[cporada]

This past week, I built a pair of super tweeters for my ACI alphas to see what they would do to the sound (the alphas roll off at about 18kHz).  I had read an article in the Journal of Neurophysiology that the human brain responds to sounds above 20kHz, and that listeners in a double-blinded study were able to consistently identify music containing information from 22kHz-100kHz (in addition to the audible range) when compared to music only containing the audible range.  It appears that the brain processes this information and alters the way the person perceives the audible sounds, and people felt the music sounded more natural when it contained the ultrasonic information, even though they could not correctly identify when only the high frequency information was played by itself.  Inspired by this article, I set out to build a pair of these to see what they would do to the sound of my system.  I used the Fostex FT17H, since it is very efficient and linear to 50kHz. I built a crossover at 18kHz with a 1uF Kimbercap and a 0.1uF Audiocap Theta, and ran the signal from my binding posts up through an L-pad so I could match the output to my Alphas. The sound is intoxicating.  Everything is very natural.  It sounds like music now from real instruments right in my house, not like a stereo.  The soundstage is considerably wider, deeper and higher, and everything sounds very smooth and nonfatiguing, which I find curious, since I usually think of high frequency sounds as being very fatiguing and harsh.  The sound seems to float in the room rather than coming from the speakers.
    Having given this background, now my question; does anyone know how to time-align the supertweeter with my main drivers, i.e., where is the acoustic center on a horn-loaded tweeter, is it at the front of the dome?  Is there some way to calculate this to get the super tweeters positioned correctly from front to back on top of my ACI alphas?
Thanks
Chris Porada

[Mike Dzurko]

Usually the acoustic center is near the middle of the voice coil. However, keep in mid that ideally you will also want to account for the angle to your ears in your preferred seating area. If you have the means of generating an impulse response, that would be ideal.

Extreme high frequencies are not usually harsh. What we usually hear as harsh is generally much more in the upper midrange and lower treble, the part of the spectrum where our ears have evolved to be the most sensitive, (think screaming baby).  Very high frequencies often add a sense of air and space, and this fits precisely with what you're hearing. Sounds like you've made a good improvement for not a lot of money, congrats!

[ScottMayo]

This past week, I built a pair of super tweeters for my ACI alphas to see what they would do to the sound (the alphas roll off at about 18kHz).

I've toyed with the idea of ultrasonics on and off. The thing that stopped me was that I don't have any sources that will generate anything above 22kHz or so. What kind of source are you using?


Time alignment at these frequencies is not much of a concern. The wavelengths are very, very short; nothing in your brain, as far as I know, is going to be able to detect and process phase information in that range. If it's working for you, I wouldn't mess with it.

[cporada]

Hi Mike and Scott,
Thanks very much for the replies.  In answer to your question Scott, I use SACD, vinyl and CD, but the strange thing is the improvement is very apparent with redbook CD as well, which leaves me scratching my head for an explanation. The only thing I can think is that the upsampling on my SACD player artificially adds high frequency info back to the redbook CD, but this is a guess. I know when Stereophile, Positive Feedback, and other mag's reviewed the Townshend supertweeters, they also heard an improvement on regular CDs and were equally puzzled with this, but perhaps this is due to marketing in this popular mag's.

I also have a question for Mike (or anyone else); if my crossover to the super tweeters (1.1uF caps) is -3db down at 18kHz (according to a cross-over calculator i found on the net) and has a 6db/octave slope, how low are the supertweeters playing at an audible level?
Thanks
Chris

[Mike Dzurko]

Hi Mike and Scott,
Thanks very much for the replies.  In answer to your question Scott, I use SACD, vinyl and CD, but the strange thing is the improvement is very apparent with redbook CD as well, which leaves me scratching my head for an explanation. The only thing I can think is that the upsampling on my SACD player artificially adds high frequency info back to the redbook CD, but this is a guess. I know when Stereophile, Positive Feedback, and other mag's reviewed the Townshend supertweeters, they also  ...

If the impedance of the tweeter is very flat, the tweeter will be down 9db at 9Khz, which is about ½ the volume level , (-10 db is ½ volume). At 4.5Khz you’ll be down 15db and so on.  Once the tweeter itself starts to roll off on the low end the slope will become much steeper. As far as audibility, you can do your own test using sine waves. Play the test tones in decreasing frequency with and without the supertweeter and let us know what you find out 

[BrunoB]

This past week, I built a pair of super tweeters for my ACI alphas to see what they would do to the sound (the alphas roll off at about 18kHz).  I had read an article in the Journal of Neurophysiology that the human brain responds to sounds above 20kHz, and that listeners in a double-blinded study were able to consistently identify music containing information from 22kHz-100kHz (in addition to the audible range) when compared to music only containing the audible range.  It appears that the brain processes t ...

Could you try to play music with the supertweeter alone?  I wonder if you would hear something because of the 6 db slope.




Bruno

[cporada]

Thanks for all the helpful replies thus far.  I will see what I can do about listening to them alone or testing them with some test tones (I have the Stereophile test CD) to see what I find out.  I was considering also seeing what happens if I remove the 0.1uF cap and thus shift the -3db point up to 20kHz instead of 18kHz, but I'll have to desolder the cap out, and I don't want to keep heating the leads on the caps too much.  I can defintiely hear material coming out of the super tweeters while listening to music, so I'm sure there is output in the audible range.  Output is most apparent in music with cymbals (brushes on cymbals are now great), bells, or Native American and Tibetan wooden flutes, which seem to have a lot of high overtones.  Some rock CDs (Tom Petty in particular) seem to have a somewhat inflated treble sound, but that was apparent with the alphas prior to adding the supertweeter, so I think those recordings are just bad.
Thanks for all the suggestions
Chris[/img]

[audiobomber]

I was considering also seeing what happens if I remove the 0.1uF cap and thus shift the -3db point up to 20kHz

The impedance graph for the F17H shows it to be around 9 ohms in the crossover region, which means you should be -3dB at about 16kHz with 1.1uF.

[ScottMayo]

In answer to your question Scott, I use SACD, vinyl and CD, but the strange thing is the improvement is very apparent with redbook CD as well, which leaves me scratching my head for an explanation. The only thing I can think is that the upsampling on my SACD player artificially adds high frequency info back to the redbook CD,

Upsampling can't add what's not there, and nothing above 20kHz is there. A CD simply cannot encode information above 20kHz.

My guess is, you've just added more output between 18-20kHz, and your ear likes it. Nothing ultrasonic about it (well, it would be for me, I can't hear much of anything about 15.5kHz.), but there's good musical info there and getting it louder might be a help in all sorts of situations.

I read somewhere that cymbals output frequencies to well above 50kHz; when a cymbal is hit we don't hear *most* of what is going on. Our cats know all about it, though.


I found this amusing:
http://www.lsu.edu/deafness/HearingRange.html
Cows hear more than we do - and don't even start on porposies.

[Russell Dawkins]

Since, according to that chart, humans hear only from 64-23,000 Hz, I guess I'm wasting my time with a subwoofer and I can't hear the low E on a bass guitar (42Hz). I must just be hearing the octave.
18-20kHz represents less than two tones in the top of the uppermost octave, the whole last octave being 10-20kHz. I can't imagine how that alone would account for what Chris reports hearing.

[SET Man]

Hi,

   Chris and other... I've been  using the Fostex FT17H for a few months now. Great tweeter thought. Excellent for super tweeter.

   Okay, right now I'm using the FT17H with my Fostex FE167E. The 167E are running fullrange and have rated efficiency of 95dB. The FT17H is rated at 98dB.

    Yes, accroding to the xover calculation with 1uf cap would give -6dB at 18khz But of cause this doesn't work with all driver. And the reason why you said that you hear improvement even with CD is becasue the FT17H still produce sound below 18K. Keep in mind that eventoght your main speaker say that is rated up to 18K or even 20k in real room with off axis that would less. So, the 17H would be filling the gap here and that might be the reason why you hear improvment even CD.

    Anyway, I fooled around with xover trying come up with the right blend with my main. Fostex recomend using .68uf-.33uf with my 167E. I started with 1uf cap and .68uf with L-pad than  than .47uf than .33uf. And amazingly .47uf and .33uf seem to blend well with my main. Currently I'm using .33uf cap without L-Pad. Doesn't sound right isn't it? According to xover calculation that would be at 60K!!!! And yes even that I could still hear lower range play through the 17H alone.

   So, you must expriment with the xover until you get the right blend with your main speaker. Try higher cap and see if sound blend well.

   And as far as the question about if adding super tweeter make any differnt? Personally I feel that it does make thing sound better, especailly with well cut LPs from analog master.... maybe they do contain some ultra sonic? Or maybe the tweeter load + the main load cause fs respond to change? Or maybe I'm loosing my hearing of high Hz? Or maybe is just my imagination? Oh! what the heck! The system sound great now, I will leave them on.

Take care all,
Buddy

[ScottMayo]

Since, according to that chart, humans hear only from 64-23,000 Hz, I guess I'm wasting my time with a subwoofer and I can't hear the low E on a bass guitar (42Hz). I must just be hearing the octave.

The rest of the article was at some pains to point out that they normalized the results to get to a common criteria. Human hearing starts to take a nosedive in sensitivity in the low frequencies; probably it fell off the end of their statistical cutoffs, at least that's my guess. Anyway, plenty of people hear frequencies above 20Hz as individual beats; I don't hear a real tone until somewhere around 23Hz and some people go higher than that. 20-20k is a rough and convenient simplification. So is the data in that chart - it's just a different set of simpifications.

18-20kHz represents less than two tones in the top of the uppermost octave, the whole last octave being 10-20kHz. I can't imagine how that alone would account for what Chris reports hearing.

I have no idea how to account for it. I know that redbook CD can't produce anything over 20kHz. (Well, that's a simplification - it does a bad job with 20kHz and some less than ideal CD players might put out some pretty harsh, sharp-edged stuff trying, which could turn into ultrasonic noise. But there are usually filters for that.)

But I know my own hearing cuts out in the 15k's, but music with sound above that range still "sounds different to me". I could speculate for days why, but I don't know. I just know I need to stick with speakers and amps that reach well above 15kHz to be happy.  I read some research somewhere that claims that people react (brain function changes measurably) to sounds upwards of 50kHz and higher, even though they can't *hear* it. Ultrasonics is an interesting field...

[cporada]

Hi Scott,
I think the research to which you refer was published in Journal of Neurophysiology in 2000 by Tsutomu Oohashi and colleagues.  They conducted a very well done double-blinded study (neither the investigators or the subjects knew what signal was being played at each given time) in which subjects were allowed to listen to music containing only information from 20-20,000Hz, the same music with information from 22,000-100,000Hz added on, or only the 22,000-100,000Hz information.  While none of the subjects could tell when only the high frequency information was playing (as opposed to silence), they could all distinguish between the music that stopped at 20,000 and that which contained the additional information above 22,000Hz.  The authors also confirmed these results by showing alterations in brain electrical activity and in regional cerebral blood flow that occurred only when the high frequency information was present.  I guess it makes you wonder what "hearing" really means.
Chris Porada

[ScottMayo]

Hi Scott,
I think the research to which you refer was published in Journal of Neurophysiology in 2000 by Tsutomu Oohashi and colleagues....

Yup, dat's der puppy. And yes, I do wonder what hearing is. My guess is, the next big thing in HE audio is going to be sources that represent 10-100k. SACD can do this, but as far as I know, no one is using it for that purpose.

Of course, with mp3 taking over the world, there's not a lot of incentive for carefully recording instruments up to 100kHz.  

[cporada]

Thnaks for all the help from everyone on this forum.  I wanted to bring averyone up to date with some additional progress I've made with the super tweeter.  I took out the 0.1uF cap, leaving just a single 1uF Kimbercap as the crossover, and am using the Fostex R80B 100W L-pad at about -8db to match the output level to my Alphas.  Now the transition appears to be seamless both on music and on treble warble test tones.  One additional thing I did which has made a tremendous leap forward in soundstaging and imaging was to use a laser pointer to attempt to angle the supertweeters down (they are on top of the Alphas, which stand about 5 feet tall) so that the signal from the supertweeters hits at exactly the same height at my listening position as the signal coming from the alpha's main tweeters.  This was a huge improvement.  I figured the drivers on the JM Lab Grand Utopias and the Wilson's were set up like this, so why not try it.  I'm going to radio shack to get a sound meter to make some measurements to get an idea what the linearity of the frequency response is, but to my ears at least, it must be pretty good.
Thanks ACI for making such a great kit that 10 years later I can add a couple of low cost tweaks and still be blown away with the unbelievable realism of the sound that I get from my system. The original cost of the alpha kit was about $400, and I've listened to speakers at local shops selling for 8 times that much, and just happily chuckle to myself and head home to listen to some real music.  Thanks Mike and everyone at ACI.
Chris

[Mike Dzurko]

Chris:

You're very welcome  Please do report with any further changes/ improvements. This has been a great discussion, and I do think that the top octave and beyond is an area in audio that could certainly use more research.

[Mike Dzurko]

Gang:

I think you'll find this interesting.  John Potis reviews the $2500 pair muRata ES103A supertweeters as add - ons to the Sapphire XLs. :

http://www.6moons.com/audioreviews/murata/murata_2.html

[Groch]

Facinating review.  It appears that supertweeters can make a major impact particularly if you listen to SACD or DVD-Audio.

I thought his comment that adding these supertweeters to the XLs "makes them competitive with speakers that cost 3 times the price" was amusing.

The reviewed supertweeters sell for $2500 which makes the combination sell for almost 3 times the price.