Apropos of amplifier noise, see the fascinating review of the Devialet D-Premier in the January 2013 Stereophile. JA does both the listening and the measuring, and although he is very enthusiastic about the product, his measurements are difficult to reconcile with the manufacturer's claims, particularly of a S/N unweighted of 130dB (no reference level given, but the max power is specified at 240W into 6 ohms, for which I calculate just under 38V rms).
If a spec is bad yet not to the point of disaster could it be a choice ? Certainly it would have a masking effect . If the spectrum was pinkish I would be especially interested . This is so fundamental as to look suspiciously like a choice . I spend most of my life measuring noise and saying to myself how good do I need this to be . For me it is impossible to beleive an engineer can make a mistake over this . Many circuits I design aspire to 78 dB ! I think someone said the electrons are counted in how many per second passing not how many million when the signal level is 100 uV . I sometimes use carbon composition resistor for inputs . They are not good on measurements . This becomes a non issue when the pre amp is connected . Strangely although the measurements become acceptable in real life with a pre amp shunt the sound is non the less changed . I do not like CC resistors . However I like them more than other cheap ones . Tyco non inductive are my favourites , $8 in bulk a piece . I am told pulse lasers keep the production of cheap CC resistors going .
I obviously do not believe specs can tell us much about the sound of anything.
Specifically, the spec on noise: I have had the opportunity many a time in life to hear devices with brilliant specs which didn't sound all that good at all, as well as devices with very humble specs which sounded brilliantly. "Humble" means at or just below par with its respective price class average, not as Nige says disasterous.
What I heard the least in my life were power or integrated amps which interfaced well with ahything I chose to call a speaker. One such item is the H/K 6550 integrated amp, despite its relatively modest power output of 50/70 Watts into 8/4 Ohms - BUT, it does its thing into everything I ever threw at it, both at home (AR94, JBL Ti600, B&M 1041 Monitor) and outside it. I am not saying it loved them all equally, but I am saying that the difference was surprisingly small, and double surprisingly in view of its relatively low price.
While some of its competitors, with much better specs, fared much worse. The fact that they had better S/N ratios didn't do them any good.
Specifically, the spec on noise: I have had the opportunity many a time in life to hear devices with brilliant specs which didn't sound all that good at all, as well as devices with very humble specs which sounded brilliantly. "Humble" means at or just below par with its respective price class average, not as Nige says disasterous.
What I heard the least in my life were power or integrated amps which interfaced well with ahything I chose to call a speaker. One such item is the H/K 6550 integrated amp, despite its relatively modest power output of 50/70 Watts into 8/4 Ohms - BUT, it does its thing into everything I ever threw at it, both at home (AR94, JBL Ti600, B&M 1041 Monitor) and outside it. I am not saying it loved them all equally, but I am saying that the difference was surprisingly small, and double surprisingly in view of its relatively low price.
While some of its competitors, with much better specs, fared much worse. The fact that they had better S/N ratios didn't do them any good.
For whatever reason, great specs by no means guarantee good sound.
The number of possible reasons for this is impressive.
Excellent specifications are only meaningful in the context of a complete, working system -- and measured over a wide range of operating conditions. Otherwise, it will just lead you astray.
It's as simple as, how strong is a chain? As strong as the strongest link, the average of the strengths of all the links, or precisely that of the weakest link(s). Strangely enough,
, for me I've found it to be always the last scenario, so the answer has been to work out which were the dodgy ones, even if they were boring as hell, no bling or pizzaz associated with them whatsoever, and sort them out ...
Frank
It's as simple as, how strong is a chain? As strong as the strongest link, the average of the strengths of all the links, or precisely that of the weakest link(s). Strangely enough,
, for me I've found it to be always the last scenario, so the answer has been to work out which were the dodgy ones, even if they were boring as hell, no bling or pizzaz associated with them whatsoever, and sort them out ...Frank
The trouble is, no-one is doing it for complete systems, which means, measuring various performance parameters for the complete path from the source file through to the output of the speakers. Under a variety of circumstances, for a range of testing material. Yes, this is all going to be difficult, good microphone capture techniques will necessary and highly capable software to crunch the results, comparing the output to input. But, if the bullet is bit, then I'm sure the results will be extremely illuminating; you will, for example, discover how the distortion varies depending on the usual things that people fiddle with, in their tweaking.
And I believe it will tell a very large part of the story ...
But of course all this is too hard, too messy, too everything, so it won't be done ... probably ever. So the audio fraternity will continue to circle each other, taking odd pot shots at each other's point of view, and no progress will be made, c'est la vie ...
Frank
A fairly reasonable definition for "good sound" is that it's of a quality capable of fooling at least some of the people, some of the time, I would have thought ...
Frank
I did these some years ago . The graphs are for my reference and not to any weighting curve etc . Note how the 7R ( a bit for wires added ) load of the Lyra Helikon is always dominant . The outcome is interesting . The Tyco foil resistors were supplied by a friend to fit . I listened and was very surprised . The foil resistors made the pick up sound as if the stylus was super clean . The phono stage had no other special parts apart from polystyrene capacitors . The carbon composition sounded better than the metal film , noticeably 1950's . I could imagine a mix of film and comp might work ( 68 K CC + 300K + ap 300K film ? ) . I did try reverse phase films as in radio amateurs book ( ty-rapped ) . I think it helped ? Better PU's like Helikon do not seem to require special loading . 22R to 47 K seems OK . As incredible as it might seem the inductance might be important for a phono stage . That seems crazy . Good advice I think is use 47 K until certain . Shure's like much more ( 68 K even 1 M , adjust EQ to suit ) . One PU specifies 845 R !
Yet, often enough, when one replaces standard carbon 5% resistors with 1% metal film in critical places (Input and NFB path, input stage), one is rewarded with better sound and more clarity, sometimes just barely audible, other times clearly audible, as the case may be.
Truly, the answer is not always in the light where the light shines.
Truly, the answer is not always in the light where the light shines.
I completely agree great specs does not mean great sound. For about 7000 posts we have been wondering which specs may actually be important, and have pretty much agreed many are pretty useless, like DF.
But, how often do you get great sound with terrible specs? Not often. Kind of like one of my favorite lines: "You don't always get what you pay for, but if you don't pay for it, you won't get it"
Just a few more furniture projects to finish up and I will get back to the amp testing I had planned. I have been so stressed from work I can't focus on studying circuits. Making sawdust seems to help. Still, I want to HEAR the differences in various CCS designs, not believe SPICE. I have a suspicion we have been over-emphasizing stiffness for consistency. Just a suspicion.
But, how often do you get great sound with terrible specs? Not often. Kind of like one of my favorite lines: "You don't always get what you pay for, but if you don't pay for it, you won't get it"
Just a few more furniture projects to finish up and I will get back to the amp testing I had planned. I have been so stressed from work I can't focus on studying circuits. Making sawdust seems to help. Still, I want to HEAR the differences in various CCS designs, not believe SPICE. I have a suspicion we have been over-emphasizing stiffness for consistency. Just a suspicion.
@tvrgeek
I am not sure just how much can one actually hear this or that CCS design, if you mean CCS as a part of a say differential circuit. In that case, a CCS is sort of an auxilliary ciruit, necessary for proper operation of the key, or primary ciruit.
If you meant the entire CCS stage of an amp, that's a different story. But even so, assuming no obvious mistakes were made, I think you'll find that their differences show up best when they are really stressed, which means it's going to be hard, even potentially deadly (because of the missus) to listen for it. And I suspect your eardrums won't be too compliant, either.
The problem there is that for a real test, you'd need one VAS stage, with which you could switch the following CCS stage. If not, you'd be hearing differences from various amps, i.e. back to square one.
And think of the problems involved. Just as an example, if you use a driver and output two stage, you might need say at least 10 mA of current drive. If on the other hand you use a triple stage, i.e. predriver -> driver -> output, you could get away with about 5 mA of drive simply because a triple stage has a much bigger current gain factor. But, a triple stage will also have a lower output impedance and might enable you to use a greater variety of output devices.
Oh man, you have your work cut out for you.
I am not sure just how much can one actually hear this or that CCS design, if you mean CCS as a part of a say differential circuit. In that case, a CCS is sort of an auxilliary ciruit, necessary for proper operation of the key, or primary ciruit.
If you meant the entire CCS stage of an amp, that's a different story. But even so, assuming no obvious mistakes were made, I think you'll find that their differences show up best when they are really stressed, which means it's going to be hard, even potentially deadly (because of the missus) to listen for it. And I suspect your eardrums won't be too compliant, either.
The problem there is that for a real test, you'd need one VAS stage, with which you could switch the following CCS stage. If not, you'd be hearing differences from various amps, i.e. back to square one.
And think of the problems involved. Just as an example, if you use a driver and output two stage, you might need say at least 10 mA of current drive. If on the other hand you use a triple stage, i.e. predriver -> driver -> output, you could get away with about 5 mA of drive simply because a triple stage has a much bigger current gain factor. But, a triple stage will also have a lower output impedance and might enable you to use a greater variety of output devices.
Oh man, you have your work cut out for you.
Specs do matter . However sometimes it is only in open loop or un-shunted we see a trend . A trend and not a pointing finger . Often trends become reliable . As shown previously shunted carbon composition looks almost the same as any other resistor . un-shunted slightly different as would be expected although not as drastic as some would imagine . The sound of a CC is decidedly pink . It is almost have your cake and eat it too if liking it . Did anyone notice on my graphs CC almost looks the best ? I did rerun the tests and found similar traits . Weird . By ear CC was no worse on hiss if shunted ( 7 R shunt load ) . By ear I would say more obvious than the scope either way if CC . Below are the un-shunted results . Again my data so forgive the choice of reference points . It is a RIAA preamp with gain of about 65 dB ( 1 kHz ) . You can see the EQ and FT of the circuit in the 47 K graphs . Note how reasonable a modern CC is ( Foil $15 , CC 30 cents , film 1 cent ) .
- Status
- Not open for further replies.