[PhilNYC]

I'm in the #4 camp; have definitely found consistency in things like matching power ratings to speaker sensitivity, etc. But beyond that, I have learned to trust my ears.  

One of the best speakers I have ever had the pleasure of hearing was the Vince Christian E9 + c12p subwoofer.  The E9's frequency response measurements said it rolled off at about 18kHz...but you would have a hard time convincing me that those speakers didn't sound as good as other speakers in the same price range (about $12K for the E9 + c12p)...

[DVV]

Someone elsewhere commented that the Halcro gear, which apparently is near-flawless in regards to measurements as sounding boring or something.  "How can that be?" I wondered  My guess is that people find distortions of various kinds to be subjectively pleasing and that perhaps a truly useful piece of gear would be one that could 'dumb down' a bit and offer "lousy specs" as an option in addition to its near-perfect textbook option. ...

Oh Nate, you have no idea just how right you are. There was a test in some university many years ago, I think it was Berkley, where it was shown that the vast majority of people actually preferred intentionally distorted sound (2. and 4. harmonic) to undistorted sound. It was also shown that odd harmonic distortion (3. and 5. primarily) was irritating to the vast majority. In effect, our ears LIKE some forms of distortion.

The smarter among the tube manufacturer crowd exploit that fact.

... My feeling is that specs probably are more valuable to you guys who actually grok electronics and have a mental correlation between the graphs and what comes out of the speakers. I envy that!

"...who actually grok electronics" ... How poetically phrased! :mr.green: You touch our deepest sentiments, Nate.

But again, I think you are at least partially right. Those who can not only read, but interpret the data, provided it is sufficient enough, can obtain a mental picture of what to expect in very general terms. Unfortunately, the key word is "general" only, I'm afraid; to actually know, you still have to sit down and audition it.

Cheers,
DVV

[DVV]

I'm in the #4 camp; have definitely found consistency in things like matching power ratings to speaker sensitivity, etc. But beyond that, I have learned to trust my ears.  

One of the best speakers I have ever had the pleasure of hearing was the Vince Christian E9 + c12p subwoofer.  The E9's frequency response measurements said it rolled off at about 18kHz...but you would have a hard time convincing me that those speakers didn't sound as good as other speakers in the same price range (about $12K for the E9 + c12p)...

Well, it depends what they mean by "roll-off". For example, if they got it down to +/- 1 dB linearity, at -1.5 dB at 18 kHz it could be said to start rolling off.

Others do other wicked things. A highly respected Scandinavian speaker manufacturer, often thought of as "airy", in fact have their treble accentuated by +3-5 dB in the 14-20 kHz range. So, while sounding "airy", they are in fact non-linear, acting as if you had boosted some tone control.

Cheers,
DVV

[PhilNYC]

Well, it depends what they mean by "roll-off". For example, if they got it down to +/- 1 dB linearity, at -1.5 dB at 18 kHz it could be said to start rolling off.

Cheers,
DVV

The E9's frequency response is 65hz to 18.5kHz +/- 3db.

[DVV]

Well, it depends what they mean by "roll-off". For example, if they got it down to +/- 1 dB linearity, at -1.5 dB at 18 kHz it could be said to start rolling off.

Cheers,
DVV

The E9's frequency response is 65hz to 18.5kHz +/- 3db.

Gee, Phil ...

I guess by roll-off at 18 kHz, they mean -4 dB at 18 kHz.

Fine for older folks with impaired hearing.

Unfortunately, at 51, I can still hear 18 kHz 10/10 times, which I can prove, of course. So to me, they could (possible, but not certain) sound a bit dull, or, er, rolled-off?

I'm kidding, of course, I haven't heard them so any comment related to their sound is pure conjecture.

Cheers,
DVV

[PhilNYC]

DVV,

That's exactly what I expected before I heard them, but as I said, I was really impressed by how they sounded.  Given the topic of this thread ("how important are measurements to you?"), I only brought this up because the "poor" measurements didn't correlate to the great sound...

(of course "+/- 3db" could also mean they were jacked up 4db at 18khz...!     But this was not the case, as Vince himself confirmed that it was a drop-off)...

Phil

ps. If you're interested in a more detailed account of my time with the E9/c12p, check this out:

http://www.sonicspirits.com/newsletter2.htm

[JoshK]

You guys have pretty much touched all the bases.  I reside somewhere in the 2)-3) camp.  I think measurements like statistics, don't lie, it is the people who use them who lie.  You (or in this case, a manufacturer) can paint whatever picture they wish to paint by manipulating measurements, or how measurements were taken.  This is quite a different perspective then using measurements to ellicit the best results.  

Measurements done right, and interpreted correctly I think can provide tons of usefulness.

[PhilNYC]

Josh,

There is a famous saying that goes: "There are 3 kinds of lies...Lies, Damn Lies, and Statistics..."  

[JoshK]

I think that was Mark Twain who said that.

[DVV]

DVV,

That's exactly what I expected before I heard them, but as I said, I was really impressed by how they sounded.  Given the topic of this thread ("how important are measurements to you?"), I only brought this up because the "poor" measurements didn't correlate to the great sound...

(of course "+/- 3db" could also mean they were jacked up 4db at 18khz...!     But this was not the case, as Vince himself confirmed that it was a drop-off)...

Phil

ps. If you're interested in a more detailed acco ...

What people don't understand in general is that saying "+/-3dB" gives you an envelope of 6 dB. In other words, you could have +3dB at say 10 kHz and -3dB ad say 15 kHz, an amplitude of 6 dB, and still be within your nominal specifications - yet you have a 2:1 voltage ratio drop, which you can't help hearing as one effect or another.

Another thing is that some manufacturers, and this could be one of them, give their frequency response at power levels much higher than the nominal 2.83V into 8 ohms, corresponding to 1W/8 ohms. In a sense, 2.83V/8 ohms could be said to be like the frequency response with amps at 1W/8 ohms, while the other spec would be more like power bandwidth with amps.  My view is that power bandwidth measurement is much more representative of a device's response, be it an amp or a speaker, even though it should be taken at nominal power which not likely to be used in real life for fear of clipping and/or overloading. It's a "pessimistic" spec, insofar that it tells you the worst case in real life.

Could be these folks have quoted what is effectively the speaker's "power bandwidth", i.e. response at high power levels. If so, I for one applaud them.

Of course, that leaves you wondering what other speakers, rated at 2.83V/8 ohms, respond like when fed 20V/8 ohms, which is just 50W.

Cheers,
DVV

P.S. I did look. I note your new location is probably like 100 yards away from where my aunt used to live in NYC way back in March 1970.

[gonefishin]

The point is, specs are MOST useful if you know how to interpret them properly. They do tell us quite a lot, perhaps even most of the story, but I know for a fact they can never tell the WHOLE story, not even electrically, let alone acoustically.

   To me, the question is just too flawed (no offense)...and the above paragraph seems to cover all the bases of way.

  First off...Specs and measurements are two different things (least to me).  Specs are something that might be supplied to a consumer in a brochure or some self written text.  This is mainly used for sale.

   The problem I have with these self manufactured specs is that I'm not assured there is any scrutiny or standards that have to met by companies that provide them.  Actually, It seems like there isn't any type of standards that companies have to adhere to when providing specs on their merchandise.  As I see it...this is a huge problem.  For this reason...manufacturer specs seem to be of very little use to me.  Most of the "supplied" specs are touted more as marketing tools rather than actual measurements.  Most of what seems to be supplied are simply convenient specs...not a full list of measurements.

    Which brings me to number two...your question.  What about measurements?  What do I think about them?

   You know what...I feel I could look at various specs and make sense of them...but I don't put much faith in the source where these specs come from.  But actual measurements and circuit design (which I would think, do go hand in hand...just like beans and cornbread).  I should really have no part in trying to decipher a complete list of measurements or the circuit design that goes with it.  My opinion...most other people shouldn't either.  It's not necessarily a bad thing to try and educate yourself thru minor experimentation and reading a couple of books or articles written on the web.  But I still don't think many of us (you and I...not Hugh, DVV or Dan B)  But I still don't think many of "us" are even in a position where we should be interpreting measurements with circuit design.  Least not me

   This (also) doesn't mean that any individual engineer or designer is always right in opinion simply because he is an actual engineer or designer.  It simply means he/she may be in a position to better interpret the whole story that the measurements and circuit design paints.  But he still (then) has to interpret this into what he feels may be right...or wrong.  



   
   it makes no difference what you think about me.  But it makes a whole lotta difference what I think about you!


      WEEEE>>>GO HAND IN HAAAAAND   Louis Jordan




     I know I may be going a wee bit off of the topic...but that's my 2 1/2 cents on the matter  



  take care,


  dan

[PhilNYC]

I think that was Mark Twain who said that.

Mark Twain made it popular in the US, but it actually originated from a guy from the British Parliament named Benjamin Disraeli

[Rob Babcock]

I'll agree I could've phrased some of the poll options better, but I think the question is valid.  My concern is not the semantic quibbles with manufacturers advertised specs vs the actual measurements- that's a red herring.  What I really want to know is:  do measurements truly correlate to sound quality?  That is to say, "in your opinion."

For the sake of this topic, I don't mean for the actual engineer or creator of the product; certainly most guys that design amps, for instance, will use some measurements as they work.  I'm primarily concerned with the end user.

For the sake of clarity, my question attempts to quantify in a rough way the ratio of objectivists to subjectivists at AC.  I don't know how to put it more simply than that!

[DVV]

... The problem I have with these self manufactured specs is that I'm not assured there is any scrutiny or standards that have to met by companies that provide them. Actually, It seems like there isn't any type of standards that companies have to adhere to when providing specs on their merchandise. As I see it...this is a huge problem. For this reason...manufacturer specs seem to be of very little use to me. Most of the "supplied" specs are touted more as marketing tools rather than actual measurements. Most of what seems to be supplied are simply convenient specs...not a full list of measurements.

Not true. You see, in the US, power delivery by power amps, for example, must meet FTC rules, standards and procedures. If not, the manufacturer risks trouble with the FTC and substantial penalty payments.

These are long and elaborate rules, but let me mention just one, power delivery. This must be measured with all active channels driven simultaneously, from 20 Hz to 20 kHz in a sweep function, at a given level of harmonic and intermodulation distortion. Nobody has to have 0.1 or 0.0001% THD, but if you say your THD is 0.1%, then it must be so 20 Hz to 20 kHz, both channels driven into a specified load, of which only 8 ohms is mandatory.

The now all but dead German DIN standards (45500 and onwards for audio) were actually extremely well specified themselves, with VERY clear and precise conditions and methods of measurement. The set values were sane at the time they were introduced, but became rather obsolete later on, which is a shame. FTC borrows quite a bit from DIN standards, a very common practice - we don't need to reinvent hot water all of the time.

One company I'd like to commend for their approach to specs is Harman/Kardon. Read their sales literature specs, then open their service manuals, and you'll see two columns listed there: the first is what you just read in the sales ads, and the second are actual service values.  The actual service values which must be met for a device to be declared fully functional are in fact rather more strict than the values in the ads.

Being sceptical as I usually am towards the big industry, I took the time and trouble to check them out, and lo and behold, the values I obtained were better than even their service limit values! All told, significantly better than their sales blurb says they are. A commendable practice, one which was normal in the 70-ies, but has become all too rare these days.

Of course, there are others doing the same, however I mentioned H/K because all of the others (except for Rotel, which is much the same) whom I know are doing the same thing are in much higher price categories. For example, TAG/McLaren's multichannel amp is declared as 5x100W/8 ohms - under the specified conditions, it delivered 5x165W into 8 ohms, at a lower distortion level than specified. It was not specified for 2 ohms at all, yet delivered peak power levels of just over 600W into 2 ohms - all five channels at full blast. Kudos to them.

Of course, these are just examples that I know of, there are surely others out there as well I am not acquainted with.

Cheers,
DVV

[gonefishin]

What I really want to know is:  do measurements truly correlate to sound quality?  That is to say, "in your opinion."

 

     I'm not educated enough to properly interpret measurements    and I don't feel manufacturer specs are always a true representation of the component.


     I understand exactly what you were saying...and I wasn't trying to just pick apart your post.  Not at all.  I just believe that reading (and understanding) just one or two sets of measurements doesn't give you a full picture of what your dealing with.  Simply because an amp has a good frequency range and response with seemingly low distortion and good power ratings tells me very little about the sound of that component.  In fact...I'm guessing a good designer could build several components to achieve and meet a certain set of "wanted" measurements, while having each component sound quite different from one another.  
   
  But really...why would they want to do that?  I suppose some designers wouldn't want to do this...while others feel it may be a good way to sell some gear to people who don't fully understand measurements but buy equipment solely on how it performs on paper rather than in air.


   Again...I wasn't just arguing about wording.  I could really care less to argue simply about how something is worded.
 
take care,
  dan

[DVV]

I'll agree I could've phrased some of the poll options better, but I think the question is valid.  My concern is not the semantic quibbles with manufacturers advertised specs vs the actual measurements- that's a red herring.  What I really want to know is:  do measurements truly correlate to sound quality?  That is to say, "in your opinion."

Gee, Rob, you could have asked us who "grok" (Nathan's term, bless his cotton socks) electronics, and we would have been the first to tell you there is no clear cut correlation at all. Good specifications simply tell you that something has been designed properly, no more than that.

For the sake of this topic, I don't mean for the actual engineer or creator of the product; certainly most guys that design amps, for instance, will use some measure ...

Of course, that's clear enough. To give you an example, just about every simulation schematic I draw will work firt time off. Before anyone thinks that's impressive, I must admit that it's really not impressive at all, given the years of work behind it; over time, you develop design hunches, which are right most of the time. However, the fact that it works means very little, all it means is that the basic circuit is essentially operational.

That took about 2 hours. The next 200 hours you spend measuring and changing, sometimes redesigning, because while it works, at times it just barely works and needs refinement. And it's only 200 hours now, 20 years ago that was much more like 1,000 hours, because there was no PC and no incredibly good simulation software, so we had to do everything by hand.

OK, after that you spend God alone knows how many hundreds of hours tweaking the @#$%^& thing, because while two resistors, for example, will measure the same, no simulation software can tell you how each will sound. As you all know, a typical power amp will have anything from 60 to 600 parts; let's assume only 10 of them are critical to the sound. The number of variations is:

1x2x3x4x5x6x7x8x9x10 = 3,628,800

assuming repetion of same places. It's here that experience saves the day, because you use it as much as you can to reduce that number to something actually manageable.

But in the end, you come up with something like this:

Time spent on simulation and ides - about 1%
Time spent on refining and ironing out - about 19%
Time spent on tweaking and fine tuning - about 80%.

So much for simulation and specifications.

Cheers,
DVV

[Rob Babcock]

That's cool, GF.   Anyway, the poll has enough responses to confirm my suspicions:  AC'ers as a group take a middle of the road approach.  I wasn't surprised to see not a single vote for total faith in specs, but I was a little surprised that there weren't more votes at the other extreme.

At any rate, the other site that I moderate is composed of mostly diehard spec worshippers, and it's not uncommon for them to chose amps and receivers based on one having .0001% difference in THD!   That's just looney to me.  I do think measurements are interesting, and I certainly put some stock in speaker measurements, but not a whole lot in amps.  IMOHO, some specs can be misleading at worst and at best don't tell the whole story.  I think there's some intangibles that don't correspond to anything we're measuring now.

The "other site" is also populated by an awful lot of guys who think all amps & cables sound the same, etc etc.

[Rob Babcock]

Again, DVV, clear to you- an awful lot of guys are still slaves to the specs.  One mans truth is another mans data point.

[hubcaps]

The following links might be interesting because they are measurements made by the National Research Council (NRC) of various audiophile amps and speakers.  I'm not sure if the measurements mean anything, but they do look interesting.

http://www.soundstagemagazine.com/amplifiermeasurements/

http://www.soundstagemagazine.com/speakermeasurements/

[DVV]

Again, DVV, clear to you- an awful lot of guys are still slaves to the specs.  One mans truth is another mans data point.

More than that, Rob.

I would propose you ask the people on your other site to tell you what are the accepted (meaning verifiable in real life) thresholds of hearing THD and Intermodulation distortion. This will let you see just how informed they are, which in turn will tell you what to expect further on.

To me, there is no sense in making something already quite inaudible still more inaudible, not least because this typically requires very substantial investments soon enough, due to the diminishing return curve.

THD audibility threshold is taken as 3% (three per cent), although a few people will go down as low as 2.5%.

IM distortion audibility threshold is much lower, and is taken as 0.3%. This is because it has been discovered by heueristic tests that intermodulation changes the "color", or "voicing" of the sound produced. Futhermore, devices prone to IM will in all probability produce still more Transient Intermodulation Distortion (TIM, TID), although this has been overcome over the last 10 years or so, as designs got better.

Then, noise. It's little known, but noise is NOT constant, and can produce bursts of what we will hear as the equivalent of distortion (e.g. shot noise).

Lastly, all of this is given as AVERAGE; this means that while some THD will be as low as 0,01%, other parts may hit 5%, but because of their short time duration, when averaged out, this will raise that 0.01% to just 0.015%.

And so forth. But knowing this, how can you trust any specification to tell you how anything will actually sound to you?

Cheers,
DVV