0 Members and 1 Guest are viewing this topic. Read 2109 times.
...wider baffles...
This is exactly what baffle step correction is for... you'd have to increase the baffle size to stop those frequencies from going around to the back. I doubt you're seeing diffraction, it's just baffle size vs frequency for the most part. As far as summing or canceling on the back side, why does it matter? It's all going to be delayed reflections. What window gate are you using for measurements? Might want to think about that. 500hz only takes 6.5" of baffle/cone to become front directional only so it makes perfect since why you're seeing an increase after 400-500hz area as everything is moving forward at that point. If you're measuring up in the air you shouldn't have floor bounce. Where are the lowers crossed?
The shape of the baffle largely controls the nature of the 2π to 4π transition.At the frequencies you ar ehaving issues with the round-over (a chamfer can affect a larger effective diameter in the same space) needs to be in the multiple inhes.dave
Finding this question from Danny about combating baffle diffraction in low mid-range and Folsom’s reply then Dave posts the baffle shape vs response text book figures I thought of this image circulating in the HiFi press. Side view of the recent refresh of B&W’s 801.Certainly apparent which route B&W took. Usually I find spherical speakers if that are bigger than little 4” desktops or Fujitsu Ten Eclipse when scaled up rather goofy looking. As some of the best industrial designers in HiFi B&W have their sphere without the oddness of a Cabasse.Back to Danny’s issue, I note in Dave’s posting the “TRUNCATED PYRAMID ON PARALLELOPIPED” is second best after the sphere. That textbook must be in B&W’s library because that truncated pyramid is very reminiscent of the first couple of generations of 801/802 midrange ‘heads’. I expect the response trough and ripples and any relief from them are related to how large the round overs or facets are vs wavelengths of interest. Danny’s measurements revealed his issue between 200 to 500 Hz. This chart reports those are wave lengths of 2 to 6 feet. Audio frequency vs. wavelength chartGiven the wavelengths involved I would tend to agree with Folsom “just baffle size vs frequency for the most part.” Facets and shapes as Dave and the designers at B&W use I would characterize as icing on the cake. Not that us audiophiles don’t like icing. And sprinkles. And buttercream roses on our cakes I mean speakers. Stupid analogy, now I want some cake.While in the experimental phase of design R&D you could try wider baffles for the cost of some MDF (or foam core board?). Here to most successful somewhat recent example being the Sonus Faber Stradivari Homage. Troels Gravesen dove deep into that rabbit hole in his clone.http://www.troelsgravesen.dk/PMS.htm
Various unrelated tidbits: As the wavelength increases with falling frequency, it starts to diffract around the box edges when the wavelength equals or larger than the baffle width. All free standing speakers (that sound good) have a compensation EQ filter built into the crossover to correct this diffraction rolloff and create a flat response down into bass frequencies.Likewise, increasing the width of the baffle will lower the diffraction rolloff frequency. But you still need EQ to make flat response into lower frequencies than the natural rolloff. But you'll need less correction, which will preserve some system sensitivity.The trend is toward narrow baffles for their appearance, and for enhancing audiophile attributes like the faux imaging illusion caused by diffracting higher frequencies into the room acoustic. There was a time when wider baffles were popular because they better supported the vocal range of the pop singers that sold the audio gear to normies. Now only audiophiles buy quality audio gear and they want imaging, even if it's fake from room distortion, so narrow baffles are popular.More technical details about diffraction compensation:https://www.sound-au.com/bafflestep.htmhttps://www.trueaudio.com/st_diff1.htmHow to make the diffraction compensation in your crossover? When you create your frd file for use in your crossover simulator, just measure the drivers in the cabinet in which they will be used. This will capture the diffraction loss, which you can then compensate for in your simulated crossover design. Alternatively, if you don't use a simulator, then use the formula on the trueaudio link article to calculate the center frequency of your baffle step compensation filter manually and apply the necessary EQ in your crossover designThe magnitude of baffle step compensation is always between 0 and 6dB. It depends on the room size, woofer size, baffle size, proximity to front wall, tonal preference, etc. 3-4dB of compensation is common in typical commercial speakers with narrow baffle. There are two variables to the compensation: the center frequency and the gain. The freq is determined by the baffle size. The gain is mostly determined by your taste. You may want more than 6dB bass boost, but that is not due to the diffraction loss, which can never be more than 6dB. All free standing commercial box loudspeakers have baffle step correction. Only in-wall speakers (and headphones) don't need it.In a passive crossover there is no way to "boost" the bass. So we accomplish the correction by cutting the high frequencies to get the same affect. This results in a loss of overall system sensitivity. So a woofer rated 91dB sensitivity, when properly compensated for edge diffraction may yield only 87dB and the tweeter must be padded down to 87dB to remain in balance with the woofer.A wider baffle should need less diffraction correction gain, and it will be centered at lower frequency. This makes vocal band sound better by shifting any phase distortion from the diffraction below the vocals and less BSC gain means less loss of sensitivity. If you're more of an audiophile type then you may prefer narrow baffle with more midrange room echo. If you're more of a music lover and want denser more relaxed male vocals or higher system sensitivity then wider baffle might be for you. Rounding the edges of a box to smooth diffraction into lower frequencies requires surprisingly large radius. The diffraction smoothing effect diminishes down to about 1/4 wavelength. So a 1 inch radius edge only effect diffraction down to about 3400Hz. A 4" radius effects diffraction down to about 850Hz. A 2 foot sphere will affect diffraction down to 280Hz. But this is only the 1/4 wavelength where the effect falls off. The strongest diffraction mitigation radius is at one wavelength, which corresponds to the frequency at which the diffraction begins. Above that the baffle is wide enough to support the whole wavelength and not diffract.
...have a compensation EQ filter built into the crossover to correct this diffraction rolloff and create a flat response down into bass frequencies.
There are more ways than that to skin the cat. https://www.t-linespeakers.org/tech/bafflestep/intro-bds.htmlA blunt EQ (as implied) with do nothing to tame the ripples that one gets in the transition as illustrated by the Olsen chart i posted. No EQ will.Everything is a compromise. Also keep in mind, that a BSC filter as suggested that raises the on-axis response to approach flat, also raises the room response above flat. The ROOM. Typically the room starts to dominate below about 250-350 Hz. OP's FR issues are certainly creeping into that territory.I would also dispute the comments on imaging. Some may well do as suggested, but it is a goal i seek with my loudspeaker designs, which have very low diffraction, the better ones both the edge manifestation of diffraction (of which Ric refers) and the space transition lower down.If the information is on the recording, a very good, solid 3D soundstage/image illusion is created.dave
Great info here. I have already planned for the baffle step circuit for this speaker which will compensate for the upper mids being louder. It's actually a pretty simple solution and is generally just an inductor and resistor in parallel with the mid-woofer. What it won't do is deal with the peaks and valleys between 200Hz and 500Hz, which are caused by the effects of the thin baffle (as you eloquently explained). The reality is that most rooms will cause more extreme phase cancellation and nodes so it's probably fine. I think the real solution would be a wider baffle that correlates with my crossover frequency between the mid-woofer and the Low woofers, but that would be a 28" wide baffle and I don't want the speakers to look like that.
Have you tried a nearfield or gated measurement?
As Dave said, there's always compromises, and new solutions!Instead of adding an additional coil with parallel resistor for the BSC, you can often use a larger coil on the woofer low pass filter to achieve a combined low pass and BSC filter. You want to minimize adding resistance to bass drivers as it will reduce electrical damping of the driver motion, reducing detail and slam. Added resistance also reduces damping of upper drivers but it's more benign, but still good idea to consider the drivers' natural sensitivities with BSC in mind from the beginning.
It makes me wonder why 2.5 way designs aren't more popular. They certainly are an elegant solution to BSC issues.
