I agree, that there might be audible differences. But...
...as an engineer I strongly believe, that what can be heard, must also be measureable.
What can be measured, can also be compared and benchmarked, and serve as a criteria to base a decision on.
As the standard test-procedures all are based on resistive-load testing, they are good to compare amplifier's characteristics. But they might not be sufficient to determine, if an amp "sounds" good. IMHO an amp should not sound, it should reproduce. If there are differences in sound with a real loudspeaker, there consequently must be a measurable difference in one ore more metrics.
I am trying to identify sound-relevant metrics of an amp with a real loudspeaker load using spice as simulation tool. I neither have the time nor the money to build and listen to all possible variations, but I can simulate all of them and try to correlate simulation results with the subjective perception of the people of this board/thread with the various IPSs and OPSs of the slewmaster rumble.
I have implemented and tested a loudspeaker-model, which can be parameterized to different loudspeakers and enclosures, and a growing set of IPS/OPS.
My project is ongoing and I am looking forward to the discussion with you.
The experience of the people of this board are highly welcome. I need an evaluation on which IPS in this thread were considered good or less good, ranking 10 (very good) to 1 (lousy). So I guess a valid range should be 5 to 10.
Please give me your support.
I have used the KND IP(CFA) stage with a mosfet OP stage. I am very happy.
I have build the Spooky IP stage as well, which is a VFA, was difficult to tell the difference but chose the KND because of good mids and highs.Nothing wrong with bass but difficult to choose !
Maybe it is just measurable, not hearable.
I have build the Spooky IP stage as well, which is a VFA, was difficult to tell the difference but chose the KND because of good mids and highs.Nothing wrong with bass but difficult to choose !
Maybe it is just measurable, not hearable.
Last edited:
+1000
I recommend to you to listen amplifier with low and high slew rate. Example: Full power bandwidth is 50kHz and more than 500kHz.
Second, try to compare amplifier with 20kHz distortion high and low. Example 20kHz THD is around 0,1% and below 0,001%.
Third, try to make H2 distortion dominant, try to make H3 distortion dominant, try to make harmonic profile monotonic.
Please tell your impression and what your taste.
Dynamic distortion like memory distortion is difficult to recognize because it happen in some condition only.
Three good points. I am going to identify metrics for all of them.
My expectation:
1. CFA should have superior slew-rate.
2. THD @20kHz might show this advantage in slew-rate (this is a good proposal)
3. H2 vs H3; I will hopefully find it in the simulation results.
4. Dynamic Disto (IMD and TIM) might need a specific procedure to be measured. Proposals are very welcome.
@Krisfr & bimo:
Personally I do not believe in private listening tests; they are based on personal perception and can differ from person to person, day to day, loudspeaker to loudspeaker, room to room.
And last not least I fear, I might not notice any significant difference under my personal listening conditions. Instead I believe in characteristic metrics.
Only what I can measure, really exists. All other mostly lead to semi-religious discussions with sometimes personal offencive behaviour. I am not part of that.
So I wouldn't recommend to build a jumper-able amp (especially not for my purpose).
In addition it might induce disturbance and noise.
Personally I do not believe in private listening tests; they are based on personal perception and can differ from person to person, day to day, loudspeaker to loudspeaker, room to room.
And last not least I fear, I might not notice any significant difference under my personal listening conditions. Instead I believe in characteristic metrics.
Only what I can measure, really exists. All other mostly lead to semi-religious discussions with sometimes personal offencive behaviour. I am not part of that.
So I wouldn't recommend to build a jumper-able amp (especially not for my purpose).
In addition it might induce disturbance and noise.
Last edited:
It possible use VFA with different compensation. Because it is difficult to get low slew rate in CFA. In VFA, it can use ordinary Miller compensation, may be slightly over compensated.
But sepp2gl do not like it.
You're assuming we know how to measure every metric that makes a great sounding amplifier. Sadly, I don't think that is the case. If it could be done, we'd have already single "perfect amplifier" design. Listening tests still have there uses. Unless you're designing a commercial product, in which case the numerical representation in the sales brochure is important.
@brian92fs:
No, not at all. I don't even assume, that we know all the metrics, that matter for sound.
I am currently spending a lot of effort to benchmark a variety of power-amps in order to find metrics, that separate "winners" from "loosers" by means of spice simulation. Within this project I question commonly known metrics and try to find new metrics, that might provide a better separation. So I am a searcher myself and it is not about any commercial amp, that you can buy anywhere. It is about THE amplifier, that I would design for myself.
And finally: yes, there must be a listening contest to correlate technical metrics with audible perception.
Coming back to this thread: if there are VFA and CFA input stages, with different technical characteristics, there should be one of those to be superior. And this should be reflected in specific metrics. This should not be a matter of like or dislike, believe or religion.
If this is the today's horizon, then I try to look beyond.
No, not at all. I don't even assume, that we know all the metrics, that matter for sound.
I am currently spending a lot of effort to benchmark a variety of power-amps in order to find metrics, that separate "winners" from "loosers" by means of spice simulation. Within this project I question commonly known metrics and try to find new metrics, that might provide a better separation. So I am a searcher myself and it is not about any commercial amp, that you can buy anywhere. It is about THE amplifier, that I would design for myself.
And finally: yes, there must be a listening contest to correlate technical metrics with audible perception.
Coming back to this thread: if there are VFA and CFA input stages, with different technical characteristics, there should be one of those to be superior. And this should be reflected in specific metrics. This should not be a matter of like or dislike, believe or religion.
If this is the today's horizon, then I try to look beyond.
That was to interface with the 21st century protection system. How to build a 21st century protection board
One thing simulations and measurements will miss is how an amplifier interacts with a speaker connected to it. This can and does change measurements. I think this is a big part of the listening / measurement argument. This is also why I don't think there ever will be an ultimate winner in amplifier designs, it depends on your speakers and personal tastes.
Perfect reply! I could not agree more.
That's why, I use a linear loudspeaker model in spice as a third load, in addition to the typical resistive loads (4/8Ohm). And yes, it does matter and it can be reflected in the metrics.
Currently I use a wide-band speaker from Omnes-Audio with a closed-box enclosure-model, so I don't need to bother (yet) about passive cross-overs.
With a real loudspeaker-load I hope to find more separation in the metrics of amplfiers.
Last edited:
You should listen to various speakers, like I did. It is not my speakers, but my friends speaker. Luckily, they have very good speakers and it is very expensive for me. Even, one of my friend have many speaker and his room treated acoustically.
when few ns of jitter are decisive for sound quality, measurements are complicated.
The main sources are:
Capacitances in non linear voltages (Vbe and others inside the circuit)
Varicap (Cob and other)
Temperature in power devices.
We need to take control about these issues in each amplification device (stage) in the signal path
Global feedback can do nothing , it can even make it worse.
BR
The main sources are:
Capacitances in non linear voltages (Vbe and others inside the circuit)
Varicap (Cob and other)
Temperature in power devices.
We need to take control about these issues in each amplification device (stage) in the signal path
Global feedback can do nothing , it can even make it worse.
BR
Last edited: