Bob Carver's designs were often contoversial. I remember the "transfer function" demonstrations. At the time he was selling only SS gear, with the M400 being the smallest at 400 watts. I got it because an audiophile friend upgraded to the 1.5 Kilowatt model. There were those in the industry who claimed that Carver's "transfer function" was nothing more than a big resistor inserted in series with the speaker to raise the output impedance of one of his SS amps. It is interesting to note that he came out with some tube amplifiers a few years later. The Silver Seven model used 7 KT-88's per channel. It costs a whole bunch of money.
The M400 was his first "Magnetic Field Power Amplifier". Basically a very early attempt at a switching power supply that runs at 60Hz. My audiophile friend who became a Carver dealer sold about 10 of these. Mine was the only one that didn't blow up. The amp has no power switch, it is on all of the time. This is not good in Florida during lightning season. I used a power strip and turned it off when not being used. The power amp used 3 sets of output transistors running from 3 different supply rails. One, two, or all three are on depending on the power needed. A whole bunch of feedback is required to make this work. I still have the schematic somewhere.
The Plastic Tiger was the only one that didn't blow up. I think I still have the remains of a Universal Tiger around here somewhere also.
OK OK , Transistorlab signing off!
tubelab.com said:
Around this neck of the woods alot more people thot it was one of the worst amps ever made....
dave
tubelab.com said:
We only ever sold 1... us salesguys couldn't get behind it after it was demonstrated that an NAD3020 played louder before clipping (and sounded better)
dave
Yes, he has most definately given the audio community many focal points for 'discussions' over the years!tubelab.com said:
Apparently this is partially true. But accoring to this interview some other tuning was purportedly involved as well.
I just mentioned it because the phrase: "transfer function" sounds like it could capture something more encompassing about an amplifiers' behaviour beyond just a set of THD, TIM and slew-rate measurements. But alas, it seems more akin to a marketing term.
Was very highly regarded by TAS magazine at the time. Also a schematic for it was published in Glass Audio if I recall. (With very kind permission of Mr. Carver)
Ooh, lucky I didn't buy one back then. I actually auditioned one of his amps to compare it with what I was using at the time (and still now, unfortunately!). I can't remember which model his was; but mine is a Luxman amp from the 80's (the Z504). Sounds -- well, ...like it has too many components in it! However it can run at nearly 90w pure class-A. Using the Martin Logans CLS as the transducers, the Carver displayed pin-point imaging and good focus. However the imaging was too literally razor-sharp. Instruments seemed to have paper-thin widths though they were well placed on the soundstage. Was quite unnatural a phenomenon quite unlike anything I've heard before or since. However, overall clarity-wise, my amp sounded like mush.
I listed to some other gear while I was there to console myself. However, just before leaving, I had another listen to compare the two. Now, after an hour of my amp being on, the sound was very different and now sounded noticably better than the Carver. Instruments had body and the sound (though still veiled, imho) was rich and more enjoyable, and now had good focus too.
So, an eye-opening comparison for many reason. And illustrated the importance of component warm-up. (At least with a class-A amp, anyway. Though again, we should remember that given class-A operation, the "warm" part was quite literal too!)
I remember in the early 80's that one of the mags..it was either Stereophile or The Absolute Sound... called carver on his claim that he could make his amp sound like any other, and that his sounded the way they did because thats how he wanted them to sound.
So, they gave him one of the darling amps of the day (I can't remember which one) and challenged him to make his sound like theirs. They described his process as connecting the two amps out of phase with each other to a common speaker, and he tweaked his over several days until he got a null with program material. They then double blind tested the two amps..as I recall the reviewers could pick out there amp..barely..but they walked away very impressed with 'ol Bob. Carver admitted that he prefered the sound of the bogey amp, and a year later the "Silver Seven" that trans..er, tubelab described was introduced...and it WAS expensive.
This does seem to indicate that objective testing/aligning can predict subjective performance.
So, they gave him one of the darling amps of the day (I can't remember which one) and challenged him to make his sound like theirs. They described his process as connecting the two amps out of phase with each other to a common speaker, and he tweaked his over several days until he got a null with program material. They then double blind tested the two amps..as I recall the reviewers could pick out there amp..barely..but they walked away very impressed with 'ol Bob. Carver admitted that he prefered the sound of the bogey amp, and a year later the "Silver Seven" that trans..er, tubelab described was introduced...and it WAS expensive.
This does seem to indicate that objective testing/aligning can predict subjective performance.
Crown DC300....I think that it was 1971, and it is hard to remember the details, but I didn't keep it long. I went back to the Fisher 500TX (SS) that I was using before I got power hungry.
Carver M400..... I still have that one because it was unique. The one thing that this amp had was volume. It was LOUD really loud, and clean, but the Voxson that it replaced sounded much better but only had 1/10 of the power. I have since rebuilt the Voxson and use it, the Carver is stashed in the warehouse with the Phase Linear 4000.
The point that I was trying to make is that I have had a lot of different equipment over the years. People's tastes change over the years because of age, musical tastes, and yes peer pressure and clever marketing.
I let myself get talked into the Phase Linear - Carver system even though I liked my Voxson playing through vintage University 12TRX-B's in homemade cabinets. The Scott reformed my thinking in 5 minutes. I decided not to follow the audiophile crowd any more. I follow my ears, even though they still like the Voxson SS amp, and several tube amps. Follow YOUR ears, whatever they like.
A different view...(long, sorry)
(I’ve been away from audio forums for a while and I just now read this lengthy thread – luckily for my sanity)
It seems to me that discussions about gobs of added 2nd harmonics, flea-power SE amps, output impedances and clipping are clouding the issue of what are the ESSENTIAL differences between tubes and transistors. If we can strip bare the comparison and remove extraneous factors, we might increase our understanding.
A case in point: Consider comparing two line stages, one SS and one tube (let’s say triode for the moment). It’s is not hard to design (or buy) them both having similar and excellent conventional measurements/specs. I can design and I could probably find commercial designs, SS and tube-based, that BOTH meet these criteria:
Bandwidth from 1 Hz to 150 KHz
Distortion (THD or IMD) in the 0.00X% range across the audio band at all volume levels of interest (no clipping is involved)
Noise well below the noise floor of any source material
Output R around 1Kohm or so
Input R 100K ohms or so
Gain the same
No signal magnetics to confound the matter
Tone bursts, square waves, impulse responses, all virtually perfect
It is my contention, and my experience, that the tube and transistor circuits in this scenario will still sound different from one another. There will still be something essentially characteristic about the sound of each technology group. Granted, there are a thousand different solid state designs that could meet these criteria, and a thousand different tube designs that could meet these criteria, all sounding somewhat different from one another. But the essential tube and transistor character will generally shine through any half-way competent design. TO MY EARS, the tube circuits will almost always sound not just more pleasing, but more accurate to what I hear in music in a live setting. Of course the sounds of these two line stages will usually be more similar and convergent than the differences between a 300B SE amp and say, a Halcro, but the essential differences are still there, and that’s my point. So many times I've heard people try to explain why tubes sound different than transistors by pointing to one of the extra factors, like output transformer behavior. Don’t get me wrong, these factors hugely change the sound, but sometimes arguments about them obscure the underlying truth that there are essential differences between tubes and transistors, even without output transformers in the picture, for example.
If this is so, that the essential differences remain, then we can no longer blame those essential differences on such things as clipping behavior, signal transformers, euphonic 2nd harmonic enrichment, output impedance, coupling capacitor blocking, etc. If you think the differences might be due to passive components that are typically associated with each technology, go ahead and put Auricaps in the SS path, and make the tube circuit direct coupled (it can be done). Ditto for resistors, etc. Yes, the sound might change a bit, but still the essential characters prevail.
So what? We are forced to realize that our conventional measurements are missing something important to the ear/brain. This is nothing new to say of course, but it is important to consider it in light of the stripped-down nature of the line stage comparison.
So now let me go down a path that may be helpful to explain the technical differences. You may have read my postings elsewhere hinting of this topic.
The clue came from digital audio, of all places. There is mounting evidence that the ear is incredibly sensitive to digital jitter. Picoseconds do matter. Who would have guessed? It seems that that the ear/brain have evolved to locate the source positions of brief incoming sounds to a high precision, even in the presence of background noises and reflections, perhaps as a survival mechanism for evading predators and finding prey. To do that the ear/brain must have incredible timing acuity.
Let me digress for a moment, and then I’ll return full circle. Let’s look at the output of an FM modulator and an AM modulator on a spectrum analyzer, both having the same carrier and modulating frequencies. If the modulation indices are small in both cases, we’ll only see one significant sideband on each side of a carrier. The two modulation types will look the same on the spectrum analyzer, because the spectrum analyzer only measures “power spectral density” (PSD) in Watts/Hertz, and it doesn’t show phase differences. One sideband will have inverted phase in narrow-band FM, but you can’t see that on a spectrum analyzer. Now look at the two different modulations on a wideband scope. Big difference!
So how do we measure IMD in audio amps? Right, on a spectrum analyzer (whether using standalone equipment or FFT on a PC soundcard - it’s the same thing). We always assume that IMD is AM intermodulation. But this is not the case at all. It seems that IMD will be due to a combination of both AM and FM modulation. The spectrum analyzer doesn’t let us distinguish between the two by itself.
Could it be that one of the most important essential differences between tubes and solid state is due to the differing amounts of AM and FM in their respective intermodulation processes?
I hope to publish a paper on this topic sometime this year, but for now let me just allude to the fact that transistors (whether JFETs, bipolars or MOSFETS) generally have interelectrode capacitances which are extremely variable as a function of electrode voltages and currents, which is also to say variable as a function of the signal waveform. The interelectrode capacitances in tubes, while not perfectly fixed, do vary much less. In SS, those varying capacitances allow one component of the signal (music) to phase/frequency modulate another component, so that at least some of the measured IMD is FM/PM in nature. What this means in the time domain is that, for example, a bass guitar will “jitter” the phasing on an accompanying voice, and vice-versa. Or that some of the spectral components of a piano chord will phase modulate the other components. If the ear/brain is sensitive to picoseconds of jitter, what can be predicted for the sound of this kind of IMD? I’m not the only one who hears similar distortions in mediocre digital sources as I hear in mediocre SS amps. Is this a mere coincidence?
Enough for now. More to follow…
(I’ve been away from audio forums for a while and I just now read this lengthy thread – luckily for my sanity)
It seems to me that discussions about gobs of added 2nd harmonics, flea-power SE amps, output impedances and clipping are clouding the issue of what are the ESSENTIAL differences between tubes and transistors. If we can strip bare the comparison and remove extraneous factors, we might increase our understanding.
A case in point: Consider comparing two line stages, one SS and one tube (let’s say triode for the moment). It’s is not hard to design (or buy) them both having similar and excellent conventional measurements/specs. I can design and I could probably find commercial designs, SS and tube-based, that BOTH meet these criteria:
Bandwidth from 1 Hz to 150 KHz
Distortion (THD or IMD) in the 0.00X% range across the audio band at all volume levels of interest (no clipping is involved)
Noise well below the noise floor of any source material
Output R around 1Kohm or so
Input R 100K ohms or so
Gain the same
No signal magnetics to confound the matter
Tone bursts, square waves, impulse responses, all virtually perfect
It is my contention, and my experience, that the tube and transistor circuits in this scenario will still sound different from one another. There will still be something essentially characteristic about the sound of each technology group. Granted, there are a thousand different solid state designs that could meet these criteria, and a thousand different tube designs that could meet these criteria, all sounding somewhat different from one another. But the essential tube and transistor character will generally shine through any half-way competent design. TO MY EARS, the tube circuits will almost always sound not just more pleasing, but more accurate to what I hear in music in a live setting. Of course the sounds of these two line stages will usually be more similar and convergent than the differences between a 300B SE amp and say, a Halcro, but the essential differences are still there, and that’s my point. So many times I've heard people try to explain why tubes sound different than transistors by pointing to one of the extra factors, like output transformer behavior. Don’t get me wrong, these factors hugely change the sound, but sometimes arguments about them obscure the underlying truth that there are essential differences between tubes and transistors, even without output transformers in the picture, for example.
If this is so, that the essential differences remain, then we can no longer blame those essential differences on such things as clipping behavior, signal transformers, euphonic 2nd harmonic enrichment, output impedance, coupling capacitor blocking, etc. If you think the differences might be due to passive components that are typically associated with each technology, go ahead and put Auricaps in the SS path, and make the tube circuit direct coupled (it can be done). Ditto for resistors, etc. Yes, the sound might change a bit, but still the essential characters prevail.
So what? We are forced to realize that our conventional measurements are missing something important to the ear/brain. This is nothing new to say of course, but it is important to consider it in light of the stripped-down nature of the line stage comparison.
So now let me go down a path that may be helpful to explain the technical differences. You may have read my postings elsewhere hinting of this topic.
The clue came from digital audio, of all places. There is mounting evidence that the ear is incredibly sensitive to digital jitter. Picoseconds do matter. Who would have guessed? It seems that that the ear/brain have evolved to locate the source positions of brief incoming sounds to a high precision, even in the presence of background noises and reflections, perhaps as a survival mechanism for evading predators and finding prey. To do that the ear/brain must have incredible timing acuity.
Let me digress for a moment, and then I’ll return full circle. Let’s look at the output of an FM modulator and an AM modulator on a spectrum analyzer, both having the same carrier and modulating frequencies. If the modulation indices are small in both cases, we’ll only see one significant sideband on each side of a carrier. The two modulation types will look the same on the spectrum analyzer, because the spectrum analyzer only measures “power spectral density” (PSD) in Watts/Hertz, and it doesn’t show phase differences. One sideband will have inverted phase in narrow-band FM, but you can’t see that on a spectrum analyzer. Now look at the two different modulations on a wideband scope. Big difference!
So how do we measure IMD in audio amps? Right, on a spectrum analyzer (whether using standalone equipment or FFT on a PC soundcard - it’s the same thing). We always assume that IMD is AM intermodulation. But this is not the case at all. It seems that IMD will be due to a combination of both AM and FM modulation. The spectrum analyzer doesn’t let us distinguish between the two by itself.
Could it be that one of the most important essential differences between tubes and solid state is due to the differing amounts of AM and FM in their respective intermodulation processes?
I hope to publish a paper on this topic sometime this year, but for now let me just allude to the fact that transistors (whether JFETs, bipolars or MOSFETS) generally have interelectrode capacitances which are extremely variable as a function of electrode voltages and currents, which is also to say variable as a function of the signal waveform. The interelectrode capacitances in tubes, while not perfectly fixed, do vary much less. In SS, those varying capacitances allow one component of the signal (music) to phase/frequency modulate another component, so that at least some of the measured IMD is FM/PM in nature. What this means in the time domain is that, for example, a bass guitar will “jitter” the phasing on an accompanying voice, and vice-versa. Or that some of the spectral components of a piano chord will phase modulate the other components. If the ear/brain is sensitive to picoseconds of jitter, what can be predicted for the sound of this kind of IMD? I’m not the only one who hears similar distortions in mediocre digital sources as I hear in mediocre SS amps. Is this a mere coincidence?
Enough for now. More to follow…
I think you might be onto something here. Timing is everything...in speaker design...in business and in...well you know.
Just a little anecdote..that might not be relevant..but I wanted to share it anyhoo. Yesterday I received the cd "Hope - Hugh Masekela"...wow what a recording...practically a concert in my living room. Turned the CD over and what do you know... AAD..apart from the fact that it had been "mixed" "analogly" they had obviously messed very very little if any with that recording.
And that ties in with Corvus corax's statements on recording quality these days...
And like you said..the brain is one big calculator..that needs to know where is what when.
The more you mess with things the more you mess it up!
Looking forward to your next post...and indeed hope you are onto something and find some proof..and/or a way to measure it.
Food for thought...
Regarding AM vs FM: That's a very interesting proposition. Many thanks for taking the time to contribute.
Regarding AM vs FM: That's a very interesting proposition. Many thanks for taking the time to contribute.
In the world of high frequency RF power amplifiers (my full time job) we deal with a phenomenon called AM to PM conversion. It happens when you pass amplitude modulated RF signals ( in my case CDMA, iDEN, SAM, TETRA and other advanced digital radio formats) through a non linear power amplifier and then clean it up with cartesian (amplitude and phase) feedback. What a coincidence.
We have equipment designed to measure this phenomenon, although I doubt that it can operate at audio frequencies. I will check as soon as I get back to week (two weeks).
We have equipment designed to measure this phenomenon, although I doubt that it can operate at audio frequencies. I will check as soon as I get back to week (two weeks).
Yes, measuring and separating the small amounts of AM or FM/PM produced by an audio amplifier is not so easy, especially when it’s in the point-zero-something-percent range. That’s something I’ve been considering for a while. An interesting related observation: either modulation class, BY ITSELF, will produce equal amplitude first-order sidebands, assuming there are no frequency response tilts in the range of interest. However, when you add AM and PM products generated by the same inter-modulating frequencies, the two partial sidebands on one side add constructively. The other-side partials destructively combine since they’re out-of-phase. So when you see an intermodulation spectrogram in a magazine review and one sideband is taller than the other, you might consider that you are witnessing the combination of both AM and PM processes. You can’t deduce much from this by itself due to unresolved ambiguities, just an interesting observation. Likewise, when you see sidebands that are close in level, you can’t be sure whether the IMD is mostly AM or mostly PM, although AM is likely the bigger culprit except in the upper registers. I’m working on test waveforms and a test methodology.
By the way, I use FM and PM terminology somewhat interchangeably here since they are fundamentally similar processes that differ only by an integral-differential relationship of the modulating signals.
By the way, I use FM and PM terminology somewhat interchangeably here since they are fundamentally similar processes that differ only by an integral-differential relationship of the modulating signals.
Brian,
Is there some way you might exaggerate the effect... only for the sake of making it more measurable?
Is there some way you might exaggerate the effect... only for the sake of making it more measurable?
There was a Malcolm Hawksford paper in the Journal of the Audio Engineering Society about ten years ago, but I can't remember any more specifically than that.
The paper (Mills and Hawksford, "Distortion Reduction in Moving Coil Loudspeaker Systems Using Current Drive Technology" JAES 37:3, 129 (1989)) can be found at Hawksford's site and, I think, at Nelson Pass's site. Excellent reading.
Poobah asked:
That’s a good question. If you mean test signals for evaluating low distortion amplifiers, then, yes, there would be a class of test waveforms that would especially provoke the FM/PM mechanism. Generally, these will have significant energy at the upper end of the audio band where the non-linear (signal varying) poles exhibit the most (varying) phase shift. But your question may imply something else I’ve considered also. It might be useful to build a special test stage for correlating listening tests to measurements where the FM/PM production can be adjusted anywhere from a decently low baseline to a horribly exaggerated condition.
To me, the older group of solid-state opamps with low bandwidths and low slew rates probably exemplify the worst FM/PM offenders. They sound terrible even well below the onset of slewing. Generally, if all else is equal (and it never really is), any steps taken to increase bandwidth will move the offending variable poles higher and higher above the audio band, with less residual phase shift left within the audio band to vary with the signal. The better recent solid-state amplifiers tend to have higher bandwidths, at least internally. Ironically, the broader the bandwidth, the softer and more rolled-off the high end sounds very often with these amps. Paradoxically, many solid-state amps that sound like they have excess top end, the ones that sound zippy and etched, may really be suffering from audible FM/PM distortion products due to inadequate bandwidth.
I think that IMD, and FM/PM in particular, contribute to a hashy harshness when music gets complex. Have you ever noticed how many amps and systems only sound good with simple arrangements of music? But when lots of things happen loudly at the same time, you have to cover your ears or turn down the volume control. Many types of distortion mechanisms are worsened by complex passages, and few recording capture these events cleanly to begin with, but I suspect that a certain hot hashy sound, to which I am very sensitive, may be the result of the ear rebelling against complex phase “inter-jitter”.
That’s a good question. If you mean test signals for evaluating low distortion amplifiers, then, yes, there would be a class of test waveforms that would especially provoke the FM/PM mechanism. Generally, these will have significant energy at the upper end of the audio band where the non-linear (signal varying) poles exhibit the most (varying) phase shift. But your question may imply something else I’ve considered also. It might be useful to build a special test stage for correlating listening tests to measurements where the FM/PM production can be adjusted anywhere from a decently low baseline to a horribly exaggerated condition.
To me, the older group of solid-state opamps with low bandwidths and low slew rates probably exemplify the worst FM/PM offenders. They sound terrible even well below the onset of slewing. Generally, if all else is equal (and it never really is), any steps taken to increase bandwidth will move the offending variable poles higher and higher above the audio band, with less residual phase shift left within the audio band to vary with the signal. The better recent solid-state amplifiers tend to have higher bandwidths, at least internally. Ironically, the broader the bandwidth, the softer and more rolled-off the high end sounds very often with these amps. Paradoxically, many solid-state amps that sound like they have excess top end, the ones that sound zippy and etched, may really be suffering from audible FM/PM distortion products due to inadequate bandwidth.
I think that IMD, and FM/PM in particular, contribute to a hashy harshness when music gets complex. Have you ever noticed how many amps and systems only sound good with simple arrangements of music? But when lots of things happen loudly at the same time, you have to cover your ears or turn down the volume control. Many types of distortion mechanisms are worsened by complex passages, and few recording capture these events cleanly to begin with, but I suspect that a certain hot hashy sound, to which I am very sensitive, may be the result of the ear rebelling against complex phase “inter-jitter”.
That's what I'm talking about... not go directly for the 0.0X measurements... but sneaking up on them slowly. It could serve to support your theory; but also give you an opportunity to refine/develop measurement methods more easily. I can see how you could make a transistor amp worse... making a tube amp more "tubey" escapes me though.
- Status
- Not open for further replies.