I agree that you can't just stick any old **** in, and I am guilty of selecting some silly low disortion amplifiers cos I can. But the rational voice in my head says that 0.01% THD or less is more than adequate and I can worry about other things.
I also embrace the fact that KNOWING I have the highest performing amps I can afford will make the system sound better. I'm not zen enough to be able to disconnect my bias so why not leverage it
. Let's face it, I have no credibility to lose!But you did not measure slew rate, and you in any event it appeared to be mistracking............
Sorry Esperado, I don't see anything from those plots as I don't know what schematics you've used and how you've simulated it. Better, if you like to proceed this discussion, we use "Unique" thread.
Of course the phono cartridge was mistracking. That was the point of the exercise, to find what the reproduce stage has to contend with. However, it was a popular vinyl record of the time, often used in hi fi comparisons, and the results were similar to what a TIM 30 test waveform would put out in terms of output, bandwidth, spectral rolloff, etc.
Then that point is moot. Mistracking is abnormal, avoidable, and who cares whether the system is capable of reproducing chaotic artefacts which aren't in the programme material, and in any event sound horrid ? This is the tail wagging the dog..........
dadod, your feedback ratio begin to decrease at 1KHz. No surpise you don't see any difference with feedback impedance.
Everything was explained about this simulations in my previous posts.
My purpose was not to criticize your design, but to help-you to understand what a CFA is.
Unless you achieve to add stages faster enough to introduce no openloop bandwidth reduction, you cannot do-it. CFA are in the spirit of Kartings, you cannot design them like a limousine (VFA way of thinking).
With the actual available device's speed on the shelf, you need to get a flat openloop bandwidth up to 10KHz. That is the goal. After this, you can try to add complexity in order to reduce distortions, taking care to not destroy original speed, or adding unwanted poles, not good for stability.
Well this is the old problem. How far do we allow an overload to go down the signal chain, and how cleanly do we clip and recover from OL if we have to? Should we consider limiting at the very input? Does the Cordell input loading approach (for MM) accomplish more than meets the eye? Should my suggestion of a negative input Z for MC be pursued, despite the challenges?Then that point is moot. Mistracking is abnormal, avoidable, and who cares whether the system is capable of reproducing chaotic artefacts which aren't in the programme material, and in any event sound horrid ? This is the tail wagging the dog..........
Of course, along with insanely great tracking, clean surfaces and proper pressings are obtainable, if requiring a lot more scrutiny than many managed in typical conditions. One guy I know who is so hollow-state-obsessed that he may banish all silicon from his environs at least agrees with me that a good record washer is essential!
Oh, and we also both enjoy listening to LPs. I don't try to defend the medium as somehow higher-Fi, but I think I get good results. And I have plenty of "legacy" material, and some of the remastered stuff that is quite impressive.
Esperado
On your opinion what types of devices is needed to get CFA flat OL-BW up to end of audible spectrum ~20Khz ?
i was never able to achieve this with decent harmonic distortion in class AB.
(I tend to prefer Laterals or HEX FETs for power devices, because they are fast, not subject to secondary breakdown, with a flat or negative tempco, and like to be run hot).
A fast as they are, and as strange it can appears, it looks like actual transitors are not fast enough for audio ;-)
Well, as 10KHz second harmonic is out of our listening range, I think 10KHz is not a too bad threshold ?
Last edited:
Christophe, careful with enough. For decades, we thought our fast new gen power transistors (by Toshiba, Motorola ON/Semi, Sanken, et al.) are more than fast enough, easily hitting 30+ MHz, and it turned out that even that is not fast enough. We actually have no clear idea what is fast enough just yet. And under which conditions.
Esperado
- Do you think that ~20Khz OL-BW can be achieved from A-class CFA ? , with decent THD .
- Yes Lat-Fet`s like to run hot , linearity wise on the first place .
- IMHO ~20Khz or wider OL-BW is better , but this my conclusion come from different OTL tube amplification world .
- Do you think that ~20Khz OL-BW can be achieved from A-class CFA ? , with decent THD .
- Yes Lat-Fet`s like to run hot , linearity wise on the first place .
- IMHO ~20Khz or wider OL-BW is better , but this my conclusion come from different OTL tube amplification world .
Last edited:
the "flat loop gain over audio bandwidth" really hasn't been show to explain anything about audio amp audible distortion - and is usually "supported" by technically incorrect assertions
and there are good feedback theory consequences weighted with well established psychoacoustics that suggest it is a poor way to structure your distortion reducing loop gain for music reproduction
high loop gain reduces all distortion at the frequencies the gain is high - human hearing is most sensitive ~ 3 kHz, most musical instrument fundamentls are well bleow that, most music content falls rapidly with frequency above 3-5kHz
so larger gain at less than 20 kHz is well justified - not flat gain to 20 kHz
40+ years and "PIM" hasn't been explored, proven more audible than "AM" IMD in any controlled listening test I've ever heard of - and several someones should have been anxious to dig any up to throw in my face over the years of my posting the above critique here
and there are good feedback theory consequences weighted with well established psychoacoustics that suggest it is a poor way to structure your distortion reducing loop gain for music reproduction
high loop gain reduces all distortion at the frequencies the gain is high - human hearing is most sensitive ~ 3 kHz, most musical instrument fundamentls are well bleow that, most music content falls rapidly with frequency above 3-5kHz
so larger gain at less than 20 kHz is well justified - not flat gain to 20 kHz
40+ years and "PIM" hasn't been explored, proven more audible than "AM" IMD in any controlled listening test I've ever heard of - and several someones should have been anxious to dig any up to throw in my face over the years of my posting the above critique here
I think the message is more that having flat gain over the audio bandwidth isn't important at all. Having a lot of loop gain is what's valuable, especially in the critical hearing region.
Given equal gain at 20kHz, I'll take the amplifier that is already behaving as an ideal integrator--lower OL bandwidth (or highest DC gain) the better--as that will give me another 20dB of loop gain in that 1k-3k region..
Given equal gain at 20kHz, I'll take the amplifier that is already behaving as an ideal integrator--lower OL bandwidth (or highest DC gain) the better--as that will give me another 20dB of loop gain in that 1k-3k region..
banat, first a personal remark. There is no 'more correct' technical justification to prefer ppm distortion numbers, increasing feedback at lower frequency and sacrificing coherency all over the listening bandwidth than to prefer the contrary.
We don't have yet sacred threshold numbers in VFA/CFA religions.
The only way I know is to try to figure-it out, with a freethinker's position , listening and listening again, choosing our poisons. As long as it is done in an honest attitude.
About class A, my personal position is it a waste of energy and heat for nothing. Not complicated to have a class AB amp, working in class A at normal listening levels and turning only into class B during very short peaks at high listening levels. I even find it give a enhanced feeling of dynamic.
Concerning the distortion numbers we can achieve with a very simple CFA, even in class AB, they are, on my opinion, yet well under the minimal requested value if correctly designed.
Some prefer impressive distortion numbers with sinusoidal signals? OK, go to VFAs.
Some prefer slew rate, extended phase coherency, extended constant feedback ratio etc... go to CFA.
I prefer class D for my subs, because it is low distortion, high efficiency, and that the evils of Class D (switching frequency) is far away from the range of the boomer. I prefer VFA for my bass/low medium, because I can benefit from both a high open loop gain, and constant feedback ratio in its range. I prefer CFA for my high medium tweeter...definitively.
As i prefer to keep very bad educated "stupid" people in my ignore list.
We don't have yet sacred threshold numbers in VFA/CFA religions.
The only way I know is to try to figure-it out, with a freethinker's position , listening and listening again, choosing our poisons. As long as it is done in an honest attitude.
About class A, my personal position is it a waste of energy and heat for nothing. Not complicated to have a class AB amp, working in class A at normal listening levels and turning only into class B during very short peaks at high listening levels. I even find it give a enhanced feeling of dynamic.
Concerning the distortion numbers we can achieve with a very simple CFA, even in class AB, they are, on my opinion, yet well under the minimal requested value if correctly designed.
Some prefer impressive distortion numbers with sinusoidal signals? OK, go to VFAs.
Some prefer slew rate, extended phase coherency, extended constant feedback ratio etc... go to CFA.
I prefer class D for my subs, because it is low distortion, high efficiency, and that the evils of Class D (switching frequency) is far away from the range of the boomer. I prefer VFA for my bass/low medium, because I can benefit from both a high open loop gain, and constant feedback ratio in its range. I prefer CFA for my high medium tweeter...definitively.
As i prefer to keep very bad educated "stupid" people in my ignore list.
Last edited:
a low bandwidth high loop gain design with typical single pole gain slope will have ~6x more loop gain at 3 kHz with the same ultimate unity loop gain frequency, singe pole compensation case usually considered compare to the flat gain to 20 kHz amp
all errors measured by the input differencing at 3kHz will be reduced ~6x more with the high, sloping loop gain amp than in the flat loop gain to 20 kHz case
"all errors" includes phase errors
and the high loop gain amp has proportionately more loop gain below 3 kHz where it is even more plausibly important
due to the only few decade old psychoacoustic result of auditory nerve pulse phase locking - the inner hair cell neurons only fire on one phase of the basilar membrane motion giving human hearing a "absolute phase" resolution mechanism - but the phase locking pattern goes away above a few kHz
interestingly "phase denial" is very common among recording engineers - mic placement, analog signal chains don't always capture "audience perspective" (close micing drums) or try to preserve polarity throughout the recording/mixing/mastering process
if "phase coherency" is a concern - we know that human hearing has actual soundwave phase encoding ability below 3 kHz - above that the ears respond almost exclusively to amplitude - the phase information for sound frequencies above a few kHz doesn't make the step into neural firing pattern at all
all errors measured by the input differencing at 3kHz will be reduced ~6x more with the high, sloping loop gain amp than in the flat loop gain to 20 kHz case
"all errors" includes phase errors
and the high loop gain amp has proportionately more loop gain below 3 kHz where it is even more plausibly important
due to the only few decade old psychoacoustic result of auditory nerve pulse phase locking - the inner hair cell neurons only fire on one phase of the basilar membrane motion giving human hearing a "absolute phase" resolution mechanism - but the phase locking pattern goes away above a few kHz
interestingly "phase denial" is very common among recording engineers - mic placement, analog signal chains don't always capture "audience perspective" (close micing drums) or try to preserve polarity throughout the recording/mixing/mastering process
if "phase coherency" is a concern - we know that human hearing has actual soundwave phase encoding ability below 3 kHz - above that the ears respond almost exclusively to amplitude - the phase information for sound frequencies above a few kHz doesn't make the step into neural firing pattern at all
Last edited:
- Status
- Not open for further replies.