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'm no expert, but the term standing waves refers to the room and are part of the room modes ( of vibrating).
To exist, those need to find the energy and the space and the frequency at which they occur. Otherwise is a matter of reflection and refraction with the peaks and dips etc.

No, standing waves are the superposition of a forward and backward travelling wave that does not propagate. It can happen in a room, as anywhere else, as the reflections lead to such forward and backward travelling waves. Sharp (compared to the wavelength) impedance changes cause diffraction and reflection that can lead to standing waves if not properly damped. Those standing waves are what causes the peaks and dips in the response as the transfer function of the system is perturbed close to a resonant mode (a standing wave).

Refraction is the bending/reflection of a wave passing from one medium to a different one. Often seen in audio as the sound passes through a mass of air with a temperature gradient or when one introduces stuffing in a pipe. In the extreme, the difference between refraction and diffraction is a bit blurry, but it is not the term I would have used here.

Regards,
Geoffroy
 
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 :(

So I ran a little simulation of the concept. The picture is the 2.00 V (1W on 4 ohm nominal load) power response in 4 pi at 1 meter on axis (same as default in hornresp) with the setup as described below. The results of my code matched up to hornresp in other situations, but as I understand, it can't handle this situation.

top2.png

The result is really encouraging, however, I remain a bit sceptical as efficiency seems way too high. I model the ports as lumped mass elements as their length is like 1/16 of the highest frequency of concern. Also, the mass loading is modelled as simply using the port input impedance as the mouth impedance. It should work in principle.

If those results can be trusted, then the initial idea seems very worthwhile! The standing waves do not seem to cause problem indeed! 4.3 cm is 1/4 wavelength at 2kHz and thus a delay of only 90 degree: the tweeter can even be "delayed" around the crossover with a simple passive all-pass when no dsp is used.

Here is a detailed description of the simulation. Two 6" drivers (Dayton's PA165 in parallel) are each fitted with a 2 liters back chamber and a front chamber of 70mL and a 1/2" long slot port of area 12 x 2.54 cm (4.7" x 1"), followed by a short 4.3 cm conical horn segment flaring to a final area of 24 x 8 cm (9.4" x 3.1"). In turn, these short horns are each mass loaded by 1/2" long slot port of area 12 x 2.54 cm tapping in the main astigmatic conical horn at area 0.011 m^2 that flares over 21.5 cm to a mouth area of 0.21 m^2.
 
Looking very promising. Please continue !!!
I'm slugging out trying to add bass-reflex ports to my synergy attempt..would be nice not to have to go that route


re crossovers

my take on the patent has been that the goal is acoustic first order.
That the 1/4 wave spacing, and acoustic port location design accommodates that goal
So, no additional first order electrical crossover needed, just whatever additional filter it takes to achieve acoustic 1st order.
Maybe I'be been misunderstanding...
 
first order xovers don't really work on a Unity horn

In order for them to work, the pathlengths would have to be equidistant

Check out "Crossovers, a Step Further" by LeCleach

Yep, all I was trying to say is that it appears something close to a net first order acoustical crossover between sections is the final design. Not electrical first order at all..

Looking at a SH-50, which is the speaker I use for a synergy reference, it looks like roughly 90 deg phase lag per crossover (assuming say 300Hz and 1000Hz cross points).

Thx for the link. I've read that excellent paper before.
I just wish it, or rather an updated version of it, had current studies of phase audibility.
Better experiments are easier to do today, i think.
And same thing with the crossover recommendations...too complicated...too dated imho.
 
It turns out something was wrong with my simulations. I was putting the "pre-horn" throat impedance in series with the main horn impedance... When I properly put the horn itself in series (that is, connect the bandpass port impedance to the throat impedance of the pre-horn and its mouth impedance to the main horn throat impedance), I still get the same increase in loading at lower frequencies. However, the mass loading causes major resonances that vary according to the length of this first horn. The more I increase the mass loading, the worst it gets. Without substantial mass-loading, I get mild peaks such as the following (pre-horn length is 1/2 wavelength at 2 kHz)

top2.png

If this length is sufficiently short, those can be made to occur outside the pass band, but HF extension is sacrificed. In the following the pre-horn is 1/4 wavelength at 2 kHz:

top3.png

It seems that the main conical horn throat does not resistively load the port sufficiently to avoid reflections. Simply using the bandpass chambers, everything seems alright, except I loose all the gain advantage:

top4.png

Moreover, the horn above cannot really be physically built... I can recover some driver loading by increasing the mass loading of the port at the cost of a somewhat peaky response. This is the direction I'm considering with some EQ to reign in the peaks. In the following, the port is twice as long:

top5.png

In conclusion, the suggested approach seems to work if one restricts the bandwidth of the driver. I am sceptical of its usability in a two-way synergy where the driver has to get in the 2 kHz range...

Another option would be to ditch the synergy concept, although an efficiency above 100dB/W @ 1m as the last simulations suggest seems enough. How much should I value constant directivity / the synergy approach versus efficiency in a portable battery powered PA context?
 
Last edited:
first order xovers don't really work on a Unity horn

In order for them to work, the pathlengths would have to be equidistant

Check out "Crossovers, a Step Further" by LeCleach

I am having great success using a 1st order crossovers on a unity horn. The mids are running full range with the polarity reversed.

The sound is very pleasant and an improvement compared to running without the mids. So to say they don’t work would be unjustified.

I searched for "Crossovers a Step Further" article but what I found looked like a power point with no real reference of 1st order crossovers.

The Klipsch type A 1st order crossover has a cult like following for a reason.

If you could please elaborate on your thoughts and reference your findings in the article.
 
Last edited:
Hi all,

So, fiddling more with simulations, the big problem seems to be the impedance/flaring discontinuity at the point where the pre-horn enters the main horn. At low frequency, this is irrelevant as this discontinuity happens over too short of a length to cause much of anything. However, at the higher end of the bandwidth, where the wavelength is much shorter, the propagation is seriously affected and causes reflections. This is really a big challenge! I see several ways forward, but they are starting to push the limits of simulations.

One way is to ditch the pre-horn loading, avoiding the discontinuity problem entirely, but sacrificing the gain.

The only other way I see is to somehow avoid the discontinuity, at least at the frequencies that are problematic. To do so, one could make use of an "acoustic crossover" and inject the high frequencies directly in the main horn while making the low frequency take a longer route through a pre-horn (or even an entirely different tap point!). This could be achieved using a combination of the following:

Acoustics_filter_imp_hpass1.jpg

and

1024px-Acoustics_filter_imp_lpass1.png

Another way to do it would be to elongate the taps along the main horn axis, as Patrick Bateman did, in order to obtain a geometry that matches the impedances. This is starting to go beyond what I can simulate reasonably. Thus I'm building a little prototype setup to measure those ideas. I am quite swamped by work now, so it may take a while for this to be completed, but I need those results for my final design!

Here is the test setup, it now has a simple tap:

20191012_221331.jpg

P.S.: I forgot to mention it: Even with the cheap piezo tweeter not properly placed, no phase alignement and a smidge of EQ, it already sounds surprisingly good and can be quite loud! Having the drivers close together really makes a difference! I just received my minidsp, but do not really have the time to play with it now...
 
Last edited:
After a lot more simulations, I arrived at the conclusion that it can be done if the bandwidth of the woofer is not extended too much. The approach I am using is to add a cavity with some stuffing between the "pre-horn" and the port to the main horn.

This cavity acts as a lossy acoustic compliance that is (acoustically) in parallel with the port and main horn inertance. This effectively amounts to a parallel LC circuit that can be tuned to increase the impedance at a problematic frequency to minimize reflections. Of course, one must damp it enough so that a high Q does not cause other problems. In practice, I can fill in the first valley and attenuate the first peak caused by the reflections in the response.

The other angle of attack is to use damping material in the pre-horn to selectively attenuate the standing waves. This can be done by placing foam at the displacement maxima of the standing waves, a bit like it is done in some quarter wave enclosures. In my case, it will be hard to compute as I'm using a kind of paraline expansion in the pre-horn, such that the propagation modes are hard to predict. There will have to be some trial and error.

I am building a prototype with these ideas and will report once done (it will be some time).

In the meantime, any ideas for a good damping material? I would need some foam that offers resistive damping.

Regards,
Geoffroy
 
Last edited:
I am having great success using a 1st order crossovers on a unity horn. The mids are running full range with the polarity reversed.

The sound is very pleasant and an improvement compared to running without the mids. So to say they don’t work would be unjustified.

I searched for "Crossovers a Step Further" article but what I found looked like a power point with no real reference of 1st order crossovers.

The Klipsch type A 1st order crossover has a cult like following for a reason.

If you could please elaborate on your thoughts and reference your findings in the article.

When you have a tweeter and a midrange and a first order crossover on both drivers, they will radiate in phase as long as they're equidistant.

In a Unity horn, the midranges are closer than the tweeter.

Due to this, you have to delay the midranges... somehow.

The LeCleach method, described in "Crossovers, a step further" works nicely because:

1) It gives us a half cycle of delay, which is about what we need

2) The phase response is nearly as good as a first order filter

3) the power handling is dramatically higher than a first order filter, like night and day. With some tweaking you can increase the power handling by around 4-8x as much.

For instance, the midrange taps in a Danley SH-50 are 3.5" from the throat of the horn. The xover point is somewhere around 1000hz iirc. 1000hz is 13.5" long.

A half cycle delay would delay the radiation of the mids by 6.75". IE, if your mids were 6.75" closer to you than the tweeter, you'd want a half cycle delay at the xover point.

In the SH-50, the reason that a half cycle delay isn't too much, is because the tweeter also has a delay.

This next part I may be wrong about, but IIRC, the delay on the tweeter is a combination of the delay that's cause by it's high pass xover, along with the inherent delay that's caused by the distance from the diaphragm of the tweeter to the exit of the compression driver.

Way back in the day, about 19 years ago, I recall that the Lambda Unity Horns were offered with two different xovers, because the TAD compression driver had a *significantly* longer distance from diaphragm to compression driver exit.

Anyways, to make a long story short, you want a high order crossover on the midrange in a Unity horn because a high order xover introduces a delay. The delay compensates for the distance between the tweeter and the mid.

Oh, and by the way, you accomplished the exact same thing by flipping the polarity ;)

But flipping the polarity isn't ideal. If you run a phase measurements on the speaker, you'll see that when you flip the polarity the phase isn't as good.

Actually, on 2nd thought, just to complicate this even more, it just occurred to me that the LeCleach topology flips the polarity, lol

Unity horn crossover and phase is far and away the most complex part of the design. I've listened to a Lambda Unity horn and a Danley Synergy Horn back-to-back, and IMHO the major reason they sound different is because of the advancements in the xover made by Danley. Everytime we've talked about the xover it makes my head hurt, I manage to absorb about 10% of what he said, and then I go back and review what he said, over and over until it makes sense.
 
Founder of XSA-Labs
Joined 2012
Paid Member
690122d1530682794-horn-bandpass-horn-synergy-screen-shot-2018-07-04-07-37-24-png


If I were to model the horn in Akabak, it wouldn’t be a horn at the end as large chamber is now huge with a small exit. The boundary condition is the exit aperture. For a horn to be a horn, the exit aperture needs to be expanding and larger than the inlet aperture. The sound waves won’t “know” that it’s a horn and what you are left with is similar to a mass loaded expanding cross section transmission line. The horn expansion is really a waste of wood and complexity for naught.
 
If I were to model the horn in Akabak, it wouldn’t be a horn at the end as large chamber is now huge with a small exit. The boundary condition is the exit aperture. For a horn to be a horn, the exit aperture needs to be expanding and larger than the inlet aperture. The sound waves won’t “know” that it’s a horn and what you are left with is similar to a mass loaded expanding cross section transmission line. The horn expansion is really a waste of wood and complexity for naught.
Well, not necessarily... You see, the expansion still allows for impedance transformation, provided the end is loaded appropriately. As I found, a compliance and an aperture, seems to be able to smoothen the transition for a restricted bandwidth.


At the end of the day, whether you model a horn or a TL, the equations remain the same. Only the boundary you set changes.


Regards,
Geoffroy
 
I've seen most of the unity / synergy builds on DIYaudio and all the measurements of the mids / tweeters that came with them.

There is no way on earth any of them could have a 1st order (acoustic ?) crossover ?

The mids come with somewhere between 4th and 6th order acoustic slopes by default, and the CD is usually on a 4th order slope naturally around 800 / 1200Hz before any xo.

Am I misunderstanding something here ? :confused:

Rob.
 
I've seen most of the unity / synergy builds on DIYaudio and all the measurements of the mids / tweeters that came with them.

There is no way on earth any of them could have a 1st order (acoustic ?) crossover ?

The mids come with somewhere between 4th and 6th order acoustic slopes by default, and the CD is usually on a 4th order slope naturally around 800 / 1200Hz before any xo.

Am I misunderstanding something here ? :confused:

Rob.
Well then, unless you use delays, you cannot obtain a linear phase speaker. You can still have somewhat of a point source, but it will be astigmatic on the axis of the horn with respect to frequencies.


6th acoustic order on the mids should work with 4th acoustic order on the highs with respect to summation of the amplitude. The phase will still do an entire 360 degree rotation at the crossover point.
 
The LeCleach method does make a great traditional xover...

I always enjoy reading that paper ......'Crossovers, a step further'

And i must admit, i particularly like how he concludes his circa 2004 paper on the last page .....'more easy' and 'more perfect' .... true for synergies too ;)
 

Attachments

  • LeCleach xover.JPG
    LeCleach xover.JPG
    148.2 KB · Views: 210
I've seen most of the unity / synergy builds on DIYaudio and all the measurements of the mids / tweeters that came with them.

There is no way on earth any of them could have a 1st order (acoustic ?) crossover ?

The mids come with somewhere between 4th and 6th order acoustic slopes by default, and the CD is usually on a 4th order slope naturally around 800 / 1200Hz before any xo.

Am I misunderstanding something here ? :confused:

Rob.

I don't think you're missing anything Rob.

I re-skimmed this thread and saw back last year, that i was thinking that an acoustic first order was the goal....but that was just from trying to dissect the patents...(which frankly i now think are a bit of a smokescreen).
And before i started trying to build any synergies on my own...

Anyway, here's the phase trace of the SH50 from the Danley site..
...note the phase scale on the right :D
 

Attachments

  • SH50 phase.JPG
    SH50 phase.JPG
    126.6 KB · Views: 208
Well then, unless you use delays, you cannot obtain a linear phase speaker. You can still have somewhat of a point source, but it will be astigmatic on the axis of the horn with respect to frequencies.


6th acoustic order on the mids should work with 4th acoustic order on the highs with respect to summation of the amplitude. The phase will still do an entire 360 degree rotation at the crossover point.

In your 1st post you said the basic synergy horn uses 1st order xo's. Can I ask where you got that information as I am having trouble understanding due to the bandpass nature of the mids and the natural roll off of the tweeters ?

(Unless you mean 1st order electrical ? )

Thanks,
Rob.
 
I don't think you're missing anything Rob.

I re-skimmed this thread and saw back last year, that i was thinking that an acoustic first order was the goal....but that was just from trying to dissect the patents...(which frankly i now think are a bit of a smokescreen).
And before i started trying to build any synergies on my own...

Anyway, here's the phase trace of the SH50 from the Danley site..
...note the phase scale on the right :D

Thanks Mark,

I find the xo the most complicated bit of the MEH style horns and there seems to be a lot of conflicting information out there. I remember something from Tom D about modelling a bunch of 4th order bandpass slopes and 'shoving' them together. Not much about 1st orders though I think the chap with the klipsch horn (cask ?) said they used 1st order slopes..

Cheers,
Rob.
 
Hi Rob, yeah, I don't know how i let myself go down that first-order rabbit hole....what a crock....all twas needed was a look at the Danley phase traces...(a bit sneaky with the scale imo ;))

I can say this though, using the exact same method for xovers I use on everything,..... steep as heck complementary linear phase,
i've had zero trouble getting excellent mag, phase, and impulse... and good off-axis too... on a bunch of synergy tries....

Cheers indeed !
mark
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.