SE distortion is higher since a single-ended (unbalanced) system produces odd and even harmonic distortion. A balanced system (i.e. Push-Pull) cancels even order harmonics. If you look at the Fourier Transform of a nonlinear system, you have a power series, half of whose components are to even powers and the other half to odd powers. Raise a term with a 'j' in it to an odd power, and you get an odd number. Raise it to an even power, get an even number. When you subtract, you lose all the even terms.
Another way to look at it is that an unbalanced system can produce an asymmetrical waveform that contains even order harmonics. A balanced system can make symmetrical waveforms only, and these are all made of odd harmonics.
THD figures for VTs are but rough figures of merit. Distortion is not created the same by any means. You can see that on an o'scope screen: add the second harmonic of a pure fundamental, and the disturbance is very difficult to see until you have in excess of 20% H2. Add higher order harmonics, and you have a greater disturbance at lower amplitudes. Seen this for myself: PP 807 finals made very little harmonic distortion in comparison to the output, but the residual after nulling out the fundamental with a Twin-T filter looked almost like a sawtooth at three times the frequency. That sounded nasty. In 1938, the developer of the type, O. H. Schade, wrote about this, and recommended employing LNFB. He was right about that, and that's the reason.
Types like the 6V6 or 6BQ6 made more distortion, but the residual looked almost like a pure sine -- much less nastiness. These types require just the assistance of gNFB to take off that "edge".
So far as SE amps, it depends. Some will make mainly the less detrimental H2, but others may not. It's a matter of VT type, loadlines, and how well the design was done.
SE is limited to low power since the OPT has to carry the bias current. This leads to DC core magnetization, and the saturation problems that brings. It is more demanding to design a decent SE OPT than it is a PP OPT for that reason. It's a nasty circle that feeds on itself like an ouroboros. Introduce an air gap to control flux density, inductance drops, replace the inductance by adding turns (and stray capacitance) but magnetization increases, widen the air gap, lose inductance, wash, rinse, repeat.
You can get away with parafeed, but that just transfers the core saturation problem from the OPT to a plate load choke. Or replace the choke with a SS gyrator, but that makes for more heat (and big heat sinks since SS doesn't like heat) and the gyrator (it "fakes" only the opposition to current changes of inductance: it can't store energy) can't allow the plate voltage to rise above Vpp, so you'll need a higher voltage source.
"I want to parallel two 6550 to produce @40 watts with a 1.5K impedance. Is there something in SE design fundamentals that makes this a bad idea?"
How it works is highly dependent on the quality of the OPT. It's certainly easier to match the load to 1K5 than it would be to match it to a much higher impedance. Still, it ain't gonna be cheap by any means if it has any claims to sonic excellence.
Please don't say that SE distortion is higher without mentioning on which power. The lower is the power, the lower are distortions, going to zero. What is attractive in SE, when some direct heated triodes are used in proper regimes people don't hear their distortions like they hear distortions of some other topologes because dynamic character of their change with level reminds change of distortions in the physical world that we filter out, ignore, don't hear. As if they don't exist. And indeed they don't exist as distortions, they are cues to the surrounding.
Hi DF96, I think that if you have distortion (harmonic) you will have music that is not in the original musical program.
In SE amps, is a phenomena with even harmonics. in jazz, solo guitar with few instruments probably give "that velvet sound" But if you listen Big orchestras and symphonic music IE Bruckner you will generate an endless of harmonics that at the end this gonna be noise,
I listen music every day that is the clue fell any difference. Correct me if I am wrong please, is my IMHO and you are a wisdom and active writer in DIYAudio.
Best Regards from Menorca.
I didn't say I believe those implications. I was just struck by how uncommon it is for an SE fan to admit them publicly. I am not an SE fan, although I accept Wavebourn's point about power level. Other things being equal, an SE amp ought to have vanishingly small distortion at low power levels. Not always true for P-P.
To add to what Wavebourn mentioned. Even order harmonic distortion is more "natural" sounding. On a string instrument we actually get even order harmonics every time a note is struck IF the string is in proper tune (fundamental tone) . Odd order harmonics happen when the string is NOT in tune (with the fundamental tone), which causes the listener to notice the out of tune string or odd order harmonics. The even order harmonics from a fundamental tone sound "natural" to human hearing and our brain can ignore the even order harmonic which some call even order harmonic distortion. Odd order distortion sounds "un-natural" to human hearing and stands out similar to a string (or person singing) that is out of tune with the rest of the music.
SE tube amps generate even order harmonics (notice I left out the term "distortion") more predominantly VS. PP amps that cancel out the even order harmonics and generate odd order harmonics which sound un-natural to human hearing/brain.
When reproducing music we dont want to add any odd order harmonics, even order harmonics as mentioned will in most cases be ignored by our brains but not the odd order harmonics which are what is usually referred as "distortion" (for the sake of discussion I call it distortion).
Of course some folks brians work different and can hear even order harmonics. I suspect those are the folks that dont like SET amp sound.
Hope that helps....
cheers,
Bob
People often talk about soundstage. What soundstage? The recording was close-mic'ed! Vast majority of modern recordings are. You're not supposed to have soundstage! "Accurate" playback simply sounds unnatural. Adding a bit of 2nd order distortion creates a sense of space, as this is exactly what air adds as sound waves transverse to our ears. A SET amp done correctly simply sounds natural. You can enjoy the music forever.
I did not mean that even order distortions are natural sounding. Try to add even mode distortions that are independent on loudness, or even worse when more distortions on low loudness than on high, and it will sound bad. What I meant, when distortions grow with loudness some transfer curves result in less audible distortions than other transfer curves.
I'm no expert on the physics of musical instruments, but if I understand you correctly then what you say is wrong. Changing the tuning of a string will not change to any significant extent the overtone structure; they will all change in frequency of course but the pattern will remain the same.bobrown14 said:
The second harmonic may be tolerable because it is simple an octave higher, but second harmonic from an amplifier will be accompanied by second-order IMD which will not be so nice.
I can organize you quite real soundstage using recording console with digital stereo reverberator. Also, I know about experimental digital console that modeled different natural environments with possibility to place sound sources in that environment, in different positions. Reflections from different digitally synthesized walls, including diffused reflections, can fool imagination pretty well.
bobrown14 said:
@DF96, I was about to post the same response.
@bobrown, You imply that when a string is put in tune (whatever musical scale is used) the harmonic structure will suddenly change from predominately odd to even. This does not make sense. Do you have a reference for this statement?
If a string is not in tune it will generate dissonant beat tones with *other* strings and notes on other instruments, which sounds bad. This has little to do with harmonic over-tone structure of a string. A musician can actually manipulate the over-tone harmonic structure depending on how a string is plucked and where along the string it is plucked.
Yes, it is a fundamental design limitation. Expect about 20 - 30% effeciency for SE Class A. The same rule applies for SS. The efficiency can be pushed a bit higher at the expense of a severe increase in distortion.
Hi Bob, Dear, This is and old story about differences between SE vs PP and all the same information that however justify use VT and SE when people compare them with Solid State. But this is the short story. The long is more complex at all. But at the end. IE one "C" in Clarinet is different at our ears with the same frequency "C" in a flute. Why, because the harmonics give the timbre. One Stainway grand piano sound different from a John Broadboard & Sons. Why because the harmonics. When an organ in an only register open the more and more tubes and sound different if compare it with an organ noising one register alone. If one "C" in a soprano voice you rerecord over by playback the same singer in the same note "C" in the same frequency, the sound going to be different. And when you put an amplifier that produce even harmonics in a determinate level you gonna listen different that the original source of sound. Imagine that in one or two instruments, is easy and "velvet" friend sound. But when you listen Stravinsky fortissimo with much dissonant arpeggios or accords that the composer decided to use an "A" with a "C#" with one pair of tubas in E. That rising the level of current in the anode of your poor WE300B that only somebody told that only produce even harmonics. I can say today that the sound of this fortissimo gonna sound ****. It contradicts nothing of the expressed for many of us or them but of any manner explain the complex and the expensive (both: money and energy) that you need for listen symphonic music across of a SE amplifier.
Now maybe you need a couple of GM70 in A2 with very good transformers and probably with intertranies too. Look and design a driver with monster amplification and without distortion. Wawac is a good answer for this question. But the pricey of that kind of amps is prohibitive for many of all. Maybe you can run your own transformers, but such skill we need. And at least in one double blind listen test against a AB1/ transistors amp, we can find the difference.
Best Regard from Menorca
As far as I can see, some basic facts have not yet emerged in this discussion.
1. The distortion harmonics produced depend on the transfer function of the tube. Triodes tend to product predominantly 2H with the other harmonics falling off smoothly. Pentodes tend to produce more 3H than 2H due to their differing transfer function.
2. Bias point makes a difference to the amount of distortion produced but has little effect on the relative amplitudes of the components.
3. The amount of distortion produced depends on how much of the transfer function is traversed. Distortion in tubes is therefore almost directly proportional to output signal level.
4. Push pull cancels much of the even ordered harmonics, 2H, 4H etc. The remaining spectrum therefore tends to be predominantly odd harmonic components.
Cheers
Ian
1. The distortion harmonics produced depend on the transfer function of the tube. Triodes tend to product predominantly 2H with the other harmonics falling off smoothly. Pentodes tend to produce more 3H than 2H due to their differing transfer function.
2. Bias point makes a difference to the amount of distortion produced but has little effect on the relative amplitudes of the components.
3. The amount of distortion produced depends on how much of the transfer function is traversed. Distortion in tubes is therefore almost directly proportional to output signal level.
4. Push pull cancels much of the even ordered harmonics, 2H, 4H etc. The remaining spectrum therefore tends to be predominantly odd harmonic components.
Cheers
Ian
Everybody says this. Is it true? It might be better to say that a pentode produces more 3rd than a triode with the same output, but in both cases there is likely to be more 2nd than 3rd. Even if the oft-heard statement is true at full output, it won't be true at lower signal levels because 3rd usually drops faster than 2nd.ruffrecords said:
Not usually. 2nd percentage distortion is proportional to signal level; 3rd more like signal^2. However, I agree that much discussion of distortion seems to completely ignore the fact that distortion is almost always strongly dependent on signal level.
Here is a document with some interesting FFTs. They are of a 300B output showing the spectral response.
Plot 1 is particularly interesting as it shows (1) second harmonic is less than 3dB greater than 3rd Harmonic, (2) there are measurable harmonics out to 24th shown, and they probably exist beyond that.
Plots 4 and 5 are more in line with traditional thinking in that 2nd harmonic is greater than third. However, the high order harmonics are still present.
The driver is a 6SN7, so there is no doubt that Triodes generate odd harmonics.
http://www.fetaudio.com/wp-content/uploads/2003/02/FFT-300B-final.pdf
Plot 1 is particularly interesting as it shows (1) second harmonic is less than 3dB greater than 3rd Harmonic, (2) there are measurable harmonics out to 24th shown, and they probably exist beyond that.
Plots 4 and 5 are more in line with traditional thinking in that 2nd harmonic is greater than third. However, the high order harmonics are still present.
The driver is a 6SN7, so there is no doubt that Triodes generate odd harmonics.
http://www.fetaudio.com/wp-content/uploads/2003/02/FFT-300B-final.pdf
Did you draw graphs learning Algebra and Trigonometry? It is the essence of understanding of distortions.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.