Unfortunately once we convert a signal for storage in whatever medium we choose we have altered the signal and set ourselves on a never ending unsuccessful journey to convert it to its original form. Whatever alteration we discover and restore begets a new alteration. We are forced to choose between a minimalist approach or a never ending battle to restore all the alterations.
What to look for in a power amp?
- Thread starter Gregadd
- Start date
Hmmm... Sorry muralman1, not sure what I did to earn your antipathy, but for the record I am not in the audio industry, but doing engineering of another type (at much higher frequencies, where ironically the simplest circuits needed are what we use due the bandwidth required). As for quantum physics, well, I understand enough to get by... And, I was pre-med in college so have some exposure to the natural sciences.
Tubes have advantages out the output of DACs, but I think it best I stay out of this thread. I need more practice tme for an upcoming concert anyway; why didn't Copland give the brass someplace to breathe at the end!
All the best - Don
Tubes have advantages out the output of DACs, but I think it best I stay out of this thread. I need more practice tme for an upcoming concert anyway; why didn't Copland give the brass someplace to breathe at the end!
All the best - Don
Sorry Don, someone else brought up the name Don and I stupidly ran with it. I was not thinking of you at all. With some thought the one person I am least agreeable with would be Ethan Winer who is a strong adherent to oversampling. I will delete the reference.
I agree. Since I find the NOS approach to have nil distortion, it can't fail to make me wonder, what it was that set engineers on an, "improvement," binge, other than profit.?
Thanks muralman, I really appreciate it. If I want abuse I can always talk to my teens...
Regarding NOS DACs, it is very challenging technically to produce a high-precision DAC that is not of the over-sampled (delta-sigma) variety. Element matching is a b* -- err, PAIN! -- and there are other technical issues. It is much easier to realize a high-precision delta-sigma (oversampling) DAC, and it's worth noting that virtually everything complex in that type of DAC (OS) is digital -- the output can be a very simple one-bit (comparator) cell. In contrast, a 16-bit analog DAC must match some number of cells with very high precision to realize 1 part in 65,536 error. When you consider typical process variation is around 1 %, with 0.1 % achievable with selection and trimming, then it is clear that getting 1/65536 = 0.0015% (16-bit) matching is pretty durn hard! Even discrete components are only good to about 0.1 - 0.05% and you will pay plenty for those. A lot of TLC goes into achieving 16-bit precision, let alone 24; thus, their high cost.
Another issue is the output filters -- it is very difficult to create a DAC that does not generate wideband noise when it switches, and the output filter needed to suppress that noise is also very difficult to realize. De-glitching circuits exist but are also difficult to implement as they must have high (e.g. 16-bit) precision (accuracy, noise level, etc.) Oversampling makes it much, much easier as the output filter can be much lower order (much simpler) and thus has much less impact in the audio band.
These are fundamental physical limitations not easy to get around, making any true 16-bit (or higher) NOS DAC a very expensive proposition*. Orders of magnitude more difficult and more expensive to develop, produce, and test than OS DACs. Oversampling solves these issues at the cost of others, including more digital complexity (but, digital ICs are cheap!) and potentially greater wideband noise and switching transients.
I find it very interesting you like your class-D amps with your Apogees; that is not a combination I would have ever guessed! I have some theories why that may be but they are best left for now. I would love to hear your system someday.
HTH - Don
* As a point of interest, several years ago I designed a custom 16-bit NOS DAC for a customer of the company I worked for (bad English, sorry). It worked very well, settled to < 1 lsb in about 35 ns (that is very fast), and had excellent linearity and noise. It could also be clocked up to around 1 GS/s. It cost nearly $1k each due to the NRE (non-recurring engineering) and extensive production trim and test time (about 10x the next nearest DAC in our catalog). Took about two years and perhaps 4 - 6 manyears to develop. A manufacturer will probably double (at least) the cost to the customer, so you can see it would be pricey... Because the production cost was almost all in the test time, an audio converter would not have been much cheaper, though I would have made design trades for a slower part that would have simplified that. Even so, modern delta-sigma (DS, oversampling) DACs are in the few dollar (USD) range for a good 16- to 24-bit DAC. Hard to compete with those numbers...
Regarding NOS DACs, it is very challenging technically to produce a high-precision DAC that is not of the over-sampled (delta-sigma) variety. Element matching is a b* -- err, PAIN! -- and there are other technical issues. It is much easier to realize a high-precision delta-sigma (oversampling) DAC, and it's worth noting that virtually everything complex in that type of DAC (OS) is digital -- the output can be a very simple one-bit (comparator) cell. In contrast, a 16-bit analog DAC must match some number of cells with very high precision to realize 1 part in 65,536 error. When you consider typical process variation is around 1 %, with 0.1 % achievable with selection and trimming, then it is clear that getting 1/65536 = 0.0015% (16-bit) matching is pretty durn hard! Even discrete components are only good to about 0.1 - 0.05% and you will pay plenty for those. A lot of TLC goes into achieving 16-bit precision, let alone 24; thus, their high cost.
Another issue is the output filters -- it is very difficult to create a DAC that does not generate wideband noise when it switches, and the output filter needed to suppress that noise is also very difficult to realize. De-glitching circuits exist but are also difficult to implement as they must have high (e.g. 16-bit) precision (accuracy, noise level, etc.) Oversampling makes it much, much easier as the output filter can be much lower order (much simpler) and thus has much less impact in the audio band.
These are fundamental physical limitations not easy to get around, making any true 16-bit (or higher) NOS DAC a very expensive proposition*. Orders of magnitude more difficult and more expensive to develop, produce, and test than OS DACs. Oversampling solves these issues at the cost of others, including more digital complexity (but, digital ICs are cheap!) and potentially greater wideband noise and switching transients.
I find it very interesting you like your class-D amps with your Apogees; that is not a combination I would have ever guessed! I have some theories why that may be but they are best left for now. I would love to hear your system someday.
HTH - Don
* As a point of interest, several years ago I designed a custom 16-bit NOS DAC for a customer of the company I worked for (bad English, sorry). It worked very well, settled to < 1 lsb in about 35 ns (that is very fast), and had excellent linearity and noise. It could also be clocked up to around 1 GS/s. It cost nearly $1k each due to the NRE (non-recurring engineering) and extensive production trim and test time (about 10x the next nearest DAC in our catalog). Took about two years and perhaps 4 - 6 manyears to develop. A manufacturer will probably double (at least) the cost to the customer, so you can see it would be pricey... Because the production cost was almost all in the test time, an audio converter would not have been much cheaper, though I would have made design trades for a slower part that would have simplified that. Even so, modern delta-sigma (DS, oversampling) DACs are in the few dollar (USD) range for a good 16- to 24-bit DAC. Hard to compete with those numbers...
Well... Don.... There you go again.... Sounded right when the Gipper said it.
I should give you a few other examples at how weird my audio world is. With my speakers able to say anything you tell it to, and a preamp/amp set that loves to amplify everything without a footprint, you can tell an awful lot about other components.
A fellow brought over his Marantz player. It was an earlier model, and he loved the way it sounded on his Duettas. Actually, come to think of it, the fellow I am talking about is a member here. Anyway, it was obvious to us that his OS player was a bit grainy, and bright. All in all, it made a respectable show of itself.
One day, a dealer showed up with two sets of speaker cables. One was the 10k Cardas Golden Reference SCs. The others were Jennas. The Cardas is one of many thick hoses. It made quite a racket sounding like an irritated teacher's shhhh. The Jennas were better and only hissed like a Persian Cat.
Paul Speltz makes a point of this in selling his good as naked magnet wire cables. My system just proves his point with enthusiasm.
That is why I kept the insulation to a thin film of Kapton on my speaker cables. This system is so hysterically sensitive, a simple thing like cutting the ends of the cable into little ribbons in order to facilitate cramming a two inch wide ribbon into the throat of a spade, makes an audible noise, as super faint it is. I only know that because I have fixed that problem. Now, with all the cables silenced, the music appears on stage all by itself.
Although I am amazed at the performance of the 47 Labs Flatfish, I am not at all surprised. The designer took great pains to minimize all circuitry. The solid state transport has something like a dozen parts. His site does say, it's spartan design will newly reveal low bass material, and harmony. Do you know what? They were absolutely right.
Audio Note DACs also has few parts. There is a simple digital board with it's own little power supply (that is where the four diode switch made such a tremendous difference). There is another tube rectified power supply for the analog stage, a seemingly ridiculous simple affair.
The music keeps all it's original character by not being battered about like a tennis ball in a Wimbledon Championship match as it is in OS DACs.
I should give you a few other examples at how weird my audio world is. With my speakers able to say anything you tell it to, and a preamp/amp set that loves to amplify everything without a footprint, you can tell an awful lot about other components.
A fellow brought over his Marantz player. It was an earlier model, and he loved the way it sounded on his Duettas. Actually, come to think of it, the fellow I am talking about is a member here. Anyway, it was obvious to us that his OS player was a bit grainy, and bright. All in all, it made a respectable show of itself.
One day, a dealer showed up with two sets of speaker cables. One was the 10k Cardas Golden Reference SCs. The others were Jennas. The Cardas is one of many thick hoses. It made quite a racket sounding like an irritated teacher's shhhh. The Jennas were better and only hissed like a Persian Cat.
Paul Speltz makes a point of this in selling his good as naked magnet wire cables. My system just proves his point with enthusiasm.
That is why I kept the insulation to a thin film of Kapton on my speaker cables. This system is so hysterically sensitive, a simple thing like cutting the ends of the cable into little ribbons in order to facilitate cramming a two inch wide ribbon into the throat of a spade, makes an audible noise, as super faint it is. I only know that because I have fixed that problem. Now, with all the cables silenced, the music appears on stage all by itself.
Although I am amazed at the performance of the 47 Labs Flatfish, I am not at all surprised. The designer took great pains to minimize all circuitry. The solid state transport has something like a dozen parts. His site does say, it's spartan design will newly reveal low bass material, and harmony. Do you know what? They were absolutely right.
Audio Note DACs also has few parts. There is a simple digital board with it's own little power supply (that is where the four diode switch made such a tremendous difference). There is another tube rectified power supply for the analog stage, a seemingly ridiculous simple affair.
The music keeps all it's original character by not being battered about like a tennis ball in a Wimbledon Championship match as it is in OS DACs.
Nice point, but damage is done even before you convert it for storage. As soon as you convert sound (mechanical energy) in an electrical signal you altered it - different microphones have different sound signatures.
Some time ago I had an interesting discussion with a sound engineer - he considered that the sound from his monitors in the control room during the recording session was the exact sound to recreate in an audiophile room. But one knowledge person remembered that every recording needs mastering, and almost all great recordings are associated with an expert mastering engineer. One more reference was killed at that dinner ...
A good way of creating a reference could be making also a binaural recording at the same time of the main recording for later comparison between the sound heard by headphones with what we have in our listening room.
yeah but nobody does it like you and me Don.
How many can actually answer this question when quizzed by the general public? What does a tube do and why is it better than a transistor?
Take a look at this. Most notable is what instruments sound best with tubes and what sounds best with transistors.
http://www.soundfountain.com/amb/tubes.html
http://www.soundfountain.com/amb/tubes.html
...And, how many people think of thermonuclear devices when hearing the term "thermionic emission"?
A vacuum tube consists of electrodes in a vacuum in an insulating heat-resistant envelope which is usually tubular. Many tubes have glass envelopes, though some types such as power tubes may have ceramic or metal envelopes. The electrodes are attached to leads which pass through the envelope via an airtight seal. On most tubes, the leads are designed to plug into a tube socket for easy replacement.
The simplest vacuum tubes resemble incandescent light bulbs in that they have a filament sealed in a glass envelope which has been evacuated of all air. When hot, the filament releases electrons into the vacuum: a process called thermionic emission. The resulting negatively charged cloud of electrons is called a space charge. These electrons will be drawn to a metal plate inside the envelope, if the plate (also called the anode) is positively charged relative to the filament (or cathode). The result is a flow of electrons from filament to plate. This cannot work in the reverse direction because the plate is not heated and does not emit electrons. This very simple example described can thus be seen to operate as a diode: a device that conducts current only in one direction. The vacuum tube diode conducts conventional current from plate (anode) to the filament (cathode); this is the opposite direction to the flow of electrons (called electron current).
Vacuum tubes require a large temperature difference between the hot cathode and the cold anode. Because of this, vacuum tubes are inherently power-inefficient; enclosing the tube within a heat-retaining envelope of insulation would allow the entire tube to reach the same temperature, resulting in electron emission from the anode that would counter the normal one-way current. Because the tube requires a vacuum to operate, convection cooling of the anode is not generally possible unless the anode forms a part of the vacuum envelope (in which case the cooling is by conduction through the anode material and then convection outside the vacuum envelope). Thus anode cooling occurs in most tubes through black-body radiation and conduction of heat to the outer glass envelope via the anode mounting frame. Cold cathode tubes do not rely on thermionic emission at the cathode and usually have some form of gas discharge as the operating principle; such tubes are used for lighting (neon lamps) or as voltage regulators.
Sometimes another electrode, called a control grid, is added between the cathode and the anode. The vacuum tube is then known as a "triode." A triode is a voltage-controlled device, in that a voltage that is applied as an input to the grid can be used to modulate the rate of electron flow between anode and cathode. The relationship between this input voltage and the output current is determined by a transconductance function. Control grid current is practically negligible in most circuits. The solid-state device most closely analogous to the vacuum tube is the JFET, although the vacuum tube typically operates at far higher voltage (and power) levels than the JFET.
Wikepedia
That will make the average tube detractors eyes glass over and get them off your back. Then you can get back your real job of drinking all the good beer at the party.
The simplest vacuum tubes resemble incandescent light bulbs in that they have a filament sealed in a glass envelope which has been evacuated of all air. When hot, the filament releases electrons into the vacuum: a process called thermionic emission. The resulting negatively charged cloud of electrons is called a space charge. These electrons will be drawn to a metal plate inside the envelope, if the plate (also called the anode) is positively charged relative to the filament (or cathode). The result is a flow of electrons from filament to plate. This cannot work in the reverse direction because the plate is not heated and does not emit electrons. This very simple example described can thus be seen to operate as a diode: a device that conducts current only in one direction. The vacuum tube diode conducts conventional current from plate (anode) to the filament (cathode); this is the opposite direction to the flow of electrons (called electron current).
Vacuum tubes require a large temperature difference between the hot cathode and the cold anode. Because of this, vacuum tubes are inherently power-inefficient; enclosing the tube within a heat-retaining envelope of insulation would allow the entire tube to reach the same temperature, resulting in electron emission from the anode that would counter the normal one-way current. Because the tube requires a vacuum to operate, convection cooling of the anode is not generally possible unless the anode forms a part of the vacuum envelope (in which case the cooling is by conduction through the anode material and then convection outside the vacuum envelope). Thus anode cooling occurs in most tubes through black-body radiation and conduction of heat to the outer glass envelope via the anode mounting frame. Cold cathode tubes do not rely on thermionic emission at the cathode and usually have some form of gas discharge as the operating principle; such tubes are used for lighting (neon lamps) or as voltage regulators.
Sometimes another electrode, called a control grid, is added between the cathode and the anode. The vacuum tube is then known as a "triode." A triode is a voltage-controlled device, in that a voltage that is applied as an input to the grid can be used to modulate the rate of electron flow between anode and cathode. The relationship between this input voltage and the output current is determined by a transconductance function. Control grid current is practically negligible in most circuits. The solid-state device most closely analogous to the vacuum tube is the JFET, although the vacuum tube typically operates at far higher voltage (and power) levels than the JFET.
Wikepedia
That will make the average tube detractors eyes glass over and get them off your back. Then you can get back your real job of drinking all the good beer at the party.
Easy - in tube electrons move in vacuum, in a transistor conductivity is due to doping with impurities. In vacuum, there is no propagation of vibrations and sound from tubes has less coloration. Impurities should sound hard and affect the sound purity ...
PS - For non technical people - this is a joke!
...And, how many people think of thermonuclear devices when hearing the term "thermionic emission"?
More or less the same number of those who believe that quantum physics can explain the effects of power supply purifiers!
I think the credit goes to David Manley for bringing high power tube amps to the market, not the son.
Tube Amplifier vs Transistor Amplifier
Things that sound better withh tubes
strings
harp
bassoon
timpani
harpsichord
trumpet
saxophone
voices
acoustic guitar (nylon strings)
mid and high registers of church organ
panflute
recorder
and: applause
Things that sound better with transistors.
piano
electronic bas
synthesizer
flute (traverso)
snare drum
horn
Linn drum
electric guitar
guitar with steel strings
tambourine
mandolin
triangle
Things that sound better withh tubes
strings
harp
bassoon
timpani
harpsichord
trumpet
saxophone
voices
acoustic guitar (nylon strings)
mid and high registers of church organ
panflute
recorder
and: applause
Things that sound better with transistors.
piano
electronic bas
synthesizer
flute (traverso)
snare drum
horn
Linn drum
electric guitar
guitar with steel strings
tambourine
mandolin
triangle
Sure - the trick is to make a *high powered* amplifier that's also super fast.
I love tubes. Too many tubes I do not love. I have heard expensive all tube systems, and the piling on of distortion through each stage softens the music. The softening is not disagreeable, but it clouds from the listener leading edges, brash sounds, and hard hit sounds and just plain detail.
By putting two driving tubes in my DAC, I set the stage for all the tube quickness, spaciousness, real time decay, and immediacy. That magic is lovingly preserved with white glove delicacy, until the sound produced by those tubes is unleashed to the speakers and now watch how tubes rule.
By putting two driving tubes in my DAC, I set the stage for all the tube quickness, spaciousness, real time decay, and immediacy. That magic is lovingly preserved with white glove delicacy, until the sound produced by those tubes is unleashed to the speakers and now watch how tubes rule.
Similar threads
