK. Good point.
But . . .
Now, play the recorded signal from the CD player through a single device that has both a high pass filter and a low pass filter. That is called a transformer. Or, you might more accurately call the transformer a bandpass filter.
Whatever the relative phase of the high frequencies versus low frequencies was from the CD player signal, after that signal has gone through the transformer, there will be a new relative phase change of the high and low frequencies.
Whether those relative phase changes can be heard or not, is up for debate (or not up for debate). But the fact remains that the relative phases of the high and low frequencies has been changed by that transformer.
That would be like slitting the paper tape along its length into 3 separate tapes, and then changing the relative position (phase) of the 2 frequencies. Changing those 2 phases would also change the shape of the 3rd pen (but the ink of that sum signal was already dry, so it would no longer be the true wave shape). Thank you Malcolm Irving for that excellent example. It provides more food for thought.
My post #34 was supposed to have a different set of examples. I basically lost my train of thought on that post, and lost lots of text, and then I sent it out. It can happen in the wee small hours of the morning.
Just ignore my post #34, there is no way I know to take it down (erase it).
But . . .
Now, play the recorded signal from the CD player through a single device that has both a high pass filter and a low pass filter. That is called a transformer. Or, you might more accurately call the transformer a bandpass filter.
Whatever the relative phase of the high frequencies versus low frequencies was from the CD player signal, after that signal has gone through the transformer, there will be a new relative phase change of the high and low frequencies.
Whether those relative phase changes can be heard or not, is up for debate (or not up for debate). But the fact remains that the relative phases of the high and low frequencies has been changed by that transformer.
That would be like slitting the paper tape along its length into 3 separate tapes, and then changing the relative position (phase) of the 2 frequencies. Changing those 2 phases would also change the shape of the 3rd pen (but the ink of that sum signal was already dry, so it would no longer be the true wave shape). Thank you Malcolm Irving for that excellent example. It provides more food for thought.
My post #34 was supposed to have a different set of examples. I basically lost my train of thought on that post, and lost lots of text, and then I sent it out. It can happen in the wee small hours of the morning.
Just ignore my post #34, there is no way I know to take it down (erase it).
6A3, I don't know why the phase should or would be possibly or could be modified by the output transformer.
In general the higher power, the better the phase in OTs!
The phase shifts in OT occurs way past 20khz, and they only are an issue with feedback at high frequencies which you possibly don't hear around 15khz. But they are a stability issue, nothing else.
This is a special type of C cores, good around 60-100 watts, depending on bias/tubes.
With only CFB I get +10 degrees 3hz , -10 degrees at 15khz -20 at 28khz -60 at 70khz,
Bandwidth 0.3hz to 80khz -3db points
On another transformer, 20 watts
Bandwidth 5hz to 100khz -3db
+14deg at 18hz, -18deg at 20khz
In general the higher power, the better the phase in OTs!
The phase shifts in OT occurs way past 20khz, and they only are an issue with feedback at high frequencies which you possibly don't hear around 15khz. But they are a stability issue, nothing else.
This is a special type of C cores, good around 60-100 watts, depending on bias/tubes.
With only CFB I get +10 degrees 3hz , -10 degrees at 15khz -20 at 28khz -60 at 70khz,
Bandwidth 0.3hz to 80khz -3db points
On another transformer, 20 watts
Bandwidth 5hz to 100khz -3db
+14deg at 18hz, -18deg at 20khz
Last edited:
gabdx,
I am not disagreeing. But there are phase effects, even though they can not be heard.
For simplicity, lets take a single ended output transformer. It is not a great one, just an inexpensive one. It is not your expensive push pull output transformer (or very expensive SE transformer). It is a 3k Ohm to 8 Ohm model. It has 15H of primary inductance. Load the secondary with 8 ohms (reflects the 8 Ohms on the secondary to be 3k Ohms at the primary). Drive the primary with an rp of 1k Ohms. We will not use negative feedback.
1k in parallel with 3k Ohms is 750 Ohms.
15H is 750 ohms at 7.96 Hz. The resultant single pole response is: the -3dB and -1 dB frequencies are an octave apart. Relative to a 1 kHz signal: The signal is -3dB at 7.96 Hz. There is a lead of about +45 degrees at that frequency. It also is -1dB at 15.9Hz. There is a lead of about +23 degrees at that frequency.
The high frequency response is at least a two pole response. The -3dB frequency will be less than 2 times the -1dB frequency. There is lots of phase shift at the - 3dB frequency. For example, it may be near 35kHz.
We can hardly hear a 15.9Hz signal, most speakers make nothing but harmonic distortion there, and no fundamental.
We can not hear the 35kHz signal, not even if our tweeter makes it to 35kHz; our ears will not get there (unless we had a bat ear transplant).
Not hearing a phase difference is not the same as not having a phase difference.
Another concept is group delay. Take a filter that has bad group delay, different frequencies that enter the filter at the same time, will leave the filter at different times. A transient from a percussive instrument will be smeared.
If you could take a square wave, and remove only the 5th harmonic, and not disturb the amplitude or phase of any other frequencies, you would see an Inverted 5th harmonic riding on the top and bottom of the square wave.
If you could take a square wave, and phase shift only the 5th harmonic, and not disturb any other frequencies, you would see a phase shifted 5th harmonic riding on the top and bottom of the square wave.
I am not disagreeing. But there are phase effects, even though they can not be heard.
For simplicity, lets take a single ended output transformer. It is not a great one, just an inexpensive one. It is not your expensive push pull output transformer (or very expensive SE transformer). It is a 3k Ohm to 8 Ohm model. It has 15H of primary inductance. Load the secondary with 8 ohms (reflects the 8 Ohms on the secondary to be 3k Ohms at the primary). Drive the primary with an rp of 1k Ohms. We will not use negative feedback.
1k in parallel with 3k Ohms is 750 Ohms.
15H is 750 ohms at 7.96 Hz. The resultant single pole response is: the -3dB and -1 dB frequencies are an octave apart. Relative to a 1 kHz signal: The signal is -3dB at 7.96 Hz. There is a lead of about +45 degrees at that frequency. It also is -1dB at 15.9Hz. There is a lead of about +23 degrees at that frequency.
The high frequency response is at least a two pole response. The -3dB frequency will be less than 2 times the -1dB frequency. There is lots of phase shift at the - 3dB frequency. For example, it may be near 35kHz.
We can hardly hear a 15.9Hz signal, most speakers make nothing but harmonic distortion there, and no fundamental.
We can not hear the 35kHz signal, not even if our tweeter makes it to 35kHz; our ears will not get there (unless we had a bat ear transplant).
Not hearing a phase difference is not the same as not having a phase difference.
Another concept is group delay. Take a filter that has bad group delay, different frequencies that enter the filter at the same time, will leave the filter at different times. A transient from a percussive instrument will be smeared.
If you could take a square wave, and remove only the 5th harmonic, and not disturb the amplitude or phase of any other frequencies, you would see an Inverted 5th harmonic riding on the top and bottom of the square wave.
If you could take a square wave, and phase shift only the 5th harmonic, and not disturb any other frequencies, you would see a phase shifted 5th harmonic riding on the top and bottom of the square wave.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.