|
Home » Sponsored » Pi Speakers » 4PI Plans Request
|
|
| Re: 4PI Plans Request [message #91243 is a reply to message #91242] |
Mon, 09 December 2019 10:47   |
 |
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
You've got mail!
I agree with Barry about the JBL 2226H. It is an excellent midwoofer. The flux stabilization ring in the motor structure reduces distortion down to the levels normally only seen in horns. When combined with a horn tweeter, you have a system that is very smooth and distortion free.
Also compare the H290C waveguide to the SEOS12. The H290C is measurably smoother. I considered using the SEOS12 at one time, but I wasn't happy with its performance. It has peaks that must be smoothed with notch filters in the crossover. So I designed the H290C waveguide specifically to allow an upgrade path beyond that.
Don't get me wrong - SEOS enthusiasts are very much like Pi Speakers enthusiasts. And their waveguides and loudspeaker designs are very good. In fact, when they started out, many of them were regulars here and on (Zilch) Evan Flavell's "Econowave" threads. Pi Speakers is sort of like the grandfather design for Econowave and SEOS speakers. The loudspeaker models here were inspirations for their designs.
That's why I considered using the SEOS waveguide when they first started out. I had a wood horn/wavguide that provided constant directivity, but it was very expensive. My other option was a radial horn that had nearly constant-directivity. This was much more popular because of its cost. So I was looking for an upgrade path for the radial horn using a horn/waveguide of approximately the same physical dimensions.
At that time, about ten years ago, the guys that are now DIYSG were talking about making a waveguide too. Back then, I thought we might work together. And like I said, many of them were regulars on this forum and on the Econowave thread. We were all very much kindred spirits.
But we diverged mostly on the points of acoustic loading. We agreed on pretty much all other aspects. I personally do not like waveguides that deoptimize acoustic (horn) loading because they become excessively resonant and require notch filters in the crossover to avoid peaks in response.
So I chose to have an injection mold made for the H290C horn/waveguide, which was very much like the expensive wood horn/waveguide I offered at the time. The H290C does everything I want it to do, at a reasonable cost. It provides constant directivity and is very smooth, not needing notch filters in the crossover.
And speaking about the crossover, I suggest that you stick with the passive crossover as shown in the plans. You could employ an active crossover, but it must have the exact same transfer function as the passive crossover or you would be degrading performance rather than improving it.
I use a digital active crossover to design the passive crossover. It uses an impedance chart in the form of a ZMA file to incorporate the actual electro-mechanico-acoustic load of the drivers in the system. See the link below for more information about the process.
Systems with active crossovers have their benefits, but first-things-first: To get amplitude and phase response right, we need either to include the driver's electro-mechanico-acoustic interaction directly with the passive components, or if isolated using an active filter, we need to be able to employ a ZMA file or some other way of including the load impedance.
It's an Nth degree thing. You can definitely make a great loudspeaker without this kind of precision. It's like making a hotrod car with a carburetor and distributor with ignition points. You can make some awesome stuff with that technology. But if you have a computer sensing oxygen and detonation, you can get even better performance.
While some might see the passive crossover to be more akin to the carburetor, it's not the case. The passive crossover has been dialed-in with the computer. So in this case, the passive crossover is the one that is the Nth degree solution.
An active crossover that didn't exactly emulate the transfer function described above would be more like the computer controlled carburetors of the 1980s. It wouldn't be exactly right. Sort of like how the computers in the 1980s cars couldn't detect detonation and their narrow-band oxygen sensors don't have the ability to provide an accurate signal. They couldn't rapidly change the air/fuel mixture or set the ignition timing. So even though those 1980s cars had computers, they didn't offer much in the way of performance. Same could be said of an active crossover that wasn't dialed in, but instead used generic filter functions.
The only way to get an active crossover to give a true benefit would be to employ a digital filter like I use for designing, one that has an impedance plot of the drivers in the system. It would then give the exact same transfer function as the passive unit.
And one last thing. Not something you asked about but worth mentioning anyway. I highly recommend flanking subs be used. They work very well with my mains or with DIYSG speakers. They are a subset of the multisub concept, which is another thing I advise. But if you want to start small, use two flanking subs and then later add another sub or two, placed further away. I call the distant subs "distributed" multisubs, to distinguish them from flanking subs. But both "flanking subs" and "distributed subs" are parts of a total multisub setup.
|
|
|
|
|
|
| Re: 4PI Plans Request [message #91245 is a reply to message #91244] |
Mon, 09 December 2019 12:25 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I think that's a good way to go. Run the flanking subs with a 100Hz/2nd-order low-passed copy of the signal sent to the main speaker each sub is flanking. Run the mains set to "large" so you blend with the flanking subs. The mains should be able to handle the LF content unless they just really can't handle the low notes or you are pushing them close to their limits. If you're pushing them hard, then high-pass them at their Helmholtz frequency.
This Helmholtz recommendation is, of course, relevant for vented systems. Above the Helmholtz frequency the vent reduces excursion, but below Helmholtz the woofer is unloaded. If sealed, it's not such a stark a change in excursion but then again, there is no reduction of excursion from the vent. So for sealed woofers, high-pass where excursion becomes a problem either because of IMD or xmax or both. The goal is to run the mains as low as possible for modal smoothing.
The improvements from flanking subs and multisubs are measurable and audible. The effect is anywhere from subtle to striking, but it's always noticeable. Flanking subs make the upper midbass and lower midrange smoother. Male voices, piano, guitar and cello sound clearer. Distributed multisubs make the deep bass smoother. You will notice bass notes that seemed light or even missing before. Sounds like a better foundation.
|
|
|
|
| Re: 4PI Plans Request [message #91247 is a reply to message #91245] |
Mon, 09 December 2019 14:59 |
mathiasb
Messages: 7
Registered: December 2019
Location: WA
|
Esquire |
|
|
Wayne Parham wrote on Mon, 09 December 2019 12:25I think that's a good way to go. Run the flanking subs with a 100Hz/2nd-order low-passed copy of the signal sent to the main speaker each sub is flanking. Run the mains set to "large" so you blend with the flanking subs. The mains should be able to handle the LF content unless they just really can't handle the low notes or you are pushing them close to their limits. If you're pushing them hard, then high-pass them at their Helmholtz frequency.
This Helmholtz recommendation is, of course, relevant for vented systems. Above the Helmholtz frequency the vent reduces excursion, but below Helmholtz the woofer is unloaded. If sealed, it's not such a stark a change in excursion but then again, there is no reduction of excursion from the vent. So for sealed woofers, high-pass where excursion becomes a problem either because of IMD or xmax or both. The goal is to run the mains as low as possible for modal smoothing.
The improvements from flanking subs and multisubs are measurable and audible. The effect is anywhere from subtle to striking, but it's always noticeable. Flanking subs make the upper midbass and lower midrange smoother. Male voices, piano, guitar and cello sound clearer. Distributed multisubs make the deep bass smoother. You will notice bass notes that seemed light or even missing before. Sounds like a better foundation.
Totally agree (not that facts need agreeing) about multiple subs, I run two now, independently.
I'd have to get a second amp to do this, so I can DSP the front speakers - however, quick question on the low pass for the subs - how come 100Hz, and not, say, 80, or 120? Is that based on Pi4 (would make sense), or just a starting point in general?
|
|
|
|
| Re: 4PI Plans Request [message #91248 is a reply to message #91247] |
Mon, 09 December 2019 16:45 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
The flanking sub low-pass recommendation of 100Hz second-order is empirical, so you can consider it a starting point and not a hard-fixed number. However, I have found that the slope is always better being gradual than it is being sharper. Fourth-order slopes are too steep. I have also found that low-pass below 80Hz is too low for flanking subs. The range of 100Hz to 120Hz and sometimes as high as 150Hz seems to work best. Much higher than that and the subs become localizeable and draw attention to themselves.
The low-pass and amplitude of flanking subs is set by simultaneous optimization of three things:
1. Amplitude level-setting the subs to the mains, so we gain bass extension at the appropriate SPL. This usually amounts to about 10dB more gain on the subs, because they are usually about 10dB less efficient than the mains.
2. Low-pass that conjugates baffle-step. Cabinets the size of my three Pi and four Pi mains tend to have baffle step in the 100Hz to 200Hz region. Instead of employing compensation in the crossover, we use the flanking subs to provide additional SPL as BSC.
3. Low-pass that mitigates higher-frequency room modes and SBIR from nearest boundaries. The worst anomaly usually comes from the wall directly behind the speakers, but the nearest side wall is sometimes objectionable too, as can be floor bounce. The modal region extends above 100Hz, but distant multisubs cannot be run this high without being localizeable. And the boundary interference notch from the wall behind the speakers seems to almost always be between 80Hz and 120Hz, probably because proximity to the wall behind the speakers is most convenient and/or popular in this range of a few feet. So this makes it useful to have a truncated array in the 80Hz-120Hz region, to counter the interference notches there. The flanking sub and the midwoofer in the mains form a simple two-element array.
|
|
|
|
|
|
| Re: 4PI Plans Request [message #91260 is a reply to message #91257] |
Sat, 14 December 2019 09:53 |
|
Wayne Parham
Messages: 18787
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
You've got mail!
Kits include the speaker drivers, the connector panel, hook up wires, "π" logo decal and a printed copy of the plans. For larger speakers that incorporate a compression tweeter and crossover network, kits also include the crossover, Zobel woofer damper, and all cable assemblies are completed and ready to install. Every kit containing a compression driver also includes the horn flare and the bolts to mount the driver to the horn.
I recommend using only MDF or Baltic Birch. You will probably want to purchase wood at a quality lumber yard, because the large chain retail hardware stores do not usually stock MDF or Baltic Birch. They only sell cheap plywood products and chipboard. Both of those are suitable for many construction projects, but not for loudspeaker cabinets.
Please peruse the Pi Speakers FAQ. I think you will find the answers to all of your questions there, as well as answers to many things you haven't thought to ask yet. The FAQ has information about upgrade options, the differences between models and upgrade choices, build and setup advice and other useful things. And as you start your build, please feel free to document it here and ask questions or make comments along the way.
|
|
|
|
|
|
| Re: 4PI Plans Request [message #91268 is a reply to message #91248] |
Sun, 15 December 2019 22:12 |
mathiasb
Messages: 7
Registered: December 2019
Location: WA
|
Esquire |
|
|
Wayne Parham wrote on Mon, 09 December 2019 16:45
The flanking sub low-pass recommendation of 100Hz second-order is empirical, so you can consider it a starting point and not a hard-fixed number. However, I have found that the slope is always better being gradual than it is being sharper. Fourth-order slopes are too steep. I have also found that low-pass below 80Hz is too low for flanking subs. The range of 100Hz to 120Hz and sometimes as high as 150Hz seems to work best. Much higher than that and the subs become localizeable and draw attention to themselves.
The low-pass and amplitude of flanking subs is set by simultaneous optimization of three things:
1. Amplitude level-setting the subs to the mains, so we gain bass extension at the appropriate SPL. This usually amounts to about 10dB more gain on the subs, because they are usually about 10dB less efficient than the mains.
2. Low-pass that conjugates baffle-step. Cabinets the size of my three Pi and four Pi mains tend to have baffle step in the 100Hz to 200Hz region. Instead of employing compensation in the crossover, we use the flanking subs to provide additional SPL as BSC.
3. Low-pass that mitigates higher-frequency room modes and SBIR from nearest boundaries. The worst anomaly usually comes from the wall directly behind the speakers, but the nearest side wall is sometimes objectionable too, as can be floor bounce. The modal region extends above 100Hz, but distant multisubs cannot be run this high without being localizeable. And the boundary interference notch from the wall behind the speakers seems to almost always be between 80Hz and 120Hz, probably because proximity to the wall behind the speakers is most convenient and/or popular in this range of a few feet. So this makes it useful to have a truncated array in the 80Hz-120Hz region, to counter the interference notches there. The flanking sub and the midwoofer in the mains form a simple two-element array.
I'm having difficulty finding a good high pass for the mains. Considering they'll be getting a full range signal I want to protect them. The tuning is around 50Hz. With a sub LPF of BU2 100Hz, is BU4/BU3 HPF at 50-60Hz a good start for the mains? After that - there's the playing around with the delays to get the sub and speaker to integrate, before even looking at EQ.
|
|
|
|
Goto Forum:
Current Time: Sun Nov 24 04:09:01 CST 2024
|
Posting Rules
Members
Search
Help
Register
Login
Get a T-Shirt!
, it'd be a challenge but has potential.
