and they have been in the loudspeaker manufacturing business for 40 years...
Well, I have to admit that I enjoy getting mired down in characterization testing. It's part of the fun! But it's also true that your panel is going to sound the same regardless of how accurate your modulus measurement is!
That said, if you decide to get mired down again, I think you would get the best results with free-free boundaries, and the closest you can get to that is hanging from light strings attached at the nodes of the fundamental (about 0.22*L) from each end, and using the fundamental frequency, and the equations you already referenced. I have done that many times and it feels pretty reliable to me.
Eric
Unless they have knowledge in DML's it dont matter how many years they have in loudspeaker manufacturing business.
Just like how most (newbies are here because of him) of you believed what Tech Ingredients told you just because he is a science type of guy.
T.I. design is basically Rich M's design, which is basic beginners entry level intro into DML technology.
This is a accurate review of Tech Ingredients design.
All this discussion around lack of HF in DMLS...
I have battled intensively trying to get decent on-axis ruler-flat measurements on all of the DML panel composites that I have tried out.
Just a quick observation:
DML are omni-directional. But in a conventional point-and-spit cone system, one would be measuring the speaker stack on-axis with the microphone aiming down the throat of the tweeter (or the tin-can or the whizzer for that matter). Such cone-speaker systems are DIRECTIONAL by default, and their off-axis readings might be showed-off if those off-axis responses happen to be unusually admirable.
But DMLs are omni's both at low frequency AND at higher frequencies below Fc... and the on-axis SPL measurement is rather misleading since the radiated power response is more important.
Whereas a conventional cone speaker concentrates all (ok, most) of its HF at a tiny little sweet spot on the couch at the apex of an equilateral triangle, the DML reckons 'bugger that' and fires the same total HF power all over the place. It might be a better idea to take multiple measurements (100's?) of the DMLs in an sphere in an anechoic space in order to assimilate the full power response of the panels, OR to set them up in a diffusely reverberant space to measure the same thing but with a single measurement. The latter implies a reverberant space that has known RTA decay times at all frequencies and which can be subtracted from the resultant measurement.
This is all precipitated by the observation that, while I'm performing live with the panels just behind or just in front of me, I usually start with the HF EQ set a few dB hot to make up for the apparent drop in top end. But by the end of the gig I discover that I've dropped the HF to a more-or-less flat curve, and the panels still sound very rich and detailed, probably more-so than any of the the cones and horns that I've used, and that the audience nevertheless comments without fail on the amazing clarity and presence of the sound.
I have battled intensively trying to get decent on-axis ruler-flat measurements on all of the DML panel composites that I have tried out.
Just a quick observation:
DML are omni-directional. But in a conventional point-and-spit cone system, one would be measuring the speaker stack on-axis with the microphone aiming down the throat of the tweeter (or the tin-can or the whizzer for that matter). Such cone-speaker systems are DIRECTIONAL by default, and their off-axis readings might be showed-off if those off-axis responses happen to be unusually admirable.
But DMLs are omni's both at low frequency AND at higher frequencies below Fc... and the on-axis SPL measurement is rather misleading since the radiated power response is more important.
Whereas a conventional cone speaker concentrates all (ok, most) of its HF at a tiny little sweet spot on the couch at the apex of an equilateral triangle, the DML reckons 'bugger that' and fires the same total HF power all over the place. It might be a better idea to take multiple measurements (100's?) of the DMLs in an sphere in an anechoic space in order to assimilate the full power response of the panels, OR to set them up in a diffusely reverberant space to measure the same thing but with a single measurement. The latter implies a reverberant space that has known RTA decay times at all frequencies and which can be subtracted from the resultant measurement.
This is all precipitated by the observation that, while I'm performing live with the panels just behind or just in front of me, I usually start with the HF EQ set a few dB hot to make up for the apparent drop in top end. But by the end of the gig I discover that I've dropped the HF to a more-or-less flat curve, and the panels still sound very rich and detailed, probably more-so than any of the the cones and horns that I've used, and that the audience nevertheless comments without fail on the amazing clarity and presence of the sound.
I like your resonant mode/damping test apparatus from the other thread. Measuring steady-state response seems more robust than capturing transients from a moderately damped system. That being written, I have captured cantilever first mode frequencies at work with the 'pluck method' using an accelerometer, a charge amp, and a digital oscilloscope, and the result was very close to the FEA model.
Last edited:
Accidentally saw your DML board
I don't know if you are still doing update testing
I provide YAMAHA information for your reference
I don't know if this can help you
Because about a while ago I found out about OB's information
I just heard about the YAMAHA NS series from my friend
Let me be more amazing about non-directional speakers like OB and DML
The link below has many pictures of YAMAHA NS20, NS30...etc.
I saw it at a friend once and it was just a big foam board
Then it has very small XMAX
Now I think YAHAMA seems to be the forerunner of DML
But YAMAHA does not seem to continue to study this area
They seem to spend more time on cone speakers
Hope this information is helpful to you
https://www.audioplanet.biz/t99985-yamaha-ns20-y-ns30
Hello André,
I can follow what you say but I can't for now come to a practical conclusion... The professional world can access to anechoic chamber; there were work to determine the expected response in anechoic chamber (for standard loudspeakers) based on the appreciation of loudspeakers in real rooms (see M Floyd works) but for DIYer no such rooms.
Steve in previous posts advised not to make single point measurements. I made some tests either with IR to see the directivity of a panel or what is called Moving Microphone Measurement where a pink noise is played with a RTA making an average FR (I think it is Steve's technique)I put it down to the larger voice coil diameter. I have no indication of directivity in "in room" measurements so, but I can be wrong, the axis single seems to be a not to bad information.
The next step seems to be more in the field of the "house curve". As far I understand, the idea of the "house curve" is not being in anechoic conditions, what is the FR target for a given loudspeaker (meaning a loudspeaker we know the directivity).
Probably collecting the experiences of EQing (I have no EQ...) will help to clarify the topic?
Christian
Hi Andre, just a thought.
In some of the literature I have read one of the claimed traits a DML has is an apparent lack of interaction with its environment. Clearly this can't be true, but I did wonder if the reason for this apparent lack of room effects is actually more to do with consistency of interaction. The spectral content of reflected sound that you get from a conventional speaker is determined by its directivity, The first reflection has far less HF and MF content than LF content in a conventional speaker whereas with DML's the first reflection has a much more balanced spectral content. I believe Its one of the reasons that people that hear DML's for the first time often refer to the DML's as being like listening to headphones, which are obviously perfectly spectrally balanced as there is no room interaction at all.
This is all a long winded introduction to say I have found measuring DML's at the listening position helpful in guiding development. Its where the first impulse, first reflection and subsequent reverberation all comes together. For PA applications that's not as helpful but maybe measuring at several points in the room can help here?
Burnt
Speaking for myself I plan to EQ and most likely use a house curve for one particular listening spot. The response in the rest of the space will be what it will be, unless there is something really nasty going on. However I am hoping the response will be flatter in the frequency domain across the space compared to the cone driver plus horn cabinets I am using now.
Speaking of which, here is the first panel on the frame with the EPDM foam surround. The exciters will be arriving this evening so I'll be testing soon. The mechanical pencil is included for scale. Dimensions are 24" x 92"
Speaking of which, here is the first panel on the frame with the EPDM foam surround. The exciters will be arriving this evening so I'll be testing soon. The mechanical pencil is included for scale. Dimensions are 24" x 92"
Attachments
Last edited:
Yes I've heard this one too. So one of the first things I did in my first experiments was to wind up the volume and stick a mic right up to the panel, just like the guy in the clip we've all seen.... And as true's nuts, it did not howl-back!! A very impressive demo. (Obviously to do with node/antinode positions.)
But take the mike a metre or so away, then the howl-back does occur, albeit not as aggressively as a piston speaker, and you really have to wind up the gain levels. It appears that wave-fronts do eventually correlate and build up to some degree.
And this might also explain why these panels are so extremely clean in reverberant environments, they somehow do not create as many coherent reflections as what a solid wave-front would do.
Yes!
Absolutely.
I think straight measurements at different points in a room will probably measure the room, even with DML panels? I think the proper way is to first measure the room itself with an impulse (two planks slapped together, or a whip-crack, or a rat-trap tripping onto a decent surface, or something like that) and then take RTA60 or RTA30 measurements to get the reverb decay curve nailed down. And then take speaker measurements and measure the difference between them. It's complicated.
Maybe it's just force of habit, but I prefer to measure outside after I've done all the nitty-gritty measuring and tweaking stuff indoors the workshop.
But what has surprised me with DMLs is the relatively small difference in measurements between inside vs outside. A conventional speaker's FR is normally unrecognisable between indoor and outdoor measurements. But DML not so. For me the outdoor measurements are so similar to indoor, it's becoming almost pointless to measure them outdoors.
At a recent outdoor venue, my wife took a recording of the sound balance with a cell-phone, and took a walk all around while recording. And through headphones, there's very little spectral difference between centre-stage at 60ft, vs way off centre 100ft away. I found this interesting.
You know, I should conduct this exercise for myself. However I find this very interesting. I am sure if I submit HQ sound clips of the speakers that are not outside, that everyone will attribute the perceived SQ success or SQ failure to the room even if its contribution is insignificant.
What I will most likely do is some quick pink noise and sine sweep measurements outside, then save all of the serious EQ tuning for when they are inside and in position.
I have almost no experience of outdoor measurements with conventional speakers but I did some with DML. The main difference I found was in the low frequencies. Some cancellations occur in room not outdoor.
Which goes in the way of a low difference in room according to the direction. In addition the SPL reduction with the distance is lower than with a cone.
+ @Veleric , @BurntCoil
Building my plywood panel, I made also the choice of the grain in the direction of the long side. The 3 ply has an important difference of stiffness according to the direction. The high geometric ratio seems to be a factor to smooth the FR but in this case, the difference of stiffness will reduce the effective ratio.
Any opinion?
Christian
Hi Christian,
From my experience there appears to be no significant difference. I built a pair of Scandi Blondes for one of my son's friends and unfortunately because I did not specify the grain orientation i received two identical high aspect ratio panels with different grain directions. This meant I had one board which was appropriately stiff for the design in structural terms and one that was not self supporting and developed a strong curve. However, the sound from each panel displayed no significant difference. I replaced the bendy panel but both were used for some weeks before the new material arrived.
This is hardly objective or definitive but might be of some help anecdotally.
Burnt
From my experience there appears to be no significant difference. I built a pair of Scandi Blondes for one of my son's friends and unfortunately because I did not specify the grain orientation i received two identical high aspect ratio panels with different grain directions. This meant I had one board which was appropriately stiff for the design in structural terms and one that was not self supporting and developed a strong curve. However, the sound from each panel displayed no significant difference. I replaced the bendy panel but both were used for some weeks before the new material arrived.
This is hardly objective or definitive but might be of some help anecdotally.
Burnt
Last edited:
Which exciters did you choose and how many per panel ?
What thickness EPDM ?
I walked around our big box store last night trying to find EPDM foam in a ~ 8 mm thickness with adhesive both sides but they had nil , thats my "theory for the day" to have more compliance / less damping from a thick softer foam compared to VHB tape.
Eucy , Was the Tasmanian Blackwood panel you used in a 3 mm ?
I recall it was available in a 1.5 mm thickness so that could offset the density / efficiency loss.
Plyco also have a Tasmanian Sassafras in 1.8 and 3.0 mm , Sassafras has a SG of 0.4 - 0.5 and density of 495 kg3M versus Poplar being 450 kg3M.
The Sassafras has better grain aesthetics than Poplar unless you happen to be living in Sveden or Norway where white bleached timber is the fashion.
TBD... I am still building these but I have four of the Dayton DAEX32U-4 ready to go. I plan to use two per panel.
I used this EPDM foam on the surround. It is far thicker than the VHB tape (which I also have) and more compliant so it should allow much more movement at the edges and provide a lot of damping.
Frost King E/O 9/16 in. x 5/16 in. x 10 ft. Black EPDM Cellular Rubber
OK... so...
After some quick experimentation with pink noise and music I am noticing a few things...
- I am using the 3/5 2/5 position as a baseline with the 3/5 along the long edge. In this position the sound is highly localized and does not spread across the panel. I need to flip this around and try the 3/5 along the short edge I suppose.
- As mentioned in previous posts, small movements make a big difference. I can hear changes in phase and frequency almost continuously as the exciter moves, as well as changes in how broadly the response is spread (or not spread) across the length of the panel.
- There is a huge difference in audible response depending on whether I am listening on the exciter side of the panel or the opposite side; the frequency range is much broader on the opposite side in a big way to the point where listening on the exciter side while moving its position might be pointless. This is, umm, an interesting problem since ultimately the listening position will be on the opposite side, but I want to be on the exciter side to move it around. I might need to use an assistant to move it around while I listen.
- Trying to keep the panel horizontal while moving the exciter basically means I am lying on my back with the DML frame sitting on sawhorses above me. This adds the effect of the space between the floor and the panel which will color the sound somewhat. I do not see a way around this however until the exciter is mounted, and then it cannot be moved easily of course.
- I am wondering if the panel might have too much damping. Right now it has 5/16" EPDM foam along both of the long edges. I think I may try removing sections of it, distributed of course (not all in one spot), and see what happens. This may be at least part of the reason why sound is localized around the exciter position since total damping will increase with distance along the long axis. I am guessing mode shapes are another part of this if not a big part.
- Starting off with a lot of damping while looking for the ideal spot might be like tuning a TL design in a simulator (Martin's work sheets or Hornresp); the damping is not added until the very end so that the effects of geometry changes while optimizing the design are much more obvious. In other words the damping may mask the changes in response with the movement of the exciter. The obvious problem though is that removing damping (edge foam) also changes the stiffness of the suspension and thus damping and edge stiffness are not mutually exclusive. With a TL adding damping does not alter the tuning (much); it just flattens the resonant peaks (and lowers the port response, killing low end extension).
Apologies for the word wall, but there is a lot to consider as I am sure the experimenters here have already noticed
After some quick experimentation with pink noise and music I am noticing a few things...
- I am using the 3/5 2/5 position as a baseline with the 3/5 along the long edge. In this position the sound is highly localized and does not spread across the panel. I need to flip this around and try the 3/5 along the short edge I suppose.
- As mentioned in previous posts, small movements make a big difference. I can hear changes in phase and frequency almost continuously as the exciter moves, as well as changes in how broadly the response is spread (or not spread) across the length of the panel.
- There is a huge difference in audible response depending on whether I am listening on the exciter side of the panel or the opposite side; the frequency range is much broader on the opposite side in a big way to the point where listening on the exciter side while moving its position might be pointless. This is, umm, an interesting problem since ultimately the listening position will be on the opposite side, but I want to be on the exciter side to move it around. I might need to use an assistant to move it around while I listen.
- Trying to keep the panel horizontal while moving the exciter basically means I am lying on my back with the DML frame sitting on sawhorses above me. This adds the effect of the space between the floor and the panel which will color the sound somewhat. I do not see a way around this however until the exciter is mounted, and then it cannot be moved easily of course.
- I am wondering if the panel might have too much damping. Right now it has 5/16" EPDM foam along both of the long edges. I think I may try removing sections of it, distributed of course (not all in one spot), and see what happens. This may be at least part of the reason why sound is localized around the exciter position since total damping will increase with distance along the long axis. I am guessing mode shapes are another part of this if not a big part.
- Starting off with a lot of damping while looking for the ideal spot might be like tuning a TL design in a simulator (Martin's work sheets or Hornresp); the damping is not added until the very end so that the effects of geometry changes while optimizing the design are much more obvious. In other words the damping may mask the changes in response with the movement of the exciter. The obvious problem though is that removing damping (edge foam) also changes the stiffness of the suspension and thus damping and edge stiffness are not mutually exclusive. With a TL adding damping does not alter the tuning (much); it just flattens the resonant peaks (and lowers the port response, killing low end extension).
Apologies for the word wall, but there is a lot to consider as I am sure the experimenters here have already noticed
Last edited: