|
Home » Sponsored » Pi Speakers » My 3π adventure
| Re: My 3π adventure [message #95930 is a reply to message #95260] |
Mon, 22 August 2022 16:02   |
ppkstat
Messages: 38
Registered: February 2022
|
Baron |
|
|
I'd like to thank everyone for their kind words, and Wayne for all the support. I cannot stress this enough especially for Wayne who answered all my stupid questions with great patience.
That being said I am back from vacation so its measurements time!
This is a nearfield measurement of the port, woofer and driver.
Seems pretty normal as far as I can understand except from the tuning frequency of the cabinet which seems to be around 24Hz. This is exactly the same on both speakers. I have no idea why they turned up like that since I followed every dimension with a 1mm tolerance. Maybe the added volume from the support rings or the less thick insulation?
In order to see if everything went well I tried to do a quasi-anechoic measurement of the speaker, indoors. I laid it on its back and measured it, my high ceiling (5m) would probably helped with this. Below is the response as I got it and then the same response with impulse response gating.
I then tried to merge this response with a combined port-woofer response but I failed completely. I could not match the SPL between the woofer and the port and the formula described in Jeff Bagby's whitepaper produces strange results. This might be due to the fact that the cabinet is tuned that low.
Ok now moving on to the in-room responses. BIG DISCLAIMER HERE. What you're about to see has nothing to do with the performance of the speakers, it's all due to my awful, untreated, listening space.
Here is the response from the listening position which is a couch. The are 3 graphs for the center, the left and the right of the couch in this order (1/12 smoothing).
RT60 is horrible averaging around 800m
and the spectro doesn't look very good either
There are several issues with the SPL response. In case you're wondering they sound as bad as they measure. The first issue is poor bass response, especially in the 30-100Hz region, which is followed by a bump. There is massive HF roll of above 2kHz and I think this is something that has to do with the space as I was measuring the same with my previous speakers. There is also a bump around the crossover region which I think also has to do with the space as it was absent on the measurements of the speaker lying flat.
What I do find strange however is that in the published 3pi response the 30-150Hz region is still 3db lower than the mid region. I don't know why this is the case, is this by design or a personal preference? 
In any case the speakers in my listening space need a very heavy amount of equalization. I used EQ APO with a +4db Harman curve in order to get a more balanced result. Below you can see the result but I did the whole procedure a bit hastily. The definitely sound better this way though. These speakers seem to have an excellent polar response so I hope that they will be equalized well in the end. Red is unfiltered and blue is equalized. Subjectively, that boost in the 30-100 region makes a huge difference 
These need quite a bit more work to sound correctly inside the room. Future attempts will include proper placement (symmetrical and away from walls), Dirac, room treatment and 4 subwoofers. The room is notoriously difficult to treat for several reasons. However, one of the first thing I will do soon is to construct some freestanding thick panels to place behind the speakers in attempt to reduce SBIR issues. I'll keep you posted!
-
Attachment: nearfield.jpg
(Size: 523.14KB, Downloaded 711 times)
-
Attachment: 3.jpg
(Size: 468.46KB, Downloaded 704 times)
-
Attachment: 2 (1).jpg
(Size: 494.17KB, Downloaded 743 times)
-
Attachment: center.jpg
(Size: 525.67KB, Downloaded 599 times)
-
Attachment: right.jpg
(Size: 541.42KB, Downloaded 704 times)
-
Attachment: left.jpg
(Size: 524.93KB, Downloaded 700 times)
-
Attachment: spectro.jpg
(Size: 573.97KB, Downloaded 684 times)
-
Attachment: rt.jpg
(Size: 769.21KB, Downloaded 707 times)
-
Attachment: equa.jpg
(Size: 532.60KB, Downloaded 705 times)
|
|
|
|
| Re: My 3π adventure [message #95932 is a reply to message #95930] |
Mon, 22 August 2022 17:01 |
 |
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
Measured in-room, I think your response curves are pretty good. This is typical in-room response. Looks to me like all is working as designed.
In-room measurements show the room more than they do the speakers. But having a waveguide speaker really helps in that regard.
What isn't helped by the waveguide is the range you're talking about - the 30Hz to 150Hz region - which is entirely in the modal region. It's why we want to employ multisubs and flanking subs. The mains can't do much there by themselves, at least not without energizing room modes, which is generally counter-productive.
This gradual rolloff below 150Hz is caused by baffle step. Above 150Hz or so, the sound radiates mostly forward because of the size of the baffle. But below that frequency, the sound radiates omnidirectionally, so on-axis SPL is reduced. The sound energy is spread over a wider area at low frequencies, and that's what causes baffle step.
Some designers might "boost" the region below baffle-step, but they actually do it by attenuating the whole audio range above baffle-step. I prefer to go the other way, to use what I call "flanking subs" to provide extension and to bring up the energy below 150Hz. You can see it as a three-way speaker with the bass driver detached, in its own cabinet.
I mean, you could boost bass going to the mains. But then you have a single sound source - or two, if stereo - energizing the room in the modal region. This small number of sound sources creates well-defined room modes. The room has strong peaks and dips in the modal region.
Adding sound sources in the modal region helps to mitigate room modes. The room is energized in a variety of locations, which tends to smooth the modes.
You'll notice the big peaks and deep notches in the 60Hz to 120Hz range. Response is usually made smoother using flanking subs. That's the whole reason for having them. They reduce speaker boundary interference response (SBIR) anomalies, which are what usually causes the roller-coaster in that region. And they also provide baffle-step compensation, which is what makes the speakers droop below 200Hz or so.
The SBIR anomaly is really pronounced in the measurement you showed with the speaker lying on its back. I see this often - it's kind of a textbook case for why to use flanking subs - because many speakers are pushed back directly against the wall behind them, or maybe just a foot or two away from the wall. This is a common placement for speakers, because it is a convenient use of space. Most people don't have rooms large enough to pull their speakers several feet out into the room. So it results in that tell-tale SBIR roller-coaster between 80Hz and 120Hz. Pull 'em out a little further and the notches shift lower, but they're still there.
SBIR is not the same as room modes, but they are both caused by reflections. Modes are standing waves whereas SBIR is self-interference between a direct wave and a reflected wave. The nasties in the 80Hz-120Hz region are primarily caused by SBIR in most cases. But higher-frequency room modes are there too.
So the point is you will really benefit by employing flanking subs and possibly one or two additional subs placed far away in a multisub arrangement. Flanking subs will mitigate baffle-step, SBIR and high-frequency room modes. Distant multisubs will mitigate lower frequency modes.
|
|
|
|
|
|
| Re: My 3π adventure [message #95935 is a reply to message #95934] |
Mon, 22 August 2022 18:33 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I like symmetrical multisubs, which is a Welti arrangement. You can use such an arrangement and also include flanking subs.
Place the mains with flanking subs outboard. They are just a little behind, beside and below the mains. They are stereo subs - as you say - with a gentle 2nd-order slope low-pass slope. With that slope, the crossover frequency is usually best around 100Hz, which then still has significant energy up into the lower midrange around 150Hz.
If you get very close to a flanking sub, you can hear muffled vocals. That's fine because it's very close to the main speaker it's flanking. It's a few feet away in all three dimensions, with your three Pi mains on 12" to 15" tall stands and your subs on the floor.
Then the distant subs are on the other side of the room. They get a traditional LFE signal, which tends to be fourth-order at 50Hz or 60Hz, something like that. It's an all-channel summed bass signal with steep slope and low crossover frequency. Place them symmetrically with the flanking subs.
You can also experiment with an asymmetrical arrangement, like Geddes prefers. Simply place the distant multisubs in a random location distant from the mains. Geddes tends to like one in a corner, and the other in a random location that isn't symmetrical or coincident with any of the other subs.
Both Welti and Geddes agree that once you get to four subs, location begins to matter less and less, provided they are all in different locations. So you have some "wiggle room" on the placement of the distant multisubs.
The distributed multisubs combined with the flanking subs and mains form a woofer array that can be arranged like what Welti or Geddes describes. Either way, the distant subs blend with the flanking subs and mains to smooth deep bass modes.
The ~60Hz crossover on the distant multisubs prevents them from smoothing modes above that point, but you really can't crossover higher without localization problems. Their main purpose is deep bass extension and low frequency modal smoothing.
The flanking subs also run down that low, so they blend as additional low frequency sound sources. But since they run much higher, they blend with the mains in the higher-frequency modal range. They provide smoothing above 100Hz, and they also provide baffle-step compensation.
It's a synergistic system approach.
|
|
|
|
| Re: My 3π adventure [message #95939 is a reply to message #95935] |
Tue, 23 August 2022 09:09 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I meant to add some comments about the Helmholtz frequency in each of my responses above, but forgot. It was more important to me to discuss the use of flanking subs and multisubs, and the reasons for needing them. So I became involved in those responses and forgot to come back to box tuning. I'll correct that now.
Your measurement of the Helmholtz frequency looks correct to me. I don't get too focused on box tuning unless someone has adjusted dimensions. And even then, I'm not as focused on the Helmholtz frequency as I am on the possibility of anomalies from internal standing waves. Of course, if the Helmholtz frequency were shifted by a large amount - particularly upward in frequency - there would be adverse effects. So if a person were to forget the duct entirely - leaving just an unvented port hole - or do something else that shifted box tuning up a half octave, that would be bad. It would show up in the response curve as an underdamped hump. But I rarely see that problem.
I designed both the three Pi and four Pi speakers to have a slightly overdamped response curve. They aren't designed for max flat extension. This is partly for size and partly because I like the behavior of a slightly overdamped vented cabinet. It cannot shift anywhere close to having an underdamped hump, even as parameters have thermal shift at maximum power levels.
Such a system has a response curve that looks a little like a sealed cabinet, but with deeper extension from the same sized box. Its overdamping makes it look less like the traditional vented fourth-order curve and more like a sealed second-order. It also becomes very tolerant of parameter shifts. And being vented, it benefits from reduced excursion even with the added energy down low.
This design approach creates a speaker that rolls off gradually and works well with the flanking sub approach. It sounds good without subs - having plenty of useful bass - but it does lack the full depth of a system that includes subs. Sometimes, at shows, I switch off the subs and people don't notice a huge difference. It's not completely lacking in bass. But the system does benefit from subs, for all the reasons I described above.
As an aside, acoustic suspension advocates would argue that the transient response and group delay of a sealed system is better, but those are very dated arguments. To argue that phase shifts from the vent are a disadvantage that sealed cabinets don't suffer overlooks the very real influence of room modes. There is no phase benefit from a sealed cabinet when used indoors, because phase in the modal region is all over the place and different in every location. There again, that's why we use multisubs.
The published impedance curve of the four Pi shows it to be tuned to around 35Hz and the three Pi is tuned to around 25Hz. That's the approximate location of the impedance minima. If I were to look closer at the raw data, I might see that the exact minima is 24Hz for the three Pi and 34Hz for the four Pi. These two speakers might drift to 26Hz and 36Hz from thermal parameter shift or different amounts of damping material. It's fine all the way up to beyond 30Hz on the three Pi and 40Hz on the four Pi. I often even round up to those values when asked. If someone is pushing them hard for prosound or high-power home theater and they want to high-pass the mains, I would suggest those values for the HPF.
The target Helmholtz frequency for the four Pi design was 37.5Hz and the three Pi was 28.5Hz. But like I said, slight shifts downward totally don't bother me and this design approach - being slightly overdamped - allows for slight upward shifts too. As I mentioned earlier, what I am equally concerned with - perhaps even more focused upon - is internal standing waves. These cabinets are large enough they develop standing waves in the lower midrange.
As part of the initial design and testing, I modeled the cabinets with Martin King's spreadsheets and then made physical models to test with, all in an effort to ensure the positions of the midwoofer, port and damping material prevented an antinode being at the port or the driver, where it would create response ripple. After all the modeling, testing and cabinet modifications to dial it in, the internal standing waves were mitigated, and the Helmholtz frequency stood where it is now.
|
|
|
|
|
|
| Re: My 3π adventure [message #95941 is a reply to message #95940] |
Tue, 23 August 2022 14:29 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I agree with you generally about almost everything you said except the part about time alignment. This is impossible in the modal region.
Outdoors, you want sound sources acoustically close together - within 1/4λ of each other - and the electrical signal phase should be within 1/4λ too. Sometimes, this isn't possible - like with hornsubs - because acoustic centers are far from the mouth. In that case, a delay line can be employed to keep everything in sync. It's not really important that they be "aligned" to the microsecond or anything, but it is important that they be within 1/4λ of each other.
Indoors, this isn't possible because of interactions with reflections that are nearly as loud as the direct sound. SBIR and room modes rear their ugly head. So since we cannot possibly have the sources and reflections in phase, we might do the next best thing. We can create a sound field at low frequencies that acts similarly to the reverberant field that exists at higher frequencies. That's created by dense interference. It's done by blending sound sources of lots of different phases.
So - no - don't plug your ports. Totally unnecessary and actually counter-productive. Just setup a flanking sub / multisub system.
To summarize:
- If everything is in-sync and there are no reflections, a coherent wavefront can be generated. This is the best case outdoors, or in very large rooms and auditoriums.
- If there are just a few sources and/or reflections, then well-defined lobes and nulls form. It looks like a checkerboard of hot and dead spots in 3D space. This is what the modal region acts like - if there are few bass sound sources - it's a worst-case scenario. You can picture it as choppy waves on water with high peaks and troughs, like what is formed from a single large item or maybe a couple of large items thrown into the water.
- If there are many sound sources and/or reflections, then the lobes and nulls become so dense you don't really notice them. Picture it as rainfall on the surface of a pond. There are lots of little impacts, but there are so many it all blends together. This is how the reverberant field acts, and it's the next best thing to a coherent wavefront. Multisubs can create this too, changing the modal region into something that acts like a reverberant field.
|
|
|
|
Goto Forum:
Current Time: Sun Nov 24 04:18:50 CST 2024
|
Posting Rules
Members
Search
Help
Register
Login
Get a T-Shirt!
