I'm sure the measurements section of many of the stereophile "high-end" speaker reviews would provide ample fodder
(It actually surprises me how often speakers that measure very un-flat still get good comments in the objective/measurements part of the review, let alone the subjective part...)
*cough* Advertising Cash Cow *cough*
Pano,
Certainly is a lot going on in the tones, your RTA shot of the test tone has better resolution than the Logic EQ RTA, interesting to see the 4, 8, 10, 12 and 14 kHz harmonics lurking below the level of the primary beat frequency tone of 2 kHz.
The beat tones only are audible where the output of the two horns converge, they are not present in either horn or driver whatsoever.
As the Maltese horns used for the test are very narrow dispersion, and were toed in, there was only a small area in space centered around the microphone where the beat tone was audible.
Several of the recordings show the small area phenomenon, in one I tried to whistle along with the beat tone, but for me to hear it well, I had to put my head in front of the microphone, so in the recording at that point you can only hear my (off pitch) whistle while the beat tone disappears, as in the “hand” recording in post #180.
At 16 & 20 kHz, a hand in front of the microphone is pretty much the same as a wall, though a tiny portion of the tones got through the gaps in my fingers.
Smaart only reads to 23.74 Hz, so we can’t read the second harmonic above a fundamental frequency of 11.87 kHz.
The BMS drivers seem constant in distortion vs. SPL, judging by the distortion at 10 kHz (and other sine wave tests):
4550 four volts (one watt) 10K 104 dB, 20K 80.5 dB -23.5 dB, (<6%distortion)
4552 four volts (one watt) 10K 105.4 dB, 20K 82.5 dB -22.9 dB (around 7.08 % distortion)
The upper harmonic distortion in the 15-20 kHz range should be similar since the SPL generated in the test was almost the same.
The output of the signal generator, amp, DSP, and microphone are quite clean, as can be seen in the twin Smaart screen shots of the BMS 4450 at 16 kHz and the BMS 4552 at 20 kHz in post #180.
I have included a screen shot below with the level raised by 20 dB so the noise floor is visible.
There is some wind noise visible in the noise floor, as well as 60 Hz line noise harmonics, and some driver sub harmonics.
The highest level sub harmonics present in the 4552 with the 20 kHz tone is 57 dB down from the fundamental, THD around 0.2 %.
The loudest sub harmonic of the 4550 with the 16 kHz tone is 60.1 dB down from the fundamental, THD in the 0.1 %. range.
Looking at and hearing these results are a real eye-opener to me, this test is the most fascinating I can recall other than seeing Tom Danley’s acoustic levitator.
The results have significance in the context of extended VHF response.
They also have me thinking off topic of what type of algorithm would be required to invert audio frequencies so the beat frequency output would be intelligible for projecting speech (or music) to a specific location in space.
I play the Theramin, and can “kind of” play melodies with an analog sine wave generator. Conducting some of the test was odd for me, as the melody heard when trying to “play” the generator is the opposite of what one is used to, turn the knob up and the pitch goes down.
One zip file attached is an example of some “playing”, and the beat tone between the fixed 20 kHz generator and the BK Precision analog generator is audible down to around 100 Hz. The "jet" zip file is a continuation of the same recording, 100 Hz can be heard over (under?) the high altitude airplane.
The BK “Precision” generator gets pretty tough to set when the frequencies are high, a tiny knob movement results in a 100 Hz change, at times the 4550 went above 20 kHz and still was able to generate beat tones.
To open the audio files, change “zip” to “mp3”.
Use caution in playback, the HF tones could burn up tweeters without you hearing them if played too loud.
Best to use headphones.
Art Welter
Attachments
Hi Art
Ok, I am not sure what your point is. That air is nonlinear? Thats not new nor argued with. When you play two tones there will be a difference tone generated if the medium is nonlinear. This tone will be at exactly the same frequency as the "beats', the beats being what are actually exciting the nonlinearity. This is all quite expected. I somehow am missing any significance in this (except the obvious fact that VHF generate lower frequencies in typical audio playback systems, the later being audible even if the VHFs are not. This implies that playing signals at VHF will generate signals at low frequencies even if you cannot hear the VHF noise. Unless this effect is very tightly controlled any tests of audiblity using VHF signals are questionable.)
Ok, I am not sure what your point is. That air is nonlinear? Thats not new nor argued with. When you play two tones there will be a difference tone generated if the medium is nonlinear. This tone will be at exactly the same frequency as the "beats', the beats being what are actually exciting the nonlinearity. This is all quite expected. I somehow am missing any significance in this (except the obvious fact that VHF generate lower frequencies in typical audio playback systems, the later being audible even if the VHFs are not. This implies that playing signals at VHF will generate signals at low frequencies even if you cannot hear the VHF noise. Unless this effect is very tightly controlled any tests of audiblity using VHF signals are questionable.)
The "beat" frequency is the frequency of the envelope of the two signals. It exists even if there is no "tone" at the difference frequency. But any nonlinearity, such as in a driver, in the air or in the ear will create an actual "tone" at the "beat" frequency. This makes testing the audibility of VHF extremely difficult.
My comments, and I believe Earl's as well were to the belief that supersonic tones need to be recorded for best fidelity. This is not true. The arguement typically goes that you may not hear tones above your normal hearing range but they can intermodulate and create something that you can hear. Your measurements show the manner of the intermodulation but do not show the need for having the supersonic elements.
The way I would explain it is to picture a live music event. The musicians are creating lots of tones, some in the audible range, a few in the supersonic range. If the air is nonlinear then some of the supersonic notes will mix and create audible subharmonic difference tones.
If I put microphones at the listener location I will capture both the priamry tones and any of the already generated subharmonics. That is what my ears would have heard at that location. Now, if I record supersonic tones as well, and if my playback system is susceptable to generating intermod products, then I run the risk of hearing at home the orriginal event plus new intermod products created by my system on my side of the fence.
That is not an improvement.
Back to beat notes. Beating is just a perceptual artifact of the ear latching on to the modulated envelope. In a linear system we might "hear" the difference between two tones but a spectrum analyzer would see nothing there. For my example of a piano string at 440 and a second string at 441, we would perceive a 1Hz beating. A spectrum analyzer would show no 1 Hz component. As soon as we add nonlinearity to the system then 1 Hz may be generated (2nd order nonlinearity required).
People who study radio theory are generally familiar with these concepts because the mixing products are both good and bad. Good in that a desired signal mixed with a second oscillator allows conversion to another intermediate frequency, the basis of superhetrodyne reception. Bad in that 3rd order nonlinearity creates intermod products (phantom stations) at adjacent frequencies next to two strong signals.
David
The way I would explain it is to picture a live music event. The musicians are creating lots of tones, some in the audible range, a few in the supersonic range. If the air is nonlinear then some of the supersonic notes will mix and create audible subharmonic difference tones.
If I put microphones at the listener location I will capture both the priamry tones and any of the already generated subharmonics. That is what my ears would have heard at that location. Now, if I record supersonic tones as well, and if my playback system is susceptable to generating intermod products, then I run the risk of hearing at home the orriginal event plus new intermod products created by my system on my side of the fence.
That is not an improvement.
Back to beat notes. Beating is just a perceptual artifact of the ear latching on to the modulated envelope. In a linear system we might "hear" the difference between two tones but a spectrum analyzer would see nothing there. For my example of a piano string at 440 and a second string at 441, we would perceive a 1Hz beating. A spectrum analyzer would show no 1 Hz component. As soon as we add nonlinearity to the system then 1 Hz may be generated (2nd order nonlinearity required).
People who study radio theory are generally familiar with these concepts because the mixing products are both good and bad. Good in that a desired signal mixed with a second oscillator allows conversion to another intermediate frequency, the basis of superhetrodyne reception. Bad in that 3rd order nonlinearity creates intermod products (phantom stations) at adjacent frequencies next to two strong signals.
David
Maybe a little bit of semantics, but my main point was that "beats" are not "tones". Words matter.
Try this web site.
Beat Frequencies
Look down the page to the section showing the waveforms.
The beating is a phenomonon of the envelope and, in the absence of nonlinearity, there is no actual tone at the difference frequency. You perceive the variation in the envelope. Again, with a 440 and 441 pairing, you hear a 440 (ish) tone with a 1 Hz warble to it. 1Hz does not exist. You can not hear 1 Hz. The piano can not radiate 1 Hz. But you will sense the periodic variation in the envelope.
The beating is the artifact of the two near waves going in and out of phase. If they start in phase, 1/2 second later they are exactly out of phase and their outputs will cancel. 1/2 second later they are back in phase and their amplitudes add to double. So the combined amplitude envelope comes and goes at a 1 Hz rate. Now if you move the frequencies far enough apart, say 440 and 900, there is no perception of beating and you perceive 2 independent tones.
440 and 441 are true tones. A spectrum analyzer would indicate 440 and 441 are present. 1 Hz is not a true tone, it is a beat. The spectrum analyzer would not detect it.
David
Beat Frequencies
Look down the page to the section showing the waveforms.
The beating is a phenomonon of the envelope and, in the absence of nonlinearity, there is no actual tone at the difference frequency. You perceive the variation in the envelope. Again, with a 440 and 441 pairing, you hear a 440 (ish) tone with a 1 Hz warble to it. 1Hz does not exist. You can not hear 1 Hz. The piano can not radiate 1 Hz. But you will sense the periodic variation in the envelope.
The beating is the artifact of the two near waves going in and out of phase. If they start in phase, 1/2 second later they are exactly out of phase and their outputs will cancel. 1/2 second later they are back in phase and their amplitudes add to double. So the combined amplitude envelope comes and goes at a 1 Hz rate. Now if you move the frequencies far enough apart, say 440 and 900, there is no perception of beating and you perceive 2 independent tones.
440 and 441 are true tones. A spectrum analyzer would indicate 440 and 441 are present. 1 Hz is not a true tone, it is a beat. The spectrum analyzer would not detect it.
David
What he said.
A "tone" is a real signal, a beat is an envelope modulation. It really is important to understand the difference here. It is not just semantic.
And Dave was right on the money when he talked about how nonlinearities of inaudible VHF signals can fold down into the audible passband yielding all kinds of things that are not what are intended. For example, we hear the folded signals and think that we are hearing the inaudible ones. Its a real can of worms.
Last edited:
David,
Obviously we can’t hear 1 Hz at less than earthquake levels :^).
However, we can clearly hear the 2 kHz difference tone created by the frequency of beating of an 18 and 20 kHz tone reproduced by two BMS drivers on efficient horns driven by about one watt each.
The spectrum analyzer clearly can and does detect the 2 kHz difference tone, it is therefore a true "tone".
In fact, the spectrum analyzer and our ears can clearly detect the beat frequency tone down to the 100 Hz range when one horn is broadcasting 19,900 Hz and the other 20000 Hz.
Perhaps you have not read my posts #179 #182, or listened to the recordings.
I welcome any questions regarding how the test was conducted.
The fact is audible tones are generated by ultrasonic tones.
You may argue that the sub harmonic tone would not be audible in a “normal” listening environment, I would not care to join in that argument, as subjective perception is difficult to measure.
As Earl says in post #189:
“Maybe a little bit of semantics, but my main point was that "beats" are not "tones". Words matter.”
I agree completely.
The audio files and RTA screen shots in my posts #179 #182 prove conclusively that the beat frequency of two ultrasonic tones create an audible tone visible on an RTA in addition to the ultrasonic tones, the lower tone generated by the difference between the two ultrasonic tones.
The fact that the ultrasonic tones are causing the audible beat frequency tone is quite obvious in the recording in post #179 when I put my hand between the microphone and the two horns, blocking the 18 and 20 kHz tones, blocking the 18 and 20 kHz tones, which kills the 2 kHz tone the beat frequency caused.
Had the 2 kHz (and 4, 8, 10, & 12 kHz) tone been generated in either of the horns, some portion of the tone would have gone around my hand and reached the microphone.
In post #182 the output of the two individual horns is shown, they are not generating sub harmonics.
The 2 kHz tone is not a “subjective tone” heard only by our ears, it is clearly visible on the RTA graph.
The 18 and 20 kHz tones emanating from the two horns were at a level of 103.9 and 104 dB SPL.
Although our hearing is nonlinear, air is still quite linear at 104 dB.
The audible beat frequency tones generated by the 18 and 20 kHz tones were 71.6 dB at 4000 Hz,
54/6 dB at 8000 Hz, 57.5 dB at 12000 Hz.
A 10000 Hz tone is visible on the RTA at 44.2 dB, in the level range of the electrical and wind noise.
The screen shot below is the B&K 4004 microphone “hearing” the two horns, the level has been raised by 20 dB so the noise floor is visible.
The wind was fairly low for this time of year hear, it is responsible for the noise under 60 Hz, most of the noise at 60 Hz and above are AC line noise harmonics visible in all my tests using the same gear.
Again, perhaps you and others don’t understand the details of my test, I welcome any specific questions.
Cheers,
Art Welter
Attachments
I have read your posts but I haven't listened to your samples yet, but will later.
I've chosen my words carefully and several times pointed out that difference tones and subharmonic tones are clearly possible if nonlinearity is present. Since you are clearly seeing real tones generated, then this has to be due to nonlinearity in the system, either in the drivers or perhaps the air itself. Of course, these are not beats but are intermodulation products.
As I have said several times, the only subharmonic tones that I would want reproduced are those that had already occured at the location of the original recording (knowing they are likely inharmonic, we could probably do without them as well). Subharmonics generated during reproduction (either in the drivers, in our amplifiers or even in the air) are clearly distortions that we should strive to eliminate. If it takes strict bandlimits to reduce in-band products then thats what we should do.
David S.
QFT.
My fullrange speakers go to about 15k like the ones of the OP and i feel a lot of harmonics, ambience and air missing from them.
At the time, i preferred to NOT listen to HF distortion of the poor digital sources that i had in the past.
Last edited:
And what high-fidelity means?
HFs are not that hard to reproduce from speakers (although hard to do it WELL), why not even trying? Nobody is saying that it's more important than doing the midrange right...
David,
From all I have read, air non linearity is not much of a problem until around 150 dB SPL.
The test level was 104 dBA (and 103.9) at the microphone, if air non-linearity exists at such a low level, please show us some documentation as to that, even moderate levels at the listening position require that SPL near a tweeter.
Sub harmonic distortion is clearly visible, albeit at a low level, when 18 kHz and 20 kHz was fed to a single driver, but none with discreet tones to the separate drivers
The RTA screen shots of the individual horn output show no non-linearity, if there was any in the tone generators, DSP, amp channels, drivers or horns they would be clearly visible, but they are clean down to the noise floor.
You seem to be missing the fact that this test, typical of any multitrack recording (the vast majority of modern music recordings) , unlike your oft-repeated field mic example in which "sub harmonic tones already occurred at the location of the original recording", sends discreet information (tones) to each loudspeaker.
The only way one can hear any sub harmonic tones generated by ultrasonics recorded with a discreet signal path is to have speakers capable of reproducing those ultrasonics.
You can argue that accurate reproduction of inharmonic aspects of recorded music is not desirable.
Everyone is entitled to their opinion, but that opinion is perhaps the first time I find myself in disagreement with you.
I prefer that a reproduction system should reproduce music as accurately as possible, regardless of the content, "if the music sucks, let it suck out loud"
.Please listen to the recordings and review the supporting graphics again, I think you will be quite surprised, I certainly was.
After decades of doing my own tests, and reviewing others, finding a real surprise is kind of refreshing.
Art
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.