Well, what i wanted to say is, if some prefer chocolate and others strawberries, it is probably because they don't feel the same molecules the same way. And the same for the colors and our eyes. We still know that the frequency range of our ears are not the same between women and men.
About test protocol, i was so often fooled by myself, mixing a tune during hours, to discover the next morning, at first listening, that it was all wrong, polishing little details, neglecting the essential... that i'm very suspicious.
First, forget about blind tests: you cannot do them when you are working, alone, on a component or a detail of an amp. Set two samples of your amp to compare, don't believe in your memory.
I'm mainly interested on transients. so, i chose sample sources chosen with care, that i know very well, which can reveal details, like nail or mediator attacks on guitar chords, sticks on cymbals, kick on drum's skins. Separation between basses and kick drums.
Two sorts of samples. Very short ones (just enough to put the detail you care of in perspective), that i listen with much attention. And long ones that i listen as much unfocused as possible. sometimes from an other room. No delay when you compare, repetitive test.
Alternate short and long samples.
Male voices are good to feel distortion, females sibilance. Cymbals you want natural, no 'treeble' or 'psss psss', but the feel of the metal size with the hit of the stick on it.
Some classical recording to judge space, some badly aggressive recordings to judge fatigue.
Are the musicians with a real presence in front of you, well situated in the space that you can follow and separate from others easily ? Did-you feel the virtual space they are playing in ? Are the 'colors' very different from a record to an other ?
Because i can fatigue myself very fast, i take for granted only obvious changes and make long pauses and repetitive sessions before to be sure of something.
But what we cannot share with others, is our way to listen (some listen in a vertical way, some horizontal): our musical culture, our knowledge of instruments, our goals can be so different.
Having spend my life recording Rock and Jazz instruments with mikes at few centimeters, what i can have in common with an opera fan, listening to classical big orchestras in huge halls 50 meters away from the instruments ?
Last edited:
Exactly.
Yes you can. This is fundamental and essential to eliminate bias.
Without this any tests of consistency and statistical analysis are pointless.
No reputable scientific journal in the world would accept such results.
You may be convinced, but it's worthless to me.
If you ever do real tests then please send me an email.
Best wishes
David
Is anything really scientific in hifi musical reproduction ?
Virtual reality, make believe, special effects. A pinch of salt, a pinch of science, a pinch of pepper 🙂
First comes theory - in science, that is usually the Math. Then design and engineering then tests and measurements. Using drug making/medicine for a moment... many tests are done long before the DBT and measurements of final product against the stated goal(s).
It is not appropriate at this stage to be talking of DBT/listening tests and statistical analysis and such.
Lets keep thinking about how we would best design a CMA to show or discover any special attribute.... then design, bulid and then test against what we thought.
[don't bother to even try to pick this apart - its a simple point is all]
Nor is this ever going to be submitted to the IEEE. So rigorous and thorough analysis isn't warranted, either. There are plenty of commercial IC makers who do that every day for a living. And they have the time, money, people and equipment to do it to the highest standards required. This is a design learning experience and eventually something to listen to music thru.
Thx-RNMarsh
It is not appropriate at this stage to be talking of DBT/listening tests and statistical analysis and such.
Lets keep thinking about how we would best design a CMA to show or discover any special attribute.... then design, bulid and then test against what we thought.
[don't bother to even try to pick this apart - its a simple point is all]
Nor is this ever going to be submitted to the IEEE. So rigorous and thorough analysis isn't warranted, either. There are plenty of commercial IC makers who do that every day for a living. And they have the time, money, people and equipment to do it to the highest standards required. This is a design learning experience and eventually something to listen to music thru.
Thx-RNMarsh
Last edited:
Hi Waly --- I Need music -- it helps keep me balanced..... And, good music reproduction does that best for me. The more real, the better it is for me.
Thx-Richard
Thx-Richard
Last edited:
Anybody ever heard of the WFA (wattage feedback)class Q with positive and negative feedback and input, with no distortion that is detectable in the audio spectrum? Professor Goldilocks designed it to be JUST RIGHT. 
Electronics is an exact science and so are power amps. It gets a little annoying when that's an oversight to people trying to conceive hypothesises and theories the scientific way. But no, that gets thrown out the window when it comes to discussing CFA subjective qualities. Remember, all the CFA's have to deal with an OPS, mooting much of the SR advantage and bandwidth.
One argument repeatedly said is how many folks have listened to and declared the VSSA to sound better. Playing the Devil's Advocate, ofcourse. It's a very easy amp to DIY and many have built it because of it. Throw in a bit of hype and the fact that each DIY'er is going to appreciate his build you have a receipt for a claim that "hundreds are convinced of CFA's superiority in all kinds of subjectivist's descriptions"/
One argument repeatedly said is how many folks have listened to and declared the VSSA to sound better. Playing the Devil's Advocate, ofcourse. It's a very easy amp to DIY and many have built it because of it. Throw in a bit of hype and the fact that each DIY'er is going to appreciate his build you have a receipt for a claim that "hundreds are convinced of CFA's superiority in all kinds of subjectivist's descriptions"/
I am sorry if the short description of my listening experiences with different CFA topologies did upset you, I was only trying to be helpful... I have a long version (about 10 pages), send me an email if you are interested.
Numbers only ?
Hey, Richard -- me too. I listen to music all day long, every day and everywhere.. home or car or garage. background - foreground. CD, download from Internet, XM, FM. Grew up with music in the home... piano, singing, records playing. Guess its important to me in some way.
But then, its really Just some sort of an art thing. Do you think? Or, is it really all a numbers thing?
-RNM
Hey, Richard -- me too. I listen to music all day long, every day and everywhere.. home or car or garage. background - foreground. CD, download from Internet, XM, FM. Grew up with music in the home... piano, singing, records playing. Guess its important to me in some way.
But then, its really Just some sort of an art thing. Do you think? Or, is it really all a numbers thing?
-RNM
Recap -- again --
Much of this is true. But it doesn't matter if it is or not... at least not to me... some can sort it out and some cannot. Just keep your eye on the ball and follow the intent of the thread and keep THAT part moving; Designing a really good CFA.
They both sound good. VFA and CFA. Each can choose one or the other or both as each wishes, if they like. Lets design and learn what and why the CFA is different and how and what to do about it. The in's and out's of it as much less has been written for audio about it.
Thanks... can we move on now? Lets boost the SR and see if we can do that and keep all else approx. equal.
-RNMarsh
Much of this is true. But it doesn't matter if it is or not... at least not to me... some can sort it out and some cannot. Just keep your eye on the ball and follow the intent of the thread and keep THAT part moving; Designing a really good CFA.
They both sound good. VFA and CFA. Each can choose one or the other or both as each wishes, if they like. Lets design and learn what and why the CFA is different and how and what to do about it. The in's and out's of it as much less has been written for audio about it.
Thanks... can we move on now? Lets boost the SR and see if we can do that and keep all else approx. equal.
-RNMarsh
Last edited:
Helas, no. So many amps, all different. Nothing in human technology is an exact science. But we try.
???
Fet power devices are far to be the most limiting part of a power amp, regarding bandwidth (>20MHz) and slew rate.
Try doubling it as a goal for CFA.
We have already gone down the THD/FFT route.
Will high SR separate the VFA and CFA differences? How hard or easy is it (and stay stable)?
Will we find any greater separation between the two topologies?
-RNM
Last edited:
A bit off topic here but a question for Richard or Christophe perhaps. Is there a point in these power amplifiers where the output power becomes the dominant factor, ie. at some point does the available output devices favor one of the topologies over the other or are they still comparable no matter the upper limit to output power? Is there a break point where one topology starts to have an advantage over the other? We have mostly been talking about 50watts @8ohms output and I am wondering what happens when we start looking at say 200watts@8ohms as a realistic level. Does a triple ops in a vfa sound the same as a triple in a cfa?
Push it harder -
Good question and to one of the main points to be discovered from this "exercise'. We need to push the topology harder than 50W.... to see quicker and easier what falls out of this. I have no idea why that number was chosen - other than to compare against already made VFA, perhaps. No amp at 50W is useful to me because my speakers are low efficiency. It is also more likely to run into compression/overload/clipping/higher thd etc. My own test data example that I put up was well over 200W at the test equipment limit of about -100dB. [however, I can now measure with accuracy and confidence to much lower levels - but don't have the amp here at the moment].
Lets find the break points.... I suggest push the SR as one of those break points. Make the 50W scalable to higher power levels and take a look.
Thx-RNMarsh
Good question and to one of the main points to be discovered from this "exercise'. We need to push the topology harder than 50W.... to see quicker and easier what falls out of this. I have no idea why that number was chosen - other than to compare against already made VFA, perhaps. No amp at 50W is useful to me because my speakers are low efficiency. It is also more likely to run into compression/overload/clipping/higher thd etc. My own test data example that I put up was well over 200W at the test equipment limit of about -100dB. [however, I can now measure with accuracy and confidence to much lower levels - but don't have the amp here at the moment].
Lets find the break points.... I suggest push the SR as one of those break points. Make the 50W scalable to higher power levels and take a look.
Thx-RNMarsh
Last edited:
Oh yes, but not easy: we have done the demonstration that we can achieve the same level of distortion in both topologies.
Dadod did the ultra low HD in his CFA by adding devices. Each one brings its own phase shifts at HF. That we need to compensate to keep the loop stable... limiting slew rate...
Speed ... distortion... where is the good balance ?
Of course not. Apart input stage, means the way the feedback is subtracted from the input signal, there is no need for any other difference. You can use exactly the same VAS and OPS in both topologies. That's why it is easy to transform a VFA in CFA. A good way to feel the subtle difference between the two. The only limitation will be power supply-side: As, nativaly, the long tailed pair in VFA provide better PSRR, you have to ensure rails are clean enough in the CFA to afford the extra gain of a high power amp without audible noise.
Most of the CFA amps i know, including mine, are in the 200W range. That we can bridge...
This said, looking for very very high powers (1000 watts or so), the driver and output devices we can find to afford both the high currents and voltages have a luck to be slower. Limiting the slew rate, so, on my opinion, the main advantage of CFA. But who care, apart exotic low efficiency speakers, it will be for PA.
Last edited:
Richard and Christophe,
Thank you for the answers. Richard I understand the need for more than 50 watts with many multi-way speakers with low efficiency and passive crossovers but 1000 watts seems to be out of the norm for any home system I can think of. If the slew rate of the output section dominates in power amplifiers what would be the advantage to chase higher SR in the first two sections of the amplifier? Would this be for more local feedback in these sections and attempting to linearize distortion values while not changing the overall gain margins?
Thank you for the answers. Richard I understand the need for more than 50 watts with many multi-way speakers with low efficiency and passive crossovers but 1000 watts seems to be out of the norm for any home system I can think of. If the slew rate of the output section dominates in power amplifiers what would be the advantage to chase higher SR in the first two sections of the amplifier? Would this be for more local feedback in these sections and attempting to linearize distortion values while not changing the overall gain margins?
There is no 'section' or 'stages' in a closed loop system. We separate and give names to them in our descriptions just for convenience. By example, the input stage can bring some gain.
Consider the finished amp like a black box with a given slew-rate. To explain the things in a very simple way, there is an inertia witch is the addition of all the inertias that the signal encounters. Mainly due to parasitic capacitances that takes time to be charged.
To give an obvious example, take a FET power device. It has a large gate parasitic capacitance. Lower the output impedance of the driver stage, faster it will switch.
The input stage ? It is where error corrections occurs. Did-you want to have delay here ? This is where the CFA is different* from VFA.
* Diplomatic enough ? 😉
Last edited:
I'm following you Christophe. I understand it is the whole we are looking at and obviously with global feedback whether voltage or current controlled everything basically works the same way. I guess I have just fallen into the habit now of looking at the amplifier as sectional parts instead of holistically looking at it. I understand about the parasitic capacitance's and imagine that the inductance's are just as likely to cause other problems. What is considered the fastest output type transistor seeing as that is the limitation with slew rate?