Bob Cordell said:
Some time back I made an analysis of impedance as a function of frequency for a 2 way high end box. The calculated impedance curve matched fairly well with the published measured one, with the addition of the phase angle (not specified).
It is sobering to see what our beloved circuits have to deal with.
In that same thread, Andy_c made a series of simulations regarding output impedance as a function of current for the crossover region under several quiescent bias conditions. They went a long way to shed light on the nasty low level wideband harmonic and IM distortion so characteristic of poorly executed solid stated designs.
Rodolfo
Wavebourn said:
Crossover distortion in amplifier output stages and coloration in the vicinity of the crossover in a loudspeaker are two very different things. You appear to be confusing the two. The first is a nonlinear distortion, the second is a (linear) frequency response aberration. The only thing they have in common is the word crossover.
Whether it makes the sound more pleasant or not, or whether it makes the sound more accurate or less accurate, the addition of series output resistance between the power amplifier and the loudspeaker will INCREASE the effect that loudspeaker impedance variations have on the net frequency response of the combination.
Bob
Nico Ras said:
Nico,
I would not make any generalizations about the sound of MOSFETs vs BJTs based on this amplifier. This design appears to have a lot of design flaws. First, it is almost certain that the MOSFETs are under-biased, although I honestly don't know how you got 100 mA to flow with a net forward bias of only one junction drop of a single 1n4148 between their gates. Lateral MOSFETs usually turn on at around 0.5 to 1.0V EACH, and here it looks like they each have only about 0.3V of forward gate bias.
Without getting into too much detail, I also note that the VAS is directly feeding the output MOSFETs, so unless you are running the VAS quite hot, it will be heavily influenced by the capacitive load of the MOSFETs. Also, your input differential pairs are un-degenerated, so slew rate is probably pretty bad. Sorry to say, but just not a good design overall.
Bob
ingrast said:
That's a nice piece of work. I agree that the result is sobering, and that we should do more amplifier testing with other than resistive loads.
One thing I should note is that one can electronically simulate an almost arbitrary speaker load using fairly small passive components by employing a second amplifier to back-drive the output of the amplifier under test through a power resistor. This is essentially what we call a synthesized active load. This technique can also be used to employ other kinds of "torture" tests on amplifiers, such as exploring protection circuit trigger points. Obviously, the "other" amplifier should be a fairly large and robust amplifier.
Bob
Dear Bob,
The forum kicked off with BJT vs. MOSFET and reading all the speculation and opinions I set out to attempt a practical comparison.
I illustrated the basic topology and offered a finding for which there may be an explanation and possible simple resolve for all audiophile speculations.
Bar the transistor types, the front end is a crib from the KSA50 in use since 1982. I added an adjustable bias but did not include de-generation because I was under the impression that matching transistors for such a comparison would be obvious.
Whether Dan had any idea of audio design when he created the KSA50 in the early 80's was not the question.
The idea was to knock up a proven design quickly that uses identical circuits but one running MOSFET and the other BJT.
The important issue was that the two amplifiers should electrically perform the same and that one may extract the differences attributed to characteristic of the output devices.
The one amplifier uses MJ15003/4 and the other uses 2SK1058/2SJ162 as output stages.
Both amplifiers was set to have the same gain for an input stimulus, both ran off laboratory standard 15 Amp regulated power supplies set to ±35V. Both amplifier idle currents were set to 200 mA and output offset adjusted to less that 1mV with input short circuit.
Slew rates on both amplifiers were measured at 10 kHz and far exceeded any audio requirements. Intermodulation products were not measured, as there was no time to do so.
The only notable difference was that the MOSFET harmonic content was predominantly second while that of the BJT was third.
This was an attempt to compare apples with bananas using identical topologies with two very different devices and whether their individual characters play any significant role in the application.
This evening four Jazz musician friends came over for a gig and and they were asked to individually listen to two well-known and appreciated recordings that incorporated string instruments and vocals while the second recording incorporated a sax instead of vocals and some excellent drum solos.
The input to both amps were mono fed from a CD player and they both connected alternately to the same loudspeaker standing in the centre of the room with a chair placed 2 meters away.
After each individual listening session each had to write a few words what they liked or disliked between the characteristics of the left hand pile of electronics and the right hand pile of electronics.
The consensus without any argument was:
BJT sounded more realistic on vocals and strings while the MOSFET sounded more realistic with sax and percussion. Keeping in mind that professional musicians can detect differences in sound character quite easily.
We all enjoyed the "test" and our gig lasted until 02:15 this morning. I will bin both amps as it appears to be an unimportant issue anyway.
The forum kicked off with BJT vs. MOSFET and reading all the speculation and opinions I set out to attempt a practical comparison.
I illustrated the basic topology and offered a finding for which there may be an explanation and possible simple resolve for all audiophile speculations.
Bar the transistor types, the front end is a crib from the KSA50 in use since 1982. I added an adjustable bias but did not include de-generation because I was under the impression that matching transistors for such a comparison would be obvious.
Whether Dan had any idea of audio design when he created the KSA50 in the early 80's was not the question.
The idea was to knock up a proven design quickly that uses identical circuits but one running MOSFET and the other BJT.
The important issue was that the two amplifiers should electrically perform the same and that one may extract the differences attributed to characteristic of the output devices.
The one amplifier uses MJ15003/4 and the other uses 2SK1058/2SJ162 as output stages.
Both amplifiers was set to have the same gain for an input stimulus, both ran off laboratory standard 15 Amp regulated power supplies set to ±35V. Both amplifier idle currents were set to 200 mA and output offset adjusted to less that 1mV with input short circuit.
Slew rates on both amplifiers were measured at 10 kHz and far exceeded any audio requirements. Intermodulation products were not measured, as there was no time to do so.
The only notable difference was that the MOSFET harmonic content was predominantly second while that of the BJT was third.
This was an attempt to compare apples with bananas using identical topologies with two very different devices and whether their individual characters play any significant role in the application.
This evening four Jazz musician friends came over for a gig and and they were asked to individually listen to two well-known and appreciated recordings that incorporated string instruments and vocals while the second recording incorporated a sax instead of vocals and some excellent drum solos.
The input to both amps were mono fed from a CD player and they both connected alternately to the same loudspeaker standing in the centre of the room with a chair placed 2 meters away.
After each individual listening session each had to write a few words what they liked or disliked between the characteristics of the left hand pile of electronics and the right hand pile of electronics.
The consensus without any argument was:
BJT sounded more realistic on vocals and strings while the MOSFET sounded more realistic with sax and percussion. Keeping in mind that professional musicians can detect differences in sound character quite easily.
We all enjoyed the "test" and our gig lasted until 02:15 this morning. I will bin both amps as it appears to be an unimportant issue anyway.
Nico Ras said:Dear Bob,
The forum kicked off with BJT vs. MOSFET and reading all the speculation and opinions I set out to attempt a practical comparison.
I illustrated the basic topology and offered a finding for which there may be an explanation and possible simple resolve for all audiophile speculations.
Bar the transistor types, the front end is a crib from the KSA50 in use since 1982. I added an adjustable bias but did not include de-generation because I was under the impression that matching transistors for such a comparison would be obvious.
Whether Dan had any idea of audio design when he created the KSA50 in the early 80's was not the question.
The idea was to knock up a proven design quickly that uses identical circuits but one running MOSFET and the other BJT.
The important issue was that the two amplifiers should electrically perform the same and that one may extract the differences attributed to characteristic of the output devices.
The one amplifier uses MJ15003/4 and the other uses 2SK1058/2SJ162 as output stages.
Both amplifiers was set to have the same gain for an input stimulus, both ran off laboratory standard 15 Amp regulated power supplies set to ±35V. Both amplifier idle currents were set to 200 mA and output offset adjusted to less that 1mV with input short circuit.
Slew rates on both amplifiers were measured at 10 kHz and far exceeded any audio requirements. Intermodulation products were not measured, as there was no time to do so.
The only notable difference was that the MOSFET harmonic content was predominantly second while that of the BJT was third.
This was an attempt to compare apples with bananas using identical topologies with two very different devices and whether their individual characters play any significant role in the application.
This evening four Jazz musician friends came over for a gig and and they were asked to individually listen to two well-known and appreciated recordings that incorporated string instruments and vocals while the second recording incorporated a sax instead of vocals and some excellent drum solos.
The input to both amps were mono fed from a CD player and they both connected alternately to the same loudspeaker standing in the centre of the room with a chair placed 2 meters away.
After each individual listening session each had to write a few words what they liked or disliked between the characteristics of the left hand pile of electronics and the right hand pile of electronics.
The consensus without any argument was:
BJT sounded more realistic on vocals and strings while the MOSFET sounded more realistic with sax and percussion. Keeping in mind that professional musicians can detect differences in sound character quite easily.
We all enjoyed the "test" and our gig lasted until 02:15 this morning. I will bin both amps as it appears to be an unimportant issue anyway.
Hi Nico,
I applaud you for what you have done here. It is the kind of comparison I like to do myself. Your result is very interesting. It just looked to me from the schematic that the amp wasn't a great design; maybe I was being too critical.
Cheers,
Bob
Hi Bob,
Thanks for the comment, maybe it would be worthwhile putting one of your top notch designs to task in a simular way and this issue may be put to bed for good.
I think there will always be different sounding amps, simply because every designer after creating the perfect amp finds that it needs final tweaks that includes his own preference of what he perceives real life sounds like.
Whether this is by means of using polypropelene or polycarbonate or ceramic caps, it does not matter - some global distortion is introduced to make the sound more enjoyable. If everything was exactly the same how boring would life be.
Has anybody thought of the miles of cables, thousands of op-amps, caps, potentiometers, diodes, etc. that the signal passes through before ending on the CD. Moreover, the recording engineer introduced his own idea of what the recording should sound like or would sell best.
With the advent of electronically generated music, who can tell what it should sound like, music are purchased because it produces an enjoyable effect. Who can tell the difference between two Japanese electronic orchestras.
I enjoy the new Madonna release. I know it is Madonna because it says so on the cover. I have never heard some of the backing instruments in my entire life, but it produces a great feeling and I find my foot tapping on the floor involentary. I heard this song both on my car radio and my home set-up, and must admit I like both versions even though it sounds very different.
Ask yourself how many times you listen to a track that you really like but you wished that you included tone controls into your design because it would have made it just that more enjoyable if you could have lifted the lower or upper end ever so slightly.
Bob, I enjoy your comments, your vast grasp of the subject was gained through years of experimentation, and not something that was taught. You tried it and it worked or it did not. Whether there was a scientific explanation makes little difference I guess.
I threw a curved ball with my previous comment on placing a series resistor in line with a speaker, there were several theories.
Could this be generalised to resistors in all circuits? Would cable lengths, crossectional area, make up, winding direction, insulation material or connector type have an impact on sound quality. I am sure there are theories to cover each of these effects.
Can the effect of copper tracks on a PCB be explained, what if aluminium sounded better, anybody tried?
How about sound quality of surface mount technoligy vs. discreet components. Does lead-free solder offer better sounding results?
In fact can anyone out there explain what sound quality is and whether this applies to all creatures and cultures.
Does age of the listener play any role. It does on my eyes, what about my ears. I think I could enjoy if treble were boosted by 20 or so dB, maybe I can then hear a triangle being tapped once again, I forget what it sounds like.
There could be a new challenge out there developing audio equipment that sounds nice to the over 50's. Think of it they made their millions and would spend a small fortune on equipment that they could actually enjoy.
Cheers Nico
Thanks for the comment, maybe it would be worthwhile putting one of your top notch designs to task in a simular way and this issue may be put to bed for good.
I think there will always be different sounding amps, simply because every designer after creating the perfect amp finds that it needs final tweaks that includes his own preference of what he perceives real life sounds like.
Whether this is by means of using polypropelene or polycarbonate or ceramic caps, it does not matter - some global distortion is introduced to make the sound more enjoyable. If everything was exactly the same how boring would life be.
Has anybody thought of the miles of cables, thousands of op-amps, caps, potentiometers, diodes, etc. that the signal passes through before ending on the CD. Moreover, the recording engineer introduced his own idea of what the recording should sound like or would sell best.
With the advent of electronically generated music, who can tell what it should sound like, music are purchased because it produces an enjoyable effect. Who can tell the difference between two Japanese electronic orchestras.
I enjoy the new Madonna release. I know it is Madonna because it says so on the cover. I have never heard some of the backing instruments in my entire life, but it produces a great feeling and I find my foot tapping on the floor involentary. I heard this song both on my car radio and my home set-up, and must admit I like both versions even though it sounds very different.
Ask yourself how many times you listen to a track that you really like but you wished that you included tone controls into your design because it would have made it just that more enjoyable if you could have lifted the lower or upper end ever so slightly.
Bob, I enjoy your comments, your vast grasp of the subject was gained through years of experimentation, and not something that was taught. You tried it and it worked or it did not. Whether there was a scientific explanation makes little difference I guess.
I threw a curved ball with my previous comment on placing a series resistor in line with a speaker, there were several theories.
Could this be generalised to resistors in all circuits? Would cable lengths, crossectional area, make up, winding direction, insulation material or connector type have an impact on sound quality. I am sure there are theories to cover each of these effects.
Can the effect of copper tracks on a PCB be explained, what if aluminium sounded better, anybody tried?
How about sound quality of surface mount technoligy vs. discreet components. Does lead-free solder offer better sounding results?
In fact can anyone out there explain what sound quality is and whether this applies to all creatures and cultures.
Does age of the listener play any role. It does on my eyes, what about my ears. I think I could enjoy if treble were boosted by 20 or so dB, maybe I can then hear a triangle being tapped once again, I forget what it sounds like.
There could be a new challenge out there developing audio equipment that sounds nice to the over 50's. Think of it they made their millions and would spend a small fortune on equipment that they could actually enjoy.
Cheers Nico
When it comes to how something sounds you have to take note
of the design, simple or complex etc...
this finding is not a general finding as it says little about
bjt or mosfet amps in terms of technical merit
all amplifiers have character i think, this is a result of the implementation of a particular design, the character should
not be judged in parts but as a whole, by saying one amplifier
sounded better in this part of the spectrum or the other etc
is confusing whereas saying the mosfet sounded better than
the bjt amp is more useful for proper evaluation
kind regards
I forgot to add, was the result from one to the other better or worse or maybe just different.
Which guitar sounds better, Gibson, Fender or Hoffnar? I think they are all basically good but produces quite different characeristic sound. Although each has a neck divided into a specified number of frets producing a definitive musical note.
Data being transferred from CD and converted to analogue signals, is it accurate? How can you hear the characteristics using a finite number of samples making up the entire spectrum of a complex stimulus? Beats me! But if it is the acceptable norm than that is what is available until someone comes with a new idea.
I personally think that there will be something appealing to every listener in every design topology and the companies that successfully sell millions of products into the consumer market got the mix right.
Their volumes have a further benefit of reducing prices and besides they make enough money to justify test equipment, investing into R&D researching new materials and ensure that their products always improve and manufacturing is consitent.
What does not sell becomes expensive audiophile or high-end equipment that has to be analysed to the n-th degree in order to justify its existance.
The other day, I listened to a friends CD that was just released playing on a ten year old Sony through a pair of Wharfdale Lintons using a Rotel CD player, I must admit it was enjoyable.
I also have another friend, who distributes only Audio Research, Martin Logan and Wadia. He wanted me to support his justification in obataining a set of Martin Logan statement 2 speakers. I must admit this too was enjoyable. Was the price of several cars more than the Sony set-up that more enjoyable? I cannot tell.
If he turned up the volume of the six monster Audio research amps would it still be enjoyable. Probably not, it would be very painful and maybe the last time that I listen to anything.
What John Curl mentioned in a previous comment may be more valid than one cares to admit. The majority of the so called esoteric equipment industry is really backyard. Those who made it and grows are those who got their formula right and if they persist, they will eventually proudly enjoy the fruits of the mass market.
Those who don't will continue to find reasons why their equipment is so expensive and only appeals to the few who really understands the art of listening to music.
Which guitar sounds better, Gibson, Fender or Hoffnar? I think they are all basically good but produces quite different characeristic sound. Although each has a neck divided into a specified number of frets producing a definitive musical note.
Data being transferred from CD and converted to analogue signals, is it accurate? How can you hear the characteristics using a finite number of samples making up the entire spectrum of a complex stimulus? Beats me! But if it is the acceptable norm than that is what is available until someone comes with a new idea.
I personally think that there will be something appealing to every listener in every design topology and the companies that successfully sell millions of products into the consumer market got the mix right.
Their volumes have a further benefit of reducing prices and besides they make enough money to justify test equipment, investing into R&D researching new materials and ensure that their products always improve and manufacturing is consitent.
What does not sell becomes expensive audiophile or high-end equipment that has to be analysed to the n-th degree in order to justify its existance.
The other day, I listened to a friends CD that was just released playing on a ten year old Sony through a pair of Wharfdale Lintons using a Rotel CD player, I must admit it was enjoyable.
I also have another friend, who distributes only Audio Research, Martin Logan and Wadia. He wanted me to support his justification in obataining a set of Martin Logan statement 2 speakers. I must admit this too was enjoyable. Was the price of several cars more than the Sony set-up that more enjoyable? I cannot tell.
If he turned up the volume of the six monster Audio research amps would it still be enjoyable. Probably not, it would be very painful and maybe the last time that I listen to anything.
What John Curl mentioned in a previous comment may be more valid than one cares to admit. The majority of the so called esoteric equipment industry is really backyard. Those who made it and grows are those who got their formula right and if they persist, they will eventually proudly enjoy the fruits of the mass market.
Those who don't will continue to find reasons why their equipment is so expensive and only appeals to the few who really understands the art of listening to music.
Nico Ras said:
While I can agree in general with the first part, it must also be reckoned mass markets are driven by sales, not technical goals.
Marketing people - rightly - must seek what large volume of consumers want or are willing to spend income in, even if that means radically departing from quality sound reproduction.
So, acknowledging a large fraction of cottage jobs pursue some form of mytical stance to lure well heeled customers, it is also a trully technically perfectionist stance not necessarily leads to success at the mass markets, even if technology and costs are made affordable.
Rodolfo
