@hifijim
When the distortion is below noise floor at 80dB, to see what’s happening at 80dB @1m, I often move my mic closer.
To get a better SNR I find it useful to measure at a closer distance. Eg. 50cm (+6dB) or 31.6cm (+10dB).
To see what tweeters are doing, one can move the mic right up to the dome eg. 2mm. On the other hand, when you move back to 3 times the length of the longest dimension of the cabinet; you get to see what the tweeter/ faceplate/ waveguide/ cabinet is doing. As you said, diffraction is destiny.
When you move the mic too close you run the risk of overloading your mic. So if need to measure what is happening for high SPL designs eg. 106dB/116dB or higher @ 1m; I sometimes take a ground plane measurement at 2m or 10m. Eg. 15-21+” bass horns/subwoofers etc
I sometimes joke that Australia’s (& America’s) comparative advantage is- we have space.
Necessity is the mother of all invention,
Sadly, we are a bit lazy with innovating…
When the distortion is below noise floor at 80dB, to see what’s happening at 80dB @1m, I often move my mic closer.
To get a better SNR I find it useful to measure at a closer distance. Eg. 50cm (+6dB) or 31.6cm (+10dB).
To see what tweeters are doing, one can move the mic right up to the dome eg. 2mm. On the other hand, when you move back to 3 times the length of the longest dimension of the cabinet; you get to see what the tweeter/ faceplate/ waveguide/ cabinet is doing. As you said, diffraction is destiny.
When you move the mic too close you run the risk of overloading your mic. So if need to measure what is happening for high SPL designs eg. 106dB/116dB or higher @ 1m; I sometimes take a ground plane measurement at 2m or 10m. Eg. 15-21+” bass horns/subwoofers etc
I sometimes joke that Australia’s (& America’s) comparative advantage is- we have space.
Necessity is the mother of all invention,
Sadly, we are a bit lazy with innovating…
Did you do a near field measurement of the woofer and investigate the burst decay plot? That should reveal any problematic standing waves inside the enclosure (and the driver).
The cone breaks up at 4kHz. That seems rather low to me, with rather down to earth drivers like the RS180 or AL170 the trouble starts at about 6k. I don’t see the breakup in your filtered response. The low pass slope of the mid is clean, which puzzled me.
The cone breaks up at 4kHz. That seems rather low to me, with rather down to earth drivers like the RS180 or AL170 the trouble starts at about 6k. I don’t see the breakup in your filtered response. The low pass slope of the mid is clean, which puzzled me.
Actually, I learnt from Andrew Jones that both mic in box and the use of a tweeter was in a 1972 Dick Small paper.
Cone break-up of these drivers can be compared to many third party measurements (incl datasheet).
Like HifiCompass, AudioXpress, Erins Audio Corner etc.
On the other hand, there aren't many other acoustic issues that will have something resonating at 4kHz, that is pretty high.
This woofer is so well documented at this point, that I would just go along with all those third party measurements.
In those other measurements, a peak around 4kHz can clearly been seen as well.
Even in the datasheet this 4kHz peak can be seen actually.
My thoughts as well. I don't have an interest, nor the capability, to measure and document the full performance of these two excellent drivers. I simply wanted to ensure that I had not done anything stupid which would diminish the performance. My distortion measurements and burst decay plots were to ensure I was getting something close to @HiFiCompass or AudioExpress. There is no way I can replicate their test methodology and results with the limited tools I have.
Burst decay plot of Near Field PTT6.5M
I use quite a bit of suppression on those upper level driver resonances. Here is my current set of DSP filters:
You also don't have to, because why would you if it's already done for you?
Unless something is broken, but I find that super rare for such a high quality product.
(plus you don't need to go fancy to figure that out as well)
Anyway, in this case this information was already available, so no need for discussion or debates.
I can see why you would think that since the 12" woofer is at y=0, z=0. In truth, the reference axis is midway between the midrange and tweeter... the tweeter is at y= +140, the mid is at y= -140.
When I properly account for the position of the woofer, including the time delay, it makes very little difference for the purpose of crossover design, and it complicates the situation when looking at phase. So I simplify the simulation by making the woofer at 0,0,0 with no time delay.
Here is how the filter would be actually implemented in the Hypex amps... this includes position and time delay. Since the woofer is 660 mm behind the mid-tweeter, the two upper drivers need to be delayed by 1934 us. This makes the phase plots goofy.
Hello,
Thanks for the tips and tools. I will give it a try. Could be a hobbyist for a long time and not learn the tricks that the pros use everyday. It is interesting how ringing the bell (enclosure) with your knuckles excites all of the resonances.
I often use a Impedance sweep and Peek Rub and Buzz plots to identify problem resonance issues. the plots are ran at 2.83Volts input to the driver.
If there is something loose it often will show up on the Impedance, FR and Peek Rub & Buzz plots. Funny how the software identifies resonance peeks as 2nd HD or 3rd HD on a distortion sweep.
Thanks DT
At this stage of simulation, my goal is to optimize the PIR (predicted in room) curve. At this point, the goal is a slope of -1 dB/octave from 100-10k. All of this is just to assess the feasibility.
When everything is assembled, I will take the final set of measurements. This will capture any interaction between the mid and tweeter, such as cavity resonances, and spillover of midrange frequencies into the waveguide [ @fluid alerted me to this phenomenon in post #103
https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7293406 ]
At that point, with final measurements, I will design the filter for maximally flat on-axis response. This is the easiest/best way to get the driver levels and delays dialed in, and the best way to do quality control checks to ensure that the intended filtered driver responses match the Vcad simulation and sum together as intended.
The maximally flat on-axis filter will probably sound a bit bright and forward, but it is a great starting point for the subjective voicing process.
@hifijim
I usually optimize for 15 degrees as a starting point. With a close look on 30 degrees followed by 0-degrees.
The reason has mostly to do with diffraction artifacts. When focusing on just on-axis, the overall tendency is to under compensate. Which very often results in (way) too much energy, in most cases around 800-3000Hz area (depending on width of baffle).
In general, the overall global response is far more important than little details. Especially when those details are dips.
When that is all set, one can start tweaking the details. In the end that's almost always subjective.
I usually optimize for 15 degrees as a starting point. With a close look on 30 degrees followed by 0-degrees.
The reason has mostly to do with diffraction artifacts. When focusing on just on-axis, the overall tendency is to under compensate. Which very often results in (way) too much energy, in most cases around 800-3000Hz area (depending on width of baffle).
In general, the overall global response is far more important than little details. Especially when those details are dips.
When that is all set, one can start tweaking the details. In the end that's almost always subjective.
I think this heavily depends on the system.
A speaker without waveguide often has a tendency to sound to bright because of the directivity miss match. So I always attenuate the tweeter a bit.
But the focus here is on "a bit", think about 2dB or so.
Also with a waveguide, I find sloping ever so slightly (keeping the diffraction issues into consideration) would give a better long term experience for most listeners.
Maybe less listening fatigue on the long run for some.
It also has a tendency to sound a little "warmer".
But this is what is extremely subjective about tweaking the sound.
Your hearing is not the same as mine.
Even my hearing sometimes drastically differs, depending on the mood, time of the day or well being.
I don't think I have ever created a system that was flat on-axis. Mostly because I focus on 15 degrees off-axis in general as well.
Last edited:
For the last 15 years, I have NEVER come across an high quality speaker where distortion was part of listening fatigue ever. Unless wrongly used.
Even the standard Seas prestige line (similar to any SB tweeter) is way beyond those limits. Guys like D&D have proven that very obviously.
The issue is just bad crossover design.
Biggest major advice I give to improve many production made speakers.
Either cheap OR expensive.
Even the standard Seas prestige line (similar to any SB tweeter) is way beyond those limits. Guys like D&D have proven that very obviously.
The issue is just bad crossover design.
Biggest major advice I give to improve many production made speakers.
Either cheap OR expensive.
Hello All,
For this entire thread we have been looking at and talking about smoothly sloping directivity curves.
Now with the SPK16 speaker we see a very flat, smooth and Constant Directivity curve.
For the listening experience with your eyes closed and focused on the sound what is the difference between Constant Directivity and smooth gently sloping directivity?
My big speakers use JBL D2 compression drivers and M2 Constant Directivity wave guides at +-60 degrees coverage. Talk about a huge sweet spot in my moderate size listening room I expect the SPK16's to be much the same.
Is it a preference thing like the slope of the power curve?
Thanks DT
For this entire thread we have been looking at and talking about smoothly sloping directivity curves.
Now with the SPK16 speaker we see a very flat, smooth and Constant Directivity curve.
For the listening experience with your eyes closed and focused on the sound what is the difference between Constant Directivity and smooth gently sloping directivity?
My big speakers use JBL D2 compression drivers and M2 Constant Directivity wave guides at +-60 degrees coverage. Talk about a huge sweet spot in my moderate size listening room I expect the SPK16's to be much the same.
Is it a preference thing like the slope of the power curve?
Thanks DT
Speaking in generalities I think see your point of view.
Crossover design is very important, and one of the last things that can be changed after the enclosure/cabinet/(or lack thereof) and drivers have been decided on.
But for the naysayers that say distortion doesn’t matter; well both linear and non-linear distortion matters to the point that they also affect the final SPL frequency and directivity.
For instance, a certain standard line 6.5” hard cone mid-driver has a fierce resonances at, 7, 8.5, 10 and 13Khz on axis. But guess what happens when one moves off axis? These resonances occur at different angles- So one cannot precisely notch them all out, even with 4 precisely tuned notched filters with DSP- the resonances will show their head at different frequencies at different off axis angles.
The alternative is to use a steep crossover point and low crossover point, to remove the hash and avoid a directivity mismatch with the tweeter. But then that stresses out the tweeter. And the tweeter can’t play that loud. (Non-linear distortion)
For drivers who are well damped and don’t have sharp resonances, for some of them the amplitude response falls off a cliff. Passive crossovers can only subtract; so there’s just no way they can do acoustic LR2 which allows a smooth power response. So you resort to using a standard acoustic LR4; but now the power response shows the typical wide narrow wide problem.
for decades people weren’t looking at what happens at wide off axis angles.
But In an indoor environment, all 360 degrees matter. In fact 4D matters; timing too.
Last edited: