KYW,
I know you don't like me quoting experts, but I do believe Bryston should be a reliable source
A/C POWER CABLES
http://www.bryston.ca/cablememo.html
"When you plug your power cord into the wall outlet you are in 'SERIES' with all the wire on the other side of the wall all the way back to the power source. The small length of power cord from the wall to the amp is insignificant compared to the miles of wire it is connected to. As long as the power cord can deliver the current and voltage required to drive the amplifier to full power it is as good as it can get."
I know you don't like me quoting experts, but I do believe Bryston should be a reliable source
A/C POWER CABLES
http://www.bryston.ca/cablememo.html
"When you plug your power cord into the wall outlet you are in 'SERIES' with all the wire on the other side of the wall all the way back to the power source. The small length of power cord from the wall to the amp is insignificant compared to the miles of wire it is connected to. As long as the power cord can deliver the current and voltage required to drive the amplifier to full power it is as good as it can get."
Found something very interesting:
http://www.prosoundweb.com/news/0204/shootout.php
DBT (ABX) for pro amps
200 people
http://www.prosoundweb.com/news/0204/shootout.php
DBT (ABX) for pro amps
200 people
What I meant was whats the point of plugging a $$$$ cable in when (most likely) your the rest of the circuit in the walls is cents per foot cable going to in many cases a rather dirty utility source.
Konnichiwa,
I am not sure what you refer to as "special powercord". In modern envoironments you many find that using shielded power cables reduces RFI getting into your amplifier. Using low inductance Mains cables and using certain types of mains conditioners (anything shunt type) can reliably reduce induced currents which form noise loops through the chassis and interconnect cables screen (one of the main reasons next to the capacitive effects why interconnect cables can sound different).
Therefor it follows that using suitably designed Mains cables (I would suggest low inductance, screened and RF absorbing as desirable) may give a material improvement. Such cables need not be significantly more expensive than generic mains leads either.
Does your associate truely understand why and how cables make a sonic difference?
Suitably designed mains cable can help reduce some of the problems acused by using interconnected, mainsconnected equipment. This goes for any number of cases where we need a wide analogue dynamic range and bandwidth and audio is one of the possible applications. IT MAY seem that Audio is easy, but ask a designer of RF Poweramplifiers to make you one with 3 Decades and a dynamic range exceeding 100db (this is a pretty good charaterisation of the requirements for audio).
Now I doubt what I have written is what you wanted to hear. BUt I do not subscribe to high end voodoo (appearances to the contrary are entierly internal to the person so percieving), I am interested and concerend only with what works (I like to the how but can live without) and if I find it to work with sufficiently reliability I use it.
Sayonara
keyser said:
I am not sure what you refer to as "special powercord". In modern envoironments you many find that using shielded power cables reduces RFI getting into your amplifier. Using low inductance Mains cables and using certain types of mains conditioners (anything shunt type) can reliably reduce induced currents which form noise loops through the chassis and interconnect cables screen (one of the main reasons next to the capacitive effects why interconnect cables can sound different).
Therefor it follows that using suitably designed Mains cables (I would suggest low inductance, screened and RF absorbing as desirable) may give a material improvement. Such cables need not be significantly more expensive than generic mains leads either.
keyser said:
Does your associate truely understand why and how cables make a sonic difference?
keyser said:
Suitably designed mains cable can help reduce some of the problems acused by using interconnected, mainsconnected equipment. This goes for any number of cases where we need a wide analogue dynamic range and bandwidth and audio is one of the possible applications. IT MAY seem that Audio is easy, but ask a designer of RF Poweramplifiers to make you one with 3 Decades and a dynamic range exceeding 100db (this is a pretty good charaterisation of the requirements for audio).
Now I doubt what I have written is what you wanted to hear. BUt I do not subscribe to high end voodoo (appearances to the contrary are entierly internal to the person so percieving), I am interested and concerend only with what works (I like to the how but can live without) and if I find it to work with sufficiently reliability I use it.
Sayonara
Konnichiwa,
Maybe they SHOULD. But are they? Have they really considered what happens with mains cables? Based on their writings, clearly not.
Well, they keep repeating the same mantra of those who wish to believe that reality is simple. Now here a little look at what is ACTUALLY involved (I mean past wishful thinking):
(from http://www.sstage.com/articles/pete01.htm)
Once you analyse the various Networks formed and consider the problems the solution suggests itself. It is really quite simple. I would go as far as to say that I would find it difficult to credit any situation where mains powered equipment is interconnected and no demonstrable noiseloops are present whose at least measurable impact on the systems output cannot be improved by selecting suitable cableing, mains and interconnection. Note that this referrs STRICTLY to measurable properties, the resulting changes may or may not be audible under blind conditions, that is a seperate issue.
Sayonara
keyser said:
Maybe they SHOULD. But are they? Have they really considered what happens with mains cables? Based on their writings, clearly not.
keyser said:
Well, they keep repeating the same mantra of those who wish to believe that reality is simple. Now here a little look at what is ACTUALLY involved (I mean past wishful thinking):
(from http://www.sstage.com/articles/pete01.htm)
Once you analyse the various Networks formed and consider the problems the solution suggests itself. It is really quite simple. I would go as far as to say that I would find it difficult to credit any situation where mains powered equipment is interconnected and no demonstrable noiseloops are present whose at least measurable impact on the systems output cannot be improved by selecting suitable cableing, mains and interconnection. Note that this referrs STRICTLY to measurable properties, the resulting changes may or may not be audible under blind conditions, that is a seperate issue.
Sayonara
Yes, I believe so.
I wish I could remember who it was on this forum who challenged me to hear the difference if 10 LM741s in series were included in the signal path - you two would get on famously.
I agree with KYW, there are good scientific reasons why some things sound better - the noise loop subject is an excellent example. Conventional measurements and techniques are often not ideally suited to audio.
Keyser - on one level I do agree with you. Pretty much any CD player/amp or whatever that is currently available is capable of a competent level of fidelity i.e. even on my daughter's Aiwa midi system I can hear most of what I hear on my main system. Many people would be happy with the Aiwa and see no reason for anything bigger/more expensive/ more inconvenient etc etc. However I think most people on DIYA are enthusiast or expert users and therein lies the difference.
But then again you know that this was not what I meant. If I were no enthousiast I would certaintly not take part in any discussion on this forum.
Kuei Yang Wang said:
Remember, sample size can also be offset by more trials, keeping the required threshold for stastical significance the same, but limiting the applicability of the results to a smaller and less well defined subset of the population. i.e., 100 trials for a single person requires the same percentage correct for 95% confidence as 5 trials for 20 different people. However, in the former case the results are less universally applicable; i.e., you only know if that one individual can distinguish differences or not, and it is unclear how representative he is of the general population.
Nonsense. See above. There is only a type 2 error possibility when you attempt to extrapolate results of the sample to results of the population (in the case of small sample sizes), or conversely when you try to extrapolate results of the sample to results of the individual (in the case of large sample sizes). Type 2 really isn't an error in the testing, but rather a misuse and misapplication of the data.
To clarify, you should use the appropriate sample size for what it is you intend to test for, and your intentions must include the applicability of the results. For instance, if you want to find out if 0.XX dB differences are nominally audible to the general population, you use a large sample size (which may mean you need less trials). It may be true that some individuals were able to detect the difference, even if the stastical result is that it is not audible to the general population. Here the goal is more along the lines of establishing "averages" or "95th percentile" figures for audibility. You can't reliably apply results good for the population to the individual when each individual did not participate in enough trials to make such determinations on an individual level.
If, however, you want to find out if anyone can hear 0.XX dB differences, you would begin by hand picking the sample from individuals who professed to have an ability to do just that. You would test each individual of the sample in a sufficient number of trials to determine, with stasitical significance, whether that individual could actually detect 0.XX dB differences. Similar to the above test, the results for one individual cannot be extrapolated to the general population. Perhaps you have found the only person who can make such determinations reliably, or perhaps your test subject is deaf. All you can say, with confidence, is whether that individual can detect the differences or not. Remember however if that was the goal of the test, such as was outlined at the beginning (to find out if anyone can hear 0.XX dB differences), and you find someone that has such an ability the test is over. You have confidently answered the question that was posed.
Similarly, if the goal was to only find out if a particular individual (you, a customer, someone making ridiculous claims, etc.) has the ability in question, you again need only the sample size of one with sufficient trials to confidently answer the question. The results are not applicable to the general population, but you know if that particular individual can detect the differences, and perhaps that's all you want to know.
In the case that the goal is to find out if anyone has a particular ability, and the small sample/large trial tests continue to come up negative, then you can begin to put some bounds on the expected frequency of such an ability in the general population based on the cumulative sample size to date. If you test ten "golden ears" or "audio engineers" and all tests are negative (using trial numbers appropriate to an individual), then that tells you something about the probability of finding that ability among the entire population of "golden ears" or "audio engineers."
In essence, if you have both large sample sizes and large trial numbers for each individual in the sample, you absolutely can make confident determinations about the audibility of a particular difference to humans, period. There will always be the remote possibility of one superhuman that you did not test, and the results will define just how remote.
We've covered this before KYW, and in the past you simply ignored the logic presented. You take faults in the interpretation of data and twist them to make claims that the technique used to gather that data has fundamental faults (as is clear in your above quotes, with comments about certain error types becoming practically inevitable when you reduce other error types... that is complete bunk). The fact is that by testing you, and only you, I could prove whether you had ability X or not, to a very high degree of confidence. And that is what this tread is really about... the applicability of DBT to the DIYer. The truth is that it is extremely applicable. You don't need a large sample size... in fact you don't want it. If the goal is only to find out what you can hear, then you only need a large number of trials and a proper understanding of the interpretation of gathered data.
The continuing irony is that the best way to make solid determinations about the absolute threshold for various abilities in the general population is to test the candidates most likely to possess such abilities... i.e., to test golden ears, producers, manufacturers' engineers, reviewers, etc. Yet those are the most resistant group to be found, and the ones that have written at length about why the statistics don't apply (like the ones you cited in Stereophile). I suspect that the reason they continue to obfuscate the facts with misleading "analyses" of the stastical methods and indeed misleading tests themselves is because they understand the issue far better than their pseudoscientific ramblings would have us believe.
Again, nonsense. The only bias that can skew results is the bias to not hear differences. A bias to hear them can in no way skew results beyond the actual abilities possessed by that individual. SY very clearly made that argument earlier, and you dismissed it with some handwaiving about gross differences overwhelming such a bias. Sheesh.
A couple of thougts. (1) The trials, whether for a single individual or cumulative depending on the test goals, should tend toward a balanced representation. If there is a really unexpected distribution in the trials (which should be done well in advance of the testing), then either new distributions should be randomly created or additional trials should be added. As an aside, you should never do what Stereophile did in one of their "blind" tests, which is hand adjusting the randomly created trial distribution to make it seem more random. What they did, in fact, completely explains the small "stastically significant" positive result they obtained. Go figure. (2) The interpretation of data should include separation of events into categories to check for such biases. i.e., you should look at an individual's and/or group's (whichver is applicable to the test goal) results for only "same" pairings and only for "different" pairings, if your test methodology was one of the AA/AB type. (3) You should be smart enough to choose a methodology that is completely immune to such a bias, such as ABX. Sheesh! In ABX, you know that A and B are different, you know that X is the "same" as one of them, and you know that X is "different" from the other. You just don't know which. Any bias that all equipment sounds the same can only skew the results towards more random, as it would promote a tendency for such subjects to simply randomly guess. Any bias that all equipment sounds different cannot skew the results at all, in any direction, period, which is why such subjects make the best sample candidates.
KYW, it's as if you simply don't understand the purpose of ABX testing at all (as opposed to other forms, such as AA/AB, which many of your comments are more applicable to). Or perhaps you do understand but have some other reason for making false statements? If you have any logic supporting such claims, instead of merely stating them as fact and telling people to read Stereophile (as if that were acceptable as evidence of your claims), we'd love to hear it.
RH- re the Stereophile tests, I will compliment them in that they published a letter from Herman Burstein exposing their little cheat.
Really? I wasn't aware of that. I've rarely seen that particular article discussed, and when it is and I have pointed out the error I usually get "blank stares" in return. Good to know that others picked up on it and that Stereophile at least had the decency to disclose it after the fact and publish a 3rd party explanation.SY said:
Konnichiwa,
It might have escaped your notice, but I pointed out that particular test as one worth noting for what they did WRONG, not for the result they got.
I'll save myself repeating other arguments, read the thread earlier on, my comments stand.
Sayonara
RHosch said:
It might have escaped your notice, but I pointed out that particular test as one worth noting for what they did WRONG, not for the result they got.
I'll save myself repeating other arguments, read the thread earlier on, my comments stand.
Sayonara
RH, thank you for your your post.
Very educational and although not easy for me to completely understand yet, easy to follow. The part that is probably of most use to me is the following!
That is exacly what I will do! Do more DBT's. It is completely true that I should be the only one interested in if I can hear a difference. If I can I can, If I can't I can't. I'll have to figure that out for myself.
As I said before, I expect the outcome (for me) to be that I won't be able to hear the difference between amps (atleast not the ones that I can get my hands on for DBT-'ing. No single ended tri- or pent-odes or such things).
I would ofcourse be nice to be able to convince my still-voodoo high-end friends of the same thing
KYW, I've heard many of your arguments against DBT's. There is however one thing on which you still DO have a strong opinion, without any clear arguments to back'em up. SY and myself brought it up a couple of times a few pages back, and now RH got back to it again:
Very educational and although not easy for me to completely understand yet, easy to follow. The part that is probably of most use to me is the following!
That is exacly what I will do! Do more DBT's. It is completely true that I should be the only one interested in if I can hear a difference. If I can I can, If I can't I can't. I'll have to figure that out for myself.
As I said before, I expect the outcome (for me) to be that I won't be able to hear the difference between amps (atleast not the ones that I can get my hands on for DBT-'ing. No single ended tri- or pent-odes or such things).
I would ofcourse be nice to be able to convince my still-voodoo high-end friends of the same thing
KYW, I've heard many of your arguments against DBT's. There is however one thing on which you still DO have a strong opinion, without any clear arguments to back'em up. SY and myself brought it up a couple of times a few pages back, and now RH got back to it again:
What escaped my notice was that you pointed out that test at all. I mentioned it because it was relevant to the topic of seemingly non-random trial sequences, apparently the same reason you mentioned it.
As for your refusal to provide any logic for your claims, that seems to be your standing rule. In the past when you made such claims about statistical errors and testing limitations I presented a thorough explanation of why that was not so. You then, as now, responded by asserting that you had covered that in previous posts in the thread. The truth is that you haven't offered any logical explanation for how your claims manifest themselves in reality; not now, not in the past, and I suspect never in the future. You obfuscate and mislead with factual sounding and athoratative sounding statements and handwaiving explanations, but when pressed offer only "read what I wrote" refutations.
So be it. I take your silence to mean much the same as the silence of the audiophile "review" industry - that they probably understand the real issues, but because of vested interests or entrenched agendas dare not address them.
As for your refusal to provide any logic for your claims, that seems to be your standing rule. In the past when you made such claims about statistical errors and testing limitations I presented a thorough explanation of why that was not so. You then, as now, responded by asserting that you had covered that in previous posts in the thread. The truth is that you haven't offered any logical explanation for how your claims manifest themselves in reality; not now, not in the past, and I suspect never in the future. You obfuscate and mislead with factual sounding and athoratative sounding statements and handwaiving explanations, but when pressed offer only "read what I wrote" refutations.
So be it. I take your silence to mean much the same as the silence of the audiophile "review" industry - that they probably understand the real issues, but because of vested interests or entrenched agendas dare not address them.
keyser said:That is exacly what I will do! Do more DBT's. It is completely true that I should be the only one interested in if I can hear a difference. If I can I can, If I can't I can't. I'll have to figure that out for myself.
As I said before, I expect the outcome (for me) to be that I won't be able to hear the difference between amps (atleast not the ones that I can get my hands on for DBT-'ing. No single ended tri- or pent-odes or such things).
I would ofcourse be nice to be able to convince my still-voodoo high-end friends of the same thing
One thing that you can do to minimize your bias to "not" hear differences is to remember the following: if you can hear differences between various types of amplifiers, you could make a ton of money! There are longstanding offers for anyone who can prove such an ability... $10,000 for one that I know is still valid.
That should be incentive enough (for me at least!) during the course of your DBT's to take seriously the possibility that you just might be able to resolve minute differences, and to do your absolute best to develop such an ability. If money is as much a motivator for you as it is for me, and you still fail to pass your DBT's, it would say a lot about your particular abilities wrt audible differences.
😉
I don't see that much difference in viewpoint between SY, KYW, RHosch, and myself, besides a lot of mostly semantic details and the weighting of what each finds most important. It boils down to
- if you want to make universal statements, you need a lot of foolprofing and rigorous testing with a lot of people. If you don't have these, better word your statements on this forum along the more careful lines of "in my personal experience" or "I myself can't hear...".
- if you want just an aid for your personal decision making, much simpler tests suffice, and with an ABX comparator you can do that easily.
The crux of the matter lies in that most amps *do* fare very similarly in the *big* things: frequency response, THD etc. . The question remains whether *minor* amp differences can make a difference in the *long term* enjoyment. Those *minor* differences will lie in factors *not* addressed by the standard specs such as THD. None of the amp designers threads here go on and on about lowering THD. They will discuss much more subtle things, such as influence of current sourceson PSRR and CMRR, overload recovery, input stage overload, stability issues etc. People calculate, model, and measure these differences all the time. No one would dream of claiming "these 2 topologies have the same THD so they sound the same".
In addition, something that no one has addressed yet in this thread:
What you can't reliably hear in a brief DBT may make a difference in the long run and on some special material, which you may not have included in your DBT. I make this point specifically because I have sometimes failed to hear a difference in a blind test, but in the long run I ended up consluding that one of the amps or wiring layouts etc *did* sound better to my ears. And this has to do with things very hard to pin down: RFI/grounding/AC quality , which may differ by time of day, and the like. A couple of DBT's may miss the problem completely.
When you say:
you go to the heart of the matter: you still need a large number of trials, statistically independent from each other - not 100 flips of the switches within 1/2 h, but stretched out over different days and times of day.
And many arguments may also come from the fact that people do have different hearing. What one hardly notices may bother the other. We're talking small differences here.
- if you want to make universal statements, you need a lot of foolprofing and rigorous testing with a lot of people. If you don't have these, better word your statements on this forum along the more careful lines of "in my personal experience" or "I myself can't hear...".
- if you want just an aid for your personal decision making, much simpler tests suffice, and with an ABX comparator you can do that easily.
The crux of the matter lies in that most amps *do* fare very similarly in the *big* things: frequency response, THD etc. . The question remains whether *minor* amp differences can make a difference in the *long term* enjoyment. Those *minor* differences will lie in factors *not* addressed by the standard specs such as THD. None of the amp designers threads here go on and on about lowering THD. They will discuss much more subtle things, such as influence of current sourceson PSRR and CMRR, overload recovery, input stage overload, stability issues etc. People calculate, model, and measure these differences all the time. No one would dream of claiming "these 2 topologies have the same THD so they sound the same".
In addition, something that no one has addressed yet in this thread:
What you can't reliably hear in a brief DBT may make a difference in the long run and on some special material, which you may not have included in your DBT. I make this point specifically because I have sometimes failed to hear a difference in a blind test, but in the long run I ended up consluding that one of the amps or wiring layouts etc *did* sound better to my ears. And this has to do with things very hard to pin down: RFI/grounding/AC quality , which may differ by time of day, and the like. A couple of DBT's may miss the problem completely.
When you say:
you go to the heart of the matter: you still need a large number of trials, statistically independent from each other - not 100 flips of the switches within 1/2 h, but stretched out over different days and times of day.
And many arguments may also come from the fact that people do have different hearing. What one hardly notices may bother the other. We're talking small differences here.
IF you are working in de car hifi industry, yes...
Overload recovery, input stage overload and stability are only relevant in the case of overdriving the amp. My ealier statement was that all properly designed amplifiers sound the same as long as they are not driven into clipping.
What are PSRR and CMRR?
Input stage overload has sparked heated debates here and in the past, even *without* output stage clipping. In essence it goes about, how well can the input stage handle a transient signal (which it then will transmit to the output stage). Various parts of an amplifier can have overload, stability etc. problems regardless of the final = output stage .
Of course then we can always say "then the amp wasn't properly designed". But this only turns into a semantic fight. Amplifiers *do* have different designs, and why then should they all react "the same" to various input and load and RFI and cable etc? I'd find nothing more surprising than if various different designs ended up showing exactly the same characteristics.
Stability refers to the absence of oscillations and has no relation to clipping. Amps can oscillate without any input. Oscillation occurs when spurious signals, and parasitics, make the amplifier try the impossible.
Finally, overdrive itself. As I remarked earlier, with typical music and SPL demaden from an amp, most systems *will* clip briefly. I'll try and find the reference where someone did a nice calculations of the amp power required for full scale reproduction. Not trivial! It goes up to 700 W peak. Since few people have that, one must conclude that in "normal" operation, the "normal" amps people have in their living rooms will briefly clip on occasion.
Konnichiwa,
On top of what is being discussed I would add a few types of distortion usually not quantified but very real:
1) Thermal Distortion & noisefloor modulation - what happens is that the signal (music) changes the temperature of the the various components in the amplifier, some of which (the active devices) are highly affected by temperature variations. The output stage bias control is a very obvious point but input and voltage amplification stages are also affected.
2) PIM or Phase InterModulation, it is largely a result of the use of looped feedback, referenes to the phenomenae are rare on the net or in common literature and the AES Thoughtpolice (Lipshitz et al) had a good laugh at Ottala for broaching the subject, despite his earlier highly topical and relevant publications on TIM (which the AES Thought police also disliked but could not brush off), poor never quite recovered from that I hear.
3) Noisefloor Modulation, in other words a signaldependent noisefloor usually in the presence of modest levels of RFI in either signal. Such RFI may occur for example through noiseloops formed between several items of interconnected equipment or through the powersupply (direct leakage of RFI through the mains supply and unscreend and not RFI rejecting mainscables or even noise produced in the supply as a result of rectification) or through the aerial action of the speakercables and direct injection into the Input/VAS stages of the Amp through the feedback capacitors.
Any of the above may show drastic variations between amplifiers having identical THD but different design/construction. Clipping behaviour as remarked may very well play a greater role than conventionally aknowleged, a good argument for making amplifiers having poor overload behaviour (in short most/all with significant levels of looped inverse feedback) as powerfull as possible AND for increasing speaker sensitivity well past conventional levels in HiFi.
At any extent, the subject is complex and not ameanable to single number analysis at all.
Sayonara
MBK said:
On top of what is being discussed I would add a few types of distortion usually not quantified but very real:
1) Thermal Distortion & noisefloor modulation - what happens is that the signal (music) changes the temperature of the the various components in the amplifier, some of which (the active devices) are highly affected by temperature variations. The output stage bias control is a very obvious point but input and voltage amplification stages are also affected.
2) PIM or Phase InterModulation, it is largely a result of the use of looped feedback, referenes to the phenomenae are rare on the net or in common literature and the AES Thoughtpolice (Lipshitz et al) had a good laugh at Ottala for broaching the subject, despite his earlier highly topical and relevant publications on TIM (which the AES Thought police also disliked but could not brush off), poor never quite recovered from that I hear.
3) Noisefloor Modulation, in other words a signaldependent noisefloor usually in the presence of modest levels of RFI in either signal. Such RFI may occur for example through noiseloops formed between several items of interconnected equipment or through the powersupply (direct leakage of RFI through the mains supply and unscreend and not RFI rejecting mainscables or even noise produced in the supply as a result of rectification) or through the aerial action of the speakercables and direct injection into the Input/VAS stages of the Amp through the feedback capacitors.
Any of the above may show drastic variations between amplifiers having identical THD but different design/construction. Clipping behaviour as remarked may very well play a greater role than conventionally aknowleged, a good argument for making amplifiers having poor overload behaviour (in short most/all with significant levels of looped inverse feedback) as powerfull as possible AND for increasing speaker sensitivity well past conventional levels in HiFi.
At any extent, the subject is complex and not ameanable to single number analysis at all.
Sayonara
- Status
- Not open for further replies.