A Bookshelf Multi-Way Point-Source Horn

[Patrick Bateman]

I don't get where you say it doesn't work. Look at the data.

Data summary here:
http://www.diyaudio.com/forums/mult...ti-way-point-source-horn-125.html#post4644501

This design achieves low distortion, flat phase, great bass extension, narrow directivity to reduce room interactions, excellent step response, and no direct radiators. This is not a thread where we just say it sounds great. We back it up with quantifiable measurments that confirm why it sounds great. If you think this concept doesn't work - show is measurements of a system with better performance for a similar box size and cost of materials. What do you suggest as better and not as you put it, a "mess". You say pushing air through "ridiculously small holes" doesn't work but it's exactly the nature of the bandpass port injection that gives reduced distortion by preventing the higher frequencies from even coming out. The constriction loads the cone so that less excursion occurs hence less distortion. This is all a balancing act and that is why we use models and simulations to first estimate what size holes and where to put them. Look at the modeling effort that happened in parallel at the early part of this thread. That's how the woofers were chosen and the hole diameters decided upon. Perhaps the biggest reason why the bandpass injection holes were used was to put the two drivers in a virtual point source with the mid/tweet horn. Hence no frequency dependent polar lobing effects.

Works nicely in car audio too. The system in my Mazda has fifteen drivers and only two of them are direct radiators! (The tweeters.)

 

[krivium]

If i where you and if you don't mind latency (and have enough power with your pc) you should have it done before the soundcards, just after the player (or inside it) with a dedicated solution.

Ok when i wrote that sentence i thoughs it could be possible to use a plug in (vst) to perform SRC realtime, but it don't work with the plug ins i wanted to use. I went wrong about it because my DAW allow for this kind of things in an 'invisible' kind of way (can load different format in a single session without having to think about it).

So SRC should be performed offline by dedicated software for best quality. Maybe realtime plug ins exist but for now i don't know them.

If realtime SRC is absolutely needed their is some harware units available. The one i know well are overpriced for home use (Mytek 192 SRC) but i've already used some Behringer SRC2496 which for the price asked did the job ok. Is it better than the one included in minidsp i can't tell, test should be done to determine.

 

[wesayso]

Ok when i wrote that sentence i thoughs it could be possible to use a plug in (vst) to perform SRC realtime, but it don't work with the plug ins i wanted to use. I went wrong about it because my DAW allow for this kind of things in an 'invisible' kind of way (can load different format in a single session without having to think about it).

So SRC should be performed offline by dedicated software for best quality. Maybe realtime plug ins exist but for now i don't know them.

If realtime SRC is absolutely needed their is some harware units available. The one i know well are overpriced for home use (Mytek 192 SRC) but i've already used some Behringer SRC2496 which for the price asked did the job ok. Is it better than the one included in minidsp i can't tell, test should be done to determine.

JRiver itself has real time SRC. For the source files as well as the FIR filters.
So bushmeister can at least compare what MiniDSP does to JRiver's method.

(example from JRiver's wiki, each sample rate can be set to up sample or down sample seperately)

And hopefully report back with his findings .

 

[krivium]

JRiver itself has real time SRC. For the source files as well as the FIR filters.
So bushmeister can at least compare what MiniDSP does to JRiver's method.

That's great if it's implemented realtime inside the software! No hassle with offline treatments and audioloopback inside the pc.

I see an other serie of test in the future to determine which one is better between chip in minidsp and Jriver own upsampler... or offline solution!

This give some clues about different solution :

SRC Comparisons

 

[BYRTT]

BYRTT, the amount of equalizing you use is an indicator that there is something wrong with concept. Bass-reflex needs port area of at least one tenth of diaphragm area in order to have low distortion. Altho bass-reflex ports have low bandwidth thus few intermodulation. Small Synergy ports transmit more than two octaves, so when air distortion starts, there occurs intermodulation, too. There occurs much Doppler distortion, too, because Doppler depends on speed only, of air gushing thru a port just as well as of a moving diaphragm.

There is at least one better concept for high directionality, one by Follgott, using Horbach-Keele FIR filters, a tweeter horn and two direct-radiating ways in a symmetrical arrangement.

Hi why the talk about bass reflex : )

The two SB23NRXS45-8 woofers bushmeister use per horn sit in a "4th order symmetric bandpass", and waves from backside of woofer stay inside enclosure in a sealed room and we only get the wavefront from frontside of cone after it passed through bandpass chamber. Natural acoustic output from"4th order symmetric bandpass" at stopbands seems 2nd order but with resonance peak as showed in post 1346. So had hard to see why you mention bass reflex and also why the more talk about amount of EQ, think it seems so logic that if low pass roll off from injection port is a 2nd order roll off with a ugly spike then it need EQ repair for spike and more EQ or electric slope massage too, so as final acoustic low pass for woofer ends up in BW4 acoustic slope that is target for a Harsch XO concept. Regarding speed in injection ports and their area can mention xrk971 had that covered in a calculation. Please be a bit more positive : ) especially because they document build with nice measurements.

 

[cookiemonster77]

Well, I finally got past the thinking stage and decided to try with the PH 4528 horns from P.audio. One big difference from the builds in this thread so far - I got a couple of FaitalPro HF 206 drivers used. These are 1.4" exit but comes with a 2" adapter. It´s probably not a perfect fit for the PH 4528, but I figured I would give it a go, and possibly do something about the adapter later to improve hf response. I did some measurements, and to me they look decent (didn´t save them - i can redo them when the house is empty and quiet). Anyway, to get a quick and dirty test using minidsp, I used some old speakers as midrange speakers, and an old 8" rel subwoofer to fill in the bass, and it sounds nice! No sign of horn shout or honk to my ears (but I don´t really trust them..). I don´t need 20khz response, as only the kids in the family will hear that anyway.

So at this stage, I´m looking for woofers that are available in Norway that will work from a 600hz crossover frequency. I think this is fine with that driver for home use right?

I was looking at 8" woofers, as this thread has focused on that. And I can certainly fit one on the top and one on the bottom of the horn. On the sides would be better if I wanted to make a small enclosure, but then the entry ducts would be quite long since due to the curvature on the sides.

One issue with the top/bottom mounted 8-inch drivers, is that they will protrude above the horn edge quite a lot - so the box in the end will be more square than rectangular. I kind of liked the rectangular look - hoping to place these on stands in the living room. So I was wondering if perhaps four smaller woofers would be just as good as two 8-inchers? Any thoughts on this?

An 8-incher I had in mind using was this one: P-Audio SN8-200F:
Re 5.0
FS 89hz
Mms 19.41g
Mmd 17.26g
Qms 5.35
Qes 0.5
Qts 0.46
VAS 14.55lt
BL 10.46 Tm
Cms 1.7e-04 m/N
Rms 2.03 Ns/m
Le (100hz) 0.26mH
Sd 241cm2

It seems to sim quite nicely in hornresp (just used a closed box and a bandpass chamber, no horn loading), but with a peak (100db at 100hz, peak at 115db at 641). I was hoping that peak could be offset by the horn reflection and crossover. This was with a throat entry of 20cm2, which should be two holes with 3.6cm diameter. EDIT: Forgot to mention: 10L rear chamber used.

What do you think about this woofer? I´m kind of limited in availability - shipping woofers from P-E doubles the price at least, so if I can get something locally its better.

 

[BYRTT]

cookiemonster77,

Nice new hardware and plans, admit myself haven't come to test my own 4528 yet with any full range driver mounted.

Have no experience how to operate Hornresp but if you want to move resonance front box peak up higher in frq then add some volume filler in front box although it will also add length to injection port so its a balancing act, think bushmeister and xrk971 used EQ to correct that peak.

You ask for smaller woofers and had myself some thoughts on using 4x Faital Pro 5FE120 as woofers because of their small print and only takes 3,88 liter volume per driver to get a sealed Q 0,7 f3 98Hz alignment. 4x will be 0,03192 sqm Sd area with a 5,25mm linear stroke. In 4th order symmetric band pass it seems possible each feeded 64 watts and the doubling of power because they inside 1/4 wave to get 112dB total output. In below black trace is the 112dB spl with f3 98Hz and grey plus blue is example if target XO is LR4 80 or 100Hz and sits at 109dB line. For home use less spl is needed but if they worth the power or are avaible in Norway don't know, but they relative cheap in UK Faital Pro :: Faital Pro 5FE120 5" Speaker Driver 80W 8 Ohm 21.78 IN STOCK (4 Apr 2016) and also available in Sweden Faital Pro 5FE120 - supersonic.

 

[USRFobiwan]

You can also check this one out, Dayton RS180*4 7" Reference Woofer 4 Ohm, placed in series, specs here: http://loudspeakerfreaks.com/datasheets/dayton/RS180-4.pdf

I don't know how it sims but with 6mm xmax and useable range of 39 <> 3.4k it is worth a try.

 

[xrk971]

I am using the paper cone RS180P-8 for this and it works well. Pretty good sensitivity - but not anywhere near the same Sd as the SB23's.

Cookiemonster,
If you can get me the horn profile (maybe 5 different axial measurement of major and minor axis vs axial position - I can run an Akabak sim for you with the driver of your choice. Most of the model is there already.

 

[zobsky]

I am using the paper cone RS180P-8 for this and it works well. Pretty good sensitivity - but not anywhere near the same Sd as the SB23's.

Cookiemonster,
If you can get me the horn profile (maybe 5 different axial measurement of major and minor axis vs axial position - I can run an Akabak sim for you with the driver of your choice. Most of the model is there already.

Shouldn't be too hard to extrapolate from a blown up printout of this (and a pencil, scale, and calculator to account for the scaling factor)
http://www.paudiothailand.com/pdf/products/PH-4528.pdf

 

[xrk971]

That's an exercise for Cookiemonster to do however he likes. The drawings are external and I am not certain how accurate they are relative to actual manufactured units. If measuring - it would be helpful to get details of the pinch point so maybe 7 data points axially in this case would be better.

 

[cookiemonster77]

I will have a go at measuring as soon as i can. The unknown is the acoustic distance to the CD. I had plans to use a horn to sacrifice to get the tap point right. Just try a few holes and see where the first reflection is. Then, just use a reasonable compression factor (the size where hornresp just starts to reduce output sensitivity) to get the hole size. And finally just reduce the cavity until the peak is manageable. Sounds reasonable?

Another thought i has was running a sweep with an object partially blocking the horn at different points. Hopefully that will show the acoustic distance as a reflection? Did anybody try this approach?

 

[natehansen66]

Couldn't you just stick the mic in the throat and do a time locked measurement?

 

[cookiemonster77]

Nate - there actually is an iphone app that sends an impulse and measures the time for it to return. I thought of doing that, except it's not very accurate the reviews said. But your idea is interesing - is it possible when using a dsp-based measuring setup with only digital inputs? If so, can you give some pointers? I have a usb-based umik-1, and optical out from my mac to a toslink to coax converter, then the nanodigi, then dacs etc.

 

[nc535]

re' acoustic distance to CD

I snooped this off the web a few years ago. If I recall correctly, JLH posted it:

The path length of the BMS 4550 is 6.7cm.
The path length of the BMS 4552 is 5.1cm. later data = 4.4cm?
The path length of the Selenium D220ti-omf-8 is 3.65cm
The path length of the RCF N350 is 3.95cm
The path length of the RCF ND350 is 2.25cm

I expect the B&C Speakers DE120 to have a path length around 2.2cm to 2.5cm. I'll measure it when they arrive. All you have to do is remove the diaphragm and slip a piece of paper through the phase plug slits. Measure the distance the paper goes from the top of the phase plug to the exit. Add about a 1mm for the space between the phase plug and diaphragm. This will give you a good enough number to model with. This number will give you the worst case in the model because the acoustical center of the driver actually occurs slightly forward of the diaphragm. To measure the actual acoustical path length you need to do an impulse test and calculate it based on the speed of sound.

 

[xrk971]

You can also do the acoustic interferometry method typically used to determine acoustic offset. Measure raw response with just tweeter then woofer then both in parallel. All with mic and speaker physically untouched. Then use a xo simulation program like PCD or Xsim and adjust physical offset until simulated combined response matches measured combined response peak for peak and dip for dip. You can get the offset down to fractional millimeters this way.

Here is an excellent and complete explanation with step by step guide by Bagby.

http://techtalk.parts-express.com/filedata/fetch?id=1149302

 

[kees52]

Do you now a gap of 1/2 inch between speaker and baffle throat works nicely for extending? For me it dampens also peaks and removes horn like sound, I dd get it from Edgar horn papers. I go experiment with it.

You need fill up the edge of the gap with foam, or scotchbrite (do not fall apart in time.)

measurement is straight on amp, no filtering at all, for dome mids/high this is a very nice way to extent.

regards

 

[xrk971]

Do you now a gap of 1/2 inch between speaker and baffle throat works nicely for extending? For me it dampens also peaks and removes horn like sound, I dd get it from Edgar horn papers. I go experiment with it.

You need fill up the edge of the gap with foam, or scotchbrite (do not fall apart in time.)

measurement is straight on amp, no filtering at all, for dome mids/high this is a very nice way to extent.

regards

I tried this trick and did not get as much success as I was able to achieve by making a very smooth 3d printed throat adapter. It may have something to do with the gap width and depth. Your response is actually very impressive for no filtering at all.

 

[kees52]

I tried this trick and did not get as much success as I was able to achieve by making a very smooth 3d printed throat adapter. It may have something to do with the gap width and depth. Your response is actually very impressive for no filtering at all.

Thanks.


The fact the responses of different speakers do cause that some do well here other not, sound is a very complicated thing coupled to speaker mass, motor strength etc.

I make a new horn because I want to try more options, just I am be curious last time about horns and the papers about it. I go make a adaptor who has the full Sd of the speaker, and no compression at all, cone breakup I try to avoid, with will work because I go als use a K tube or HF horn in horn with pointy back to prevent waves be disturb. For now learning is my believe to do first to make nice things, I do slowly understand more and more.. Th slit on the troat is not a open slit, there is some kind of leak with damping material, in my case polyester, even make the throat smaller and use damping I can maybe find a nice smooth spot be more precise.

PS Volvotreter makover his software so you can now easy play with akabak without so much typing using more sections, in this case 100 because of the throat compression part.

Maybe you can put a look at it.

regards

 

[perceval]

Yes thanks and lucky me they flew far east from US without any sign of jet lag, a 10F into that enclosure have nearly same main resonance as in free air as seen below where blue is free air and yellow is DCR.

Those came to me?
I still can't thank you enough for that!

Sorry, I'm late to the party, having spent a few days to get passed page 100 now... almost there!