The full spectrum aspect of white is why I suggested using pink isn't a good idea. Pink mean checking that the speaker is producing a true pink spectrum.
Dayton mention moving the exciter around by hand and listening to the tone to find the best position. That needs pressure and isn't the same as they are "supposed" to be used glued on. The pressure will deflect the panel. When simply glued on the excitation forces work against the mass of part of the body of the exciter. I've been wondering about some sort of mechanical hand to move the exciter around but suspect a better arrangement would be to move the panel around which I suspect will finish up edge constrained in a frame. Not keen on panels needing to be off the wall as well. That may fit in.
I like the mount on a disk and fit in a hole in the panel - which I think was your idea??? Not sure.
1. Difference between pink noise and white noise, and it's not just an arbitrary thumb-suck:
White noise consists of equal energy per frequency and pink noise consists of equal energy per octave.
https://audiouniversityonline.com/white-noise-vs-pink-noise/
2.Moving the exciter around by hand:
The (very light) hand pressure, a few grams?, against an exciter on a horizontally-mounted panel presupposes that you are using a small enough amplitude signal that the coil does not bounce against the panel (check your distortion readings!) Higher signal-levels mean that you will have to press harder against the exciter to prevent it bouncing.
But having said that, I have never seen a difference in frequency response, found by manually positioning the exciter, and the response when the exciter is glued in place. I'm sure there would be more difference caused by atmospheric temperature and humidity than by the microscopic difference that a few extra grams of pressure would.
3. The disk mounted in a hole in the panel is also not arbitrary. Different combinations of materials will need different compliances between the connecting legs between the materials. A lot of careful measuring must be done to get it just right.
Best rethink the use of polyurethane glue. It's moisture cured and bubbles as it cures, forcing surfaces apart unless they're very well clamped.
Gorilla makes/markets glues of several types including modified PVA with varying weather resistance
Eucy
A disc the same diameter as the magnet case, but my exciters are just simple cylinders so it's easy to do. I don't think it needs to be complicated, it's just a damping pad
Eucy
My goal is to build a pair of 3-way tower speakers - here is a part of my journey so far.
I have set up and taken spl readings of these different monolithic materials with no cuts or special shapes, damping, or composite layering.
I have a stack of 6 Dayton exciters that I have burned out, or damaged over the last few weeks.
I have 5 functional exciters remaining, I now use a shoe horn to pry the foot off the panel, instead of twisting from the top, and i have a little 1amp fuse on the red wire to catch high currents.
I reviewed these materials to get a snapshot of what their strengths were.
TLDR: I wouldn't recommend a monolithic material for anyone wanting a full range flat frequency response with 80db+ SPL.
for the Americans who need a quick comparison of the mm numbers from the results below:
1/4th inch is ~5mm
1/8th inch is ~3.7mm
1/16th inch is ~1.6mm
In the size of 30x30cm(12x12inch) with free edges I have tested, with brief notes, and thicknesses:
I have set up and taken spl readings of these different monolithic materials with no cuts or special shapes, damping, or composite layering.
I have a stack of 6 Dayton exciters that I have burned out, or damaged over the last few weeks.
I have 5 functional exciters remaining, I now use a shoe horn to pry the foot off the panel, instead of twisting from the top, and i have a little 1amp fuse on the red wire to catch high currents.
I reviewed these materials to get a snapshot of what their strengths were.
TLDR: I wouldn't recommend a monolithic material for anyone wanting a full range flat frequency response with 80db+ SPL.
for the Americans who need a quick comparison of the mm numbers from the results below:
1/4th inch is ~5mm
1/8th inch is ~3.7mm
1/16th inch is ~1.6mm
In the size of 30x30cm(12x12inch) with free edges I have tested, with brief notes, and thicknesses:
- polystyrene low density - loud, high distortion
- polystyrene high density (like what you get for holding a flat-screen tv in a box) - loud, peaky resonances
- XPS (the pink panel) - loud, peaky resonances
- Polyethylene open cell foam (1mm, 2mm, 15mm) - quiet, flat bass response
- Polycarbonate core flute board (5mm) - loud, peaky resonances
- Carbon fiber twill sheet (0.5mm) - loud, shrill resonances
- Picture frame UV glass (3mm) - medium, shrill high frequency resonances
- Neoprene closed cell (1mm, 2mm, 3mm) - quiet, flat bass response, upper frequency around 1000hz (this is much better when held taught, but thats not what i was measuring yet)
- Neoprene solid rubber (1mm, 2mm) - medium, flat bass response, upper frequency around 3000hz
- ceiling acoustic panel - medium, the loss of decibels as it reaches the higher frequencies was inexcusable, and it also is heavy.
- Cardboard - corrugated (3mm, 4mm) - loud, great clarity and flat freq between 100 and 1500hz
- Cardboard - honeycomb core (20mm) - loud, great clarity and increasing SPL for freq response from 80 to 5000, drops after that
- Cardboard - solid (2mm,3mm,4mm) - loud, lower spl from 80-1500hz, then increases and is flat to 20khz
- Vinyl - Flooring board (15mm) - quiet, too heavy
- Cutting board plastic (2mm) - loud, spl 90 and flat between 150hz and 2000hz, then dips to 70 decibels between 2000-3000hz, then jumps back to 80 decibels and is flat from 3-20khz (a small dip of 5db 6khz to 6.5khz)
- Acrylic (1mm, 2mm, 3mm) - loud to quiet based on thickness, the SPL average is OK, freq response average drops by 10 decibels between 1k and 20khz, but this is the material with the most hedgehog spiking (30db up and down) for the length of the graph.
- Walnut wood (0.5mm, 3mm) - 3mm is medium loudness, 0.5mm is loud, the 0.5 is good response but full distortion until over 1000hz (its my favorite material when its used with other things though), the 3mm has a flat response in two ranges 150-1000hz anf 6000-16000hz, and massive dips between.
- Basswood (1.5mm, 3mm) - loud, perfectly flat response from 80 to 580hz at SPL 80db, 580-700hz drops by 30db, then from 700hz to 6000hz the SPL jumps to 90db and is flat??!, then another flat plateau from 16khz to 20khz, but at spl 70db
- Balsa wood (1mm, 2mm) - loud, SPL 90db from 100-1000hz, then has a linear drop in SPL from 1000-6000 from 90 to 70 decibels.
- MDF (3mm) - loud, sloped off too fast
- Fiberoid Fish Paper (very strong fibrous and hard paper) (0.5mm) - loud, flat from 500-10000hz, not too spikey, much distortion below 600hz
- Trivek Paper (un-rippable thin light flexible paper) (0.2mm) - loud, distorted until 1khz, then a double hump of a graph with peaks at 1.8khz and 8khz
- Aluminum plate (0.5mm, 0.1mm) - loud, has a graph like a fuzzy caterpillar, the median has a gentle slope down from low to high frequencies, but it has so many resonant coincidences it sounds like a robot is whispering in your ear.
- Aluminum foam sheet (5mm, 10mm) - loud, 80-decibel average (spikey but stays at that spl) from 1k to 10khz, drops off by 10 decibels before and after that band. (great!)
- Titanium plate (0.5mm) - medium, spikey with peaks and troughs all over
- Brass plate (0.5mm) - loud, spike at 1.8khz and from 7-8khz, spikey graph
I have been building composites and metamaterials (components wiithin the panel), and have tested through hundreds of those.
I have a list of best results which I will share when I get another few hours spare!
As a TLDR and teaser for that post.
For composites these materials layer well
outside materials that are good:
filling layer:
Binding agents I've used:
Some things end up opposite of what you expect when you laminate them together:
Weight mitigation:
You can use a small part of a denser material (garolite, aluminum, carbon fiber, cardboard for example) on a surface to get a portion of its sound profile without covering the whole panel and weighing it down but to get the most audible resonance profiles from the material, it needs to be where the exciter is attached to, as in exciter then aluminium then your composite panel.
For example, a normal panel might be:
20x30cm panel made from fibroid fish paper then 3mm aramid honeycomb then walnut veneer sealed together using rubber cement.
To add some of the higher frequency resonances from garolite, or some of the flat midtones of basswood, you can make a flat disk (5-10cm), or a tri-armed brace (pictures provided). Attach that to the panel, and then attach the exciter to that.
example: Tri-arm of aluminum on walnut veneer on basswood with epoxy
I have a list of best results which I will share when I get another few hours spare!
As a TLDR and teaser for that post.
For composites these materials layer well
outside materials that are good:
- basswood - 2mm
- veneer - single ply
- paper - stronger the better, like trivek and fibroid fish paper, and tracing papers too (A+ for lightness, flexibility, and no compression)
- cardboard - 2mm thick
- garolite - 1mm
- fiberglass - 1ply
- carbon fiber - 0.5mm
filling layer:
- honeycomb cardboard - 2cm thick
- EVA foam rolls - 1mm and 2mm
- EVA non-foam rolls - 1mm and 2mm
- polyethylene open cell foam - 1mm and 2mm
- aramid honeycomb - 3mm
- aluminium honeycomb - 4mm and 10mm
Binding agents I've used:
- 2 part pouring epoxy and thin fiberglass
- 2 part epoxy wood adhesive
- 2 part epoxy metal adhesive
- rubber cement
- cyanoacetate
- weld-bond
- expanding foam crack filler
Some things end up opposite of what you expect when you laminate them together:
- Trivek paper (super floppy) glued to 2mm non-foam EVA (super floppy) you get a hard plank.
- basswood (hard) on the outside and polyethylene foam sheet (soft) you get a more flexible plank than I expected.
Weight mitigation:
You can use a small part of a denser material (garolite, aluminum, carbon fiber, cardboard for example) on a surface to get a portion of its sound profile without covering the whole panel and weighing it down but to get the most audible resonance profiles from the material, it needs to be where the exciter is attached to, as in exciter then aluminium then your composite panel.
For example, a normal panel might be:
20x30cm panel made from fibroid fish paper then 3mm aramid honeycomb then walnut veneer sealed together using rubber cement.
To add some of the higher frequency resonances from garolite, or some of the flat midtones of basswood, you can make a flat disk (5-10cm), or a tri-armed brace (pictures provided). Attach that to the panel, and then attach the exciter to that.
example: Tri-arm of aluminum on walnut veneer on basswood with epoxy
What a super and valuable effort... Well done.. look forward to more info. 👍👍👏
Eucy
Does this "composite board" have video or sound (Youtube, FB)? Is there an SPL curve?
Thank you so much for the data you shared
These data are very complete and referential
@xsuper9988 thanks for asking.
No I’ve not done any YouTube videos but I could give it a try.
I have an SPL curve the composite and it’s sub components.
But for examples sake it’s probably cleaner if I make a new panel tomorrow and do measurements/photos/videos as I do it.
That photo of the triarmed thingie above (does that shape have a name?) was part of my tweeter quest. Sadly its strength is not as a tweeter. But the aluminum on the veneer did have a great sound and volume, however it needed to have its edges fixed to stop the distortion from the thin wood flapping at lower frequencies. Attaching it to the basswood panel like that allowed it to emit sound from the basswood with any extra energy it had.
I totally got too messy with the epoxy on that specific experiment and so the resulting sound is more influenced by that than it is meant to be.
Ill do a fresh one with carbon fiber (2 swelling peaks and dips) and cardboard (spl starts high and tapers down as freq increases with a couple of plateaus).
It’s 6am here, so once I shoo my family out the door to school and work by 9 I’ll go do the experiment.
No I’ve not done any YouTube videos but I could give it a try.
I have an SPL curve the composite and it’s sub components.
But for examples sake it’s probably cleaner if I make a new panel tomorrow and do measurements/photos/videos as I do it.
That photo of the triarmed thingie above (does that shape have a name?) was part of my tweeter quest. Sadly its strength is not as a tweeter. But the aluminum on the veneer did have a great sound and volume, however it needed to have its edges fixed to stop the distortion from the thin wood flapping at lower frequencies. Attaching it to the basswood panel like that allowed it to emit sound from the basswood with any extra energy it had.
I totally got too messy with the epoxy on that specific experiment and so the resulting sound is more influenced by that than it is meant to be.
Ill do a fresh one with carbon fiber (2 swelling peaks and dips) and cardboard (spl starts high and tapers down as freq increases with a couple of plateaus).
It’s 6am here, so once I shoo my family out the door to school and work by 9 I’ll go do the experiment.
Hi Jamie
I'm wondering about boundary effects/interferences brought about by all the interfaces involved in these multilayer/multi-join panels.
We know that density and thickness changes cause issues with reflection, velocity differentials etc within the panel..Seems like you have a rich mix of all of these here.
And I still can't get to grips with how the waves are transferred evenly past the slots to the outer surfaces of the panels in the case of the ortho planar springs.
Any thoughts/explanations?
Eucy
I'm wondering about boundary effects/interferences brought about by all the interfaces involved in these multilayer/multi-join panels.
We know that density and thickness changes cause issues with reflection, velocity differentials etc within the panel..Seems like you have a rich mix of all of these here.
And I still can't get to grips with how the waves are transferred evenly past the slots to the outer surfaces of the panels in the case of the ortho planar springs.
Any thoughts/explanations?
Eucy
@Eucyblues99 well yes it’s kinda crazy right?
The composite lamination panels have all of the layers vibrated from the same point, even if each of those layers transfers the waves at different speeds, when they are layered you end up seeing an averaging effect.
This is noted in one of the papers I posted a link to recently. So that’s part of the effect I suppose.
From observation, A portion of why the orthoplanar spring works to flatten the response is because it is reducing the transfer of the low hz frequencies by converting it to heat via the spring much like damping does. Thus it puts the lower frequencies on a similar spl to the higher frequencies.
It’s possibly obvious But I don’t have a background in audio engineering or physics. So I have few fundamentals to work from. I am reading a lot of research papers right now to catch up with the state of the art and help guide my experiments.
So I don’t have an answer for how it’s all working specifically.
I can only describe what I am seeing.
So far most of these discoveries have been a surprise.
I have a umik-1 for measuring and I have a 6 way audio switcher with an exciter on a cable connected to each of the 6 channels. And so I can connect and disconnect several things at once, listen/measure to them individually or together. So i guess it made it easier to notice the changes between runs and therefore patterns that came up.
The composite lamination panels have all of the layers vibrated from the same point, even if each of those layers transfers the waves at different speeds, when they are layered you end up seeing an averaging effect.
This is noted in one of the papers I posted a link to recently. So that’s part of the effect I suppose.
From observation, A portion of why the orthoplanar spring works to flatten the response is because it is reducing the transfer of the low hz frequencies by converting it to heat via the spring much like damping does. Thus it puts the lower frequencies on a similar spl to the higher frequencies.
It’s possibly obvious But I don’t have a background in audio engineering or physics. So I have few fundamentals to work from. I am reading a lot of research papers right now to catch up with the state of the art and help guide my experiments.
So I don’t have an answer for how it’s all working specifically.
I can only describe what I am seeing.
So far most of these discoveries have been a surprise.
I have a umik-1 for measuring and I have a 6 way audio switcher with an exciter on a cable connected to each of the 6 channels. And so I can connect and disconnect several things at once, listen/measure to them individually or together. So i guess it made it easier to notice the changes between runs and therefore patterns that came up.
Concerning the use of white or pink noise that was recently discussed, I don't think one other is "better". For me, the important thing to understand is what to expect the analysis curve to look like like with each of them. If you are plotting frequency response using a spectrum analyzer, white noise should appear as a flat horizontal line, while pink noise should have a response that falls with frequency. On the other hand, if you are plotting frequency response using an RTA, pink noise should result in a flat line, while white noise should have a response that rises with frequency. When I'm doing tests using noise, it's simpler to look for a flat line than to look for the correct slope, so I tend to use pink noise with an RTA analysis mode.
Below is a snip from the REW help file:
Below is a snip from the REW help file:
Eric, I find that I can use white PN, live, and get exactly the same FR as a straight forward log sweep (given the same smoothing.)
I generally use pink noise for bandwidth-limited sub-woofer tests, or for background noise masking.
How did you manage to get a flat response from pink noise in speaker tests?
I generally use pink noise for bandwidth-limited sub-woofer tests, or for background noise masking.
How did you manage to get a flat response from pink noise in speaker tests?
I have a sticker on my pc screen display with these written on it:
Linear sweeps and white noise:
flat on FFT
rising at 3dB/oct on RTA
Log sweeps and pink noise:
dropping at 3dB/oct on FFT
flat on RTA
FFT: bins of constant width in frequency
RTA: bins of constant width in octave
White noise: equal energy per Hz bandwidth
Pink noise: equal energy per octave bandwidth
George
Linear sweeps and white noise:
flat on FFT
rising at 3dB/oct on RTA
Log sweeps and pink noise:
dropping at 3dB/oct on FFT
flat on RTA
FFT: bins of constant width in frequency
RTA: bins of constant width in octave
White noise: equal energy per Hz bandwidth
Pink noise: equal energy per octave bandwidth
George
3-way??
I can only assume you want to run the DML panel from, say, 100hz on the low end up to, say, 7khz on the top.
I world further assume that you want to cross over to a cone sub on the low end and to a planar on the high end.
DML action is of no use for subs, so a nice cone in a nice box would work well.
And open-back planar super-tweeter might retain some of the advantages of DML on the high end.
One of the many advantages of a DML panel is that you don't really need cross-overs for a wide-band response, because a single driver on a properly-designed panel should give you massive bandwidth.
I have noticed that when I play my gig panels at full bandwidth, then they definitely sound a lot better than when I have to wind them up bit and then cross over to a sub bass for higher SPLs and power handling.
Fully agree on all of that.
Digital signal processing being based on a way or an other on FFT, making the signal periodic avoid any question of window or what is known as window leakage. The secure way to understand FTT is to imagine it sees the signal as if it is periodic. So if the signal ends at a different value it finish this create a jump (see below) and creates artifacts (not real spectral content)
2 solutions :
- a window function that forces to 0 the beginning and the end
- a signal having a periodicity in an integer relation to the FFT length