HiFiCompass provided the following details about their test conditions (here, click "Measurement setup description"):Do you know what conditions HiFiCompass took those measurements? Notably, what baffle size?
HiFiCompass provided the following details about their test conditions (here, click "Measurement setup description"):
Test baffle size: 1650x850 mm
Measuring equipment:
- measuring microphone Brüel & Kjær 4133
- microphone preamplifier Brüel & Kjær 2639
- preamplifier power supply Brüel & Kjær 2804
- sound card EMU-0202USB
- DSP Xilica XM2040
- low THD power amplifier
If this derails the thread PM me and I'll edit, but I thought some people might be interested in what you can achieve if you have a lot of room for ports, say in the bottom 2/3rds of floorstander, you can suppress about 20 db of port noise using resonance traps - as has been demonstrated in the DIYaudio thread.
Here are my rudimentary results from one nights tinkering. First pic was using a 60mm port for a baseline (didn't record length, you could work it out, looks tuned a bit too low). The second pic and all plumbing/port tinkering thereafter was done with 75mm ID piping. The FR's are not directly comparable, but you can compare the resonant peaks of a simple straight port vs. the spaghetti port I ended up with. The traps are stuffed quite generously with pillow stuffing.
Purifi long stroke woofer 6.5" in a 15L test box (as per directiva). Cabinet walls lined with foam. Port installed to base of box as if it were leaving the woofer chamber and running through the leg of a floorstander or stand i.e. there is no port volume within the woofer cabinet.
…
Hmm, no I thought 30L the woofer would end up being too peaky at Fb, the upper limit of this woofer being like 22L imo. 15L seemed like the sweet spot for tuning in a bass reflex cab. Would still be interesting to see the resonance effect, but I don't have any cabinets near 30L at the moment.
Rick, this kind of tuning rarely sound good. Slow gradual rolloff is what you should strive for. Acc to my sims with Purifi drivers, ~15l seems optimal.
It is not about placement in the room, though it could mitigate it. Ok try it, trials and errors....
Just wanted to express that any speaker designer would most most likely stay away from this tuning.
How about no ports because they're stupid and make the phase messy and makes subwoofer integration a nuance?
What my intention was was to build a segregated floor stander cabinet, that is a cabinet that looks like a normal floor stander but is separated by an internal partition . 15L section at the top with the woofer and tweeter, and the port + traps occupies the space in the bottom. So it’s still a 15L speaker, just with a floor standing cabinet.Think I may have misunderstood as the smallest floorstander cabinet I have is about 30L. You suggested just using the space to fit the port with the traps I gather?
If so, seems useful to compare anyway. With a 3 inch diameter port, you get an f3 of 45 Hz in a 30L cabinet. Here is a look if you modeled in VCAD...
View attachment 151250
As best as I can judge from looking at Rick’s pictures of the baffle, the center of the woofer is about 8.3” from the top edge, and about 7.8” from the bottom edge. (Please pardon my use of archaic units of measure.) For the strongest, 1st peak in the edge diffraction, the distance from the center of the driver to the baffle edge is 1/2 wavelength, such that the resulting 180 degrees of phase change, combined with the 180 degrees you get with a “soft” reflection at the edge, yields 360 degrees of phase offset vs. the wave direct from the driver. The two wavelengths (top and bottom edges) are 16.6” and 15.6”. The corresponding frequencies (using v = 1125 ft/s which means that 1 foot corresponds to 1.125 kHz) will be 12/16.6 x 1125 kHz = 813 Hz, and 12/15.6 x 1125 Hz = 865 Hz. These two frequencies are so close together that we will anticipate a fairly broad peak at about 840 Hz, allowing for the inaccuracy due to the crude way I obtained the two distances. The 1st dip will occur one octave higher, at about 1.7 kHz, and the 2nd peak will be at about 2.5 kHz.
When I look at the 2nd graph in post #660, these peaks and dips are unmistakable in the red trace (Amir NFS), but not in the blue graph (Rick gated). The same in post #672, which I assume is the same graph as in #660. Thus, the implied hypothesis is that the difference in the two response curves is due to one of them (Amir NFS) fully revealing the effect of baffle edge diffraction for the top and bottom edges whereas this effect is masked in the other response curve (Rick gated). Possibly there is something about the rig that Rick is using that allows the on-axis measurement to pick up an off-axis reflection, possibly a ceiling or floor reflection arriving within the gating window. Or, the explanation for why the two measurements are different in this respect might be that the woofers were oriented differently in the speaker that Amir measured vs. the one that Rick measured. If perchance this happens to be the true explanation, it then seems likely that in the speaker that Rick measured, the woofer was oriented such that the proud regions of the surround were at the North, East, South, and West compass points whereas in the speaker that Amir measured, these regions of the surround were not at these compass points (shifted from these compass points by 60 degrees I expect). It will likely be useful to confirm whether the two speakers were in fact different in this respect. If it turns out that the woofer orientations were the same, then my guess would be that in Rick’s setup, the diffraction effect was present and that the reason it doesn’t show up is that he was picking up an off-axis reflection.
While it will be useful to confirm whether the two speakers were in fact different in this respect, I think that it is indicated (regardless) to perform reflection-free measurements of the vertically on-axis and off-axis responses (if it hasn't already been done). It should only be necessary to do this for one edge, either at vertical angles above the top of the speaker or else below the bottom edge. As we have previously observed and noted, there is a conspicuous signature of baffle edge diffraction, which is evident when you look simultaneously at the on-axis and off-axis response curves. This signature occurs due to the fact that the affect on the on-axis response and the affect on the off-axis response are mutually opposite. At a diffraction dip in the on-axis response, there will be a peak in the off-axis response, which may be observed in a typical style of plot as a sort of pinch in the group of curves that includes the 0-degree off-axis response (i.e., the on-axis response). At a frequency corresponding to a diffraction peak in the on-axis response, there will be a spreading effect (instead of a pinch). I think that this may in fact be evident in the first two plots in post #650. (This signature effect - the pinch and spread at wavelengths separated by full octaves - is more plainly evident to me when the off-axis responses have not been normalized to the on-axis response.)
For anyone who might not be following what I'm saying, I ask that you look at the second graph in post #660, the red trace in particular, and take note of the frequency locations of the major peaks and dips you see in the red trace, then go back and look at the first two plots in post #560, with those same frequencies in mind. If the frequencies for the peaks and dips don't match up exactly with the frequencies I identified, the reason is almost certainly attributable to error in my crude measurements of the baffle, which I took off the screen from Rick's pictures.
If it turns out that this diffraction effect is appreciable and that the reason that it does not show up in Rick’s measurement is that the woofers were oriented differently, then the evident conclusion is that with this speaker it is important that the woofer be oriented such that the proud regions of the surround are at the N, E, S, and W compass points, such that the surround serves to prevent the baffle edges from being strongly illuminated. This is the more fortunate possibility, because the solution is trivial. The less fortunate possibility is that the reason for the difference in the two measurements has nothing to do with the woofer orientation, and is due to off-axis reflections being picked up in Rick's measurements. In this case, it may be that the true on-axis response, measured free of reflections, will always look like the response that Amir obtained, i.e., regardless of the woofer’s orientation. In this less fortunate possibility, we may have gained knowledge as to the explanation for the difference in the two measurements, yet be left in want of an easy solution.
I hope that this is helpful and that I haven’t just created confusion.
One strong possibility is what was suggested previously by McFly, which is to say, the woofers are oriented differently in the two speakers, such that the top and bottom edges of the baffle were in the shadow of the woofer surround with the speaker that Rick measured, but not with the speaker that Amir measured.
Another strong possibility is that this diffraction effect is present in the speaker that Rick measured but that it didn't show up in the measurement, the presumptive reason being that his on-axis measurement was picking up a reflection of the vertically off-axis response, likely from the floor or ceiling.
As explained in post#667 and post#675, the protruding surrond of the woofer as a secondary sound source affects the axial frequency response, with the first minimum at 5.3kHz and the first maximum at 2.6kHz. So a direct influence can be excluded - we all agree on that.
This means that it is only a question of whether the protruding surround "shadows" - i.e. attenuates - the sound propagation to the vertical edges of the loudspeaker cabinet.
View attachment 151420
Source link
If this were the case, the secondary sound sources produced by the vertical cabinet edges would weaken.
This means that the first minimum affecting the axial frequency response around 1.7kHz, caused by the vertical edge diffraction, would be less pronounced - we all agree on that too.
In Amir's measurement, the protruding surround, was therefore not in the way - in the case described by @MrPeabody. The simulation did not consider the protruding surround either.
Therefore Amir's on-axis measurement should follow the frequency response of the simulation better than Rick's measurement does.
In green you can see the simulated on-axis frequency response, which deviates greatly overall because the Purifi driver has a paper cone and this breaks down into eigenmodes early and thus deviates from the ideal.
However, we are only looking at the course of the frequency response, i.e. where are the humps and dips.
View attachment 151449
We can see a flat dip in the frequency response around 1.5kHz in the simulation and a hump around 2kHz followed by a wide dip up to about 3.2kHz. Rick's measurement shows approximately such a curve progression, Amir's NFS measurement not at all - so exactly the opposite as expected.
Now let's look at the vertical frequency responses of Amir's NFS measurements and the simulation.
I smoothed Amir's measurements with 1/3oct, frequency responses are normalized to on-axis FR.
Vertical up
View attachment 151452
Vertical down
View attachment 151462
Vertically upwards, the frequency response curves of NFS and simulation fit together quite well.
Vertically downwards, however, it looks bad, especially at larger angles.
For those who have doubts about the accuracy of the simulation, here is the comparison of the horizontal NFS measurements with the simulation for the tweeter with waveguide (measurements with 1/3oct smoothing, frequency responses are normalized to on-axis FR):
View attachment 151463
What the hell does it all mean?
Well, that's the million dollar question. Rick's measurements seem to be okay, anyway.
Spontaneously I would have guessed, but I know that Amir doesn't make such a mistake, that the measurement platform was larger than the base area of the speaker. This would then explain why measurement and simulation fit better vertically upwards than vertically downwards.
I now believe differences are more likely to be in variations to the woofers due to production date. Purifi have tweaked the woofer over time making it slightly more sensitive. I compared my woofer in a similar baffle to ricks measurements and my woofers a different response again.