These chambers are not big, you get pressure chamber behaviour at low frequencies and can go down lower as 10Hz without problems. 100kHz is more tricky ... Level after the membrane resonance does not fall to zero but you have a lot of influences from every little thing on the front panel and from the speakers themself. With quite some tuning you can get 94dBSpl at 50cm up to 100kHz. (of course the speaker get's EQd for a linear measurement)
Had a look into that KA formular.
At 2kHz the 1" driver has a ka of 0,46 - but we have a wide baffle, there is some directivity in this frequency range from the baffle.
The 74mm membrane has a THEORETICAL ka of 1,7 - when it would be a flat disk of that diameter. It isn't. While this formulars are good to estimate things, real membranes don't behave exactly like that.
https://hificompass.com/en/speakers/measurements/bliesma/bliesma-m74a-6
There is only 1,5dB less spl at 60° and 2kHz - it rises above. So I still would say there is nearly no beaming from the 3" when you cross at 2kHz. (as you cross with a very wide radiating speaker and get mixed up in the transition zone ... you probably can cross a little higer)
p.s.: Seems I'm less familar with american formular as I thought - a for radius is normal?
At 2kHz the 1" driver has a ka of 0,46 - but we have a wide baffle, there is some directivity in this frequency range from the baffle.
The 74mm membrane has a THEORETICAL ka of 1,7 - when it would be a flat disk of that diameter. It isn't. While this formulars are good to estimate things, real membranes don't behave exactly like that.
https://hificompass.com/en/speakers/measurements/bliesma/bliesma-m74a-6
There is only 1,5dB less spl at 60° and 2kHz - it rises above. So I still would say there is nearly no beaming from the 3" when you cross at 2kHz. (as you cross with a very wide radiating speaker and get mixed up in the transition zone ... you probably can cross a little higer)
p.s.: Seems I'm less familar with american formular as I thought - a for radius is normal?
What a nice view! Your foto precisely shows nothing but the average loudspeaker tinkerer's secret dream ...
May I comment about the tapes on the baffle: Well done! It matches my experience that it is worth to even out any irregularity of the ideally plane baffle, as long as possible. And while proceeding in this logic - did you testwise try to even out also the 2 x four centermost screwpoints (centermost related to the center point of the baffle) of the bass drivers? And also the micro-small circular notch between the rim of the drivers and the baffle cutout? These structures are in the direct projection lanes of the Mids and the Tweeters on the baffle plane, and they eventially, as acoustically small as they are, might contribute to tiny, but in search of perfection avoidable diffraction artefacts.
The algebraic structure of the Ka equation is a bit odd, and to me, a bit confusing.
It is often expressed as something like "the Ka=2 frequency for that driver is 1000 Hz". If we work through the math, that driver would have an Sd of 375 cm^2. This is an odd way of expressing an equation, and it is overly complicated.
To the best of my understanding, the "a" in Ka is the wavelength which is equal to the apparent circumference of the diaphragm. So if the Sd is 375 cm^2, the apparent diameter is 21.8 cm, and the apparent circumference is 68.6 cm (pi * diameter). A wavelength of 68.6 cm has a frequency of 500 Hz
So Ka=1 is 500 Hz. Ka=2 is 1000 Hz. As I said, it is an odd way of expressing this math... "Ka=2 is 1000 Hz" is NOT a mathematical expression, it is some kind of jargon shorthand.
I hope this helps your understanding.
j.
It is often expressed as something like "the Ka=2 frequency for that driver is 1000 Hz". If we work through the math, that driver would have an Sd of 375 cm^2. This is an odd way of expressing an equation, and it is overly complicated.
To the best of my understanding, the "a" in Ka is the wavelength which is equal to the apparent circumference of the diaphragm. So if the Sd is 375 cm^2, the apparent diameter is 21.8 cm, and the apparent circumference is 68.6 cm (pi * diameter). A wavelength of 68.6 cm has a frequency of 500 Hz
So Ka=1 is 500 Hz. Ka=2 is 1000 Hz. As I said, it is an odd way of expressing this math... "Ka=2 is 1000 Hz" is NOT a mathematical expression, it is some kind of jargon shorthand.
I hope this helps your understanding.
j.
http://www.gedlee.com/Papers/directivity.pdf
KA represents a unit/degree of directivity for generic piston.
Last edited:
Thanks @camplo, I had misplaced my documentation and was working from memory.
So Ka is a dimensionless number, like Reynolds number or Prandtl number or Mach number.
Ka = 2*pi*f*a/c, with a=radius and c=velocity of sound
From my point of view, the equation "Ka = some constant" is most useful when we know what value of Ka we want, and we need to find the frequency f.
So making a useful algorithm from this, the first thing I would do is substitute r for a... It is much easier to associate "r" with radius.
Ka = 2*pi*r*f/c.
f = Ka*c/(2*pi*r) = Ka*c/circumference
If we know Sd, we can calculate either circumference or radius... our choice.
edit: since we most often know Sd
Just keep in mind that the units for C and Sd have to match... if Sd is in cm, then C = 34300 cm/s at room temperature.
So Ka is a dimensionless number, like Reynolds number or Prandtl number or Mach number.
Ka = 2*pi*f*a/c, with a=radius and c=velocity of sound
From my point of view, the equation "Ka = some constant" is most useful when we know what value of Ka we want, and we need to find the frequency f.
So making a useful algorithm from this, the first thing I would do is substitute r for a... It is much easier to associate "r" with radius.
Ka = 2*pi*r*f/c.
f = Ka*c/(2*pi*r) = Ka*c/circumference
If we know Sd, we can calculate either circumference or radius... our choice.
edit: since we most often know Sd
Just keep in mind that the units for C and Sd have to match... if Sd is in cm, then C = 34300 cm/s at room temperature.
Last edited:
OR just what type of directivity to expect per KA value. 2 is what I remember as a bare minimum for the lower crossing woofer. So a formula where you enter the ka nd it outputs what frequency this happens, is what I use. 5460(KA/Radius in Cm), I don't know where I got this formula from.
Correct me if I am wrong, but to see meaningful H3 at 20 khz, we need response out to over 60 khz, and a microphone and measurement tools to see this as far as I am aware. H3 at 30khz, would need response out to 90 khz in the same way.View attachment 1271170
H3 goes down from 2kHz. The tweeter is producing even less H3 as the 3" midrange. And we are talking of about 0,03% of H3 ... most setups can't even measure that
So in THIS case the tweeter is producing some more H2 and less H3 as the midrange at 2kHz. Both produce exceptional low THD!
Most home gamers are only able to get meaningful H3 up to about 7khz. Also depending on the windowing applied to the measurement, there may be not enough resolution, in other words the bins may be too far apart.
This is still a prototype setup of the installation - I can not show the complete setup with the ref & DUT mic setup installed. But there is an absorber addition to further minimize interaction between chassis. I will post a complete pic in a few days if thats interesting.
The lf drivers have no influence to the tweeter any more (which was a positive result) and the midranges don't run to high frequencies. I make bigger roundovers of the corners with the actual version of the speaker to reduce baffle step a bit more. Tweeter front mesh produces a notch at 45kHz which is not nice cause we have to compensate fir that and it pushes a lot of power in the tweeter.
There is a lot of tinkering to get these details right - nothing you need for listening at all cause most of these effects are at frequencies out of the hearing range. But it's a really cool project and you learn how to do it right.
The audio precision measurement system runs up to 80kHz and my Earthworks M50 is very linear up to 50kHz. For measureing higher frequencies I use 1/4" mics from GRAS or B&K, depending on the measurement setup.
That' not a home gamers play to do measurements up to 100kHz
But windowing - only makes problems at low frequencies. You always have plenty resolution at high frequencies.
In the measuring chambers you don't need windowing at all and in my semianechoic setup I normally also don't gate and accept the influences at lower frequencies, higher ones are clean.
It's the front grid (sorry, no native speaker and it's pretty late here
) - it doesn't vibrate but there is some chancellation, maybe the trapped volume or size of the openings ... wavelength is so small, there are interesting things happenig. I could never see an influence at smaller frequencies so it's nothing to concern about when you use ears instead of measurement microphones I tried alternaitve grids but always got worse results. It helps a little to move the grid more in front and away from the membrane what is very unpractical in a normal speaker. But I for sure recommend to stick with the original setup! Damaged Beryllium membranes are nothing you want to deal with and repair is not far from the price of a new tweeter.
I guess you know about the REW wavelet spectrogram feature? If not, then give it a try. You won't regret it. It's a fantastic and well appropriate tool for assessing and analyzing even micro kinds of speaker irregularities caused by reflections or diffraction artefacts.
As an example, the two pictures show the effect of the original metal grid in front of a Quad ESL63. To Grid or not to Grid only, e.g. testwise without the additional sock and the dust protection fitted of any regular ESL63. AAA is all air only, AGA is with the grid. In AGA, you see the first distinct unwanted grid effect at 400uS along with a blurred signal augmentation >600uS. I think a vavelet analysis for a metal grid protected beryllium driver would show a similar result.
Despite of intrinsic bad manners of any kind of metal grid partially obturating acoustic pathways, I would never ever place neither an ESL, nor a beryllium dome without theirs protective metal grids outside of a lab/testing/protoyping setup. And by the way and a bit off-topic here, my ESL63' are of course (re-)fitted with the complete set of dust protections, grids & socks for theirs safe and hopefully longterm daily use.
As an example, the two pictures show the effect of the original metal grid in front of a Quad ESL63. To Grid or not to Grid only, e.g. testwise without the additional sock and the dust protection fitted of any regular ESL63. AAA is all air only, AGA is with the grid. In AGA, you see the first distinct unwanted grid effect at 400uS along with a blurred signal augmentation >600uS. I think a vavelet analysis for a metal grid protected beryllium driver would show a similar result.
Despite of intrinsic bad manners of any kind of metal grid partially obturating acoustic pathways, I would never ever place neither an ESL, nor a beryllium dome without theirs protective metal grids outside of a lab/testing/protoyping setup. And by the way and a bit off-topic here, my ESL63' are of course (re-)fitted with the complete set of dust protections, grids & socks for theirs safe and hopefully longterm daily use.
Attachments
Last edited:
Just a tip…..if your desire is to have this system as the core of an ATMOS installation with overheads, you’ll need to voice this with a -3db dip centered at 8kHz. 8khz is our height cue and a peaked response in the overheads allows for a much more convincing procession of object based content. You could of course manage this with a little PEQ as well but the phase relationship isn’t as nice.
In all my years as a recording engineer, we struggle to choose the right microphones, positioning and room environment to avoid as much destructive equalization as possible. As such, I’ve become very sensitive to phase and a better listener and engineer for it. While I understand the current use of PEQ, it’s not the clean up tool that folks think it is…….it’s more liken to paint of a different shade…….not as bad a blemish, but noticeable for those with the notion to take notice. Food for thought.
In all my years as a recording engineer, we struggle to choose the right microphones, positioning and room environment to avoid as much destructive equalization as possible. As such, I’ve become very sensitive to phase and a better listener and engineer for it. While I understand the current use of PEQ, it’s not the clean up tool that folks think it is…….it’s more liken to paint of a different shade…….not as bad a blemish, but noticeable for those with the notion to take notice. Food for thought.
This is - by far - the lowest distortion speaker on the internet to date!
And it covers the entire frequency range, without any Helmholtz magic!
No seriously lol! These numbers are typical amongst high efficiency design or no? My horn/driver combo has similar performance if I stay above 2x cutoff like I'm supposed to. I would think the same is true for anyone else using compression drivers on horns. The low end would suggest that anyone using Sd=(2 12" drivers) or more are also in the same boat. People don't share these types of measurements often enough so...
Interesting thought! But all ATMOS reference systems are measured and set up linear. Therefore this dip/peak should be made during mixing? I can imagine to put a peak in the ceiling speakers but to be honest want my mains to be linear.
It doesn't matter if I do this dip with a passive crossover network, active in my dsp crossover or with Dirac in the goal curve - it's a minimum-phase system as long as it acts linear and phase has to follow the frequency response.
That's different at recording! Comb filters from reflections, Instrument behaviour, microphone off axis response etc. can't be EQt wihtout issues, it's not a minimum-phase system!
For linearisation of the drivers you NEED the phase shift of the EQ cause it counters the phase shift of the driver. So it strongly depends on the use of filters.