What beats the R2904/7000 ring radiator?

[smellygas]

You can't swap a tweeter with another just like that. The crossover need to be optimized, it needs to be tweaked until it sums perfectly and the levels need to be matched perfectly.

Your profound improvement could be just because the modified 7000 and the 9800 are slightly louder than CDM-1 and 6600. About 0.5 dB louder can play with your mind. Some even say 0.1 dB...

You didn't play with the L-Pad, nor the x-over, your tests are pointless, sorry. Please calibrate everything and report back.

My test with the 48db/oct filter is good enough. I've been using the D2904/6000 (not 6600) in the same setup for >1 year. I've swapped in the modified 7000 and optimized it the best I could for about a week. There is no question, even when the "l-pad" adjustment isn't perfect, that there are a) differences and b) definite improvements with the 7000. The example with the CDM-1 (with SS 9800) was just anecdotal. Like I said. If all tweeters sound just as good at 4kHz+, then every loudspeaker mfr would be doing it - unless you happen to be smarter and know more than every professional design team out there.

SG

 

[454Casull]

Is there something you're trying to prove, then?

 

[simon5]

There is no question, even when the "l-pad" adjustment isn't perfect, that there are a) differences and b) definite improvements with the 7000. The example with the CDM-1 (with SS 9800) was just anecdotal. Like I said. If all tweeters sound just as good at 4kHz+, then every loudspeaker mfr would be doing it - unless you happen to be smarter and know more than every professional design team out there.

SG

First, a) yes there's differences, the L-Pad is not adjusted, the crossover may not sum perfectly and isn't optimized. b) definite improvements can be caused by that. I'm not saying it's that, but it could be that for sure. Then you just said your SS 9800 is anecdotal, so it's not valid.

Some people here forget easily that loudspeakers manufacturers hire people in marketing instead of actual engineers. Best example is Bose, very profitable company that will live forever. Loudspeakers manufacturers build speakers that sell well, the sound quality is not important.

By the way, like I said to Mario Pankov, I'm not impressed by the use of Argumentum ad populum in your argumentation.

 

[smellygas]

First, a) yes there's differences, the L-Pad is not adjusted, the crossover may not sum perfectly and isn't optimized. b) definite improvements can be caused by that. I'm not saying it's that, but it could be that for sure. Then you just said your SS 9800 is anecdotal, so it's not valid.

Some people here forget easily that loudspeakers manufacturers hire people in marketing instead of actual engineers. Best example is Bose, very profitable company that will live forever. Loudspeakers manufacturers build speakers that sell well, the sound quality is not important.

By the way, like I said to Mario Pankov, I'm not impressed by the use of Argumentum ad populum in your argumentation.

Let me put it another way because I don't think I'm quite getting through to you. If it is so widely known and accepted that harmonic distortion products from 1-2.5kHz are the definitive determinants of tweeter sound quality, then CERTAINLY it would be well-published in peer-reviewed journals and/or textbooks. I challenge you to find just ONE reputable published source (i.e. not "bob" on XYZ internet forum) that supports such a notion that is apparently so obvious and self-evident to you and jAy here.

Otherwise, you two are apparently incredible geniuses who are privy to information that none of the thousands of audio engineers out there could figure out, despite the availability of equipment to easily measure and verify this assertion. Or perhaps you are merely acquiring and perpetuating this information from internet websites and discussion forums???

SG

 

[Jay_WJ]

But 6600 still is more nonlinear which leads to more distortion playing wide band material. Third order only means a more symetrical suspension and motor. At 2kHz the 6600 has about 10dB more distortion.

So the third order products are simply a ghost to you that aren't audible? Doesn't make sense to me. The C23 has about 20 dB higher 3rd order products at 2 kHz. Below that it's even worse considering that Zaph's measurement doesn't compensate for the tweeter's natural rolloff.

How do you know that?

Did you miss this? "The ECM8000's distortion at high SPL is commonly known. It's not only you."

Even I knew it. Ran across this thread a couple of years ago:

http://www.diyaudio.com/forums/showthread.php?threadid=96133

But you are mistaken when you believe that this is the case and that a closer placement of the mic makes the situation better.. when it's the opposite.

Right, I was confused between the contribution of mic and that of room acoustics. The ECM8000 (in fact, including other similar mics) is known to have much more nonlinear distortion at lower frequencies than at higher. This is why Zaph sets his mic at 92 to 96 dB at 1/2 meter for his midwoofer tests. At this SPL, the ECM8000's distortion is a non-issue (less than 0.1 % according to a test with 200 Hz input). For the tweeter test, he sets it at about 110 dB at 4 inch for fundametals between 1.5 k - 3 kHz. I don't know how much the ECM's distortion in this frequency range is, but believe that it is reasonably low so that the measurements provide useful information. Otherwise, he wouldn't have been reporting nearly 50 tweeters' test results over the past several years. BTW, I agree with ShinOBIWAN's comment and Svante's reply I found:

Run a 200hz test tone at 120dB and see if the mic at just over 1% THD or the driver has more distortion Comparatively it is very little which was my point all along.

The issue of THD contributed by the ECM8000 being a factor that would call distortion measurements into question is mute, there is an error margin but if your sensible about this then you realise that its all relative. If you take issue with that then you better email Zaph and John K to tell them they're misinforming the DIY community with their efforts.

My point is that the data captured by the ECM8000 is more than good enough to be informative.

Well, ok, let's agree on the numbers, the distortion is about 1% at 120 dB. Really, that is all the information needed.

Thanks for participating in this interesting discussion.

-jAy

 

[Jay_WJ]

I challenge you to find just ONE reputable published source (i.e. not "bob" on XYZ internet forum) that supports such a notion that is apparently so obvious and self-evident to you and jAy here.

Speaker building is science plus engineering. Textbooks and articles can't (and don't intend to) cover all the practical skills and knowledge acquired and even commonly shared by real-world designers in their experience. For example, drivers' relative acoustic offset in a 2-way on a flat baffle (i.e., midwoofer's acoustic center is usually 15 to 30 mm behind the tweeter's) is usually handled by what we call "asymmetric" or "relaxed" rolloffs in LR4 designs. We extensively discussed this topic at PE TT board in the past.

Find just ONE reputable source that discusses this particular technique that is now common to many experienced people around here.

-jAy

 

[Pan]

So the third order products are simply a ghost to you that aren't audible?

Of course not but third order harmonics just as second order are found on most instruments and basically it's the higher order AND IMD that must be avoided. You will get more IMD with a high level of 2nd order and in teh drivers we are discussing that one will dominate the nonlinearities.

Doesn't make sense to me. The C23 has about 20 dB higher 3rd order products at 2 kHz.

Yes but about 10dB lower HD total, which is significant.

Did you miss this? "The ECM8000's distortion at high SPL is commonly known. It's not only you."

But still you say the distortion is of no consequence?
Doesn't make sense.

Even I knew it. Ran across this thread a couple of years ago
http://www.diyaudio.com/forums/showthread.php?threadid=96133

Yes and according to the graphs in that thread the ECM8000 has about 0.2%THD at 100dB. That matches teh SPL of Zaph's tweeter measurements well... which I estimated to 102dB SPL.

Obvisouly (according to that graph) the mic distortion at that frequency is aprox. same magnitude as the drivers measured and in such case it WILL give false results.

Right, I was confused between the contribution of mic and that of room acoustics. The ECM8000 (in fact, including other similar mics) is known to have much more nonlinear distortion at lower frequencies than at higher.

Graphs I've seen and measurements I have done do not support that. Do you have something that support that claim?

I have mesured HF IMD of a bunch of mic's and ECM8000 had about 20dB higher ditstortion than Earthworks QTC1. Also typically distortion increases at higher frequencies according to my measurements.

Sennheiser have made measurements on many mic's and distortion typically increases in the highest octaves according to their findings as well.

You propose the opposite but I have not seen anything that indicates that is true.

This is why Zaph sets his mic at 92 to 96 dB at 1/2 meter for his midwoofer tests. At this SPL, the ECM8000's distortion is a non-issue (less than 0.1 % according to a test with 200 Hz input).

I'm not sure abou that. Typical mic distortion as can be seen in the ARTA app noted pdf document in that other thread shows singnificant distortion at those levels.

For the tweeter test, he sets it at about 110 dB at 4 inch for fundametals between 1.5 k - 3 kHz. I don't know how much the ECM's distortion in this frequency range is, but believe that it is reasonably low so that the measurements provide useful information.

Again, according to the pic in the other thread aprox. 0.4%THD if the midrange dist. translates to mid/high which I see no reason not to believe until I see evidence on the contrary.

Otherwise, he wouldn't have been reporting nearly 50 tweeters' test results over the past several years.

People are humans and humans do mistakes and make wrong assumptions sometimes, welcome to reality. ;-)

BTW, I agree with ShinOBIWAN's comment and Svante's reply I found:

Run a 200hz test tone at 120dB and see if the mic at just over 1% THD or the driver has more distortion Comparatively it is very little which was my point all along.

But 1% at 120dB in the nearfield is not little. You must realize the difference between measuring a speakers output at 1-2 meter @ 120dB SPL and putting the mic there, then the speaker distortion will likely swamp the mic distortion but you can not take that for granted measuring at high SPL in the nearfiled.

The issue of THD contributed by the ECM8000 being a factor that would call distortion measurements into question is mute,

We have strong indications that you are mistaken.

there is an error margin but if your sensible about this then you realise that its all relative. If you take issue with that then you better email Zaph and John K to tell them they're misinforming the DIY community with their efforts.

Ah you stole that comment from Shin or some other guy in the other thread. Yes I will ask Zaph about it and of course it would be a shame if all those measurements are flawed.

My point is that the data captured by the ECM8000 is more than good enough to be informative.

But you don't know that unless you have verified the mic's distortion at those levels. If the mic has -60dB HD and the tweeter has the same then the third harmonic may be doubled and the 2nd harmonic may be doubled or cancelled depending on the phase of the signal. This is significant with a big "S".

If a tweeter has -80dB HD then the readings will be that of the mic. Only if the tweeter has significantly higher levels of distortion the results are reliable.

Svante and I have measured different ECM8000 in different registers and even thoiugh some measurements remain everything so far indicates that teh ECM8000 has about 20dB higher distortion across the range than for example an Earthworks mic.

I would not have raised this question and argue with you if I haven't been thinking about this and performed measurements.

I'll see if I can find some pic's.


/Peter

 

[Pan]

I forgot to mention.

I measured IMD with 9k+10k and 19k+20k at aprox. 85dB SPL.

ECM8000 gave a 1kHz product aprox. -57dB... that's at 85dB SPL.

That should be enough to raise some question don't you think?

Now, you can't take it for granted but such a high level of IMD suggests similar levels of HD (which is trickier to measure in the high range but we are working on it) and imagine what another 20dB SPL would do.


/Peter

 

[Jay_WJ]

Yes but about 10dB lower HD total, which is significant.

Don't have time to reply to all the points, but just one here. It is now known and accepted by many that THD calculation often times gives limited information. It has been validated that human ear/brain has a different level of sensitivity in peceiving harmonic products of different orders. Do you want to start discussing this topic? Let's bring in Geddes!

 

[Pan]

No for heavens I have been thru that with Geddes allready!

And I'm very well aware of what you bring up and I have done some testing as well, blind of course, and I do not like high levels of 2nd order I like low distortion..


/Peter

 

[pedroskova]

Don't forget the contributions of the (tube) amp. Guess how I know that.

 

[smellygas]

Speaker building is science plus engineering. Textbooks and articles can't (and don't intend to) cover all the practical skills and knowledge acquired and even commonly shared by real-world designers in their experience. For example, drivers' relative acoustic offset in a 2-way on a flat baffle (i.e., midwoofer's acoustic center is usually 15 to 30 mm behind the tweeter's) is usually handled by what we call "asymmetric" or "relaxed" rolloffs in LR4 designs. We extensively discussed this topic at PE TT board in the past.

Find just ONE reputable source that discusses this particular technique that is now common to many experienced people around here.

-jAy

jAy, thank you. To be honest, I am skeptical of some information that circulates on internet forums. While I recognize that not every aspect of speaker design has been published, something as profound, important, and easily/cheaply verifiable as your assertion (that 1-2.5khz distortion products are the primary determinants of tweeter SQ, and something that I still find silly: that virtually any tweeter crossed over at 4kHz has the same SQ), should definitely be at least MENTIONED in a reputable journal or textbook. Unless, of course, this idea is not universally accepted by professional audio engineers, and is instead, just a theory circulated and perpetuated by a group of semi-anonymous amateur DIY speaker builders who post on an internet forum, some of whom have websites.

SG

 

[simon5]

jAy, thank you. To be honest, I am skeptical of some information that circulates on internet forums. While I recognize that not every aspect of speaker design has been published, something as profound, important, and easily/cheaply verifiable as your assertion (that 1-2.5khz distortion products are the primary determinants of tweeter SQ, and something that I still find silly: that virtually any tweeter crossed over at 4kHz has the same SQ), should definitely be at least MENTIONED in a reputable journal or textbook. Unless, of course, this idea is not universally accepted by professional audio engineers, and is instead, just a theory circulated and perpetuated by a group of semi-anonymous amateur DIY speaker builders who post on an internet forum, some of whom have websites.

SG

I kinda like this reply. Being skeptical is great and needed, especially with those many EnABL threads in this very forum... so I respect that greatly.

Well, let say something here, that one would take a good cheap tweeter and a good expensive tweeter, let say a 22TAFG versus a 7000, if and only if they had similar frequency response and similar distortion above 4 kHz, which is near but I don't know if it's near enough...

If we compare them at 4 kHz and over, with matched levels, and the results are inconclusive which is the best, why someone would get a 7000 ? Most guys with 7000 wouldn't want that information to be published, will argue to death that it's wrong, since they spent good money. Why a 240$ tweeter could be worse than a 40$ tweeter they would say... You probably heard that one before elsewhere. The manufacturer will step up on this and blackmail everyone. I mean even Bose managed to shut up Stereophile.

Then, another example. Zaph tested the mighty Thiel and Partners, Accuton D20-6 Pure Diamond dome tweeter at 2600$ per tweeter, or 5200$ per pair. The owner gave it to him for testing only if he had an embargo on the data produced. He never lifted the embargo. Strange, no ?

You probably know that the subjectivity is very important in audio. Even after a double blind test conducted by the reputed Tom Nousaine, some still argue about the sound of audio cables.

If I was a good designer, why would I publish a book to give my secrets ? Why would I publish a book that might get me in trouble ? Not every people like an open war, some prefer to do their thing and let those people argue in vain about things he doesn't care about and know it's pointless and had no effect on performance. Some might see this as a waste of energy. Trying to convert someone is hard. Just look at what religion is doing, but that's another subject.

I will put some water in my wine too. Not every tweeter sounds the same over 4 kHz, but you can probably easily find one that will beat the 7000 over 4 kHz for under 50$, maybe even under 20$.

 

[MisterTwister]

Don't forget the coolness factor. It is important maybe some tweeters match 7000 in performance, but none of them look as sophisticated. So I'd say even without listening, 7000 is the best tweeter.

 

[tomat]

in my opinion Focal berrylium tweeter beating the scan R7000 in all the way

 

[simon5]

Don't forget the coolness factor. It is important maybe some tweeters match 7000 in performance, but none of them look as sophisticated. So I'd say even without listening, 7000 is the best tweeter.

Exactly, great point you make there.

If the loudspeaker have no speaker grille, the look is often very important for the marketing department, even if it doesn't affect the sound. It must look sophisticated, excellent build quality and tolerances, it must be shiny and all...

That's one of the reasons many people prefer the Millenium over the 27TBFC/G, even if on the performance side, they are about equal and the 27TBFC/G even better on some points.

 

[Pan]

Then, another example. Zaph tested the mighty Thiel and Partners, Accuton D20-6 Pure Diamond dome tweeter at 2600$ per tweeter, or 5200$ per pair. The owner gave it to him for testing only if he had an embargo on the data produced. He never lifted the embargo. Strange, no ?

The mic Zaph use has probably more distortion than that tweeter.

Madisound had measurements up on D20 earlier and it was like 0.1%THD in the low range and 0.01%THD in most of the range (don't know at what SPL).

The bigger D30 is bettter yet at aprox. 0.03%THD @ 90dB, 0.07%THD @ 100dB and 0.3-0.5% @ 105-107dB.


The best sounding and best measuring and sounding dome tweeters I have seen and heard are the Accuton diamond tweeters and the tweeters used by Ino Audio and Guru Pro Audio (same designer, more or less identical drivers).

Unfortunately when it comes to speakers, quality cost.


/Peter

 

[Jay_WJ]

Yes I will ask Zaph about it and of course it would be a shame if all those measurements are flawed.

But you don't know that unless you have verified the mic's distortion at those levels. If the mic has -60dB HD and the tweeter has the same then the third harmonic may be doubled and the 2nd harmonic may be doubled or cancelled depending on the phase of the signal. This is significant with a big "S".

The mic Zaph use has probably more distortion than that tweeter.

Don't think you can so easily spoil someone's work, which has been carefully thought, executed, revised, accumulated over many years, simply by using your crude speculation backed by limited experience. I think I can run a computational model that demonstrates that your 'mic distortion' theory is not sophiscated enough to explain what you claim to be true. Bear with me. I'll get back to you.

-jAy

 

[Pan]

Don't think you can so easily spoil someone's work, which has been carefully thought, executed, revised, accumulated over many years, simply by using your crude speculation backed by limited experience. I think I can run a computational model that demonstrates that your 'mic distortion' theory is not sophiscated enough to explain what you claim to be true. Bear with me. I'll get back to you.

-jAy

I'm not spoiling anything.. why so defensive? I'm only raising a legitimate question.

Crude speculations is what you are spending time with, not me and you don't know what experience I have so why you call it limited I don't know.

I base what I say from measurements made by me and Svante and also the links you gave me.

I really don't get what you're about.. but have a nice Friday evening anyway!


/Peter

 

[Pan]

Here's some of the material from Sennheiser. The blue is one of their MKH mic's with symetrical capsules. Most mic's have non symetrical capsules which gives more even order distortion and also more IMD. By making the capsule more symetric the distortion is decreased.

Leaving Sennheisers claims aside, what's interesting here is that we can see the increased distortion for all mic's towards higher frequencies.


/Peter