Single full range drivers

[Matt_Holland]

Sure. I owe a better explanation re my preferences..

The single driver speakers are great imo, in comparison to conventional full-range cones.
BMR's ability to go low, and in my opinion sound about the same up high, as small cones...makes then a great small single-driver speaker choice.

When I've heard arrays of them, they sound dull in comparison to more point-source like speakers.
But in fairness, any type multiple-driver HF/VHF array sounds comparatively dull I think. (I've built large line arrays and CBT's with small cone drivers.)

My overall feeling about single full-range drivers, is that they are excellent when used singly and only used for applications they fit...
when low SPL, breakup mode highs, and usually severely truncated bass extension....all are OK.
It's really cool though, how BMR's help mitigate the bass extension part.

Yes a line array of any type of full range driver has a huge midrange boost due to the coupling of the drivers at wavelengths close to the line length. They need corrective EQ.

 

[gnarly]

You are right that phase is a component of frequency, but it is also a component of time. Since phase shift is increasing with time, it is impossible to only have a phase rotation problem (as you state) and not a time problem as well.

That's a good example of why I think phase and time are not substitutes for each other.
There are two sources of apparent phase rotation whenever we make a measurement.

One is constant time. And yes, it will show phase shift increasing with increased time.
But it is irrelevant when thinking about phase, imso.
Because any constant time needs to be removed from phase measurements.
Constant time's effect on Phase vs Frequency is simply a straight line when frequency is on a linear scale.
The straight line's slope equates to the fixed time delay in the measurement.
Get rid of the slope by getting rid of the fixed time delay, and the effect of constant time goes flat level.
Then all that's left is true phase rotations due to crossovers, or driver min phase roll-offs.
Time really has nothing to do with phase rotations

Phase rotations can only be corrected/countered with offsetting rotations.
Understanding that is the key to advanced xover design imho.

The idea of time used to offset phase rotations is a compromise kludge of varying effectiveness.
Most often it's pseudo time (degrees of phase rotation translated to constant time at a particular freq).
I think using pseudo time has become common historical practice, because the difficulty in working with IIR, or perhaps better said... passive-xovers.

fwiw, I know I would have never come to my conclusions if not for years of work with linear-phase xover design.
It's is so much easier to see how phase works, what phase means, in the absence of xover induced phase rotations.
There is no question in my mind anymore that phase belongs only to the frequency domain.....
.....picturing it as in the time domain too, is an immense source of confusion for about everyone it seems. Was for me too for a loooong time

(or should i say for a huge amount of phase rotation ...just kidding LOL)

 

[gnarly]

Yes a line array of any type of full range driver has a huge midrange boost due to the coupling of the drivers at wavelengths close to the line length. They need corrective EQ.

For sure.
The home-audio straight lines and CBTs i built both used 24 3.5" drivers (venerable TC9's).
The straight lines could EQ with global on all the drivers. But CBT's need magnitude shading of individual drivers as well.
Fun stuff. Beautiful mid-range.

I've heard lots of arrays / lines here and there..
I just don't think any type of line or array can compete with the HF/VHF of a good point-source like speaker. (other than maybe electrostats if you think of them as an array)

I've come to prefer constant directivity speakers ...with compression driver on horn/waveguide.

 

[Matt_Holland]

For sure.
The home-audio straight lines and CBTs i built both used 24 3.5" drivers (venerable TC9's).
The straight lines could EQ with global on all the drivers. But CBT's need magnitude shading of individual drivers as well.
Fun stuff. Beautiful mid-range.

I've heard lots of arrays / lines here and there..
I just don't think any type of line or array can compete with the HF/VHF of a good point-source like speaker. (other than maybe electrostats if you think of them as an array)

I've come to prefer constant directivity speakers ...with compression driver on horn/waveguide.

I sell line arrays for a living

K-array

Typically 150Hz - 18KHz (-6dB).

 

[gnarly]

I sell line arrays for a living

I thought that might be the case! K-array makes some great gear. Saw some at Infocomm a few years back.
I try to stay up on prosound speakers and processing....seems to be best place to find practical science and innovation. Never heard anything negative from live sound guys about K-array (other than price Lol)....and heard lot's of good talk.

Hey, you might appreciate some of the DIY line experiments I ran. Like said, twenty-four 3.5" TC9's in both straight and CBT cabinets. CBT arc has a 21ft radius.
I delayed and amplitude shaded the staight-line build, to try to match the CBT as best as possible. Worked close, but only half a cigar. I had to work with only 6 amp channels driving groups of 4 drivers per speaker....couldn't do a discrete 24 channel per speaker setup, or rather wouldn't do.
Also frequency shaded the straight-line to try to clean up the HF/VHF some, as well as moved apparent acoustic center up and down the line with just amplitude shading.
Fun stuff..lines ! Huh?

 

[norman bates]

shall i really warp you guys ?

Most people hate the very idea of these, covered a couch easily at 7'.

 

[ampguy]

I've got a couple of kanspeas on my desk. 3" Fostex full range, about 100Hz (extended with cabinet) to 25khz.

 

[norman bates]

A flat array has combing (cancellations in the hf due to mismatching distances).

Kuze3201 Line Array | Parts Express Project Gallery

projectgallery.parts-express.com

(from website) "The sound is very large.......comb filtering problems that definitely were occuring. However, this also could be a result of the high-frequency capabilities of this particular driver, it is hard to say for sure. Whatever the reason, for pure audiophile use I would consider these lines to be slightly less “refined” and “articulate” than many standard speakers."

Included is a step response of a 32 x 2" straight line array, along with its un-eq'd freq response.

My subjective opinion of flat arrays sound I've heard (non concert) is they sound "tall", similar to the closeup of a pair of lips on a movie screen.

Right / wrong, who cares.

measurement (unknown distance).

 

[nowonas]

That's a good example of why I think phase and time are not substitutes for each other.
There are two sources of apparent phase rotation whenever we make a measurement.

One is constant time. And yes, it will show phase shift increasing with increased time.
But it is irrelevant when thinking about phase, imso.
Because any constant time needs to be removed from phase measurements.
Constant time's effect on Phase vs Frequency is simply a straight line when frequency is on a linear scale.
The straight line's slope equates to the fixed time delay in the measurement.
Get rid of the slope by getting rid of the fixed time delay, and the effect of constant time goes flat level.
Then all that's left is true phase rotations due to crossovers, or driver min phase roll-offs.
Time really has nothing to do with phase rotations

Phase rotations can only be corrected/countered with offsetting rotations.
Understanding that is the key to advanced xover design imho.

The idea of time used to offset phase rotations is a compromise kludge of varying effectiveness.
Most often it's pseudo time (degrees of phase rotation translated to constant time at a particular freq).
I think using pseudo time has become common historical practice, because the difficulty in working with IIR, or perhaps better said... passive-xovers.

fwiw, I know I would have never come to my conclusions if not for years of work with linear-phase xover design.
It's is so much easier to see how phase works, what phase means, in the absence of xover induced phase rotations.
There is no question in my mind anymore that phase belongs only to the frequency domain.....
.....picturing it as in the time domain too, is an immense source of confusion for about everyone it seems. Was for me too for a loooong time

(or should i say for a huge amount of phase rotation ...just kidding LOL)

I have read this a couple of times trying to understand what you mean. I think I agree with a lot of what you say, but it seems to me that you are mixing minimum phase with linear phase without phrasing specific what you mean, and just refer to it as phase. In the first section I think you are referring to minimum phase, and I agree that minimum phase is affected by frequency. I.e by linearizing the frequency also reduces minimum phase problems.

However, it does not change the fact that phase is a component of frequency and time. My interpretation of your answer is that you can remove phase errors by utilising FIR filters ( i.e for linear x-overs like you mention). By using FIR it is possible to modify both the time and frequency component independently, thus making it possible to remedy the phase (timing) errors. But it does not change the fact that phase has a timing component, it only states that you have a tool to fix it.

I also agree with you that using linear x-overs using FIR is a better way to develop a x-over compared to a passive x-overs, and is in my opinion an under-communicated advantage many active speakers have.

 

[gnarly]

I have read this a couple of times trying to understand what you mean. I think I agree with a lot of what you say, but it seems to me that you are mixing minimum phase with linear phase without phrasing specific what you mean, and just refer to it as phase. In the first section I think you are referring to minimum phase, and I agree that minimum phase is affected by frequency. I.e by linearizing the frequency also reduces minimum phase problems.

Thanks for taking the time to think about and reply to, my post.
The easiest way to explain my position, is that constant time is that phase is not constant time dependent*
*when all constant time is removed, which it should be to glean any sense of what the phase curve is.

Trying to use time and phase as substitutes for each other in aligning multi-ways, is very suboptimal ime.

However, it does not change the fact that phase is a component of frequency and time. My interpretation of your answer is that you can remove phase errors by utilising FIR filters ( i.e for linear x-overs like you mention). By using FIR it is possible to modify both the time and frequency component independently, thus making it possible to remedy the phase (timing) errors. But it does not change the fact that phase has a timing component, it only states that you have a tool to fix it.

I'd say we describe phase at a given frequency as having a time component....but that's just because we have no other easy way of describing it.
Phase is a continuum....to describe it in terms of time (which requires at a given frequency) is akin to describing an entire highway by what's at one particular mile-maker.

I wasn't really talking about using FIR to correct phase. I was talking more about how FIR can give crossovers that have no phase to correct, to begin with.
It should be noted that for the same constant time delay, zero phase crossovers of any order and at any frequency, can be generated. To time is immaterial.

I also agree with you that using linear x-overs using FIR is a better way to develop a x-over compared to a passive x-overs, and is in my opinion an under-communicated advantage many active speakers have.

Yes sir! Maybe the most overlooked advantage.

 

[ParametricPi]

Interesting thread, guys. Not enough personal experience and measurements though, kind of surprising om ASR. ;-) Attached is a pair of in-room measurements of a backloaded pipe horn I designed for the Fane SOVEREIGN 15-300TC. It is/was a 15 inch fullrange driver with a double whizzer cone. (Hence TC for tri-cone in the product name.) It's the big brother of the 12 inch Fane already discussed here.

The enclosure was designed to shift the cancellation between driver and horn output relatively high, so it would be smeared out and cancelled by path length differences. The blue line was with the speaker in a corner and measured 1 meter distance on-axis. The green line is from the listening position about 4 meters away. No gating and no recollection of how much smoothing, but going from memory and looking at the graph, probably 1/12th smoothing. No EQ.

My design target was about 30 Hz. I added damping to roll-off its bass response to match room gain. My listening position is on a sofa up against the rear wall, so there is additional gain there. I suppose I could have stuffed that out as well, but I accepted it as free headroom and used DSP to get rid of it again. As you can see, I met the 20Hz goal of the single driver 20-20,000Hz challenge here. Easily. 20kHz is down 5 or 6 dB, which is not dramatic and fixable with DSP. So I met that as well. I can't find the 0-10-20-30 degree off-axis measurements I did from 1 meter distance, but the beaming starts pretty low and the off-axis response usually follows the peaks and dips of the on-axis response quite well. It's a bit like a horn or waveguide performance in that sense.

This is not an exotic driver, it cost me 140 euro a piece when I bought them. They are OEM now (I thought they were discontinued, but apparently there are limited vendors with stock and Fane still makes them/has a crate they are still selling). These drivers have no problems with power handling at all. I enjoy reggae music and for my living room, I can turn them up to nauseating levels and still hardly feel the cone moving. They have a very dynamic and clean sound. They don't sound like a fullrange driver at all in the sense that they have the frequency extremes all there and startling dynamics. I bought them after some experiments and reflection on room sound. I have harsh reflections between ceiling and floor, for instance. I usually build one or two pairs of loudspeakers a year. I have had these running for 5 years and did not feel the need to change at all. I have replaced with something else just because I wanted to make something new, not because I expected to improve on them.

Some highly speculative thoughts based on anecdotal experience:
- I am sensitive to changes in tonal character. I have less problems enjoying a loudspeaker with a consistent sound, than a system where I hear two drivers with a different signature working. Whether that is a bad crossover design at play, or drivers that have different distortions and colourations near the crossover, I don't know. All I know is I don't have that issue with fullrange drivers.
- I have heard horn-based systems that obviously benefited from controlled dispersion. I have had a similar experience for 280 euro for two drivers, instead of who knows how much for two 15" woofers, compression drivers, horns, more elaborate filtering etc. Pretty sweet.
- I have heard excellent systems based on open baffles, transmissionlines, backloaded horns, frontloaded horns, fully hornloaded, hybrid of horn and bass reflex woofers, lots and lots. The better the system design, the more they sound alike.

In the end I was looking into acoustic lenses to improve dispersion of the treble and upper midrange. To have more precise control. It was fine in the listening position, and the overall reflected sound seemed OK as well, but I grew tired of not getting the same sound when standing up and walking to the kitchen. Given some more time, I think that can be solved. Treble level will suffer, but that can be solved with DSP.

I think this has gotten me fairly close to the requirements of the challenge. It is not the most refined frequency response plot, I will admit.