I recently saw a post where it was pointed out that ATE (Automated Test Equipment) could require a tolerance of 10ppm and that audio doesn't even come close. Now, as an engineer, I'd regard 10ppm error as being seriously good engineering, so why don't the techniques and standards from other disciplines translate into "perfect sound for ever"? Why do we prefer to use a technology deemed obsolete decades ago?
Much has been made of the distortion/harmonic argument. It has been argued that solid state achieves low measured distortion at the expense of subjectively unpleasant distortion. Conversely, SE amplifiers are notorious for their high measured THD at full power.
It's difficult to make a high power SE valve amplifier, so most are <10W. You need efficient loudspeakers to go loud with 10W, so is it true (as the SS pundits would have us believe) that the SE enthusiasts are listening to distortion?
Similarly, it's difficult to make a PP valve amplifier with a "nice" distortion spectrum. Are we all lovers of grunge?
Much has been made of the distortion/harmonic argument. It has been argued that solid state achieves low measured distortion at the expense of subjectively unpleasant distortion. Conversely, SE amplifiers are notorious for their high measured THD at full power.
It's difficult to make a high power SE valve amplifier, so most are <10W. You need efficient loudspeakers to go loud with 10W, so is it true (as the SS pundits would have us believe) that the SE enthusiasts are listening to distortion?
Similarly, it's difficult to make a PP valve amplifier with a "nice" distortion spectrum. Are we all lovers of grunge?
I have nothing to add technically, but I share your frustration. I am also an engineer and deal with metrics and design targets all day, but when a "sound of cables" thread breaks out I invariably find myself on the "yes something is going on" side because the "bullspit" side arguments are so weak. Argh!
This is true, but how about thier distortion at clipping? Of course a PP amp can be well behaved under clipping too. How much of the expense of high feedback is higher order distortion and how much of it is ugly clipping? Related is SE amps low damping factor, they are often coupled with speakers that like this. An SE amp is inherently pure class A and not very sensitive to regulation, how much does this matter?
Some objective (based on real experience) opinion from a design engineer. Dashed this off in a hurry so I hope it make sense.
I have designed, built, purchased and upgraded about a dozen amps over the last 10 years or so.
Opinion 1.
The relative distortion spectrum of solid state vs tubes is important.
- My least favorite amp was a top of the line 210W per channel Rotel. this was a claimed 0.001% distortion amp (at the levels I used it) and it was just plain cold sterile and boring to the point I thought there was something wrong with it amd purchased a service manual and did a full rebias etc. - I chucked it in favour of a 10W per channel "Bervois Valley" EL84 Ultralinear I built.
- I do have a solid state amp I enjoy a lot. A Hugh Dean AKSA 55 Nivarna plus. It is incredibly revealing to look at the AKSA design in conjunction with Doug Self's "Distortion in Power Amplifiers" paper to see what aspects of Doug's "Blameless" Amplifier were incorporated into Hugh's design and what were left out. Two things stand out - Hugh uses a RESISTIVE biased diff amp (not current sourced) with a small (10%) amount of emitter degeneration and a capacitor coupled bootstrapped load on the VAS. Both of these things give the AKSA a very "Tube like" sound. Measurements showed a VERY tube amp like harmonic distortion profile even thought the total harmonic distortion was an order of magnitude lower than most tube amps.
That is the relative amount of 2nd, 3rd etc harmonic distortion is more important than the total harmonic distortion. You need some 2nd harmonic and a smaller amount of 3rd harmonic distortion (for prescence) with no harmonics above the 5th (Actually if you look at the maths and the music scale the 7th and particularly the 9th and 11th harmonics are the really nasty musically "quint" ones).
There is a limit to acceptable Total harmonic Distortion however. My 845 Single Ended Triode is stunning up to a certain listening level, above this level it is just a sickly sweat (bloated) 2nd harmonic distortiion generator and its like that 3rd or 4th cream cake - you just can't take it any more.
Opinion 2
For great detail you need ultra clean supplies. Higher frequency noise, such as diode switching noise is particularly objectionable. It is easier to generate clean high voltage/low current supplies to suit a tube amp than it is to generate clean low voltage/high current supplies to suit a solid state amp. Replacing the power diodes in the AKSA with Ultrafast Soft recovery diodes and replacing all electrolytic caps with Blackgates (the Nivarna Plus upgrade) did exactly the same as it does with a tube amp. It peeled away a "Veil" of noise and exposed a whole layer of detail which was always there but had just been masked by brodband noise". In my tube amp designs I use high voltage polypropylene caps wherever possible. On bias supplies I use Blackgate Electrolytics and return the output tube grid bias resistors to the bias supply via emitter followers with Polyphenylene Sulphide (PPS) caps from the base to 0V. That is - the bottom of the bias resistors are very tightly coupled to AC (audio) ground. I also use solid state cascode current sources for all tube diff amps although a single transistor current source can be nearly as good as long as a high reference voltage is used (a reverse connected BC547B with the base left open makes a good temperature compensated low noise approx 7 volt "zener" for this purpose - 0.5 to 1 mA through it).
Opinion 3
ANY global feedback destroys time coherency and destroys stereo image. Keep all feedback loops short (over a single stage where ever possible). Use just as much feedback as is necessary to achieve your purpose and never any more. This can be done in a Solid State Amp but to my knowledge is done by ONLY one manufacturer of solid state amps. I've never heard one, since I can't afford the Aussie$50,000 price tag BUT their amps are often touted as the best in the world. Tube Amps are nearly always low feedback affairs. My philosophy is to apply local feedback in all its forms, as required and never use more than 5 or 6 dB of global feedback and then only over two stages. To get a decent damping factor (my expeience is that a DF = 1.5 to 2 is suffiecient with good speakers and anything more than 4 is a waste of time) I use and RECOMMEND Ultralinear in combination with either shunt feedback from the output tube anodes OR cathode feedback. With minimal effort a low gain low feedback tube amp will always leave a high gain high feedback solid state amp for dead in the imaging department.
Both harmonic content and time coherency can have major influences on the leading edge of transients and on waveshape, be it the musical note itself or the "envelope" of the music content. I believe that these things influence not just Pace Rhythm and Attack (the PRATT people talk about) BUT also the "emotional" content of the music.
I keep my lovely AKSA 55N+ as a Reference Amp but for everyday listening its tube amps for me.
BTW - The "Bervois Valley" has been redesigned/rebuilt to use a current source diff amp front end and shunt feedback from the EL84 outputs. It now even more stunning. I'm using it while working on PPP EL34 Ultralinear plus cathode feedback (Menno's "Super Triode") monoblocks. There is also a KT88 version on the drawing board.
As for tubes - EL84s are seriously gorgeous, 0.7% THD in Ultralinear before applying any other feedback and Intrmodulation Distortion figures to match. A X5 headstart on most other tubes.
Just ducking off to don my flameproof vest.
Cheers,
Ian
I have designed, built, purchased and upgraded about a dozen amps over the last 10 years or so.
Opinion 1.
The relative distortion spectrum of solid state vs tubes is important.
- My least favorite amp was a top of the line 210W per channel Rotel. this was a claimed 0.001% distortion amp (at the levels I used it) and it was just plain cold sterile and boring to the point I thought there was something wrong with it amd purchased a service manual and did a full rebias etc. - I chucked it in favour of a 10W per channel "Bervois Valley" EL84 Ultralinear I built.
- I do have a solid state amp I enjoy a lot. A Hugh Dean AKSA 55 Nivarna plus. It is incredibly revealing to look at the AKSA design in conjunction with Doug Self's "Distortion in Power Amplifiers" paper to see what aspects of Doug's "Blameless" Amplifier were incorporated into Hugh's design and what were left out. Two things stand out - Hugh uses a RESISTIVE biased diff amp (not current sourced) with a small (10%) amount of emitter degeneration and a capacitor coupled bootstrapped load on the VAS. Both of these things give the AKSA a very "Tube like" sound. Measurements showed a VERY tube amp like harmonic distortion profile even thought the total harmonic distortion was an order of magnitude lower than most tube amps.
That is the relative amount of 2nd, 3rd etc harmonic distortion is more important than the total harmonic distortion. You need some 2nd harmonic and a smaller amount of 3rd harmonic distortion (for prescence) with no harmonics above the 5th (Actually if you look at the maths and the music scale the 7th and particularly the 9th and 11th harmonics are the really nasty musically "quint" ones).
There is a limit to acceptable Total harmonic Distortion however. My 845 Single Ended Triode is stunning up to a certain listening level, above this level it is just a sickly sweat (bloated) 2nd harmonic distortiion generator and its like that 3rd or 4th cream cake - you just can't take it any more.
Opinion 2
For great detail you need ultra clean supplies. Higher frequency noise, such as diode switching noise is particularly objectionable. It is easier to generate clean high voltage/low current supplies to suit a tube amp than it is to generate clean low voltage/high current supplies to suit a solid state amp. Replacing the power diodes in the AKSA with Ultrafast Soft recovery diodes and replacing all electrolytic caps with Blackgates (the Nivarna Plus upgrade) did exactly the same as it does with a tube amp. It peeled away a "Veil" of noise and exposed a whole layer of detail which was always there but had just been masked by brodband noise". In my tube amp designs I use high voltage polypropylene caps wherever possible. On bias supplies I use Blackgate Electrolytics and return the output tube grid bias resistors to the bias supply via emitter followers with Polyphenylene Sulphide (PPS) caps from the base to 0V. That is - the bottom of the bias resistors are very tightly coupled to AC (audio) ground. I also use solid state cascode current sources for all tube diff amps although a single transistor current source can be nearly as good as long as a high reference voltage is used (a reverse connected BC547B with the base left open makes a good temperature compensated low noise approx 7 volt "zener" for this purpose - 0.5 to 1 mA through it).
Opinion 3
ANY global feedback destroys time coherency and destroys stereo image. Keep all feedback loops short (over a single stage where ever possible). Use just as much feedback as is necessary to achieve your purpose and never any more. This can be done in a Solid State Amp but to my knowledge is done by ONLY one manufacturer of solid state amps. I've never heard one, since I can't afford the Aussie$50,000 price tag BUT their amps are often touted as the best in the world. Tube Amps are nearly always low feedback affairs. My philosophy is to apply local feedback in all its forms, as required and never use more than 5 or 6 dB of global feedback and then only over two stages. To get a decent damping factor (my expeience is that a DF = 1.5 to 2 is suffiecient with good speakers and anything more than 4 is a waste of time) I use and RECOMMEND Ultralinear in combination with either shunt feedback from the output tube anodes OR cathode feedback. With minimal effort a low gain low feedback tube amp will always leave a high gain high feedback solid state amp for dead in the imaging department.
Both harmonic content and time coherency can have major influences on the leading edge of transients and on waveshape, be it the musical note itself or the "envelope" of the music content. I believe that these things influence not just Pace Rhythm and Attack (the PRATT people talk about) BUT also the "emotional" content of the music.
I keep my lovely AKSA 55N+ as a Reference Amp but for everyday listening its tube amps for me.
BTW - The "Bervois Valley" has been redesigned/rebuilt to use a current source diff amp front end and shunt feedback from the EL84 outputs. It now even more stunning. I'm using it while working on PPP EL34 Ultralinear plus cathode feedback (Menno's "Super Triode") monoblocks. There is also a KT88 version on the drawing board.
As for tubes - EL84s are seriously gorgeous, 0.7% THD in Ultralinear before applying any other feedback and Intrmodulation Distortion figures to match. A X5 headstart on most other tubes.
Just ducking off to don my flameproof vest.
Cheers,
Ian
I used to work with the engineer who re-designed the Dyna Stereo 70. It received wonderful reviews in TAS. It was remarked that the re-issue was "very fast". He told me that the reason they sounded veiled was that the feedback compensation was very heavy. Most of what he did was speed the amplifier up.
As for SS stuff, transistors love lots of current. It is much easier to get away with loose power supply design in hollow state stuff IMO. Sagging supply rails and slow recovery time on the supply really drives up transient intermodulation distortion, which is quite objectionable. It usually rears it's ugly head on those big dynamics. High frequency bypassing in the supply can really help shave that stuff off.
Good, low impedence, star grounding helps as well. All of that current has to go somewhere! The quicker it moves, the better I like it.
As for the ATE units, we are talking about machines that do more reading than driving. A cheap one starts at around 500K$. They are typically under rigorous maintenence and calibration. Most of the measurements they make are DC and low level.
Rant over, thanks for listening,
Al
As for SS stuff, transistors love lots of current. It is much easier to get away with loose power supply design in hollow state stuff IMO. Sagging supply rails and slow recovery time on the supply really drives up transient intermodulation distortion, which is quite objectionable. It usually rears it's ugly head on those big dynamics. High frequency bypassing in the supply can really help shave that stuff off.
Good, low impedence, star grounding helps as well. All of that current has to go somewhere! The quicker it moves, the better I like it.
As for the ATE units, we are talking about machines that do more reading than driving. A cheap one starts at around 500K$. They are typically under rigorous maintenence and calibration. Most of the measurements they make are DC and low level.
Rant over, thanks for listening,
Al
Tube Dude,
The reasons are "Pure listening" - I've kept out of the "Listening Impressions of the AKSA 55N+" thread over at DIY Audio SS. Why - because I was one of the major contributors to the Nivarna Plus upgrade for the AKSA (I post as Ginger on the AKSA Forum at AudioCircle). I bought a 55 Nivarna before there ever was a Nivarna Plus and worked with Hugh on development of the Nivarna Plus Option. This took about 9 months as it involved extensive listening tests and letting new mods settle in before making SUB-jective evauation of improvements or otherwise. (i.e. the same time it takes cows and countesses to produce new offspring - although we know some blushing young brides can manage it in less than 9 months). Many ideas which my pure engineering analysis suggested would improve the amp distortionwise did NOT correlate to improved sonics. We gradually converged and arrived a a consensus as to what should go in and what should'nt although minor differences remain in my final 55N+ and what Hugh sells as the 55N+ (I use slightly less emitter degeneration of the diff amp and use "Super E" connected Blackgate Ns as the bootstrap cap rather than a single Blackgate Standard).
I can tell you that I have tried current sourcing the diff amp on the AKSA as well as tried current source loading the VAS in lieu of the capacitive bootstrap. In both instances the lovely "tubelike" warmth of the AKSA was lost and I abandonned those mods.
My view is that the resistive biasing of the diffamp in the AKSA introduces enough 2nd harmonic distortion to complement the amount of 3rd harmonic which exists anyway and current source biasing tube diffamp front ends reduces the amount of 2nd harmonic to again complement the amount of third harmonic. i.e. in both cases its attempting to get the 2H to 3H balance correct for best listening. With the SS diff amp we had to add a bit of 2H (by the resistor tail instead of a current source). With the tube diff amp we had to reduce the amount of 2H (by the current source instead of a resistor tail). The resistive biasing of the solid state diff amp is NOT technically "superior" - its just what gives the best sonic results. In the case of the tube diff amp I found that the better i could make the current source tail the better it sounded - hence cascode transistor current source with relatively high reference voltage (either green or blue led or the reverse connected transistor as the reference element). Unfortunately this will also depend upon the degree of matching of the 2 devices in the diff amp. In the AKSA the trasnsistors are selected to be matched - with a tube diff amp you get what you get.
I take your point thou' - I should have labelled the post SUBJECTIVE rather than OBJECTIVE
Weird Huh??
So what am I running this week - final listening test on a new amp I've built for a niece - Ultralinear + Cathode Bias (from 4 Ohm secondary windings) 6V6Gs with cascode current source biased 6SL7 diff amp front end. Zero global feedback. Harmonic balance not quite right yet (may need to degrade the current source a bit as I'm using some special Philips NOS 6SL7WGT which were intended for use in ElectroCardiograph machines and the 2 sides a re well matched)and also trying to establish if the 2.5 Ohm Output Impedance is too high for the nominally 6 Ohm speakers it to be used with. I use an Audio generator and an oscilloscope etc during set to work and in qualifying final designs BUT in getting to the final design I use just my ears.
Cheers,
Ian
The reasons are "Pure listening" - I've kept out of the "Listening Impressions of the AKSA 55N+" thread over at DIY Audio SS. Why - because I was one of the major contributors to the Nivarna Plus upgrade for the AKSA (I post as Ginger on the AKSA Forum at AudioCircle). I bought a 55 Nivarna before there ever was a Nivarna Plus and worked with Hugh on development of the Nivarna Plus Option. This took about 9 months as it involved extensive listening tests and letting new mods settle in before making SUB-jective evauation of improvements or otherwise. (i.e. the same time it takes cows and countesses to produce new offspring - although we know some blushing young brides can manage it in less than 9 months). Many ideas which my pure engineering analysis suggested would improve the amp distortionwise did NOT correlate to improved sonics. We gradually converged and arrived a a consensus as to what should go in and what should'nt although minor differences remain in my final 55N+ and what Hugh sells as the 55N+ (I use slightly less emitter degeneration of the diff amp and use "Super E" connected Blackgate Ns as the bootstrap cap rather than a single Blackgate Standard).
I can tell you that I have tried current sourcing the diff amp on the AKSA as well as tried current source loading the VAS in lieu of the capacitive bootstrap. In both instances the lovely "tubelike" warmth of the AKSA was lost and I abandonned those mods.
My view is that the resistive biasing of the diffamp in the AKSA introduces enough 2nd harmonic distortion to complement the amount of 3rd harmonic which exists anyway and current source biasing tube diffamp front ends reduces the amount of 2nd harmonic to again complement the amount of third harmonic. i.e. in both cases its attempting to get the 2H to 3H balance correct for best listening. With the SS diff amp we had to add a bit of 2H (by the resistor tail instead of a current source). With the tube diff amp we had to reduce the amount of 2H (by the current source instead of a resistor tail). The resistive biasing of the solid state diff amp is NOT technically "superior" - its just what gives the best sonic results. In the case of the tube diff amp I found that the better i could make the current source tail the better it sounded - hence cascode transistor current source with relatively high reference voltage (either green or blue led or the reverse connected transistor as the reference element). Unfortunately this will also depend upon the degree of matching of the 2 devices in the diff amp. In the AKSA the trasnsistors are selected to be matched - with a tube diff amp you get what you get.
I take your point thou' - I should have labelled the post SUBJECTIVE rather than OBJECTIVE
Weird Huh??
So what am I running this week - final listening test on a new amp I've built for a niece - Ultralinear + Cathode Bias (from 4 Ohm secondary windings) 6V6Gs with cascode current source biased 6SL7 diff amp front end. Zero global feedback. Harmonic balance not quite right yet (may need to degrade the current source a bit as I'm using some special Philips NOS 6SL7WGT which were intended for use in ElectroCardiograph machines and the 2 sides a re well matched)and also trying to establish if the 2.5 Ohm Output Impedance is too high for the nominally 6 Ohm speakers it to be used with. I use an Audio generator and an oscilloscope etc during set to work and in qualifying final designs BUT in getting to the final design I use just my ears.
Cheers,
Ian
The "tube sound" is largely due to harmonic distortion in the low registers caused by the output transformer. This gives the bass response an almost resonant, instrument-like quality. It's not actually unpleasant to listen to, but it sure ain't accurate!
Clipping behavior may play a small part, but I have an EL84 pentode amp with 26dB of NFB, and I assure you it's not harsh or grainy at all. But I took the time to stabilize it thoroughly at both high and low fequencies, and under ANY load conditions - including no load at all. Is this done by most companies? I doubt it.
Most amp makers do not measure, or give the consumer, distortion figures at 20Hz and 20kHz. If they did, you'd have a much better picture of what's going on. FFT Spectrums at 1kHz are meaningless, IMO, as are voltage respose graphs taken at microwatt output levels.
I'm convinced that if all audio amps had the same thorough, detailed specifications and test results, there would be no mystery about why some sound better than others. And it would end up having little or nothing to do with whether the underlying devices used were tubes or transistors.
Joel
Clipping behavior may play a small part, but I have an EL84 pentode amp with 26dB of NFB, and I assure you it's not harsh or grainy at all. But I took the time to stabilize it thoroughly at both high and low fequencies, and under ANY load conditions - including no load at all. Is this done by most companies? I doubt it.
Most amp makers do not measure, or give the consumer, distortion figures at 20Hz and 20kHz. If they did, you'd have a much better picture of what's going on. FFT Spectrums at 1kHz are meaningless, IMO, as are voltage respose graphs taken at microwatt output levels.
I'm convinced that if all audio amps had the same thorough, detailed specifications and test results, there would be no mystery about why some sound better than others. And it would end up having little or nothing to do with whether the underlying devices used were tubes or transistors.
Joel
Thanks to everyone who responded to my query (especially to Gingertube for taking the time to write a pair of essays!) The thoughts expressed so far seem to boil down to reducing power supply switching noise (quiet diodes, star earthing) and manipulating the relative levels of 2nd and 3rd harmonic distortion.
I agree that reducing power supply switching noise is essential. Regarding the distortion spectrum, I feel that it's important that it should remain constant with level, time, and if possible, frequency.
Level: By this I mean that as level falls from full power, the harmonic structure should either remain constant or higher harmonics should decay gently with level. The last thing I want to see is a reversal of relative harmonic amplitudes whereby at one level 2nd predominates over 3rd but at another 3rd predominates over 2nd. Differential pair drivers seem to avoid this problem.
Time: Blocking causes momentary overload to linger and is particularly objectionable. The cure is to DC couple the output stage from the driver.
Frequency: Making the distortion spectrum unchanging with frequency implies low global feedback. High feedback requires more compensation of the amplifier's open-loop response and therefore a varying feedback factor (and consequent distortion reduction) with frequency. At LF, we also require an output transformer that doesn't saturate at the slightest whiff of 20Hz.
I see distortion as a lace pattern through which we "view" the music. So long as we don't disturb the pattern, we can see through it and learn to ignore it.
I agree that reducing power supply switching noise is essential. Regarding the distortion spectrum, I feel that it's important that it should remain constant with level, time, and if possible, frequency.
Level: By this I mean that as level falls from full power, the harmonic structure should either remain constant or higher harmonics should decay gently with level. The last thing I want to see is a reversal of relative harmonic amplitudes whereby at one level 2nd predominates over 3rd but at another 3rd predominates over 2nd. Differential pair drivers seem to avoid this problem.
Time: Blocking causes momentary overload to linger and is particularly objectionable. The cure is to DC couple the output stage from the driver.
Frequency: Making the distortion spectrum unchanging with frequency implies low global feedback. High feedback requires more compensation of the amplifier's open-loop response and therefore a varying feedback factor (and consequent distortion reduction) with frequency. At LF, we also require an output transformer that doesn't saturate at the slightest whiff of 20Hz.
I see distortion as a lace pattern through which we "view" the music. So long as we don't disturb the pattern, we can see through it and learn to ignore it.
I have no problem accepting that our hearing is sensitive to many tiny nuances that we might not fully understand. It's true that test equipment is well developed and capable of somewhat amazing resolution. Similarly (though in a different way) our hearing has undergone many millions of years of evolution to help us identify tiny nuances in sound. Our ability to pinpoint the location of sound amazes me. Even though there have been lots of studies to help us understand how we detect directionality I think we're a LONG way from a full understanding of how our brain works together with our ears to process all of that information.
RE: the harmonic spectrum issue. I've come up with an imperfect analogy to vision. Imagine looking at a 'perfect' visual recording in a mirror. Adding 2nd order distortion would be something like putting a single bend in the mirror. The image would clearly be distorted (pun intended) but it would not be difficult to see things in the bent mirror. Higher order distortion would take the form of more bends. Low level very high order distortion would show up as many small ripples. Now add time related distortion: ripples that are moving around in ways that are unrelated to the image. Even though the ripples are small it would be pretty hard to see what's going on in the that mirror. Most of our depth perception would be lost... Give me back the mirror with just a single bend even if it is a big bend!
-- Dave
RE: the harmonic spectrum issue. I've come up with an imperfect analogy to vision. Imagine looking at a 'perfect' visual recording in a mirror. Adding 2nd order distortion would be something like putting a single bend in the mirror. The image would clearly be distorted (pun intended) but it would not be difficult to see things in the bent mirror. Higher order distortion would take the form of more bends. Low level very high order distortion would show up as many small ripples. Now add time related distortion: ripples that are moving around in ways that are unrelated to the image. Even though the ripples are small it would be pretty hard to see what's going on in the that mirror. Most of our depth perception would be lost... Give me back the mirror with just a single bend even if it is a big bend!
-- Dave
EC8010 said:
That partially captures my intent, however to be sure a quick clarification. It's generally accepted that when two stages, each generating a degree of 2nd harmonic distortion, are cascaded it's possible to reduce the THD on the output of the cascade pair by tweaking the 2nd harmonic of each stage. The 2nd harmonic distortion of the second stage sums destructively with the 2nd harmonic of the first, at the expense of new higher harmonics albeit at a much lower level. The theory is the same interaction occurs between lousdpeaker and amp, the second harmonic of an SE amp partially cancels the 2nd harmonic of a loudspeaker. The paradoxical result of cascading a non-perfect amp with a non-perfect loudspeaker is acoustic reproduction with a lower numerical THD. A kind of distortion pre-emphasis/de-emphasis I suppose.
I don't know if this has been tested and demonstrated. It is intriguing. Many SE amps appear to have 1 watt 2nd harmonic distortion figures in the range of common loudspeakers. It's more than just balancing harmonic number, it's reducing the total.
Again, I do think there's more going on. For example I wonder/suspect/believe the greatly reduced complexity of most tube circuits plays a part. (The super-simple 6C45PE spud headphone amp I'm playing with is an ear opener.)
I hadn't considered the possibility of distortion cancellation between an amplifier and a loudspeaker.
Another possibility is that the high output resistance of a typical SE amplifier reduces distortion in the loudspeaker. There are papers that show a significant reduction (20dB or more) in distortion using current drive rather than voltage drive.
Another possibility is that the high output resistance of a typical SE amplifier reduces distortion in the loudspeaker. There are papers that show a significant reduction (20dB or more) in distortion using current drive rather than voltage drive.
The cancellation thing has been brought up before. I have asked for some data but so far, the only thing I've seen was a link to some guy who did a pretty casual measurement that didn't show much; that didn't prevent the enthusiast from claiming that the test confirmed his hypothesis...
Of course, for this to work, phase and relative amplitudes of distortion must track with level of the two devices.
Of course, for this to work, phase and relative amplitudes of distortion must track with level of the two devices.
I've achieved cancellation between a CCS-loaded common cathode stage and a DC coupled cathode follower. Adding a variable resistor in series with a capacitor across the input of the cathode follower, I was able to achieve distortion of 0.01% at 20VRMS. Whether the same thing would work between an amplifier and a loudspeaker is another matter...
Hi SY. Do you mean casual tests investigating the cancellation between amplification stages or between amp and speaker? Somewhere in the pile-o-papers at home is a theoretical Spice-based analysis of the former from some publication or other, I think checked against a few rudimentary measurements and listening impressions. The author concluded it worked in reducing the total HD but didn't sound as good. I wouldn't mind seeing a more formal study as well.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.