tuga
Legal Alien
Stereophile uses a simulated load created by Ken Kantor (Acoustic Research, NHT):
https://www.stereophile.com/reference/60/index.html
Valve distortion - What is it and what does it do to?
- Thread starter Greyandy
- Start date
Stereophile uses a simulated load created by Ken Kantor (Acoustic Research, NHT):
https://www.stereophile.com/reference/60/index.html
Jim Audiomisc
pfm Member
(ahem) You omitted his "well designed" qualifier. 8-]
My view is almost the direct opposite of what you wrote because it does include the "well designed" expectation. For pre-amps it is much easier to get to a decent level of "well designed" as they can work into a better defined load/output requirement than the power amps.
This isn't the same as saying "all amps sound the same". Because some designers may deliberately (or unintentionally) develop an amp that is *not* "well designed" in the above terms, but "tweaked" to have a "sound" which departs from being "indistinguishable from a piece of wire with gain".
In the absence of user selectable tone controls, etc, though I'd expect "well designed" to mean a specific set of behaviours like a flat response, etc, allowing for specifics like RIAA on a relevant input.
If an amp - pre or power - gives a 'nice sound' by adding distortion or fiddling with the frequency response, then the makers should say so. Makes it easier for punters to make an informed choice.
Tony L
Administrator
It is one of those ‘how long is a piece of string’ things. What you mean by “well designed” is likely different from many other amp designers. I’m also curious by what you would personally assess and measure to arrive at the conclusion you had designed it “well”, and how that would differ from say a Mark Levinson, Krell, Naim, Pass, Quad etc. They all sound different and we haven’t even got beyond solid state yet!
As I suggested in the post above I’m far from convinced the mainstream measurement criteria of a flat frequency response and low THD tell the whole story here. Devices that measure very well on these metrics still manage to sound different, hence my suspicion much of what we hear as differences exist in the dynamic headroom and phase/timing domain. The number of respected amp designers that deep dive this aspect with hugely over-spec multi-rail PSU designs that enable huge headroom etc and are able to demonstrate the subjective benefits of doing so (e.g. Naim’s PSU hierarchy) suggest there is something to it. There are certainly many legitimate ways of thinking about what superficially appears a simple problem. It isn’t as clear cut as some would like to believe.
OK, specifically for a pre-amp designed for CD/Streamer high-level input, for "well-designed", substitute "not badly-designed", i.e. measured very low distortion (over the voltage levels encountered with real signals), near-flat frequency/phase response over the audio band, high input impedance and low output impedance. One could put number bounds against those, but you get the general idea, and I would hope most products meet this broad description as its fairly trivial to achieve. (And the frequency response criteria implicitly bounds the time-domain impulse response).
If they can sound significantly and reproducibly different - what's going on?
Tony L
Administrator
Exactly. In any situation where the measurements suggest similarity yet subjective results are different implies the wrong things are being measured.
I defy anyone not to be able to tell a subjective difference between say a high-quality stepped attenuator passive, a Quad 34, and a £6k Mark Levinson, Krell or whatever. In fact I’d go further and suggest most people will be able to tell the difference in a truly good system between multiple contenders in the £2k+ solid state line stage market, i.e. items that are not compromised or cost-cut to a price point the way say the aforementioned Quad 34 was.
PS I’ll deliberately leave valve line stages out as there is a character aspect here between specific valves (e.g. a 1950s Mullard long plate ECC83 sounds different from a mid ‘60s short-plate one, let alone say an RCA), plus vast difference between products. I like/choose valve kit, though I accept I am rolling my system to my taste rather than someone else’s abstraction of the term “accuracy”. I’m perfectly happy doing that.
John Phillips
pfm Member
As @Tony L writes there are - from long observation - different definitions of "well designed". Mine does start with the technically conventional definition and the extra word "sufficiently". However, I suspect there could be technically explicable effects that fall between the conventional testing cracks, but I don't know if I am just being paranoid.
Transient headroom is mentioned. I have seen amplifiers where the output sticks to a power rail for a short while after a voltage overload. I have seen amplifiers that recover from voltage overload slowly and sometimes via short periods of oscillation. I have seen - long ago - published tests on audio power amplifiers that showed such effects but no such testing recently. Music is very dynamic and ISTM a system's headroom for transients may not always be enough. So at least clean behaviour on overload seems to be needed but no longer tested.
I have long wondered about this as a subtle but real audible problem. I suspect this is to do with feedback. It's not IMHO possible to have too much feedback. But IMHO it is possible to not have good enough open loop behaviour for when the feedback loop - designed typically assuming a simply characterized open loop - "unlocks" on transient overload. With either voltage limits or current limits or gain changes due to loudspeaker load impedance torturing a power amplifier into revealing its less than simple open loop behaviour.
I may be off at a tangent here but it's something I have always wanted to test but never had the opportunity.
Tony L
Administrator
To my mind ‘valve distortion’ is a phrase that (outside of the guitar amp world where it has a whole other meaning) is always delivered as part of an objectivist/crusader agenda, and usually from a salesman of some kind. I don’t accept its legitimacy in the high-end audiophile world.
I do think valve amps can sound remarkably good, and maybe more convincing in some respects, and I suspect this is actually due to a lack of distortion etc in the key midband areas we are most sensitive at as almost all valve preamps will I assume be class A devices. I suspect this may help explain their ability to create a believable three dimensional soundstage a lot of solid state kit seems to flatten to some degree. That said some (usually expensive) solid state kit can produce many of the things I like about valve amps, so I’m clearly comparing design ideology more than component type here really. There is a lot of kit from many almost contradictory design schools I like which is why I try to avoid absolutism or groupthink on this one (or anything in audio, it actually really irritates me).
I accept I don’t understand what is missing in measured specification and why it doesn’t tell the whole story, and will therefore select by taste. I also have no issue if I end up preferring stuff that measures less well on paper, e.g. I’d audition say Accuphase vs. Shindo with a totally open mind and take home the one that subjectively matched my taste and system better.
IMHO its not difficult or especially expensive to produce a fully class-A low-distortion semiconductor pre-amp with good open-loop gain linearity (& very much better closed-loop linearity of each stage via feedback) and avoiding overload effects over the expected signal voltage range, or any of the other effects mentioned above, i.e. ticking all the boxes on what "good" means in terms of traditional specification and measurement. Can also make fully balanced if that helps.
So I'd be genuinely interested in any input from the design community (especially professional, as I know there are some on this forum) as to what "traditional" measurements are missing, and what makes a great-sounding pre-amp (keeping to straightforward line-level input stages rather than the more onerous design of disc pre-amplifiers)
Tony L
Administrator
Likewise, though I doubt you will even find consensus regarding negative feedback levels yet alone anything else. We like to think we understand everything, yet there is so much diversity in thinking really all one can ever do is pick one’s own “expert” based on their own proven track record, e.g. if you say prefer the sound of a Pass to a Krell you’ll likely pick Nelson Pass over Dan D’Agostino, or vice-versa. There certainly are a lot of design professionals with solid product portfolios and design ethos to assess!
Audio perception is complex and sensitive to tiny effects, and clearly real subjective differences do exist which are hard to rationalise. But IMHO the whole picture is coloured by a vast quantity of unreliable information due to hasty/non-rigorous subjective impressions in magazines & online (I don't mean you, Tony), perpetuation of technical myths (e.g. some argue high-levels of feedback is a bad thing, per-se), and an industry that understandably has a vested interest in selling more kit. So my personal take is that there is something perhaps not understood, but there is a such large quantity of BS around in the world of audio, even from established vendors and dealers, that it takes the focus off any likelihood of uncovering it.
The fact that we can happily listen to kit made 50 years or so ago shows that not much progress has been made in terms of fidelity (at least in analog audio).
IMHO good imaging is very beneficial for musical enjoyment, making it easier for the brain to identify the separate strands of a piece of music, and making better sense of the ambiance. One of the biggest enhancements I made recently was moving the speakers forward by 12inches (they are so heavy that I havn't experimented enough with placement) which really made the image snap into place.
But in a (wide-band) line-level pre-amp there isn't much to go wrong surely (not to say some designs might not be good)? At least not compared to differences in speakers and possibly the power-amplifier.
Tony L
Administrator
Agree entirely. I’d go further and suggest that many blind alleys have been followed since. I’d certainly argue that you have to go a hell of a long way with modern kit to equal say a pair of Quad IIs driving ESLs, a McIntosh MC40 driving Altec horns or even the humble Quad 303 driving 15” Monitor Gold loaded Lockwoods I’m listening to right now.
Bizarrely preamps are an area that genuinely has evolved. There are very few in even the hardcore retro audio community who would choose an old Quad 22 or Leak Varislope over say an Audio Research, Conrad Johnson, Mark Levinson, Shindo or whatever on sound quality grounds, but they’ll very happily keep their vintage turntables, power amps and speakers. Ok, some of this is down to the huge change in sources with modern 2V outputs clipping vintage kit, but even on an equal playing field I’d not swap my Verdier valve pre for a 50s Quad or Leak no matter how much I like the matching power amps. FWIW I’d not swap it for an all singing and dancing modern digital preamp either, but that’s for a magnitude of other reasons (Right To Repair etc).
PS To my mind ‘peak preamp’ probably occurred in the 1980s. Around the time valves came back, but before all the modern remote control and blue LED or LCD display bling-fi nonsense we suffer today. No matter how high my budget I’d likely look back to that era. It’s good honest serviceable kit that sounds great and is good value now second hand. If I didn’t have the Verdier I’d likely have an ‘80s Conrad Johnson or maybe an SP3 or 8.
Jim Audiomisc
pfm Member
Agree with you last (quoted) comment. There is more to it than the basic FR and THD. But the list is finite. e.g. it will include inaudible levels of hum and crosstalk. (The hum point matters as any ground loops, etc, need to be taken into account.) For a *preamp* it also includes points like not having any problem driving its output load. Must also reject mains or air-carried RFI to ensure it has no audible effect.
Amps like the ones you mention may well sound different because they *are* different in terms like the above. No mystery or magic. e.g. The QUAD power amps having a tailored FR, and differences in power amp output impedance, levels of dynamic rail effects, marginal stability, etc.
There was a time when mags showed results for, say, how the added mains buzz varied with signal level and loading. Dynamic changes in this due to poor rail rejection may mean that music is particularly affected as loud transients may generate crap - sometimes on the *other* channel.
Now they don't, and few people check for this sort of thing. Takes too long and needs too much awarness for most 'reviews'. So they tend to just use a subset of measurements and then waffle about what they (think they) heard.
Add in details like marginal stabilty so a given design may give bursts of HF oscillation and distort some waveforms when driving some loads. (mainly a power amp risj). Hence - in my view - a need to ensure unconditional stability.
As a designer/developer you can spend weeks or months doing such measurements and checking, tweaking. But then it gets ignored by reviewers. Who may then prefer some other design that *didn't* get this work, because in their system it sounded OK.
cf examples on my Armstrong pages.
And, of course, some people *like* "creamy valve distortion" anyway. Just as some others prefer a *non* flat response as it makes their taste in music/speakers 'better'.
So the above is my view, but then others would choose something different. That's fair enough, but better in my view that people be aware of the above as it helps inform the choice.
Jim Audiomisc
pfm Member
Yes. And experienced designers will generally know why/how they may vary - in a way the often depends on the conditions of use, etc. Take the most obvious example of the FR of the 34.
Jim Audiomisc
pfm Member
Most of the required measurements to fully characterise an amp are known and have appeared *at times* in magazines. The problem is that a full and well understood use of them all is rare. Reviewers and mags just choose a subset they are familiar with and think 'best'. Saves time and money and avoids drowning readers with data they can't then make use of. At least, that's the argument.
Jim Audiomisc
pfm Member
Yes, happens. I think it means a stage has gone well above the output sufficient to drive the following one. So loses loop and takes time to 'come down again' and let the output unstick from the ceiling.
FWIW I added a sensor to the 732 that lights up if the amp clips. (or senses RFI). Generally didn't come on as the amp was > 200wpc continuous and fair bit more for transients. No light, no clipping or slewing or rfi.
Jim Audiomisc
pfm Member
One factor that I suspect sometimes affects imaging in some systems is when the amps generate some mains buzz/hum when the music is loud. Background LF seems to affect imaging for me. One of the reasons I'm wary of letting an amp produce dynamic hum/buzz.
Tony L
Administrator
It is far more than frequency response though, in comparison with preamps at this level the 34 sounds flat, two dimensional and lacking in dynamic range. You couldn’t get it to sound like a top end CJ or whatever just by altering the EQ, it is way more profound than that. A whole extra level of resolution, scale, depth, slam etc.
PS I’m not slagging the 34, it is far better than its reputation suggests IMO, but it can’t compete against highly regarded classic preamps at many multiples of its price point.