Something stirs in the Undergrowth

[simon]

I've never felt the urge to get a CNC machine, but wow, I do now. That horn's a thing of beauty. Just loved it when he started to rub the finish on at the end and the wood changes colour and lustre.

[IslandPink]

Yes, i remember that with the Yuichi !
Now, this is all well and good, and watching a CNC machine do that must be very satisfying.... but when are we going to see an achievable horn solution for the next 2 octaves down ?
If you put an Onken-style mass/mid cabinet below that, it's just the same old problem that I had, maybe 10 years ago - the horn-loaded compression driver will just outclass it to where it sounds plain wrong.

[IslandPink]

1. Mark, the phase wiggles here are orders of magnitude smaller than those created by a typical crossover network. I suspect that some people are more sensitive to time domain problems (ie, phase shifts) than others, and I'm happy to accept that you might be much more sensitive than most.


- FIR processing of a multi-way speaker
2. Given where you are at the moment with your cabinets etc, I'd love to bring the Powersoft amps over (10x channels of FIR, variable output impedance <400Hz and power) and see how good we can make it. Once the lockdown is over, let's see if we can arrange something.

Thanks for the input, Chris.
1. Fewer wiggles in that area as shown by Troels do seem to give more natural, believable vocals, though. I mean - the compression driver on the Yuichi is more or less perfect from about 800Hz up, and it shows.
With crossovers, they are often higher, and I suppose although the phase shifts are there, one driver or the other is always falling away in output level, so that helps.
But in any area where you put in a crossover, even 1st-order, there will be a loss of fidelity, particularly in tone and dynamics. On the current OB, that's around 200Hz or 180Hz, and that's where I have gripes about the sound !

2. The FIR-based system will be very educational to hear, I'll look forward to that.

[SimonC]

.... but when are we going to see an achievable horn solution for the next 2 octaves down ?

The further I get into Bjorn and Thomas's epic book the more I'm appreciating the size of the challenge at these frequencies. Literally size... and folding then adds another layer of complexity to juggle. One advantage we have now even over 10 years ago is vast amounts of computing analysis power for virtually nothing (in relative terms), but for horn development how to interpret this analysis is still in its infancy.
Most of the traditional ways of simplifying the maths required (Thiele Small, lumped parameter models, 1D wave equations etc) just don't cut it in horns; at best they give you a starting place to begin your journey from. The next steps along that journey are then mainly empirical: listen, make a guess about what to change, tweak your design, listen, make another guess, tweak, listen etc until you get get bored, run out of money or have a breakdown.
FEA style analysis can already give a much more accurate approximation of what the air in the horn is doing at different frequencies, which is a big step forwards in understanding, but what it won't do is tell us what it sounds like. This is the missing link at the moment...
Onwards

[Nick]

Maybe its worth asking a few naive questions then. What is that the horn that is doing that makes you want to use it? Does the problem(s) it solves be solved in any other way?

I assume the size of the horn is controlled by the wavelength of the sound? Could you make something that works like a horn but is made of material that means that sound travels slower in it? Would that let you make a smaller horn? Would you still have the problem of interfacing it with the air at the end?

[Cressy Snr]

Maybe big horns belong in cinemas and music venues, and just aren’t meant to go in people’s living rooms, which is why nobody can get anything of domestic size to work

Not helpful. I’ll get my coat.

[IslandPink]

1. What is that the horn that is doing that makes you want to use it? Does the problem(s) it solves be solved in any other way?
2. I assume the size of the horn is controlled by the wavelength of the sound?
3.Could you make something that works like a horn but is made of material that means that sound travels slower in it? Would that let you make a smaller horn?
4. Would you still have the problem of interfacing it with the air at the end?

1. I think it's the fact that the horn loading equalises the amplitude ( FR ) while at the same time keeping the phase response flat across the same band. This is most obvious in the very long horns where this can be achieved over more than 4 octaves. I am however a little less sure about this since Hornresp changed the way they show phase, allowing you to choose the 'delay' value that unwraps phase. On the plus side I think Danley made this point about phase so he's a reliable source.
In addition, you are typically using a smaller driver on a horn, for a given frequency range, compared to a direct radiator - so less likelihood of breakup.
I think you can do this another way, but whether is sounds any good is something I'm not sure about. If you take a bigger-coned driver, with low Qts ( powerful magnet ) , unloaded ( ie. no box ) then you already have low phase shift across the band of interest ; but also a very slanted, rising response from bass to mids. Then you equalise the hell out of it with a digital crossover - and this must be FIR, to prevent re-introducing a phase rotation. Now, almost everybody who's doing this for bass is using big beefy PA drivers, because they have enough Xmax to cope with the equalisation. Whether it's possible to find a compromise driver, like an Altec 416B or something, that has lower Rms, but could cope with the equalisation, at least over say 70Hz to 600Hz ? - that might work. There is always the danger of cone breakup in the upper mids with any 12" or 15" driver though.

2. Yes, that's the big problem, for the lowest frequency you want to work, you need a horn whose length is at least half of that wavelength, and the horn mouth circumference is at least a quarter, preferably half, of the wavelength corresponding to that frequency. That's already a beast at 100Hz, and the length means you have practically problems getting it time-aligned to the upper-mid horn without using a 1 or 2-wavelength disjoint at crossover.

3. It's an interesting point, but I had a think about this and realised it's unfortunately not like optics. In liquids and solids, although they are more dense ( like optical glass is, vs. air ) the speed of sound through them increases dramatically - eg. 10x or more. So given that it's the wave frequency that remains constant in any transition from medium to medium ( as in optics ), the wavelength, instead of getting smaller, gets dramatically bigger in the solid or liquid. I can't think of any material in which sound travels slower than it does in air.

4. Yes, though an academic point after (3) , it would be another problem, yes.

[chris661]

Maybe big horns belong in cinemas and music venues, and just aren’t meant to go in people’s living rooms, which is why nobody can get anything of domestic size to work

Not helpful. I’ll get my coat.

I find horns can work in a domestic setting, but only for the mid-high range so far. Current setup uses a B&C DE250 on an XT120 horn from about 900Hz upwards. Below that is an 8" driver.

I can see some benefits of horn-loading to a lower frequency, but that mostly involves keeping sound off the side walls etc for a better stereo image. The Danley Synergy horns, for instance, are reputed to sound excellent in domestic spaces. Wouldn't want them here, though - the lower you go, the bigger the horn needs to be. Some of the Danley ones are considered big by PA standards, let alone home HiFi.

There's also the issue that complex multi-way systems are a real PITA to get running correctly, and that's me trying with a measurement mic and almost-unlimited DSP. I wouldn't attempt it with simple passive crossovers. It would be possible, I suppose, but even more difficult.


Nick, those are good questions. IIRC people have considered similar before - things like using a particularly heavy gas inside a sealed enclosure to drop the system resonance. I'm not sure how well that would work for horns, though, and of course you still have to interface that with the listening room - probably requiring an airtight membrane which would then act like timpani.

Mark, I'll be in touch once we're allowed out to other people's houses. The car could do with a proper run - I'm going to the shops once a week and taking the long way around to get the engine up to temp and the A/C run through.

Chris

[chris661]

Post #2, since I posted at the same time as Mark.

In response to point 1, I think it's worth noting that direct radiators also have a flat phase response across their linear range. You seem to be avoiding the obvious solution, then - what's wrong with putting drivers in a box?

In theory, you could grab a set of nice drivers that are each flat through and outside their wanted bandwidth, apply first-order filters, and have a resulting speaker that's linear in phase and frequency response from the treble until it rolls off in the bass.

Something like a good 1" dome tweeter, crossing around 3kHz to a 4" mid, and 500Hz to a 8-12" bass driver would be one way of doing it. I'd use the 1" dome that was used on the latest Edingdales (IIRC, Scott said that was happy going pretty low), the ScanSpeak 10F full-range driver (smooth out to ~10kHz), and pick a bass driver that gives the desired LF performance while also having a nicely-controlled rolloff at the top end.

NB - the acoustic rolloffs must be well away from the electrical crossover frequencies, or they'll combine and some phase shift will come through.

You could go a step further, I suppose, and use a good coaxial for the mid-high range, which would improve the vertical off-axis performance.

Just thinking aloud here. I get that some people don't seem to like box-based loudspeakers, but IMO if they're done well (adequate bracing, internal absorption, making sure no midrange comes out the port if there is one) there's not a lot wrong with them.

Chris

[SimonC]

Maybe its worth asking a few naive questions then. What is that the horn that is doing that makes you want to use it? Does the problem(s) it solves be solved in any other way?

I assume the size of the horn is controlled by the wavelength of the sound? Could you make something that works like a horn but is made of material that means that sound travels slower in it? Would that let you make a smaller horn? Would you still have the problem of interfacing it with the air at the end?

There's a lot to unpack there, it is going to take me a few posts and some pondering to get through them all.

What does a horn do - its an attempt to make a more efficient impedance transformer between the speaker diaphragm and a body of air than than the diaphragm on its own.
Why does this matter? You can make low frequencies by taking a large cone and throwing power at it. kW are cheap and easy to do, add in some DSP and you have a system that in theory does everything you could ever want. In reality physics kicks in, to shift a large volume of air you need a large cone, and the larger the cone the greater its mass. Transient response (which is some of what Mark is chasing) is governed by that mass, it takes time to accelerate it and decelerate it again, and the more mass you have the more power you need. It quickly reaches a point where it doesn't matter how many kW you throw at it, the mass is just too large to respond quickly enough. You can play around with magnet materials etc to increase the efficiency but at the end of the day Newton always wins.
The horn acts as an acoustic impedance transformer, it lets you take a driver with a much smaller cone (=less mass = faster transient response) and let it produce a pressure wave at the exit of the horn thats equivalent to a much larger cone. It comes with its own set of compromises, relatively narrow frequency bands where it works effectively (pesky physics again) and for low frequencies needs a lot of real estate.

I have no idea if its better or worse than any of the other ways of producing low frequencies (although Mark assures me it is when you get it right). My interest is more about exploring it for the hell of it, I have access to 3D cad and analysis software that could potentially help get Mark further along his journey and its a fun area to play in.

[steve s]

I've heard a few big bass horns. To me they are coloured to some degree and only as good as the driver..
Horn drivers should have large powerful motors. I always see the horn as a load on the driver. They increase efficiency at a cost.
I think it's easy to confuse the horn sound and the sound of high efficiency drivers.

[IslandPink]

In response to point 1, I think it's worth noting that direct radiators also have a flat phase response across their linear range. You seem to be avoiding the obvious solution, then - what's wrong with putting drivers in a box?

Not so fast, young fella .
( I really must find Tom Danley's Joelist posting, and extract some of the gold from it to post on here.
Now, this may have been Tom quoting knowledge that came from people like Harry Olson) :
A cone driver when it's operating linearly ( amplitude ) is balancing the cone mass against increasing radiation resistance as the dispersion angle reduces.
One key feature of this mode of operation is that the cone's motion lags by 90 degrees ( phase ) from the voltage. So you might think that the phase is flat over that bandwidth, too ?
No, because the time for one cycle ( wave ) varies by a factor of 3 or 4 across the flat band of a cone driver. Hence the time lag for one quarter of a cycle ( 90°) is also NOT a constant value. So, there's a slow phase shift from high to low frequency, the lower frequencies lagging behind slightly in time. This implies the transient response of a good driver on a sealed box, within its flat band, is not as good as a horn.

Lots of good comment here, will take me a while to catch up.

[IslandPink]

I've heard a few big bass horns. To me they are coloured to some degree and only as good as the driver..

Are you sure they were bass horns, Steve ?
They tend to be about half the size of a room... or the size of a room.

[steve s]

They where mid bass horns mark, guy Sargents.. most people where impressed
I heard them at least twice.
Years ago there where quite a few horn systems at Scalford, used to make an interesting show.

[SimonC]

1. What is that the horn that is doing that makes you want to use it? Does the problem(s) it solves be solved in any other way?
2. I assume the size of the horn is controlled by the wavelength of the sound?
3.Could you make something that works like a horn but is made of material that means that sound travels slower in it? Would that let you make a smaller horn?
4. Would you still have the problem of interfacing it with the air at the end?

Mark has already answered number 2,
3. This is an interesting one... (I'm assuming that this is to make it fit in Mark's house, otherwise the easiest answer is for Mark to buy a bigger house. Not necessarily the cheapest answer, but compared to what is about to follow it may be.) To reduce the size of the horn without changing the frequency it works over you need to make the wavelengths shorter and this means the reducing the speed of sound in the medium they travel through.
Assuming we are working with gasses then the choices are fairly limited. Either change the gas, or reduce the temperature, or a combination of the two. For changing the gas probably the most practical one is to fill the horn with carbon dioxide. This drops the speed of sound down from 330m/s to 259m/s with the corresponding decrease in wavelength. You'd have to fit a membrane of some kind to stop the CO2 falling out and suffocating the spiders living under the sofa, or tip the speaker on its back and listen on the ceiling. Another contender is propane (258 m/s) but this could get more exciting than Mark's house insurance is prepared to tolerate. Slightly more aggressive (in almost every sense) is sulphur dioxide (201m/s), but the obvious winner is filling the horns with Xenon at 178m/s.
The other option is to cool the gas in the horn. Cooling the air down to -100degC drops it to ~260m/s, and the same effect happens with the other gasses. Surrounding the horn with a jacket of liquid nitrogen would drop the internal temperature considerably, and would definitely be a talking point at most hifi shows.
Switching to filling the horn with something thats not a gas brings a whole different set of challenges. Pure liquids all seem to have a speed of sound far in excess of air. There is a bit of research that water/steam admixtures are significantly slower than air, but the conditions required to create these are not very practical in Mark's front room. Solids are in excess of air, until you get to squishy stuff like rubber (60m/s). It looks like solids with a slow speed of sound also act as very good attenuators/absorbers and just turns the horn into an effective way of generating silence.
OK, enough pondering on that for now...