Just read the Fremer review of the Ypsilon Hyperion...interesting
- Thread starter morricab
- Start date
No. You might find more under "memory distortion" in a google search.
They don't show up in THD because all THD is non-linearity. They are not as easy to measure, and most equipment won't work for taking it. But from device to device can vary a lot, and it is easily audible if the distortion is high (to me, I don't know if others notice as much).
They don't show up in THD because all THD is non-linearity. They are not as easy to measure, and most equipment won't work for taking it. But from device to device can vary a lot, and it is easily audible if the distortion is high (to me, I don't know if others notice as much).
Not the old Luxman. And I did not say pleasurable, I said it did drive and bass and stage and Midbass well in a direct compare
This seems to relate mostly to input stages which is clearly inapplicable in the case of Ypsilon. Of course one may argue that any signal related bias instability produces memory distortion. With class A bias the entire concept becomes meaningless.
KeithR said:
KeithR said:
Hello Keith,
I’ve been reading and pondering your comments and wonder if you can perhaps clarify your position.
Are you objecting to the measured performance of the Hyperion; or, the fact that MF preferred it in spite of JA’s measurements; or, that its measurements seem to suggest a deficiency of performance relative to its price; or, that it has a “completely voiced sound”?
I ask because both pairs of Zu’s that were measured by Stereophile according to JA had “awful... cumulative spectral-decay (and) waterfall plot(s)”, “the drive-unit (of the Essence) is in breakup at and above 2kHz, with a large suckout evident in the presence region” and the Essence's anechoic frequency response looked like this:
Though the Zu’s are certainly lower in price to the Hyperion, among other speakers of a similar price bracket (or less) it measures rather poorly in comparison. If it’s possible Zu Audio owner's subjective enjoyment of them is not hampered by their objectively poor performance, wouldn’t it also be possible that MF could also derive significant subjective enjoyment from the Hyperion despite their objective performance?
After all, in conclusion, JA stated thus: “I suspect that Zu's designer has carefully balanced the individual aspects of the Essence's design so that the musical result is greater than the sum of its often disappointingly-measuring parts.”
Perhaps to a degree that’s true that all gear, irrespective of whether it retails for $15,000, or in the case of the Wavac, 10 x that price.
Best,
853guy
Last edited:
Sorry to pick nits here but memory effects in amplifiers are non-linear in nature - these are not a source of linear distortion.
You are correct in saying memory effects vary device to device and are typically difficult to correct for. Digital pre-distortion techniques are typically used to correct for these in RF circuits but I am not aware of anyone doing this with audio amplifiers. Since this distortion is bias related, a solid (low source impedance) and wide bandwidth supply circuit is the most straightforward way to minimize this type of distortion. There is also a thermal component to memory effects which is typically more difficult to deal with.
Cheers, Joe
It is applicable. But using tubes is often an inherint way to have it be lower. But then it has SS outouts. Like I was saying, I'd like to hear it.
While the term "memory" might be applicable to other things, I have used it correctly. You seem to be mixing phase and memory distortion (which many do online that may be a source for you). I am exclusively not talking about phase, because I am talking about linear distortions. DAC filters struggle with phase. You can try to reduce problems with caps etc, but adjusting for memory distortion is about the act of controlling the active pieces (tubes & transistors) because you can control them so much.
Linear distortions are almost exclusively thermally related. But just running high input impedance and bias is not a gurantee of performance.
These guys have a decent laymen description. They have a more technical PDF somewhere as well (with mediocure transistor models, however).
I probably should not have brought this up. It is a bit advanced for the non-circuit designer. It is just what was on my mind.
Are you addressing the work of Gérard Perrot on thermal induced distortion? He wrote some very complete and interesting articles on it, unfortunately most of them are ignored as he wrote them in french. Unfortunately his patents do not cover the why's if this subject.
How can anything NOT be considered voiced, since nobody can agree on the "absolute sound", the "recording studio sound", or a flat measured frequency response in an anechoic chamber. It is all a tower of babel. The idea of "voicing" ceases to have any derogatory meaning, or any meaning at all, if there is no prior standard as to what voicing even means.
The Lamm ML3 and the Wavac SH 833 share one thing in common: an intense rendition of the power supplies. The ML3 has an enormous power chassis with multiple tube rectifiers, which I would surmise, has a lot to do with the smoothness of the sound.
The Wavac SH 833 has four options: a basic chassis for about 120K. The upgrades add power supply modules only, up to three additional, each at about 80k apiece to create the final four chassis monster.
I have not come across any discussion of WHY designers choose such elaborate and expensive power supplies, or what exactly the power supplies contribute to the sound aka how does a single chassis Wavac compare to dual channel Wavac vs. the ultimate four channel model etc., although it is clear that the designers consider this of utmost importance to the ultimate sound these big SET amps. The amplification stage is regarded as a gate for the power supply. Most audiophile discussions obsess bass ackwards from this view and consider the amplification portion the end all and be all of the device.
The Lamm ML3 and the Wavac SH 833 share one thing in common: an intense rendition of the power supplies. The ML3 has an enormous power chassis with multiple tube rectifiers, which I would surmise, has a lot to do with the smoothness of the sound.
The Wavac SH 833 has four options: a basic chassis for about 120K. The upgrades add power supply modules only, up to three additional, each at about 80k apiece to create the final four chassis monster.
I have not come across any discussion of WHY designers choose such elaborate and expensive power supplies, or what exactly the power supplies contribute to the sound aka how does a single chassis Wavac compare to dual channel Wavac vs. the ultimate four channel model etc., although it is clear that the designers consider this of utmost importance to the ultimate sound these big SET amps. The amplification stage is regarded as a gate for the power supply. Most audiophile discussions obsess bass ackwards from this view and consider the amplification portion the end all and be all of the device.
Oh man that ZU freq plot looks awfull , one the nastiest i ve seen
The mid / tweeter combo looks like a complete mismatch, definetively a coloured design which will sound lifeless in the 4 kHz area.
Nothing to do with voiced imo Just bad design.
How would you feel if your car had 20 Percent less power output at 4000 revs
The mid / tweeter combo looks like a complete mismatch, definetively a coloured design which will sound lifeless in the 4 kHz area.
Nothing to do with voiced imo Just bad design.
How would you feel if your car had 20 Percent less power output at 4000 revs
Last edited:
Vinyl and tape are extremely voiced media and sound great, attracting the preference of most of us. Curiously they show the same trend of increasing distortion versus level of the Hyperion's.
Left - Studer B77 - Third harmonic distortion vs. record level at 1 kHz and Ortofon OM40. Right - Ultimate Analogue Test LP, track 1 on side two, sweep 315 Hz from -2 dB to +12 dB .
Attachments
My B 62 stays within +- 1 db from 60 Hz - 15 kHz.
+- 2 db 30 Hz - 18 kHz i have the original testresults with the machine .
Well digital has other problems lol, they fooled me once with an expensive player never again
+- 2 db 30 Hz - 18 kHz i have the original testresults with the machine .
Well digital has other problems lol, they fooled me once with an expensive player never again
If you read some of the solid state designers, they treat second and third order harmonics like Colonel Sander's herbs and spices, a dash of one or the other seems to provide a bit of brain dither that makes the sound nicer.
Nobody seems to like higher order harmonics, and the consistent choice of lesser or no feedback by listeners seems to support that as far as real world performance of amplifiers is concerned, some kind of noise/harmonics dither is preferred and even aids in the perception of the sound.
Nobody seems to like higher order harmonics, and the consistent choice of lesser or no feedback by listeners seems to support that as far as real world performance of amplifiers is concerned, some kind of noise/harmonics dither is preferred and even aids in the perception of the sound.
Well, my true reference is our own proprietary work. It's hard to point a "source" for something that is mostly a scattered thing on the net, with a few discussions here and there. For that reason I will point at Perrot's articles since I don't have one of my own.
EVERYTHING is voiced in some way or another. Electronics and speakers are simply too imperfect to possibly pretend like that isn't the case. There are still a lot of mysteries between why some things sound different.
The 2HD and 3HD are useful tools. Using either can make up for other problems in such a way that still gives a world class product, that otherwise might have not hit the right "feel" for audiophiles. I don't like a ton of either, but depending on the product a little purposeful amount might be just what something needs. It is an error to think this in some way is a sure "defiance" of the artists intention or anything like that. The reality is the artists and studio engineers/masters may have their stuff tuned with some different 2HD or 3HD amounts, in electronics or speakers, so you might be getting closer to what they intended by having a little of either - and it varies from album to album.
No I am not mixing phase and memory and am not getting my information from online - I am an EE and have worked with RF power amplifiers and memory effects for some time. You may be using the term correctly but it’s not clear as you claim it is a form of linear distortion. My main point was that it is not - it is inherently a non-linear effect - see here as one technical description - https://www.lintech.com/PDF/min_pa_memory_fx.pdf
Intuitively this makes sense as memory effects dictate that the current state of an amplifier is dependent on its previous state (hence memory is involved) - this cannot possibly lead to a linear response. A linear amp’s state would be completely and uniquely defined by the parametric conditions at any point in time, independent of its previous state.
I can’t be certain but you seem to imply that only harmonic distortion is non-linear distortion and all others are linear but there are other sources of non-linear distortions in circuits. Please excuse me if I’ve misconstrued your meaning.
As you say, this is probably not interesting to most so we can drop it - as an engineer it’s sometimes difficult to let these topics go by
Cheers, Joe
RF and the audio band are very different. There is hardly anything linear in RF amplification. Even a linear act in RF equipment can end up not being linear due to the wackiness of phase. However the thermal memory in the signal stages for the audio band of an amplifier is overwhelmingly linear.
I'm sure no one cares...
I'm sure no one cares...
You can do this how you like but if you do not take into account the much more sonically detrimental high order harmonics, you are basically seasoning crap, IMO. You cannot mask the negative sonic consequences by putting in more or less 2nd or 3rd harmonics. By definition on how the hearing works the higher orders are outside the masking zone...not to mention they don't follow a natural monotonic pattern (or as Cheever says the "aural harmonics").
going to lower or no feedback helps, but you still need to operate in Class A or you will get a host of high order harmonics anyway (see Boyk and Sussmann's simulation paper). Also, the inherent non-linearity of all devices is different. Triodes are less non-linear than Mosfets, which are less non-linear than bipolar transistors...this is the basic physics of the devices. These device you like (VFET or SIT) are different again but seem to be more in the direction of a triode.
The consequences of this inherent transfer function leads to a different pattern of distortion that will impact the outcome. Yes, it can be tailored to some degree but it cannot be circumvented and turn a Mosfet character into a triode, for example just by playing with bias.
The more complex the circuit (number of active elements) the worse the problem will get, as will the circuit type itself (single enede , push-pull, differential etc.)...once the circuit gets complex it HAS to have feedback to measure acceptably (also perhaps to sound acceptably). Cheever defeated the feedback on a Hafler amplifier that measured wonderfully with feedback and had 25% THD without it. An extreme case? Perhaps...however, Hafler amps were quite popular in their day.
You should read a bit more on the studies that have been done looking into why things sound differently...there is quite a bit of work out there. While all is not known (and probably never will be), there are some good guidelines already available to those who would be interested in applying them.
RF and the audio band are very different. There is hardly anything linear in RF amplification. Even a linear act in RF equipment can end up not being linear due to the wackiness of phase. However the thermal memory in the signal stages for the audio band of an amplifier is overwhelmingly linear.
I'm sure no one cares...
A linear distortion is only impacting the amplitude of the signal. If phase or frequency are impacted by the distortion (such as new harmonics or sidebands etc.) then it is by definition non-linear as these new bands are being created by higher order effects. I also cannot see how these memory effects could be purely linear and not affecting the frequency domain. Temperature effects in transistors are never linear...as transistors themselves are also highly non-linear...why would secondary effects suddenly be linear??
Similar threads
