Na Harbeth forumu nađoh interesantnu duskusiju Administratora (konstruktora i vlasnika Harbeth) gospodina Alan Shaw, o razlikama između cevnih i pojacavaca sa tranzistorima, prilagođenu rekao bih laicima. ————————————- Amplifier specifications - our very best friend. An explanation for the non-technical teenager. Rather than return to the office after our break, I'm going to have one, hopefully final attempt to convey the essential point about amplifier sonics. It's a fair point reported earlier thin this thread that 'ordinary folk can't understand the Stereophile technical graphs', so let see if we can demystify them. CAVEAT: I have used as data the two links that were posted by ordinary contributors in the thread above. I could have gone hunting for any number of alternatives, but this saves time. So I cannot say, and will not say, whether these are typical of what's available in amplifier land nor pass comment on their designer's philosophy or commercial acumen; I'm just recycling what went before. We start with cleaning up the printed technical graph of the tube amplifier (Graph A) to simplify for the non technical reader. It looks like this now: I've added underneath a piano keyboard for those more familiar with musical notes than technical frequencies. Ok so what? We see a line (a graph trace) that wiggles up and down as it traverses the piano scale, and beyond. Remember: there are many pure tones (fundamentals) beyond the highest keys on the piano, and a few below the lowest leys, so in terms of pure tones, the piano only covers a range of about 20Hz to 4kHz, when the audio band has long been established as 20Hz to 20kHz for high fidelity sound. Other instruments 'take over from' the piano's upper range and use the higher ranges of the audio spectrum (violin, brass etc.) which is just as well as it gives the composer a wide tonal tool box to work from. So let's look at the vertical axis, marked on the left as being in dB (decibels). (I'm a bit confused by the little A shown; I doubt that it means that axis is in dbA weighting when measuring an amplifier, so suspect that it refers to the audio analyser's channel A). It's really handy that audio graphs almost invariable are scaled in dBs in the vertical axis (and frequency in the horizontal axis) because regardless of the make of audio analysis equipment that was used in the lab, we can make prefect visual comparisons of what is actually columns of numbers that lie behind the graph data. Sometimes though, we have to visually squeeze or stretch the graph to be able to compare apples with apples, but the beauty of working in Y axis dB and X axis frequency is that it is 100% legitimate to do so. But ONLY when we are working in dBs and Hz. Now we have our Graph A of our tube amp, we can present the graph of the solid state amplifier mentioned above, Graph B. If we take a snapshot of the graph, it looks like this: We have a problem. The visual appearance of Graph A and B is so radically different that we are in grave danger of misinterpreting any useful technical information contained therein. Even though, as you will see from the little Ap symbol in the top right of the graphs that the same Audio Precision analyser has been used. What we are witnessing is two different generations of Ap equipment with very different graph presentational capabilities (over twenty years part). So, to be in a position to safely compare these two, we need to stretch and squeeze one of the graphs relative to the other to make them exactly visually comparable. Remember - this is only legitimate if we work in dBs and Hz, as we are. It is not a cheat - it does not change the underlying data, but it corrects the different aspect ratio and scaling of the graphs for the human eye to make sense of the information. So having done that, I can now place Graph A and Graph B side by side, confident that they are displaying information that is visually cross-comparable, here: Now that I have carefully adjusted the axis of B to be identical to A (as best as I can, visually) we can present them together. If we are more curious about the way the amps cover the horizontal frequency range, we might wish to place A and B one over the other like this: If we are more curious about how the dB (loudness) of the two amps v. frequency, then we might present their graphs side by side like this: Whether we prefer the one over the other or the side by side presentation is a matter of choice - I prefer the side by side graph immediately above. And yes, Graph B does now have a visual aspect ratio that is very much different from how it was originally but we can prove that the scaling is exactly comparable with Graph A, here: (I'm showing the vertical scale cross-comparison - you can verify the horizontal one for yourself from the above graphs): So now we are at the point that we can make proper visual comparisons between the performance of these two amplifiers. Is it immediately apparent that as far as B is concerned, it's output is exceptionally flat across the audio band and far beyond it, and that A's output is greatly effected by the speaker load? So now we can consider for A, which frequencies (or octaves, or part-octaves) are going to be boosted or attenuated across the musical scale. The graph tells us that unambiguously. Again, lucky for us that the vertical scale is in dB, and by convention, such graphs are usually presented centred around the 0dB line. That makes it really easy for a human to see clearly what the deviation is across the audio band. So I've drawn in the zero dB line in orange, here: Now we have this line - and as you can see, the entire response of SS amplifier B traces the ideal 0dB line across the audio band - we can colour in which groups of notes will be made louder to the listener and which notes quieter like this: Now the picture is clear. Some, but not all, musical octaves will have enhanced loudness when this amplifier drives a typical speaker load, and some will have reduced loudness. 'Loudness' is always associated in the human brain with predictable and long understood subjective sensations, so knowing those, we can reliably anticipate the subjective sensation that this amplifier would create in the listener's brain when it is driving the typical speaker load. When driving another speaker load, or even no speaker at all and just a straight resistance, then very different results may occur. We can attribute some generic audiophile adjectives (plenty to chose from) and identify them: Considering that the subjective experience is so well understood by audiologists, it could be anticipated that a similar subjective outcome could be synthesised under lab conditions to guinea pig students by routing the audio through a graphic equaliser and into Amp B. Of course, as with all matters in hifi one man's preference is another's detestation, but Stereophile have given us the tools to make an educated stab at how any amplifier is likely to modify the audio spectrum when driving a real-world load. Which you chose is entirely up to you. This has taken a morning - a hope you find it can guide you better to neutrality or otherwise in your amp selection.
