Nice work. Do you have a dedicated thread for your amp? I like low group delay on speakers. It’s important that group delay produce minimum phase shift in amps up to 300kHz. That’s where human auditory system can detect phase shifts and why amps with flat phase up to 300kHz have great image and soundstage.
The CFH9 could be better but it’s not bad at 300kHz.
The CFH9 could be better but it’s not bad at 300kHz.
I did not create any branch on this amplifier. The history of the creation of this amplifier is as follows. I was contacted by my friend who bought a vintage amplifier and would like to replace his circuit with something modern with a supply voltage of + -25 V, but not very complicated. Without hesitation, I offered him this option.
Seems it is based on Apex AX17 front end. Looks nice - your FFT distortion analysis should plot excitation at F=1kHz and look at harmonics at 2F at 2kHz, 3F at 3kHz, 4F at 4kHz etc…
At 1W behavior is Class A and looks wonderful.
Not excite at 10kHz, and 2F at 20kHz, 3F at 30kHz which are beyond human hearing.
You might consider starting your own thread as this is the CFH11 thread and a different amp.
At 1W behavior is Class A and looks wonderful.
Not excite at 10kHz, and 2F at 20kHz, 3F at 30kHz which are beyond human hearing.
You might consider starting your own thread as this is the CFH11 thread and a different amp.
It is easy to see that in order to fulfill David Hafler's SWDT criterion at the highest frequencies of the audio range, the signal propagation delay time must be equal to only 8 ns.
Most amplifiers covered by negative feedback have a signal propagation delay time from 200 ns to 1.5 ... 2.5 μs. With such a signal delay, the sound quality does not depend on THD.
The effectiveness of this test was proved in practice by Bob Carver. See The Carver Challenge article (https://www.stereophile.com/content/carver-challenge)
Most amplifiers covered by negative feedback have a signal propagation delay time from 200 ns to 1.5 ... 2.5 μs. With such a signal delay, the sound quality does not depend on THD.
The effectiveness of this test was proved in practice by Bob Carver. See The Carver Challenge article (https://www.stereophile.com/content/carver-challenge)
Attachments
Back in the 50s of the last century, it became clear that THD does not correlate with sound quality. Then they began to pin their hopes on the measurement of intermodulation distortion. But they didn't live up to expectations either.
In 1954, M. Sapozhkov defended his dissertation on the measurement of distortions by the compensation method. In fact, this is the method I use today, only instead of delaying on RC elements I use an ideal delay line. In my opinion, this is the only way to identify all kinds of distortions.
Another original test was proposed by David Hafler - the SWDT test ("straight-wire" differential test). As a quality criterion, Hafler proposed a difference signal of no more than -60 dB at the highest frequencies of the audio range, and no more than -70 dB in the rest of the range. However, this test also has a weak point that Hafler himself did not pay attention to. To reveal real distortions, it is necessary to use signal bursts, a triangular signal with a frequency of 10 kHz, or a square wave signal as test signals. What does -60 dB SWDT mean at 20 kHz. With an output voltage amplitude of 30Vpk, the SWDT signal should not exceed 30mVpk.
The figure presented in the previous post shows that in order to fulfill this condition, the delay should be only 8 ns (coincides with the conclusions of Kirill Hammer).
It is strange that so far no due attention has been paid to these methods of measuring distortion as the most fully correlated with sound quality.
In 1954, M. Sapozhkov defended his dissertation on the measurement of distortions by the compensation method. In fact, this is the method I use today, only instead of delaying on RC elements I use an ideal delay line. In my opinion, this is the only way to identify all kinds of distortions.
Another original test was proposed by David Hafler - the SWDT test ("straight-wire" differential test). As a quality criterion, Hafler proposed a difference signal of no more than -60 dB at the highest frequencies of the audio range, and no more than -70 dB in the rest of the range. However, this test also has a weak point that Hafler himself did not pay attention to. To reveal real distortions, it is necessary to use signal bursts, a triangular signal with a frequency of 10 kHz, or a square wave signal as test signals. What does -60 dB SWDT mean at 20 kHz. With an output voltage amplitude of 30Vpk, the SWDT signal should not exceed 30mVpk.
The figure presented in the previous post shows that in order to fulfill this condition, the delay should be only 8 ns (coincides with the conclusions of Kirill Hammer).
It is strange that so far no due attention has been paid to these methods of measuring distortion as the most fully correlated with sound quality.
It is easy to see that in order to fulfill David Hafler's SWDT criterion at the highest frequencies of the audio range, the signal propagation delay time must be equal to only 8 ns.
Most amplifiers covered by negative feedback have a signal propagation delay time from 200 ns to 1.5 ... 2.5 μs. With such a signal delay, the sound quality does not depend on THD.
The effectiveness of this test was proved in practice by Bob Carver. See The Carver Challenge article (https://www.stereophile.com/content/carver-challenge)
As for the measurement of the spectrum at a frequency of 1 kHz, here again we can get a false result. Some amps have a lot of loop gain at this frequency and due to the deep feedback they can give good results. But this result does not say anything, since it does not take into account other types of distortions, in particular, high-speed ones.
Most amplifiers covered by negative feedback have a signal propagation delay time from 200 ns to 1.5 ... 2.5 μs. With such a signal delay, the sound quality does not depend on THD.
The effectiveness of this test was proved in practice by Bob Carver. See The Carver Challenge article (https://www.stereophile.com/content/carver-challenge)
As for the measurement of the spectrum at a frequency of 1 kHz, here again we can get a false result. Some amps have a lot of loop gain at this frequency and due to the deep feedback they can give good results. But this result does not say anything, since it does not take into account other types of distortions, in particular, high-speed ones.
Attachments
Well you seem to have your opinion on speed distortion as being the most important. I myself know that psychoacoustic effects of dominant second order and monotonically descending higher orders (like a SE Class A amp) are very audible and will make an amp sound great or if it is odd order dominant will sound fatiguing.
Our ears can hear maybe 2usec phase shifts. This corresponds to delays introduced by asymmetric folds of the ear (pinna) and helps us to locate sound position even with one ear. This phase delay corresponds to sounds up to 300kHz. I don’t think an amp needs to have 8ns propagation delay time. That’s how far signals in cables travel in about 8 ft. Speed of light is 11in per nano second. You can’t hear that effect - it’s in the microwave / RF region circa 125MHz.
The reason we look at FFT with 1kHz excitation is because the ear is extremely sensitive to those frequencies and any distortions in the 2kHz - 7kHz range are audible. Even at -60dB if odd order - it sounds fatiguing over long periods of listening.
Our ears can hear maybe 2usec phase shifts. This corresponds to delays introduced by asymmetric folds of the ear (pinna) and helps us to locate sound position even with one ear. This phase delay corresponds to sounds up to 300kHz. I don’t think an amp needs to have 8ns propagation delay time. That’s how far signals in cables travel in about 8 ft. Speed of light is 11in per nano second. You can’t hear that effect - it’s in the microwave / RF region circa 125MHz.
The reason we look at FFT with 1kHz excitation is because the ear is extremely sensitive to those frequencies and any distortions in the 2kHz - 7kHz range are audible. Even at -60dB if odd order - it sounds fatiguing over long periods of listening.
The measurement of the spectrum of a signal with a frequency of 1 kHz is carried out in the steady state, when there are no additional distortions associated with transients, with Group Delay.
I hope you are aware of Graham Maynard's first cycle distortion measurement (FCD) idea. In such a primitive way, he came to the conclusion that amplifiers that introduce less distortion already in the first period sound better. And this is understandable - such amplifiers either have a low group delay (do not introduce high-speed distortion), a high frequency of the first pole, which gives a short spectrum, or there is no negative feedback in the amplifiers and the amplifiers also do not introduce distortion associated with group delay. For such amplifiers, group delay is no longer of great importance.
As for the group delay in amplifiers, Hafler designed the XL-280 amplifier with adjustable group delay down to a negative value. But a low group delay value can be obtained in a limited area within the sound range. However, outside the audio range, the group delay grows and already at a frequency of 100 kHz it reaches 300 ns, and at a frequency of 350 kHz it is 1.3 μs. On stationary signals, this amplifier passes the SWDT test, but fails burst, triangle, or square wave tests.