Measurements and the Correlation of What We Hear

[Phelonious Ponk]

specs are results of measurments in my book and always have been IMO

I've been struggling with this myself, and was about to bring it up. Mark you seem to be saying that specs are design "goals," not measured performance. I've never understood that to be the case. I always thought, while they're often admittedly pretty limited, arguably inadequate measurements, that they are measurements. Am I wrong about that one?

Tim

 

[JackD201]

As a quick side note, I note the reactions to the Halcro amps: some have praised them to the high heavens, others have considered them to have a sterile sound, whatever that means: I presume in the truth and tonality debate they figure heavily in the "truth" camp. Halcro may in fact have had subtle, audible faults, I can't say either way. Of course, any component that is truly low distortion will perfectly highlight, ruthlessly expose every distortion defect that exists elsewhere in the chain that makes up the audio system, rather than hide it under a soft cloak of "harmonic enhancement".

People will say that the recording is frequently badly distorted: yes, but the types of distortion there are at a far lower level of effective audibility if the replay system does its job properly. Once you've experienced what's possible, you realise that the talk of "bad" recordings is only that, talk ...

Frank

Oh Halcro Amps. Sterile? It doesn't matter to me but in an AB comparison with another amp, the other amp made the Kodo drums sound like the skin was made of leather. The Halcro made it sound like the skin was made of corrugated roofing. Distortion figures be damned. It was not accurate. If it wasn't in an AB situation one could have looked at the pretty distortion numbers and assumed that the recorded skin really sounded like metal and not leather. Well, the other amp did make the drum whacks sound more real it so we know it wasn't the recording.

Another example of marketing using numbers in place of corny poetry. The proof will only always be in the puddin'.

 

[MylesBAstor]

Oh Halcro Amps. Sterile? It doesn't matter to me but in an AB comparison with another amp, the other amp made the Kodo drums sound like the skin was made of leather. The Halcro made it sound like the skin was made of corrugated roofing. Distortion figures be damned. It was not accurate. If it wasn't in an AB situation one could have looked at the pretty distortion numbers and assumed that the recorded skin really sounded like metal and not leather. Well, the other amp did make the drum whacks sound more real it so we know it wasn't the recording.

Another example of marketing using numbers in place of corny poetry. The proof will only always be in the puddin'.

Yeah I thought they were sterile too Never understood the hubbub about these amps! They did have that cardboardy sound too Did someone say they're out of business now? They parted ways with Philip years ago here and not sure they got another distributor in the US.

 

[mep]

I've been struggling with this myself, and was about to bring it up. Mark you seem to be saying that specs are design "goals," not measured performance. I've never understood that to be the case. I always thought, while they're often admittedly pretty limited, arguably inadequate measurements, that they are measurements. Am I wrong about that one?

Tim

It’s certainly a fair assumption to say that specifications are based on measurements. It’s also fair to say that as a designer, you could start off by listing the specifications you wanted your design to meet before you ever built it. How many times have you read the measurements section of a review written by JA from Stereophile and saw flaws in the measurements vice the specifications? Lots. It happens all the time. Sometimes one channel is wacky and the other channel is fine. Sometimes both channels are wacky. But the bottom line is that often times gear does not meet its published specifications or only one channel does. And there are other parlor tricks. Sometimes the specification for a given measurement is only taken at a specific frequency where it measures best like channel separation for example. You assume that specification holds true across the audio band, but it might not. Measurements will tell you this, specifications won’t. See the difference?

I just reached up and pulled out the Stereophile issue from this August. JA measured the Primaluna Dialogue Three and found numerous areas where it did not meet its published specifications and varied from channel to channel. The maximum gain was specified as 12dB, yet the left channel measured 9.67dB and the right channel was 9.9dB. The output impedance was specified at 2500 ohms. It measured 2400 ohms at middle and high frequencies but 4000 ohms at 20Hz. Channel separation was good at low frequencies with a measured 90dB. At the top of the audioband, separation had decreased to 45dB. Primaluna doesn’t even list the specification for channel separation on their specification sheet. The signal to noise ratio specification (weighted) was >100dB. JA measured 97dB weighted.

This is just one example but I hope you get my point. Specifications are not measurements and shouldn’t be blindly trusted to be the same.

 

[DonH50]

A design starts and end with specifications, from design goals to measured results. Hopefully they are the same, or close...

Of course, that still does not mean we measure what we hear, or vice versa.

 

[Phelonious Ponk]

It’s certainly a fair assumption to say that specifications are based on measurements. It’s also fair to say that as a designer, you could start off by listing the specifications you wanted your design to meet before you ever built it. How many times have you read the measurements section of a review written by JA from Stereophile and saw flaws in the measurements vice the specifications? Lots. It happens all the time. Sometimes one channel is wacky and the other channel is fine. Sometimes both channels are wacky. But the bottom line is that often times gear does not meet its published specifications or only one channel does. And there are other parlor tricks. Sometimes the specification for a given measurement is only taken at a specific frequency where it measures best like channel separation for example. You assume that specification holds true across the audio band, but it might not. Measurements will tell you this, specifications won’t. See the difference?

I just reached up and pulled out the Stereophile issue from this August. JA measured the Primaluna Dialogue Three and found numerous areas where it did not meet its published specifications and varied from channel to channel. The maximum gain was specified as 12dB, yet the left channel measured 9.67dB and the right channel was 9.9dB. The output impedance was specified at 2500 ohms. It measured 2400 ohms at middle and high frequencies but 4000 ohms at 20Hz. Channel separation was good at low frequencies with a measured 90dB. At the top of the audioband, separation had decreased to 45dB. Primaluna doesn’t even list the specification for channel separation on their specification sheet. The signal to noise ratio specification (weighted) was >100dB. JA measured 97dB weighted.

This is just one example but I hope you get my point. Specifications are not measurements and shouldn’t be blindly trusted to be the same.

Perhaps this is semantics, but I don't think you've made the case that specifications are not measurements. I think you've made the case that sometimes (often?), manufacturers' specifications do not represent thorough, complete or honest measurements meant to to cast anything but the most flattering light on the performance of their products.

To which my scientific and highly intellectual response is...duh.

Tim

 

[mep]

Tim-If I didn't make the case and you don't "get it," I give up. I don't know how much more clear I can make it. Really, just read your second sentence and you have made my case for me. What part of "thorough, complete, or honest" don't you get? If you want to continue to think that a set of speciifcations from an OEM actually duplicates exactly how the gear would measure, take that leap of faith my friend. Those specifications are not verified until they have been measured (and there, I said it again).

I'm going to pose another question: If all good OEMs are taking plenty of measurements of their gear, how come they don't publish them at least on their website? And don't get me wrong, I believe they are taking plenty of measurements. But I also know they are not sharing them with us. They just give us the specification sheet. You can infer whatever you want to infer as to why OEMs don't provide measurements. I don't think it's because measurements might reveal some trade secrets to their designs. ARC publishes the schematic for all of their designs so they could be copied by their competitors if they were so inclined. There are other reasons as to why they are not sharing their actual measurements and only giving you specifications that they may or may not meet.

 

[Phelonious Ponk]

Tim-If I didn't make the case and you don't "get it," I give up. I don't know how much more clear I can make it. Really, just read your second sentence and you have made my case for me. What part of "thorough, complete, or honest" don't you get? If you want to continue to think that a set of speciifcations from an OEM actually duplicates exactly how the gear would measure, take that leap of faith my friend.

You presented this elaborate argument about specifications not being measurements, but design goals, and seemed to be saying that was somehow a position against the validity of measurements. I didn't "get it" because specifications are measurements (however good or bad), not design goals, and bad measurements do not erode the validity of good ones.

Then there was another point you seemed to be trying to make - that this revelation you were sharing regarding manufacturers' specs was somehow an argument against the "if it can be heard, it can be measured" position. It's not. It's an argument against the validity and thoroughness of some manufacturers' specs, nothing more. It has nothing to do with the validity of measurements in general and whether or not they can reveal all that is audible.

Now there seems to be a third point, you seem to saying that some of us with whom you disagree "think that a set of speciifcations from an OEM actually duplicates exactly how the gear would measure." While I'm not ruling it out, I've never seen a set of specifications from an OEM that even comes close to that criteria, and I haven't noticed anyone here trying to make that point. What on earth gave you the idea that anyone here thought that?

In summary, I get everything you're saying, but I only see one accurate point in all of it: Manufacturers' fudge, short-change, and cherry-pick measurements so they can publish specs that make the marketing materials look good. At the risk of being repetitive: Duh.

ON EDIT: My apologies. You also made the point that there is much that has not been measured. That one is accurate as well, though I'm not at all sure what it has to do with the rest of the discussion.

Tim

 

[RogerD]

Is "perfected" sound possible? Some think that studio recorders can be perfected meaning when fitted with ME heads and modified electronics these machines will all sound the same. They might all sound darn good,but I don't believe they can all sound the same,no two designs are the same.

Relating to Mark's question. If you AB a Boulder 1050,ML 53 and a Soluution amplifier comparing one to the other do they all sound the same? just asking does the quoted distortion level vary by more then .001+or-? hmmmm. Anyway there is a logical answer to Mark's question and my 2 cents are well known,carry on.

 

[mep]

You presented this elaborate argument about specifications not being measurements, but design goals, and seemed to be saying that was somehow a position against the validity of measurements.

I never said or implied that specifications were a position against measurements. Quite the contrary. I believe in measurements taken by qualified people with great test gear.

I didn't "get it" because specifications are measurements (however good or bad), not design goals, and bad measurements do not erode the validity of good ones.

You assume specifications are measurements. And do tell me how you can tell the “validty of good ones” when you have no idea which ones are fairy tales, which ones are telling part of the story, and which ones actually are a reflection of their measurements unless you have the measurements Tim? You can’t and that’s my point.

Then there was another point you seemed to be trying to make - that this revelation you were sharing regarding manufacturers' specs was somehow an argument against the "if it can be heard, it can be measured" position. It's not. It's an argument against the validity and thoroughness of some manufacturers' specs, nothing more. It has nothing to do with the validity of measurements in general and whether or not they can reveal all that is audible.

Again, you are misstating my position and what I said. Telling you that specifications are not measurements has nothing to do with the oft cited phrase, “If you can hear it, it can be measured.” I don’t have an issue with the validity of measurements taken by competent people as I stated above. My point is that everything we hear is not being measured and I have previously listed lots of examples. Do I believe that we can measure everything we hear? Yes, I think we can. Do I believe we have devised measurements for everything we hear and this is being done on a routine basis? Hell no.

Now there seems to be a third point, you seem to saying that some of us with whom you disagree "think that a set of speciifcations from an OEM actually duplicates exactly how the gear would measure." While I'm not ruling it out, I've never seen a set of specifications from an OEM that even comes close to that criteria, and I haven't noticed anyone here trying to make that point. What on earth gave you the idea that anyone here thought that?

I’m staring with disbelief at the above paragraph. You and Tom have both argued that specifications and measurements are the same thing because specifications were derived from the measurements. I have gone through great pains on numerous posts to try and make you understand why specifications are not the same as measurements. Now you are telling me you never believed what you said?

In summary, I get everything you're saying, but I only see one accurate point in all of it: Manufacturers' fudge, short-change, and cherry-pick measurements so they can publish specs that make the marketing materials look good. At the risk of being repetitive: Duh.

I find this tone a little off-putting. If it’s a “duh” fact that manufacturers’ “fudge, short-change, and cherry-pick measurements so they can publish specs that make the marketing materials look good,” why on earth have you been arguing that specifications are the same as measurements and should be trusted? And if these fudged, short-changed, and cherry picked measurements are all you have to go by because you have no real measurements, what are you really left with?

 

[DonH50]

Apropos amplifiers sounding the same, in-situ amplifier measurements will reveal much more than bench testing, unless you use an elaborate load system for the bench testing. At least IME.

 

[Phelonious Ponk]

Maybe we're both communicating poorly, Mark. We certainly seem to be misunderstanding each other.

I’m staring with disbelief at the above paragraph. You and Tom have both argued that specifications and measurements are the same thing because specifications were derived from the measurements.

A good example: Manufacturers' specs are measurements. THD, IMD, FR...measurements. That is not arguing that they are the same as all measurements.

Another good example:

I find this tone a little off-putting. If it’s a “duh” fact that manufacturers’ “fudge, short-change, and cherry-pick measurements so they can publish specs that make the marketing materials look good,” why on earth have you been arguing that specifications are the same as measurements and should be trusted?

I haven't been making that argument. I have been pointing out the fact that specs are measurements. I don't think that is even in dispute. The quality and usefulness of those measurements is a separate issue. And if you think, when I've said "if it can be heard, it can be measured," I was saying that manufacturers were measuring everything that can be heard and publishing it in their spec sheets, you were making a pretty big assumption.

Tim

 

[mep]

Tom-Thank you for backing up everything I tried to say from the get-go. I hope everyone else really understands the points I was trying to make.

 

[DonH50]

Uh, specs are not measurements, though the final data sheet specifications should reflect mesaured data. You can specify something but find it wasn't made to spec...

 

[Phelonious Ponk]

Uh, specs are not measurements, though the final data sheet specifications should reflect mesaured data. You can specify something but find it wasn't made to spec...

OK...so I could say my amp should have a THD of .006%, publish it as such, never measure it, and if it is actually 1%...oh well? News to me...

Not that it matters much. When someone states that if it can be heard, it can be measured, it has little to do with manufacturers' specifications.

Tim

Tim

 

[mep]

Maybe we're both communicating poorly, Mark. We certainly seem to be misunderstanding each other.



A good example: Manufacturers' specs are measurements. THD, IMD, FR...measurements. That is not arguing that they are the same as all measurements.

Another good example:



I haven't been making that argument. I have been pointing out the fact that specs are measurements. I don't think that is even in dispute. The quality and usefulness of those measurements is a separate issue. And if you think, when I've said "if it can be heard, it can be measured," I was saying that manufacturers were measuring everything that can be heard and publishing it in their spec sheets, you were making a pretty big assumption.

Tim

Just when think you can't amaze me anymore, you amaze me some more. I guess we are now full circle with you from specs are the same as measurements, to the "duh" statement that of course people know that specs are full of shenanigans and half-truths back to specs are measurements. Which is it Tim? You can't have it both ways. It's one or the other. I will say it one more time: SPECS AREN'T MEASUREMENTS. They are words and numbers written on a piece of paper unless and until they are verfied through measurements.

 

[mep]

The CJ unit you link to has a fully regulated power supply...that is 100% feedback to name just one example. And for those not so technical about all this audio mumbo jumbo we have all surely heard the the phrase that the amplifier modulates the power supply...in other words, the power comes from the power supply, the audio signal comes from the power supply.

Tom

Tom-You are seriously trying to mislead people here by what you said. You made a blanket statement last night that tube preamps use 100% negative feedback. I asked you to name one. Now you have switched gears and made the statement that if a power supply is regulated, that means the entire preamp now has 100% negative feedback and that is nonsense. There is global feedback and local feedback, but both are discussed outside of the power supply. The amount of feedback is always discussed in terms of the preamp circuit or the power amp circuit, not the power supply. By your definition, many SS and tube preamps and power amps would be considered to have 100% feedback and it's just not true. You need think about what you just said.

 

[mep]

I think it's quite clear Tom. Feedback is always discussed and measured as it relates to the circuit in question, not the power supply. If you think otherwise, please name one product that the feedback numbers include the power supply.

 

[mep]

If the one example you listed was the CJ preamp, you were wrong. That is your leap of faith from saying if the power supply is regulated it means the entire preamp or poweramp has 100% feedback. I don't think any EE would agree with you. We have some on here so please chime in.

 

[mep]

Tom-this stuff about having a regulated power supply is a smoke screen to cover up the fact that you are simply wrong about how feedback is measured in preamps and power amps. The feedback taking place inside of the power supply circuit is not adding feedback to the preamp or power amp circuit. The power supply feedback circuit is a closed loop that is not operating in the preamp or power amp circuit. There is no audio signal in the power supply that is being fed back to the input of the power supply and then being added to the audio signal in the circuit. The regulation in the power supply is measuring voltage in the power supply and not the actual output of the preamp signal and measuring it against the input of the signal and feeding back the difference.