Good suggestion. I lodged an inquiry with one of the suppliers I found, no response yet. The aircraft plywood suppliers offer several different wood species -- birch, basswood, mahogany and poplar in various veneer/core combinations. Basswood appears to fall between poplar and birch, in terms of density. Mahogany is fairly light but costs more.
Their plywood thicknesses start at 1/16". I wouldn't go that thin. The 3/38" or 1/8" might be worth looking at. Not cheap, a 2 x 4 piece of 3/32" basswood plywood costs $47.00, not including shipping.
The insulation material I used is bonded on both sides with aluminum. This bonded foil is probably 3x 'kitchen foil' > but by no means heavy.
PS.
The need for using a tweeter/super tweeter is reflected in ALL the frequency response graphs I have seen of DML's >
you can hear it with your ears. The moving mass of a good tweeter is a fraction of that realized with a DML panel.
For good HiFi, some form of low frequency woofer is also required.
I would consider 3mm wood to be quite light weight > I was really referring to people using heavier thicker wood.
If I had the time, patience and inquisitive inclination, I would love to hear the difference between what I built and 3mm ply.
The DML's I built are not my main speakers and they do require a combination of both passive & active EQ for nice sound.
Quote:
The insulation material I used is bonded on both sides with aluminum. This bonded foil is probably 3x 'kitchen foil' > but by no means heavy.
PS.
The need for using a tweeter/super tweeter is reflected in ALL the frequency response graphs I have seen of DML's.
I would consider the foil in a cigarette pack (with the paper removed) as very light and the cheapest kitchen foils as light. Forum and thread member
Spedge (Steve) has on numerous occasions demonstrated that DML response out to 20 Khz. is indeed possible and he has done so with an assortment of different materials including card stock EPS and XPS along with various panel coatings. I do agree with you that DML panels require a sub for good low end response. Thank you for your response.
This is a 'classic' frequency response of a good DML. You can see why I recommend the use of a tweeter + crossover network.
There is another interesting thing about expecting 'true sounding' high frequencies from a DML panel >
The amount of 'moving mass' creates a lag in response time / attack time / rise time.
If you feed the average size DML with a 10Khz square wave and use a high-quality measurement microphone + oscilloscope
you won't see a square wave, but something resembling a sine wave. This means that although you hear HF, it is not accurate or true.
PS.
If you want to see what I built > go to post #11,870 on page 594 of this Thread.
Cheers 🙂
Thank you ,Moray .
I am glad that someone noticed that my eps panels can reach 20k with a little care and attention 😀
And I agree that a good low frequency driver is essential for the successful realism of a dml panel.
Steve.
I am glad that someone noticed that my eps panels can reach 20k with a little care and attention 😀
And I agree that a good low frequency driver is essential for the successful realism of a dml panel.
Steve.
Personally I really do not want a regular tweeter, I really like the non-fatiguing smoothness of the treble with DML, especially when using at loud levels.
Here is the curve from my pervious design after EQ:
It aligns pretty well with my desired target curve, and using a regular tweeter would not improve the response but create problems with directivity.
Regarding recording a square wave at 10kHZ I'm not sure what you where expecting? With a 20kHZ playback and recording system you cannot reproduce more than one harmonic, so will not look like a square wave. If you have less response at 20k it will look pretty much like a sine, but even if you are a youngling with perfect hearing it will also sound like a sine, so how does it matter how it looks? Why does it have to be accurate and true above the human hearing range?
@Leob
funny phase response still being constant but jumping a bit around.
With a fullrange driver I measure linear phase response in the mids after applying EQ.
But the measurement shows that the phase is not walking away several hundred degrees in the mids like usually with 99% of all multi way speakers
funny phase response still being constant but jumping a bit around.
With a fullrange driver I measure linear phase response in the mids after applying EQ.
But the measurement shows that the phase is not walking away several hundred degrees in the mids like usually with 99% of all multi way speakers
I would add the question of how both levels decrease with the distance. We can expect a -6dB by distance doubling with the tweeter while I remember measurements from a DML more like -3dB by distance doubling.
So for the woofer, no question... for the tweeter... I am probably already too old!
Christian
Obviously I was shooting for a more mellow than sharp sound, and was very happy with it. But to get it a bit sharper I would flatten it out between 4-10k and then let it slope sharper after that.
I'm using them for an outdoor festival, so did measurements outdoor and about 2 meters away in this case I think. This is the on-axis response, which is what I EQ:ed for, but the off axis response is a bit different.
Not sure if tools like REW perhaps get a bit confused when measuring phase of DMLs?
Hmmm...yes I might not be the right person to ask about HF performance any more either 😀
Let's take my 10Khz square wave as an example of the 'sharp edges' that are present as harmonics with various recorded sounds.
Now let's bring this down to 4Khz and look at a violin sound > accuracy of this requires fast response to 'sharp harmonics' which
a DML will once again 'round' like a sine wave. This obviously creates the smooth / warm / mellow / relaxed sound that many prefer.
I was strictly talking about accuracy & fast HF transient response.
One thing I have never fully understood, is why many people with some age-related HF hearing loss simply 'throw in the towel' saying
"I can't hear that - so why bother". Many have HF hearing loss that is just 'relative sensitivity' and NOT an actual CUT-OFF frequency of say 10Khz.
So, why not customize a speaker system, using tweeters, to compensate? There is the possibility that many 'don't know what they're missing'.
PS.
Are you building speakers for YOUR ears, or someone else's ?
I been doing audio quite many years, and it is a recent thing that I cannot hear over 16kHz, but I'm not too old to remember having perfect hearing 🙂
The target curve I use for a PA has not changed, and many engineers will use a similar curve like the "Michael Lawrence target trace", where it also slopes off above 10k. But among amateurs it seems to be this idea 20-10kHz is half the spectrum, and 10k-20k the other half, and if it is not flat all the way you are missing out on lots of information, often coming with examples like the fact that a square wave will not be a square at a certain frequency unless you have more bandwidth. Our ears are not oscilloscopes, and no-one can hear a 4kHz square wave as a perfect square either. If that limit is 16k or 20k makes quite a small difference. In loosing the top 4-5k I would say it seems to me like I lost 1% of my hearing, similar in impact to my impression of music as if I lost the ability to hear 20-25Hz.
I do agree that the ideal speaker should cover human hearing range, but in reality a good speaker capable of 35Hz to 16kHz can sound really amazing, so I don't think it is worth obsessing about that very top of the spectrum since not even teenagers will actually appreciate it.
The target curve I use for a PA has not changed, and many engineers will use a similar curve like the "Michael Lawrence target trace", where it also slopes off above 10k. But among amateurs it seems to be this idea 20-10kHz is half the spectrum, and 10k-20k the other half, and if it is not flat all the way you are missing out on lots of information, often coming with examples like the fact that a square wave will not be a square at a certain frequency unless you have more bandwidth. Our ears are not oscilloscopes, and no-one can hear a 4kHz square wave as a perfect square either. If that limit is 16k or 20k makes quite a small difference. In loosing the top 4-5k I would say it seems to me like I lost 1% of my hearing, similar in impact to my impression of music as if I lost the ability to hear 20-25Hz.
I do agree that the ideal speaker should cover human hearing range, but in reality a good speaker capable of 35Hz to 16kHz can sound really amazing, so I don't think it is worth obsessing about that very top of the spectrum since not even teenagers will actually appreciate it.
Last edited:
I agree with this, but what about speakers that are 3dB down @ 5K , 6dB down @ 10K and 10dB down @ 16K ???
Tweeters can fix this 😎
PS.
With my Satellite Foxtel home AV system, MTV music channel has a 15 KHz "BRICK WALL FILTER", but music still sounds quite reasonable.
It is the myriad of compression / maximization processors used with TV & Radio that play quite a significant role in what we hear.
With my Satellite Foxtel home AV system, MTV music channel has a 15 KHz "BRICK WALL FILTER", but music still sounds quite reasonable.
It is the myriad of compression / maximization processors used with TV & Radio that play quite a significant role in what we hear.
The method I use to reduce the spike in the 10k area on my eps panels also has the effect of reducing the shallow dip before the spike and the large dip roll off above the 10k spike.
Which can give a flat response to 20k
I see no point in trying to restrict the frequency above the 10k point up to 20k and beyond.
Using a mass such as a small blob of blu-tack in the central area of the coil ring will reduce the 10k spike and will also roll off the frequency above the 10k spike.
The amount of spike reduction and roll off would depend on the amount of mass applied.
Steve.
Which can give a flat response to 20k
I see no point in trying to restrict the frequency above the 10k point up to 20k and beyond.
Using a mass such as a small blob of blu-tack in the central area of the coil ring will reduce the 10k spike and will also roll off the frequency above the 10k spike.
The amount of spike reduction and roll off would depend on the amount of mass applied.
Steve.
Yes, but there really is no need for debate. This is because both stiff and flexible are desirable, but they are (surprisingly?) not mutually exclusive.
Stiffness (rigidity) is good mainly because it helps improve sensitivity, that is, the ratio of SPL output to the power input. Panel sensitivity is mainly determined by the ratio of stiffness/(weight-cubed). So a panel doesn't really need to be extremely stiff to provide good sensitivity, it needs only to be stiff relative to its weight. And because it is "weight cubed" in that relationship, low weight what mostly contributes to sensitivity, but high stiffness helps too.
Flexibility, on the other hand, is good because a more flexible panel has (a) a lower fundamental frequency and (b) a higher density of natural frequencies in any given frequency range compared to a stiffer panel. This means generally that a flexible panel will have a lower cutoff frequency and a smoother overall frequency response than a stiffer panel.
And the reason that you can have the benefits of both a "stiff" and a "flexible" panel at the same time is this: If you start with a material that has a good ratio of stiffness/weight-cubed, you will get good sensitivity. Then (at least in principle) you can get the benefits of flexibility with that same panel material by "simply" making the panel thinner or larger, without losing any of the original sensitivity. Of course, in reality, it is not always so simple. The material you chose my not be available in a thinner version, and/or can't be made thinner by simply sanding or slicing without changing it completely. And making it larger could be equally difficult, due to WAF, or cost, or space, etc. But at least in principle you can have both stiffness and flexibility in the same panel, basically by starting with a material with a good ratio of stiffness to weight-cubed, and then making it sufficiently thin, or large, for it to be flexible.