I agree, but it won't likely happen that way. While speaker impedance follows the design, it isn't part of the design. A crossover assumes a voltage source, one less thing to worry about. A magazine reviewer will come along and say the amp did or didn't spit cogs when they connected it.
Designing for anything else doesn't make too much sense if you consider why the impedance is non-flat, and the logistics of requiring a specific finite source impedance but leaving the resultant impedance as it is, well, what a load to dump on an amp (and amp designers
).
Designing for anything else doesn't make too much sense if you consider why the impedance is non-flat, and the logistics of requiring a specific finite source impedance but leaving the resultant impedance as it is, well, what a load to dump on an amp (and amp designers
).
Last edited:
It is not too difficult for an amp designer to provide any desired output impedance; it just requires the right mix of current and voltage sensing feedback. Of course, amps without feedback will just have to present whatever output impedance happens to emerge from the circuit. A crossover does not need to assume a voltage source, although this does simplify the design.
Given that designing for a voltage source is fairly straightforward for the speaker designer, and for the amp designer, it seems strange that some people prefer something different.
Given that designing for a voltage source is fairly straightforward for the speaker designer, and for the amp designer, it seems strange that some people prefer something different.
Those who build amps without global feedback seem to prefer simplicity. Some say they have no choice in some respects, I think that's beside the point eg: that simplicity means ease of design, I think it means a sensible load even if that means more components. As simple as possible but no simpler, understand the beast.
They are smart to go for simplicity because in no feedback s.e.t. design,(I believe this is the only amp which can work 'ok' without feedback) everything in the amplifier will be audible. The more components the more distortions.
You can hear the quality of cathode bypass caps and coupling caps to a lesser extent. Anode resistors and power supply resistors can be heard so the madness never stops with s.e.t.
The irony is that s.e.t. aren't good with micro-dynamics and since the rest of the sound is excellent, the owners can modify and spend $$$ on amplifiers interminably.
I read everywhere that the are better at micro detail. Sure they play better at lower volume levels but my personal experience was not what I can read on the websites.
imo:
s.e.t. have too much harmonics to blur the remaining low levels sounds during dynamic passages and high intermodulation. Dynamics comes at the expense of high harmonics and an uneven negative and positive phase (lot of 2nd harmonics). Low bandwidth reduce the resolution of the music too and you cannot hear any difference with poor audio sources and hi-res recordings.
s.e.t. transformers needs a bigger perimeter of lamination and this increases the hysteresis proportionally. As to if the standing class A current helps or not this is a complicated discussion. Lots of eddy currents because of the bigger mass. By comparison a class AB transformer runs very cold and is small.
Extreme inefficiency of the anode power transmission to the output. Usually high impedance tubes (the bigger the bottle also the higher) need also transformers with lot of inductance and non-linearity.
I believe that a good class AB has more details contrary to the opinion of many.
It's been a little while, a person has to look at things like the curves of Va versus Rp, gm and mu and select good quiescent conditions, as well as provide good dynamic conditions. I don't read much into interweb stereotypes. I don't particularly like distortion.
This is a popular but completely false myth.gabdx said:
The reason SET works at all is the inherent feedback of a triode, which serves to partially linearise it despite the relatively heavy load seen by the output stage. Feedback is absolutely essential and largely unavoidable; the choice is between built-in feedback or extra feedback.
This is a popular but completely false myth. :
A: I don't agree, there is many trials of tube and solid state topologies where additional stages or components only add up distortion components to the final output which null any benefit from adding such devices. One of the most discussed are simple triode input stage vs. SRPP ; addition of cathode followers to drive output tubes and many more.
They measure worst in most topologies.
--------------
The reason SET works at all is the inherent feedback of a triode:
A:
The only positive advantage that I see in S.e.t. is the loading of the anode by a transformer which maximize the tube voltage swing, like a choke loading. This is to me the most contributing factor to why the s.e.t. can be liked.
A triode for example a 6BL7 PPP loaded with 5k has a gain of 12.3 instead of 15. A pentode prevents that loss of amplification/local feedback. A triode is always inferior to a pentode working with global feedback.
The mechanism of local feedback in the triode is a poor way to correct non-linearity.
Triode natural feedback climb with the (mu) parameter but never enough to properly linearise (as you said) the output.
Feedback is avoidable , it is called a properly designed tube : pentode.
Triode feedback can be reversed by positive feedback.
In s.e.t. the bandwidth is so poor that proper amount of feedback cannot even be used and makes the amplifier clips hard.
I only see disadvantages in the triode in terms of local feedback vs a pentode with more gain and power!
Triode strapping any pentode is a wrong doing and completely miss the reason why pentode were invented.
A: I don't agree, there is many trials of tube and solid state topologies where additional stages or components only add up distortion components to the final output which null any benefit from adding such devices. One of the most discussed are simple triode input stage vs. SRPP ; addition of cathode followers to drive output tubes and many more.
They measure worst in most topologies.
--------------
The reason SET works at all is the inherent feedback of a triode:
A:
The only positive advantage that I see in S.e.t. is the loading of the anode by a transformer which maximize the tube voltage swing, like a choke loading. This is to me the most contributing factor to why the s.e.t. can be liked.
A triode for example a 6BL7 PPP loaded with 5k has a gain of 12.3 instead of 15. A pentode prevents that loss of amplification/local feedback. A triode is always inferior to a pentode working with global feedback.
The mechanism of local feedback in the triode is a poor way to correct non-linearity.
Triode natural feedback climb with the (mu) parameter but never enough to properly linearise (as you said) the output.
Feedback is avoidable , it is called a properly designed tube : pentode.
Triode feedback can be reversed by positive feedback.
In s.e.t. the bandwidth is so poor that proper amount of feedback cannot even be used and makes the amplifier clips hard.
I only see disadvantages in the triode in terms of local feedback vs a pentode with more gain and power!
Triode strapping any pentode is a wrong doing and completely miss the reason why pentode were invented.
If extra components or stages cause distortion then try omitting them from an existing competent circuit. You will quickly find that the result is far worse.
Pentodes are less linear than triodes because they have no internal feedback mechanism. The feedback mechanism in triodes can provide excellent linearity under two conditions:
1. the triode is designed so that the electrodes are all the same shape (e.g. concentric, or planar) - older triodes may be better in this respect;
2. the anode is only lightly loaded, that is high impedance - achievable in small signal stages using a sufficiently high supply rail voltage (or CCS - more components!) but not in a power output stage so SET will always have distortion at significant output power.
Pentodes are less linear than triodes because they have no internal feedback mechanism. The feedback mechanism in triodes can provide excellent linearity under two conditions:
1. the triode is designed so that the electrodes are all the same shape (e.g. concentric, or planar) - older triodes may be better in this respect;
2. the anode is only lightly loaded, that is high impedance - achievable in small signal stages using a sufficiently high supply rail voltage (or CCS - more components!) but not in a power output stage so SET will always have distortion at significant output power.
My own experience I could hear the difference with silmic II and auricap.
In my phono stage I can hear the difference with sonicap and realcap,
Output stages there is a little difference I could tell with many types of capacitors and confirm some capacitor tests others did and published in books and internet.
If you happen to have a device with components outside feedback loop try to change resistors to carbon ones, you will hear a different sound.
You probably know that the majority of boutique parts customers own s.e.t amps.
In my phono stage I can hear the difference with sonicap and realcap,
Output stages there is a little difference I could tell with many types of capacitors and confirm some capacitor tests others did and published in books and internet.
If you happen to have a device with components outside feedback loop try to change resistors to carbon ones, you will hear a different sound.
You probably know that the majority of boutique parts customers own s.e.t amps.
That may be more an observation about the psychology of SET fans than about any intrinsic qualities of "boutique" parts.
Chris
How was the test procedure set up so that the aural memory wouldn't be compromised and visual / expectation bias wouldn't affect the listener's perception? Also, how did you ensure to listen at matched volume levels?
haha, I swapped capacitors from 0.1cent to 150$ and convinced myself I could hear something different.
I once saw a tube amp designer on diyaudio who argued that all tubes sounded the same. All el34 should sound the same, people make it in their heads that tube rolling it improves the sound.
Well I have no argument for that except invite you home for tea.
I don't believe in blind tests because our mood influence even more our taste. Critical listening takes years of experience and hundred of hours of listening.
I must not be the only one who found cheap wine very good surrounded by excellent peoples and interesting discussions.
Last edited:
This doesn't help the validity of your claim.haha, I swapped capacitors from 0.1cent to 150$ and convinced myself I could hear something different.
I once saw a tube amp designer on diyaudio who argued that all tubes sounded the same. All el34 should sound the same, people make it in their heads that tube rolling it improves the sound.
Well I have no argument for that except invite you home for tea.
What is your belief based on, another belief or confirmed test data?
Are you sure it's not months of experience and tens of hours?
Having acquired these skills, would you then be willing to have them assessed in a double-blind listening test?
Chris
Norman Crowhurst (perhaps not the first) showed that negative feedback which reduces the
percentage of THD, also causes higher harmonics to appear that were not there before the
feedback was employed.
Take a pure 1 kHz tone, send it through an amplifier (*) that has only 2nd, 3rd, and no other harmonic distortion products present.
Now, take a portion of the amplifier output, and apply it to the feedback input point.
This already distorted signal goes through the amplifier again, where it is additionally distorted.
2nd harmonic x 2 = 4th harmonic
2nd harmonic x 3 = 6th harmonic
3rd harmonic x 2 = 6th harmonic
3rd harmonic x 3 = 9th harmonic
Of course these new, higher order harmonics are at reduced levels, since they are created
by the original harmonic distortion products which are already less than the fundamental.
Remember, the amplifier will take whatever is presented to the input, and to the feedback input
as well, and will distort it.
Two or more tones that are passed through an amplifier will generate Intermodulation Distortion
products (IMD). And just like the harmonic distortion products that are re-applied as negative feedback, the IMD creates even more tones.
IMD is usually measured by using 2 steady state tones.
Now, I am not suggesting that negative feedback is a bad thing.
I am just saying that any design or topology does have some sort of tradeoff.
Later than Norman Crowhurst’s articles, Matti Otala (and also a Finnish person) wrote about Transient Inter Modulation distortion (TIM).
It showed that too much negative feedback could cause other problems, not just those
of steady state IMD.
Now, back to the original amplifier (*) that had a 1 kHz tone applied and only created 2nd and 3rd harmonics.
Take the family of curves of the output device(s), be it a Triode, Pentode, BJT, MOSFET, etc.
Apply a straight (resistive load line). Before you apply negative feedback, that is usually the
lowest numbers of harmonics you will get.
After negative feedback, you can reduce the size of those 2nd and 3rd harmonic distortion products. But now you will get higher order distortion products too.
Now apply a reactive load line. With or without feedback, there will be different distortion
levels. The reactance can be transformer primary inductance, distributed capacitance, or leakage reactance. It can also be a loudspeaker load as its reactance is reflected to the primary winding.
The resultant elliptical load will cause different distortion effects depending on whether the amplifier is single ended or it is push pull.
Traditional vacuum tube push pull is one thing; Totem Pole BJT and Totem Pole MOSFET are quite another thing.
(*) you can discuss whether any such amplifier exists, but from a practical sense, and with
small to medium signal levels, such amplifiers do exist.
The general case for triode, pentode, BJT, MOSFET, etc. is that in the well behaved region (small to medium signal levels) the distortion products are reduced by the following rates as the signal level is decreased:
Reducing the Fundamental, c, by 10 dB will cause the distortion products to reduce by the following amount relative to the new fundamental level
2nd order -10 dBc
3rd order -20 dBc
4th order -30 dBc
… and so forth
Isn’t it nice how upper order harmonics, and upper order intermodulation products fall off
rapidly as the signal level is reduced?
At some point as we reduce the signal level, we have to worry about hum and noise, they are larger than the upper order distortion products.
Examples:
An amplifier putting out a signal level that in its well behaved region has 2nd harmonic distortion
of -45 dBc. Now the signal level is reduced by 10 dB. The new 2nd harmonic distortion is -55 dBc.
An amplifier putting out a signal level that in its well behaved region has 3rd Intermodulation distortion of -50 dBc. Now the signal level is reduced by 10 dB. The new 3rd Intermodulation distortion is -70 dBc.
By the way, loudspeaker distortion also is lower when the signal level is reduced.
And our ear’s distortion also acts similarly.
Regardless of what amplifier topology, and what speaker, etc. you like, hopefully you are doing like I so often do, I just put on a good FM station, good CD, or good record, and enjoy the music. If you build either your own amp or your own loudspeaker, you might
enjoy it even more.
percentage of THD, also causes higher harmonics to appear that were not there before the
feedback was employed.
Take a pure 1 kHz tone, send it through an amplifier (*) that has only 2nd, 3rd, and no other harmonic distortion products present.
Now, take a portion of the amplifier output, and apply it to the feedback input point.
This already distorted signal goes through the amplifier again, where it is additionally distorted.
2nd harmonic x 2 = 4th harmonic
2nd harmonic x 3 = 6th harmonic
3rd harmonic x 2 = 6th harmonic
3rd harmonic x 3 = 9th harmonic
Of course these new, higher order harmonics are at reduced levels, since they are created
by the original harmonic distortion products which are already less than the fundamental.
Remember, the amplifier will take whatever is presented to the input, and to the feedback input
as well, and will distort it.
Two or more tones that are passed through an amplifier will generate Intermodulation Distortion
products (IMD). And just like the harmonic distortion products that are re-applied as negative feedback, the IMD creates even more tones.
IMD is usually measured by using 2 steady state tones.
Now, I am not suggesting that negative feedback is a bad thing.
I am just saying that any design or topology does have some sort of tradeoff.
Later than Norman Crowhurst’s articles, Matti Otala (and also a Finnish person) wrote about Transient Inter Modulation distortion (TIM).
It showed that too much negative feedback could cause other problems, not just those
of steady state IMD.
Now, back to the original amplifier (*) that had a 1 kHz tone applied and only created 2nd and 3rd harmonics.
Take the family of curves of the output device(s), be it a Triode, Pentode, BJT, MOSFET, etc.
Apply a straight (resistive load line). Before you apply negative feedback, that is usually the
lowest numbers of harmonics you will get.
After negative feedback, you can reduce the size of those 2nd and 3rd harmonic distortion products. But now you will get higher order distortion products too.
Now apply a reactive load line. With or without feedback, there will be different distortion
levels. The reactance can be transformer primary inductance, distributed capacitance, or leakage reactance. It can also be a loudspeaker load as its reactance is reflected to the primary winding.
The resultant elliptical load will cause different distortion effects depending on whether the amplifier is single ended or it is push pull.
Traditional vacuum tube push pull is one thing; Totem Pole BJT and Totem Pole MOSFET are quite another thing.
(*) you can discuss whether any such amplifier exists, but from a practical sense, and with
small to medium signal levels, such amplifiers do exist.
The general case for triode, pentode, BJT, MOSFET, etc. is that in the well behaved region (small to medium signal levels) the distortion products are reduced by the following rates as the signal level is decreased:
Reducing the Fundamental, c, by 10 dB will cause the distortion products to reduce by the following amount relative to the new fundamental level
2nd order -10 dBc
3rd order -20 dBc
4th order -30 dBc
… and so forth
Isn’t it nice how upper order harmonics, and upper order intermodulation products fall off
rapidly as the signal level is reduced?
At some point as we reduce the signal level, we have to worry about hum and noise, they are larger than the upper order distortion products.
Examples:
An amplifier putting out a signal level that in its well behaved region has 2nd harmonic distortion
of -45 dBc. Now the signal level is reduced by 10 dB. The new 2nd harmonic distortion is -55 dBc.
An amplifier putting out a signal level that in its well behaved region has 3rd Intermodulation distortion of -50 dBc. Now the signal level is reduced by 10 dB. The new 3rd Intermodulation distortion is -70 dBc.
By the way, loudspeaker distortion also is lower when the signal level is reduced.
And our ear’s distortion also acts similarly.
Regardless of what amplifier topology, and what speaker, etc. you like, hopefully you are doing like I so often do, I just put on a good FM station, good CD, or good record, and enjoy the music. If you build either your own amp or your own loudspeaker, you might
enjoy it even more.
By all means. Enjoying one's audio gear is good. However, people post claims about those gears on forums and that's a different activity.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.