[Rudolf]

The ongoing discussion of cavity resonances in not-so-open baffles encourages me to show some measurements I did with my new H frame woofers:




They measure 40x40x40 cm inside with a chinese 15" PA-style woofer (A&D 15308) in the center. If this all-dimensions-equal philosphy doesn´t provoke standing waves - what else will?

So the graphs show what everybody expects - or don´t they?
First with the mike in the middle of the front and back opening plane:



In the front response (black) everybody will see a high Q res at 800-900 Hz and a next peak at ~1700 Hz. This perfectly corresponds to 40 and 20 cm wavelength. Bingo?
What bothers me is the total absence of a half wavelength resonance at ~430 Hz which I would have expected as perhaps the strongest one (and the only one which could be a source of trouble in the envisioned passband up to 200-300 Hz). And what puzzles me is the completely different response pattern to the back (red) from 800 Hz up, which should be due to the basket, filling almost half of the back space. No prominent cavity resonance there.

Next step was to step back and measure at 0.8 m distance to the driver mounting baffle:



Almost the same situation as before, albeit with starting room influence.

Somebody might ask if I might have measured some cone resonances instead of cavity resonance. The very same driver mounted in a more open alignment (left frame, not the M-frame) did not show any resonance of that kind:



(Both are nearfileld measurements in front of the driver, but with the driver at differenct locations in tne room)

What got me really crazy was when I slid a 55 cm wide and 25 cm deep book shelf diagonally into the front opening of the H-frame - dividing the frame from the basket up to 5 cm beyond the opening plane of the frame:


Absolutely NO difference, whether the front opening is divided by that plank or not!

Can anybody explain what that 800-900 Hz and 1700 Hz resonances are?

It´s a pity I need to find sleep now. But I am looking forward to some enlightening responses in the morning.

Rudolf
www.dipolplus.de

[JohninCR]

Rudolph,

Nice graphs, thanks for sharing.  Aren't the resonances you see based solely on the depth of the cavity?  That would explain why inserting the diagonal piece had no effect.  I assume that the cone area and opening are too large to stimulate a resonance where the depth = 1/4 wavelength, and instead the air column is still behaving as a lumped mass in the 430hz range and also preventing the resonance between the parallel panels.

It's interesting that the triangular shaped cavity eliminates the sharp resonances.  This indicates that the shape of our baffle foldings can be effective in solving resonance problems.

Also, I find the Fpeak at 300hz perplexing.  I expect it to be up well above 400hz.  The only explanation is the floor effectively increasing "D".  Can you put the cab up on a pedestal to simulate freespace and do another nearfield measurement.  I've been thinking that the floor benefit was primarily as boundary reinforcement, but if it also lowers Fequal, that is very good news in terms of bass response.

I hope Florian sees these graphs.  They clearly show how helpless we are in trying to fight against dipole bass cancellation without using EQ.

[MJK]

Rudolf,

My new H frame MathCad worksheet shows the fundamental peak at 282 Hz (best guess at actual box dimensions and driver T/S parameters) with the enclosure sitting on the floor similar to your 0.8 m measurement. The response then drops and levels out at 10 dB below the rounded peak's level. Since the worksheet only calculates up to 1 kHz, I cannot comment on the higher frequency data. I believe the sharp spike-like results to be typical of measurement generated ragged response and not a generic property of the speaker enclosure, it is never perfectly smooth in real life.

Martin

U, H, and single driver OB worksheets are being final checked and should be available in a couple of weeks (maybe sooner).

[Rudolf]

Aren't the resonances you see based solely on the depth of the cavity?

I don´t think so. Resonances along the cavity depth must be uneven multiples of a quarter wavelength. Deriving from the response peak the basic effective 1/4 wavelength should be ~0.28 m
No measured resonances seem to correlate to 1/3 or 1/5 of that.

It's interesting that the triangular shaped cavity eliminates the sharp resonances.  This indicates that the shape of our baffle foldings can be effective in solving resonance problems.

I still don´t know exactly where that sharp resonance comes from. So I would not (yet) speculate, how it was eliminated.

Also, I find the Fpeak at 300hz perplexing.  I expect it to be up well above 400hz.  The only explanation is the floor effectively increasing "D".

It is known from BR tubes and TMLs that the acoustic length of those "cavities" reaches well beyond the physical length. Somewhat surprisingly this effect is bigger for wide openings than for small ones AFAIK. Added boundary planes will extend that acoustic length even more.

Can you put the cab up on a pedestal to simulate freespace and do another nearfield measurement. I've been thinking that the floor benefit was primarily as boundary reinforcement, but if it also lowers Fequal, that is very good news in terms of bass response.

Nice idea to try out. I will get it up from the floor, but I won´t risk a hernia by lifting those 30 kg up to ear height.

I hope Florian sees these graphs. They clearly show how helpless we are in trying to fight against dipole bass cancellation without using EQ.

Good comment. I got another diagram handy. It shows how easily that compensation can be done:

Black line is the dipole near field response as shown above. For the red response curve I just added a 6dB/oct. LP filter (PLLXO), consisting of a small cap and resistor only. At 1 m distance and with an added 300 Hz LP this "levels out" to the purple response. Notice how that 850 Hz resonance has been completely subdued.
Since those measurements are not gated or smoothed, they already show much room influence at 1 m distance.

[johnk...]

FYI,

SoundEasy simulation of a driver with Qts = 0.52, Fs = 18 Hz in a 40 cm x 40 cm x 40 cm H frame. Driver centered at 20 cm.

Front response alone. (equivalent to near field result at front opening).

[JohninCR]

Also, I find the Fpeak at 300hz perplexing.  I expect it to be up well above 400hz.  The only explanation is the floor effectively increasing "D".

It is known from BR tubes and TMLs that the acoustic length of those "cavities" reaches well beyond the physical length. Somewhat surprisingly this effect is bigger for wide openings than for small ones AFAIK. Added boundary planes will extend that acoustic length even more.

I like that idea because it gives us some free ride using folded alignments.

Since JohnK stepped in to clearly sim the resonances, here's another idea.  Damp the rear cavity and measure the front.  This should make the rear perform like a U and give you increased bass response (added effective D) without changing the resonances in the front.  Maybe instead since the basket seems to kill the resonance, the front needs the damping and measure from the rear to see the U effect.

[Rudolf]

@ johnk
that SoundEasy sim is very interesting. Thank you very much for the diagram!

Since it is "front response alone" those peaks cannot be interpreted as front/back wave comb filtering, right? Because they correspond so well with my measurements - are they really resonances between the H frame boundaries? If they are, why does no 1/2 wavelength resonance show up? And why does a diagonal obstacle in the frame leave them unaffected

Your opinion/insight about it would be most appreciated.

BTW: My driver has Qts= 0.5  Fs=38 Hz

[johnk...]

You basically have an open ended TL for which there is no resonance at WL = 2L.

If you consider that you cut the duct in half lengthwise with the diagonal board, you have two ducts with 1/2 the volume and 1/2 the cross sectional area, but the ratio V/A = L stays the same so nothing changes. 

There are other factors, but these are the basics.

[Paul W]

Rudolf,
The shape of the low-end rolloff curve for the "mini" is interesting.  Do you have any SPL measurements directly comparing output of the mini to the H-frame?  In other words, same drive level, same woofer and microphone positions etc. 
Paul

[Rudolf]

Paul,
I still got the "mini" baffles, so a comparison is within reach. Just exchanging a driver.
But I am not sure what you would call same microphone position, regarding distance from the driver. 20 cm from the cone obviously would deliver results you can´t really compare. Could you define the distance you are interested in (excluding real farfield - I dont want to haul those heavies outside )
I already can assure you that the efficiency of the mini would be almost 10 dB below the H-frame.

[Paul W]

Microphone maybe about 1 meter height and 1 meter away...or as far as you can get reliable readings.  Microphone directly on floor might work too.  I am curious about the shape of the low end curve and the loss "over the top" of the baffle vs floor reinforcement.  Don't care so much about HF/on-axis.

I know what you mean about carrying large woofers outside...much clearer measurement results, but sore back .  So I bought a small 2-wheel handcart!

[JohninCR]

I already can assure you that the efficiency of the mini would be almost 10 dB below the H-frame.

10db seems like a lot, but I'll take your word for it.  This does, however, get me back to a question I may have asked poorly before.  What draws you to build the small H's when a U essentially half the size delivers the same bass performance?  Do the front and rear resonances of an H somehow cancel each other?  With that small depth of only 20cm, I have to believe that the shift away from a pure dipole radiation pattern would be fairly small and having free flowing output for the front wave would have to be a plus (no resonance).  From my experience the rear wave doesn't seem adversely affected by damping, so it's not like a damped U is like some kind of aperiodic box.

[johnk...]

On axis sensitivity goes like 20 Log(L1/L2) for U's and dipoles where L1 and L2 are the effecting path length difference for front and rear radiation. So 10dB = an effective path length difference ratio 3.16.
H frame resonances do not cancel but for typical woofer application and crossover points the first resonance should be well above the crossover point and well attenuated by the crossover so that direct treatment isn't required. But adding a notch filter in the crossover won't hurt.

[Rudolf]

This does, however, get me back to a question I may have asked poorly before.  What draws you to build the small H's when a U essentially half the size delivers the same bass performance?

I remember. Your question didn´t go unnoticed, but somehow slipped from my agenda.
There are two reasons:
First I´m (regrettably) not in the same league as you when it comes to the practical implementation of my ideas. So I have just not found the time to delve into U-frame praxis.
Second I explicitely wanted to keep the dipole 8 radiation pattern - for better attenuation to the sides and for active usage of the (low frequency) front wall reflection.
Your comment "From my experience the rear wave doesn't seem adversely affected by damping" somehow makes me doubt whether you are talking about a real U-frame. AFAIK the output to the rear in a U-frame should be almost annihilated by proper damping. But what do I know.
Essentially I don´t regard H-frames any better or worse than U-frames albeit they definitely don´t give "the same bass performance". They are just different solutions for different problems/targets.

Regarding the 10 dB loss I believe that johnk already has explained it all.

[JohninCR]

Thanks Rudolf and John,

I see H's all over, but almost no U's, and I'm trying to figure out why.  I keep thinking that there's something I'm missing.  While the damping might slightly attentuate the top end of the operating range, the rear wave gets through pretty much unfettered.  It's only to damp the resonance and make the driver the rear wave source, not the terminus.  The more cardioid radiation pattern of a near perfect U is the result of the back being where the front and rear waves have travelled the same distance, so that's where the greatest null exists.  The reason is not because the rear wave has been attenuated in some way with a U.

Regarding the shape of your mini, if we can come up with what the real world effective path length is for such a shape, then we'd be well on our way to designing resonance free U-type shapes (ie no damping needed).

[FlorianO]

I hope Florian sees these graphs.  They clearly show how helpless we are in trying to fight against dipole bass cancellation without using EQ.

  aa

[FlorianO]

Regarding the shape of your mini, if we can come up with what the real world effective path length is for such a shape, then we'd be well on our way to designing resonance free U-type shapes (ie no damping needed).

Enter Visaton NoBox .

John, I know you already dismissed that one as a "off the cuff design", but I reckon sloped wings do have some appeal: Larger wings around the woofer, smaller around the FR, added stability...

[JohninCR]

Regarding the shape of your mini, if we can come up with what the real world effective path length is for such a shape, then we'd be well on our way to designing resonance free U-type shapes (ie no damping needed).

Enter Visaton NoBox .
John, I know you already dismissed that one as a "off the cuff design", but I reckon sloped wings do have some appeal: Larger wings around the woofer, smaller around the FR, added stability...

Those are definite pluses.  I guarantee the thing vibrates like mad.  The extra width than necessary for the woofer is unnecessary, and damping is probably required in the lower section for better bass performance.  Around the full ranger, you have to experiment with shape for the best sound.  I don't doubt that they sound good, but I'm sure with minor changes the improvement could be quite audible.  I favored the Quasar from that group only due to the significant baffle mass.  I'd venture to guess that a flat piece of plywood with a support brace centered over the driver would sound better than the as-is NoBox, although it wouldn't deliver nearly the bass extension, some of which is almost assuredly panel vibrations from those big unbraced side panels.

It's easy to stop with incomplete designs with OB, because they sound so different in very seductive ways than boxes.  There's still a lot of unexplored territory with the enclosure shape, since the scientific types like JohnK and SL went the route of designing and using electronics to tailor response however they desired.  I don't even like changing lightbulbs, so I'm forced to go a more physical route.

For a first shot, the NoBox may be the way to go, since it can easily be tweaked.  If in fact, the small cavity behind the main driver on the Quasar is a problem, it would be difficult to correct afterward. but big radiused roundovers would be a piece of cake.  If someone donated the wood, I'd build something similar in a lot of respects to the Quasar.  I just haven't been willing to burn up that much material while I'm still learning and trying new things.

[FlorianO]

Well, talk about hijacking a thread and going OT. Sorry Rudolf

For a first shot, the NoBox may be the way to go, since it can easily be tweaked.  If in fact, the small cavity behind the main driver on the Quasar is a problem, it would be difficult to correct afterward. but big radiused roundovers would be a piece of cake.  If someone donated the wood, I'd build something similar in a lot of respects to the Quasar.  I just haven't been willing to burn up that much material while I'm still learning and trying new things.

Sounds like a collect to me

To be very honest with you, I'm torn between two alternatives:

First is taking Dick Olsher OB2 project "hook and line" (Augie, Feastrex, customed PLLXO, the lot) and go with it. Maybe later tweak it a bit like above.

Alternatively  build a Quasar clone with two Augies hooked in parallel, drive them by a plate amp XOed them smth like 200Hz. For FR buy some affordable 8 incher  and drive it with a flea powered SET  aa

Yet I should first complete my A126 BLH for FE126-E. That's been dragging for a month now... (well, it's my first BLH, alright ?!?! ).

Sorry again Rudolf. Back to the program. Really intresting thread, but have to go through JohnK's material on Hs and Us first

[JohninCR]

Florian,

Here's another alternative.  Buy the Feastrex pair and send them to me, and I'll see what I can do with them while you finish the BLH's.  Then bring your A126's with you on a vacation to Costa Rica to pick up your drivers, and you get a free cab out of the deal.  I get to hear the Feastrex and A126's without building it.  I have a bunch of stuff for a couple of days worth of comparison listening (Frugels, Metronomes, BIB's, OB-RLH's, dipole waveguides, U-baffles, W-baffles, ripole hybrids, and plenty of more typical OB's), and an extra set of interested ears is always welcome.  My wife, kids, and employees have had their fill.  This disease has progressed to the point that speakers have even become a storage problem.