Midrange chamber depth and shape and it's effect on audio quality

[blue934]

I came across this statement from JonMarsh in the Modula MT thread. It's thought provoking and brings two questions to mind.

"Keep in mind when considering reducing the box size that this also can impact the midrange quality- I find a certain amount of enclosure depth is necessary to keep a clean midrange (absorption of the back wave). Shallow boxes may look cool in some cases, but I don't think they sound good. But that may just be my own prejudice, so feel free to experiment." - JonMarsh 2006

1 - closed back midrange drivers?
2 - Midrange chambers or plastic midrange cups/chambers built into a 3-way box ? I try to angle back walls in mid chambers but haven't experimented with chamber depth.

Any thoughts on this topic or experience to share?


David

[JonMarsh]

Since you already have some of my input, I won't carry on too much here, but just mention a few questions, a few recent efforts... and some general points other designers love, going even further in that direction of "depth"....

• First, a philosophical design point- what is midrange? What frequency range do you consider applicable or of interest in describing these issues? This depends a lot on the drivers you want to work with, but also on the enclosure plan and overall design intent. For example, updating my ten year old Isiris design, the midrange is covering 250Hz to 2.25 kHz nominal. In other designs, with different woofers, like the Wavecor Ardent, with the C79, the crossover points were 600Hz and 2500Hz.
• Sealed back drivers can be an "interesting" can of worms. They tend to be popular for some professional applications, but the very limited rear chamber volume pushes the Fs up pretty high (for example an 8" B&C 8NSM64 = 250Hz, and the functional F3 is 400Hz, with a sharp 2nd order roll off, so where do you put your crossover point and at what acoustic/electrical slope? For HiFi drivers, plastic backs may not be very structurally stiff, because it's often used for a low cost solution for simplifying the main cabinet construction. Some handle this by making the steel frame a sealed back, like the PRV 6MR300SEALv2. But if you look at the result, I think you'd still think it's a nasty part to try to work with. Contrast that with the open back Faital 4FE32, which is very easy to work with as midrange drivers go...

BTW, the small dip between 300-400Hz is present in all their cone driver measurements, and is not present in my measurements of their parts, so it's an issue with their lab setup. Another Faital part I can recommend is the 5FE120, very smooth out to 5kHz, no discernible breakup mode (uses what I call the "classic" pro paper cone design, as do many others, including Eighteen Sound and PuriFi), and good 45 degree response to 3kHz.

• Other approaches? Consider some of the B&W speakers, which use a long tapered conical midrange enclosure to damp the back wave- these are well regarded in high end circles. Even the baby model, the 805, uses this technique just for the tweeter. But I have to say that my favorite dome tweeters sonically all have what I consider to be carefully done rear chamber designs.
• Consider also the popularity of what amounts to open back enclosures, such as the Bordeaux by Curt Campbell. That topic has seen a revival recently, though with the inconsistency of the tweeter response in recent production, I think it's not a recommended design anymore- but you might find it interesting to review with regards to the midrange design, though the C158 is not one of Accuton's best parts in my experience.
• Now, something completely different- enclosure loading of planar midranges and planar tweeters...like the parts originally developed by B-G, now owned by GRS in China. Some of the drivers are sold only as open back parts, some are sold both as open back and closed back. Contrast the response curves ...

GRS PT2522 dipole IB loaded SPL





GRS PT2522c with rear cup SPL





GRS PT5010 dipole IB loaded SPL






Now, the lower frequency behavior of planar drivers can be analyzed using classic T/S if you have access to some parameters, but the behavior is not so similar to cone drivers because of the differences in the ratio of parameters. What can be readily seen is how similar the dipole IB mode is to what you would get with say, a Magneplanar driver mounted in IB instead of free air. You can detect more about the differences between the PT2522 dipole versus the PT2522 open back by examining the impedance curve, and verify where diaphragm resonances occur. It's certainly a valid question to ask, for a specific speaker application, which version is better optimized? One might even question whether either is "best optimized"- for example, I have a custom closed back construction for the dipole version with better damping materials than the 2522C, after a few trials, which hits the SPL response and what I believe (perhaps erroneously!) the right impedance curve "signature". But it has smooth response to about 1400Hz and should be easy to use with a 2-2.2kHz crossover point. Consider that what I do is basically using a rear enclosure size that pushes the Fs resonance up to where I want it, then using high performance damping materials to damp the rear wave as throughly as possible.

The midrange panel, the PT5010, has its Fs resonance down about 240Hz. as the graph above shows. I've done several design rear enclosures, and have minimized reflected resonances by using a half circle rear wall and adjusting the total volume to get the desired Fs and overall transfer function for the desired crossover point (450Hz). Beyond that, I'm not talking at this point, as this work is being done for a potential project for Steve's SMJ operation. I will let you in on one interesting point- I can model the LF response of this driver using Unibox. And with the right rear cab design, I can get a smooth slightly humped response between 275Hz and 1800Hz, with a gradual rise starting above that. Fairly easy to work with, and amenable to my design targets.


I suggest applying a similar methodology to using conventional cone based drivers- think about the range you want to cover, what your practical crossover targets for you implementation is (I'm an advocate of my "home grown" LR3 concept due to better power response in the crossover region and being reliant on the lower frequency driver having an acoustic origin behind the HF driver) and chose the driver and develop the cabinet system based on that.

[blue934]

Thanks for that, lot's to ponder. Quite surprising how much the back cup affects the response in that example. I imagine this is the primary driver in higher fs for dome mids.
Regarding your half round back cup design, I would've thought that shape (concave) would direct the back wave towards the driver? Or am I imagining wrong and it is convex which would direct the waves away from the driver?
And back to you're original quote in the Modula thread, in preferring deeper mid chambers, how would deeper compare to shallower but angled back wall for back wave reflection?
I know many questions
thanks for you're time.
David

[Reet]

1 - closed back midrange drivers?

No thank you.

2 - Midrange chambers or plastic midrange cups/chambers built into a 3-way box ? I try to angle back walls in mid chambers but haven't experimented with chamber depth.

The primary cabinet reflection in midrange box will be the rear wall, so making this as far away as possible from the driver allows for more damping material in between the driver and wall to absorb the reflection. These reflections will be present in the impedance response, especially for high sensitivity drivers, as well as the nearfield acoustic measurement. So first step to analysis of the cabinet reflections is to build that measurement jig with the parts I gave you

[duvixan]

midrange is covering 250Hz to 2.25 kHz nominal.

Faital 4FE32, which is very easy to work with as midrange drivers go...

Don't want to derail the thread, but i have a few questions:

Is it desirable in speaker design to strive for the tweeter to cover the whole treble range (which starts at ~1280hz, from my understanding) ?

Is using a smaller 3"-4" to cover the ~800~3000hz range offer a significant improvement in sound?

Is there a design using the 4FE32 in the pipeline?

Thank you.

[JonMarsh]

OK, let's toss in or clarify a couple of other ideas, focusing a bit on the impedance curve as a diagnostic.... and a likely key design consideration for the midrange driver.


Here's an example that is not at all pretty, and it's the closed back midrange mentioned earlier:




One thing you can figure out quickly is that the Fs impedance hump will need to be dealt with by a shunt LCR network, and in this frequency range, it isn't cheap, but it's certainly going to be necessary, EVEN if you chose to cross this part over at 700Hz on the bottom, because that rising impedance will really mess up your high pass zeros. Then, too, the rising impedance curve is a design issue, and as another post on this topic alluded to, you may see reflections or other anomalies from the interaction of the cone and sealed rear chamber.



OTOH, for a midrange driver, the designer will usually prefer a part that looks like this: (I sure would...)



You've got an octave below 400Hz that is pretty flat and smooth to handle the initial part of your high pass network. If you cross at 500, LR3 or LR4 you can probably omit a shunt LCR for impedance control.

This, BTW, is the Faital 5FE120. This one IS in the design queue for a project or two, (I have a half dozen on hand) and if you check out the on axis and off axis response, you can see why....






This part is has been on tap for the Calliope CC project, which is temporarily on hold due to the CC Supreme being prepared for the Chancellor, which is a CC on serious steroids based on Dolby Theater specs, not home specs. That system will use the Faital 6RS140. That project developed a bad case of feature creep, based on the concept of: "If some's good, more is better, and too much is just enough!"

[JonMarsh]

Shifting gears a little bit, I'm going to share some of my investigations from mid 2022, which have been set aside in order to focus on finishing a 2021 project. Yeah, I need to get someone like ET who could round up a passle of Sith Appresntices to do development grunt work under his direction- could get a lot more done a lot faster than just me on my own! I have a large backlog of project concepts and parts on the shelf for them... part of my deferred gratification issues!



Let's look at the impedance behavior on one of the GRS drivers, the PT-5010, which is an open back dipole as delivered from the factory. My impression is that B&G had quality and consistency issues and rather poor documentation. And when claims are made that these drivers don't have any panel resonance, well, Father Physics tells me otherwise. You just have to look a little closer...

When GRS took over these parts, and did new production engineering and data sheets, I got interested again. Of course, PS Audio bringing out the FR30 with their own customized versions of these drivers raised the interest level even further- but this hasn't been something I could devote the time resources to that I wanted last year, given the other higher priority projects in the queue. Maybe 2023...


PT-5010, seen here with the GRS mounting plate accessory (very recommended, at least to start)






Recall from my earlier post in this topic the factory curves for the PT-5010:



I had some expectations about the impedance curve and what it would reveal about the panel resonances (based on having measured magneplanars back in the 70's). And I wasn't surprised.





I haven't measured this particular sample for IB open back SPL, so with some variability of diaphragm tension this one seems to be a bit lower than the sample used for the factory curves. Checking the half dozen I have on hand, they all look very similar to this. For a planar driver, this is a pretty significant resonance, which shows up in the SPL plot as a broad triangular bump in the response. Neither being what I'd prefer to see...

Along with my experiments with the 2522 tweeter panel and modeling this 5010 in Unibox, I developed a notion that it might be possible to push that resonance up in frequency and dampen it, so that we had something more like a plateau like function with the Fs resonance meeting the rising upper range response and having a relatively flat and usable area in the "midrange", which is what this driver is for, right?


As an example, this is a plot of the tweeter panel after a couple of rear enclosure iterations, both modifying the 2522C, and starting with the 2522 dipole version; I ended up liking the custom back cap and damping I came up with for the 2522 dipole best.




This is an easy to work with reposes and extends the low end by about half an octave over the 2522c factory closed back version. And I learned things from using the impedance curve as a diagnostic.


Here's the impedance curve of the dipole version 2522 tweeter panel, showing the primary resonance peak at 400Hz:






And the measured response looks a lot like the factory curves above.


My final modified versions of the dipole version look like this:







Having this broad relatively smooth plateau between 1kHz and 2kHz corresponded to the nicest impulse response and the response plot shown above. BTW, the damping materials in the back cup of the 2522c version are wholly inadequate, IMO.


Getting back to the midrange, I did several experiments and analysis with Unibox, and I have one more planned which I believe will be fairly optimum.

First one is in a 3.5L rear volume. Was tested in combination with the SB AT260 AMT tweeter, which wasn't at all up to the task.




This was a step in the right direction, but didn't get me there by any means... Next up, 2.0 L rear volume.





This is a LOT closer to what I was looking for, and had the general characteristics more like what I felt was the optimum 2522 solution.

There's another aspect to this, and that is distributing the rear reflections over a wider frequency range and enabling a better variable density absorption strategy.


This is a hint-






And this is the result of the first experiment adding this into the mix; one more POC iteration is planned based on lessons learned. Even this result is quite easy to work with crossover wise...

[Reet]

Deleted.

[Reet]

OK, let's toss in or clarify a couple of other ideas, focusing a bit on the impedance curve as a diagnostic.... and a likely key design consideration for the midrange driver.


Here's an example that is not at all pretty, and it's the closed back midrange mentioned earlier:




One thing you can figure out quickly is that the Fs impedance hump will need to be dealt with by a shunt LCR network, and in this frequency range, it isn't cheap, but it's certainly going to be necessary, EVEN if you chose to cross this part over at 700Hz on the bottom, because that rising impedance will really mess up your high pass zeros. Then, too, the rising impedance curve is a design issue, and as another post on this topic alluded to, you may see reflections or other anomalies from the interaction of the cone and sealed rear chamber.



OTOH, for a midrange driver, the designer will usually prefer a part that looks like this: (I sure would...)

On thing to mention on this comparison, is that the top impedance is shown on log scale, while the bottom is shown on linear scale. I'm sure the 5FE120 is a better driver, but one of these plots is doing a better job at hiding the blemishes simply due to the chart scaling.

[Reet]

Don't want to derail the thread, but i have a few questions:

Is it desirable in speaker design to strive for the tweeter to cover the whole treble range (which starts at ~1280hz, from my understanding) ?

Is using a smaller 3"-4" to cover the ~800~3000hz range offer a significant improvement in sound?

Is there a design using the 4FE32 in the pipeline?

Thank you.

There is no "one size fits all" answer. Balance between directivity, crossover slope, driver separation, and driver distortion characteristics are all considerations for what is "best". Crossover point, distance between drivers, and crossover slopes all have effects on the overall speaker directivity / power response characteristics. Balance all those factors against the driver distortion characteristics and you will achieve the best results. It's not as simple as "drive the tweeter as low as possible" or "avoid crossover below 3kHz", or "place drivers as close together as possible", IMO these statements are bad advice given the tools we have available today for speaker performance analysis and optimization.

[duvixan]

Balance all those factors against the driver distortion characteristics and you will achieve the best results

I emailed Troels a few times, trying to find out which in his opinion is the best driver complement and covering which ranges gets the best result

Is it a 6"+3/4"+0.75" for ~200hz/~800hz/~3500hz?
Is it a 8"+4"+1" for ~600hz/~3000hz?
Is it a 8"+5"+1" for ~400hz/~2500hz?
Is it a 8"+ CD horn for ~1200hz?

But he's always elusive and dodgy...
I'm sure all of them competently designed will offer excellent results, but I guess i'm curious if there's a somewhat objective way to get closer to the truth.

[JonMarsh]

I”m sure you could do very nicely with a PTT8.0 and a top of the line 1.4” driver (like the Eighteen Sound NSD1095N with an XT1086 CD horn, for just a two way, but the best way to optimize that is up for grabs, and if you want two of the woofers, then you’re likely to also want a midrange…. Just saying, I suspect what Troels is telling you is it all depends…including on what you’re specifically seeking. A big advantage to the right elliptical horn is limiting ceiling and floor reflections. So why not toss in….

2x 8” + 6 midrange + CD horn, but go for a minimum phase asymmetric alignment and crossover frequency that works with the Z axis origin differences? Well, one reason not to might be budget! How much do you want to weight budget in there? And how much of your budget are you going to allocate on room treatement and design? I’m really not trying to be snarky, but these are all real factors that one might want to consider… I say that having been involved in studio design and construction and coming from a very different place, I think, from a lot of HTG members about the need and desirability of actual room setup. You might get the nominal RMS values right at a specific location just by using DIRAC, but it won’t fix time domain stuff.

Nice discussion…

[blue934]

Well there's an example of assumptions gone wrong. Jon provides built and measured evidence of a half round concave back cup which gives excellent results. Looking at the pic without the measurements, I would expect the backwaves to be focused directly back to the driver giving negative results even with chamber stuffing.

[blue934]

1 - closed back midrange drivers?
No thank you.

What about the Morel dome you just purchased?

[Reet]

What about the Morel dome you just purchased?

It is a chambered mid, let's say "factory supplied box" Key is to have enough absorption of the back wave, and even those Morel's are a bit borderline on that, but they also don't operate that low in frequency either.

By closed back mid "no thanks" I was referring to normal cone drivers with closed frames, like Eminence Alpha-8MRA for example.

[Reet]

I emailed Troels a few times, trying to find out which in his opinion is the best driver complement and covering which ranges gets the best result

Is it a 6"+3/4"+0.75" for ~200hz/~800hz/~3500hz?
Is it a 8"+4"+1" for ~600hz/~3000hz?
Is it a 8"+5"+1" for ~400hz/~2500hz?
Is it a 8"+ CD horn for ~1200hz?

But he's always elusive and dodgy...
I'm sure all of them competently designed will offer excellent results, but I guess i'm curious if there's a somewhat objective way to get closer to the truth.

Keep in mind that Troels is also an "old school" designer, full directivity map, CTA-2034 power & DI chart is not really part of his design process. Even diffraction on nearfield doesn't appear to be part of it. He's been designing for many years though so I'm sure there's plenty of subjective knowledge and lessons learned from past trial and error applied to his process.

[JonMarsh]

Fully agree with your comments about Troels... he has his own path and methods, and a lot of folks seem happy with the results. He is doing it as a living or part of his living, so one shouldn't expect him to be too forthcoming about "spilling the beans".

[duvixan]

Well, one day he designs a speaker with 10" midrange and cd horn and calls it THE LOUDSPEAKER, after praising the use of 3-4" upper midranges for years.
Then he makes one with a 3" mid-dome and 1" tweeter and calls it his best design ever... I guess there's more than one way to nirvana.

[Steve Manning]

You may have already done this, but take a look at what the big boys are doing these days, Magico, Rockport, Wilson. There all doing separate midrange enclosures of some sort. Might give you some ideas.

[technodanvan]

My plan for an upcoming project was to either carve small (<1") pyramids on the back wall or create a 'wavy' back wall with pointed waves, the chamber would be at least eight inches deep. Are either of those choices ill-advised? Is a concave shape 'better' since only one central point would reflect straight back to the driver?

[blue934]

Jon, I'm curious what other shapes and sizes of back cups you have experimented with before coming to the current iteration on the GRS planar?
David