Very interesting, but, a question that lot of people want to ask : what about the group delay vs cut off frequency/size of the horn ?
Being a linear system, the group delay is going to be almost invariant across all horns/waveguides. (For all the same reasons that I gave above.) It will be the distance from the diaphragm to the listener regardless of the horn (some small variations with horn shape at cutoff, but nothing significant.) If you mean "excess group delay" - ignoring the propagation delay - then you have to specify what the origin is. If it's the woofer, which is logical, then it's basically the displacement of the CD from the woofer along the line to the listener. This ends up being a huge advantage since one usually wants to delay the HFs relative to the LFs because of the LF LP filter delay. Hence if the horns mouth is in the same plane as the woofer then the "excess group delay" will be near minimum (some forward and backwards relative movement might be required to get the absolute minimum.) It turns out that in my system the LF and HF have almost identical path lengths, within a few degrees (as verified by measurements.)
See for example post 913 about group delay and cut off frequency: https://www.diyaudio.com/forums/multi-way/140190-jean-michel-lecleach-horns-92.html#post2430347
I find this interesting! I can't imagine this being a controversial topic...because this has been proven with testing...right?
This would be a good time to enter some research showing the group delay of the same driver on many various horns....
Earl you are going to have to forgive me, I'm doing my best to play devils advocate to you, and at the same time, the most informative documentation I've read on polar response is from a paper written by you.......lol!
Last edited:
Hi,
The graph plot from Jmlc 1000 is to me ,just another way to display acoustic cut off. The group delay displayed is just another aspect of this high pass filter. It could be db or phase, too, this are just the representation of same thing seen with different angle, all are related, change one others will follow.
The advice given by Jlmc corelate with what i've seen as rule of thumb: minimum usable freq is around 1,5x fc of horn.
Feels free to correct me if i'm wrong.
Wow. Ok it wasn't obvious to me at all but your message kept flying in my thoughts as this could be a multicell.
Your analysis does make sense to me.
That said now i've read the previous message from Earl i don't know if 'loading' is correct term (but i must confess i never really understood it the way it was explained to me). Maybe better 'control' over the high range is more apropriate?
Now i wonder if the thing that bothered me with the horn is not a combo of break up and diffraction?
I wonder who introduced the fins' in the evolution of the design and the motivations which might seems obvious but who knows... Danley was studying other things when he thought his meh designs and i wouldn't be surprised if Earl began maturing his design from observation of other things too.
The way science works usually.
Why the minimum? Group delay? As you said yourself, it's all in the HP filter, so if I EQ the responses to be the same between a .75x fc and a 1.5x fc, then the group delay will be the same. As I said, its a linear system so almost everything is possible. It's the impositions imposed by nonlinear limiting constraints that pose the problem (like excursion, etc.). As long as that doesn't happen, EQ away.
I wasn't accurate enough.
About the TH-4001/compression driver i've heard/seen the lowest xover freq was 480/440hz as lower made sound quality degrade.
From my ( fairly basic) understanding of horn/drivers combo the limiting factor was the horn acoustic behavior ( cd could have withstand to go lower with moderate spl ).
I've since then seen this 1,5 fc 'rule of thumb' used on other kind of horns too as minimum recommended fc for best quality. I just found the advice from Jlmc to be inline with this.
I agree with eq you could compensate the system overall response and actualy use it... often.
Last edited:
To what profile do you exactly refer? Many things are possible but not all are interesting for meWhy copy a thing that already exists?
Maybe because it is the best way to do it!?
The specific concept I've been contemplating isn't exactly similar to these 18Sound horns. However, it's inspired by an existing "solution"... and presumably requires a separate thread.
To me this would likely be excursion related. It would be tough to accurately determine what the problem was.
To me horns and waveguides are linear systems and behave that way until very very high SPL. If something goes haywire, I'd look to the driver, first, if the problem is nonlinear (i.e. always sounds bad.) If its linear then the horn could have internal resonances or diffraction that you are hearing, but these too are linear (diffraction perception of course is not linear.) But, as I said above, the resonances do sound bad.
To repeat for the umpteenth time, I do not see anything special about the so-called "cutoff frequency." It has no relevance to me.
Last edited:
Ok i think i get your point.
Let me give an example to see if i got what you are saying right: if i could use a linkwitz transform on a cd/horn combo ( an 'ideal' cd without physical limitation) i could equalise it to whatever low freq extension i could want ( let's say two octave lower than 'native' acoustic freq response of system) and there would be no artefacts from that?
I agree resonance in horn sound bad to me.
"...Horn theory is still a work in progress. In the mid 1970s Benade described the wave motion in axi-symmetric horns of conventional shapes in
terms of spherical waves possessing both radial and axial modes. A large body of experimental measurements illustrating the usefulness of this approach accompanied his theoretical work.
Recently Geddes has proposed new horn shapes based on oblate spherical coordinates..."
terms of spherical waves possessing both radial and axial modes. A large body of experimental measurements illustrating the usefulness of this approach accompanied his theoretical work.
Recently Geddes has proposed new horn shapes based on oblate spherical coordinates..."
Hello Camplo
Similar but not sure where you are going as I was comparing CD vs Non CD horns and systems not different driver types.
Well in this case it is due to CD vs Non CD as frequency response and bandwidth are essentially identical. Main differences are system polar responses.
I knew exactly what they were before I built both speaker systems. They are essentially technological and design snap shots in time spanning 40 years . One from the early 70’s using traditional horn/ring radiator with non CD system design characteristics and the other from 2009 or so designed as a CD system using CD horns.
Rob
Thanks for your replies. I ask about the ribbon tweeters because I've sat before a few and am familiar enough to the beaming they present, and wondered if the polar is similar. Guess I could just look up some measurements of ribbon tweeters. The one thing I can think to ask is if this unbalanced feature to the narrow polar was from an overall average of listening positions or within the 0 axis of the system.
This is a large horn.
It loads quite low and has a nice bias for the horizontal plane. Minor horizontal polar roughness is from the roundover mesh resolution. The vertical would need a larger roundover it more off-axis smoothness were needed.
Add - just noticed the horn name was wrong is the original graph title - fixed now.
Looks good!!!, This horn is big, over 35" wide, and that is a lot but this is the largest of the 3 horns we picked to sim.
The PACMAN horn was brought up as a good horn, I think the Pacman and the last sim'd horn (drba_jmlc_150_hvdiff) are very similar. If I were to chose either I'd scale it down some and deal with the outcome of the loading.
Ty Don and Doc! Ty so much for your time, and ty for lending your expertise!
Lower cone excursion means less intermodulation distortion with wide bandwidth signals like music. This is easily verifiable with software, an inexpensive mic and sound card. Mix 35 or 40 hz with 500 hz, and turn up the amplitude to see the results.
Per Klippel, driven to 82% of rest Bl results in 10% distortion.
Per Klippel, driven at Fs to 75% of rest Cms results in 10% distortion.
Source: https://www.klippel.de/fileadmin/kli...inearities.pdf
On the other hand, a poor suspension system can suffer from stiction.
Speaking of poor suspension systems, here's a link to Klippel test of a Peerless unit that isn't even close to it's ad copy.
https://stereointegrity.com/wp-cont...W-350F-188PR01-04-large-signal-parameters.pdf
Going by Klippel limits, you'd have to keep this driver at less than 6 mm excursion.
Fs shift vs drive level is quite large as well, from 16 hz to 26 hz Fs at 18 mm.
Without a doubt. The question is: does increased im distortion resulting from using a cd lower than conventionally thought wise actually result in reduced subjective sound quality? I believe camplo indicated that he is concerned about im distortion, but I am delighted to hear Earl is in my camp. When I experimented with crossover points on my horn system, I found that using the cd down to 1.3fs, slightly below where the horn has lost pattern control and providing eq, resulted in increased sound quality with all 4 woofers I mated it with. People commented that the crossover point disqualifies my system as "high spl", but for use in the home, the spl capability is extreme.