Increasing loading of midbass in synergy horn

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am designing a PA mid-top with efficiency as an important criteria.

The synergy horn is a nice approach and I'm inclined to try it. However, in a two-way synergy, the use of wide and short (to attain the directivity goal) conical horns implies rather poor loading at bass frequencies (say, 100-350 Hz). At first, it seems there is no way out of this, short of narrowing the coverage angle. Even going three-way, the horn loading (for reasonably sized cabinets) drops for the woofers and efficiency seems to only be recovered with the use of a large number of rather expensive drivers. As a different approach, I'm considering the use of a radial horn flare between the drivers and the port in the main conical horn. A few questions:

Phase:
The basic synergy design uses first order crossovers so that the phase shift induced is countered by the time shift between the drivers. Adding a length of horn between the driver and the port would break this balancing act, but I have seen designs that seem to use higher order crossovers (without dsp delay). Is the phase coherence lost in such cases?

Are most of the benefits of a synergy horn still available (without dsp) if, say, the signal at the ports is in phase with the 1/4 period delayed tweeter output at the crossover frequency, but with the mid output delayed by a fixed path delay that amounts to an integer multiple of 360 degree of phase at the crossover?

Reflections:
The junction between the mid ports and the main horn seems to constitute an impedance mismatch. This is not what the patent prescribes, but most actual designs do not seem to have the impedance matched at the port (and it does not even appear to be a design criteria). I can see how this might work using bandpass enclosures between the drivers and the horn as these acts as lumped components and, properly designed, will not allow resonant standing waves (but will demonstrate less than optimal power transfer / present reactive impedance to the driver). However, if a horn is substituted to the bandpass enclosure, then a significant impedance mismatch will lead to reflections at the port and problematic resonant peaks in the frequency response. Am I off the track with this consideration?

If the some of the benefits of the synergy horn could be retained, then the design may be worthwhile. I have simulated some rough ideas and I can drastically improve the bass frequency loading. Something similar might have been done in one of Danley's designs. However, I admit not being sure about what are the design considerations in this case.

Regards,
Geoffroy

P.S. : The simulations are done with code I did myself, but it is harder for me to simulate integration with the tweeter in a synergy as I first need to think of how to approach the problem. Yes, I know hornresp exists, but I don't have a windows machine and wine doesn't want to play with me. Those simulations where verified with hornresp results on a borrowed laptop. On that note, if anybody knows the details of the multiple entry horn and offset horn model in hornresp, I'm all ears! You don't have to sugarcoat it, I play with maths for a living.
 
Last edited:
A few questions:

Phase:
The basic synergy design uses first order crossovers so that the phase shift induced is countered by the time shift between the drivers. Adding a length of horn between the driver and the port would break this balancing act, but I have seen designs that seem to use higher order crossovers (without dsp delay). Is the phase coherence lost in such cases?

Are most of the benefits of a synergy horn still available (without dsp) if, say, the signal at the ports is in phase with the 1/4 period delayed tweeter output at the crossover frequency, but with the mid output delayed by a fixed path delay that amounts to an integer multiple of 360 degree of phase at the crossover?

Reflections:
The junction between the mid ports and the main horn seems to constitute an impedance mismatch. This is not what the patent prescribes, but most actual designs do not seem to have the impedance matched at the port (and it does not even appear to be a design criteria). I can see how this might work using bandpass enclosures between the drivers and the horn as these acts as lumped components and, properly designed, will not allow resonant standing waves (but will demonstrate less than optimal power transfer / present reactive impedance to the driver). However, if a horn is substituted to the bandpass enclosure, then a significant impedance mismatch will lead to reflections at the port and problematic resonant peaks in the frequency response. Am I off the track with this consideration?

I'm no synergy expert by any means, but I'll offer a few thoughts...


If folks are using higher order crossovers (like I'm currently doing with dsp before beginning experiments with 1st order),
phase coherence can be maintained by varying delays, but load sharing across drivers gets sacrificed.

re: having "the mid output delayed by a fixed path delay that amounts to an integer multiple of 360 degree of phase at the crossover?"
That would put the drivers in phase at crossover frequency, but out of time alignment. IOW, only the exact xover freq would be in phase.
The HF driver would need to be time delayed that amount too, to get back to full phase alignment.


re: reflections.
I share your questions about impedance mismatch between mid ports and horn. I saw in one of the patents it says the port area should equal the horn mouth area, but afaict ports are invariably smaller.
Hopefully, others will chime in here...

Have you seen the new Danley J7-95? Looks like horn loaded low drivers firing into horn ports....

One final thought/question... can any synergy design actually provide horn gain down to the lower mid bass (say 100-150Hz) without being physically huge? As in 11ft circumference and requisite length?

other than maybe horn loading before going into main horn ports...like you're contemplating I think ?? :)
 
If folks are using higher order crossovers (like I'm currently doing with dsp before beginning experiments with 1st order),
phase coherence can be maintained by varying delays, but load sharing across drivers gets sacrificed.

Hmm, but then a dsp would be required. I thought the some of the Danley's, say, where using higher order crossovers without requiring a dsp. Moreover, thinking about it, the acoustical lowpass on the mid ports should add another 180 degree of phase which would also be problematic for preserving the time alignment, but I guess this could be taken care of by switching the polarity of one driver.

re: having "the mid output delayed by a fixed path delay that amounts to an integer multiple of 360 degree of phase at the crossover?"
That would put the drivers in phase at crossover frequency, but out of time alignment. IOW, only the exact xover freq would be in phase.
The HF driver would need to be time delayed that amount too, to get back to full phase alignment.

Indeed, but my hope was to get away with only having smooth phase throughout the crossover and not perfect time alignment. One should still retain the benefit of point source behaviour.

re: reflections.
I share your questions about impedance mismatch between mid ports and horn. I saw in one of the patents it says the port area should equal the horn mouth area, but afaict ports are invariably smaller.
Hopefully, others will chime in here...

Then it might be time for some experiments! Maybe a setup as simple as replacing the port chambers with a long plane wave tube would be enough to highlight impedance mismatch as peaks in the FR that would shift with the tube length. What do you think?

Have you seen the new Danley J7-95? Looks like horn loaded low drivers firing into horn ports....

Indeed! It is hard to say, but it looks similar to the J6-42 I mentioned.

Regarding your final question, this is precisely what bothers me. I want to be able to run off battery and remain portable. I might have to look at a design more like msibilia's mid top if the "pre-horn loading" approach does not work out.
 
Hmm, but then a dsp would be required. I thought the some of the Danley's, say, where using higher order crossovers without requiring a dsp. Moreover, thinking about it, the acoustical lowpass on the mid ports should add another 180 degree of phase which would also be problematic for preserving the time alignment, but I guess this could be taken care of by switching the polarity of one driver.



Indeed, but my hope was to get away with only having smooth phase throughout the crossover and not perfect time alignment. One should still retain the benefit of point source behaviour.



Then it might be time for some experiments! Maybe a setup as simple as replacing the port chambers with a long plane wave tube would be enough to highlight impedance mismatch as peaks in the FR that would shift with the tube length. What do you think?



Indeed! It is hard to say, but it looks similar to the J6-42 I mentioned.

Regarding your final question, this is precisely what bothers me. I want to be able to run off battery and remain portable. I might have to look at a design more like msibilia's mid top if the "pre-horn loading" approach does not work out.

I don't know of any Danleys using anything above 4th order crossovers...maybe there are?
I have the Danley SC-48 DSP processor that has presets for all their speakers, but haven't looked at every speaker...... if you have one in particular you want me to check, happy to do so.
The presets are mainly EQ's that ride on top of the passives in the speakers, or add linear phase 4th order crossovers if the speaker allows/is used, in biamp mode.

Having smooth phase alignment through the entire crossover region requires time alignment.
I think you'll loose a lot of point source behavior without full alignment.
Check out page 117 https://www.rationalacoustics.com/download/Smaart-v8-User-Guide.pdf

BTW, thx for your comments in the 'speaker cone mechanical energy' thread.
You give very technical but clearly stated explanations, which I still often don't understand :):eek:
 
Ugh I really don't need another project, but I DO think that what you propose is valid.

Here is how I would do it:

DeWRJBM.jpg


We all know what Unity horns look like, here is my current project.

If you are prepared to use DSP in your Unity horn project, I think it is VERY possible to use *extremely* long midrange taps. Possibly as long as a foot, maybe even two feet!

The key to making this work is that you want the midrange taps to expand continuously.

ybgIKcJ.jpg


Easiest way to do that is with very narrow midrange taps that expand from the throat up to the mouth.

Ok, now here's The Secret Sauce:

To make this work with a very long midrange tap, you'll want to mass load the tap.

jx92_dwg_view.GIF


This is a mass loaded transmission line.

Now, you're probably wondering, why mass loading?

Normally, we use mass loading to SHORTEN a horn or transmisssion line.

In this scenario, we'd use mass loading because it allows you to make the exits of the midrange tap much smaller.

For instance, you might have midrange taps that measure 12" x 1/2", but at the very last moment, at the exit of the Unity horn, the taps would 'neck down' to 6" x 1/4".

The idea is that we get the effect of a horn mouth that measures 12" x 1/2" but with a much smaller mouth, just 6" x 1/4".

You'd need Akabak to sim this by the way.
 
  • Like
Reactions: 1 user
To make this work with a very long midrange tap, you'll want to mass load the tap.

Interesting idea. I would have thought that such a mass loading would remove a lot of the horn gain. On first thought, a port is a drastic change of impedance: I can see how it helps load the driver, but I would expect a lot of reflected power at the port. However, it seems bass reflex ports manage to get good efficiency from the port, although for the frequencies of concern in this case the box/port can be taken to be lumped elements.

I don't quite see why those reflections from the port are not a problem when the box/MLTL/ML-horn are too long to be considered lumped. Yes, damping can be used to remove some of the standing waves, but that would be at the cost of efficiency.

I'll try to run a few simulations to shed light on that.

You'd need Akabak to sim this by the way.

Well, I code my own simulations in python from scratch... infinite versatility! I think I can do this case. It really is the multiple entry horn with more then one drivers that causes me problem as much of the code needs to be rewritten. I have to use multiports devices or figure out a way around that.
 
Interesting idea. I would have thought that such a mass loading would remove a lot of the horn gain.

Nope.

690122d1530682794-horn-bandpass-horn-synergy-screen-shot-2018-07-04-07-37-24-png


Because of Hoffman's Iron Law, the efficiency of a system is determined by it's volume and bandwidth.

The entire system basically behaves like a conventional front loaded horn, but with a segment in the middle that creates turbulence.

One way to visualize it might be like visualizing a freeway with six lanes that 'necks down' to two lanes, the expands back to six lanes.

That turbulence might seem like a big problem, but keep in mind that we really only care about the velocity at the mouth, not the velocity in the middle of the horn.
 
What I describe could be used to horn load the midranges, the woofers, or both.

The "trick" is that you're putting the midranges or woofers into a horn, and then using DSP to line up the wavefronts as they enter the Unity horn.

Similar to what was described in the first post:

690122d1530682794-horn-bandpass-horn-synergy-screen-shot-2018-07-04-07-37-24-png

Thx.
And sure, makes sense. That's what we are seeing in Jerichos like the J7, isn't it?
 
Having smooth phase alignment through the entire crossover region requires time alignment.
I think you'll loose a lot of point source behavior without full alignment.
Check out page 117 https://www.rationalacoustics.com/download/Smaart-v8-User-Guide.pdf

I understand the concern. One would ideally have time and phase alignement throughout the entire crossover. However, what lead me to consider having the waveform at the midrange taps in phase but with a path delay equal to 360 degree of phase at the crossover frequency is that a similar concern should arise in the synergy horn as per the patent.

Indeed, in the typical synergy horn, one has the midrange taps at 1/4 the wavelength at the crossover frequency from the apex of the horn where the tweeter is located. This corresponds to a phase shift of 90 degree at the crossover. When used with a first order crossover, that path based 90 degree shift brings the two drivers in phase at the crossover. We know that in such a first order crossover, the signals are 90 degree apart at all frequencies, but the shift caused by the path difference will only equal that difference at the crossover point. So, one could have the two signal in phase and in time at the crossover, but not exactly so below and, specifically, above. This would be similar to the first image in figure 102, p. 117 of the document you attached. However, the end result seems to work out fine.

Now, imagine that there is a first order crossover on the tweeter, but a third order crossover on the midrange. At crossover, there is now a phase shift of 180 degrees, of which, 90 degrees is corrected for by the path delay between the apex and the taps. That leaves 90 degrees of path delay to be used for "pre horn-loading". Admittedly, the situation is worse than previously described as the signals will not be in time at the crossover, but I'm curious as to what part of the synergy behavior is retained.

A better option, regarding time alignment, would be to all-pass the tweeter so that the two signals are back in time at the crossover. This would be very close to the situation described above. However, passive all-pass filters seem like a bad choice. Of course a DSP resolves all the above with delays.

Thank you for that link. That user guide is useful. Also, thank you for checking the settings files.
 
If you are prepared to use DSP in your Unity horn project, I think it is VERY possible to use *extremely* long midrange taps. Possibly as long as a foot, maybe even two feet!

Yes, a DSP would solve part of my conundrum. However, I'd prefer not to HAVE to have one. It's not that I don't like DSPs, and I will definitely have one to be used with the system. However, I'd like to retain the option of operating the system completely passively and limit the number of amplifiers in some situations. If I can figure out a way of operating such a system acceptably without DSP, albeit with some features lacking, then that would fit!

To make this work with a very long midrange tap, you'll want to mass load the tap.
[...]
In this scenario, we'd use mass loading because it allows you to make the exits of the midrange tap much smaller.

So, in this case, "mass-loading" really refers to inserting a constriction in the middle of the horn, with the constriction such that it do not affect the horn response too much. I see how that can be possible, provided the constriction is much less than a wavelength of the highest frequency desired in length.

One concern I would have would be reflections of that constrictions leading to standing waves in the first horn segment with accompanying peaky response. Stuffing might be sufficient to tame this, but the associated loss of efficiency worries me. From a video of your Metlako project, I saw you used stuffing in the mid taps expansion. Was it precisely for this reason? Did it worked enough without loosing too much efficiency?
 
axMd9CW.png


In this project, the impact of the stuffing was fairly big, like 3dB IIRC

ybgIKcJ.png


But in Metlako it almost unmeasurable. About 1dB, and only at a small set of frequencies.

I think what's going on, is that the midrange taps are so far from the tweeter, the tweeter doesn't "see" them.

For instance, the midrange taps are spaced about 10cm apart. So the tweeter doesn't "see" the taps until 3.4khz. (3.4khz is 10cm long.)

The taps are 1.25cm across, which is a small fraction of 3.4khz.

So, that's my theory on why the polyfill has a noticeable impact on the older waveguide, but not much difference on the new one.

I'll still leave it there, because it lowers distortion, but it's not required.
 
In this project, the impact of the stuffing was fairly big, like 3dB IIRC

But in Metlako it almost unmeasurable. About 1dB, and only at a small set of frequencies.

Ah, you're talking about the standing waves in the main horn. Indeed, that should not be a problem with the Metlako. My concern was with standing waves inside the flared ports themselves (especially if they are a foot long!), much the same as the peaky response from a truncated horn as it would be the same phenomenon (reflections of the mouth because of impedance discontinuity).

Indeed, stuffing in the older project might have helped more as the cavity formed by the ports would be better coupled to the tweeter output. I would think not because of the distance from the tweeter, but rather because of their orientation with respect to the horn axis.
 
To be honest, I don't think standing waves will be a problem. As long as the horn is expanding, the reflection from the transition from the midrange tap to the horn bell should be fairly benign.

In horns that load the front AND the back of the driver you get out-of-band peaks and dips. But that's not caused by standing waves, that's caused by geometry. Basically there are frequencies where the front and back are out-of-phase.

That's why tapped horns and back loaded horns have limited bandwidth.

Of course, the easiest way to find out would be to make an Akabak model. If I wasn't working so much this week I would :(
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.