Hi Jaxboy
The keywords are "3M VHB" (Very High Bound)
In the history file, I stored the post from Veleric mentionning a substitute 3M 9589 for testing. See #4959
See also #2698 from Aagas and next #2699 from Stimper who linked #2624 from Jarrodhass (3M 9473 and thinner substitute 3M 9460) which is in fact the best post about this tape as Jarrodhass had information from Dayton Audio.
Christian
Hello,
I hope it will answer to your question...
The table below shows different configurations of series/parallel exciters and gives the resulting resistance, the increase in SPL, the increase in power demand.
The cells in grey are possible too low impedance or lower power demand.
Attached is the pdf version.
Christian
@homeswinghome Thank you very much for your help and sharing
These low power exciters have a beautiful bandwidth
But after reading the data, I found that it seems to need to increase a lot to significantly improve efficiency
2*2 only increases 6DB...XD
The question is, can the bandwidth and curves be similar to the original after putting so much on the panel?
Maybe the peaks and troughs will be more prominent after putting more exciters?
It feels like doing it this way leads to more complexity... XD
This question was discussed here in the past with I think no final conclusion; the discussion being around the existence or not of comb filters. The secure way adopted by DIYers designing room applications is to stay with one exciter as there is no real question of power with the standard exciters. For DIYers designing PA applications, the power is the main topic, not the bandwidth so a 4 exciter configurations is used.
Using several exciters might be a possibility to excite more modes. There are papers about that.
Somewhere before, I posted a proposal of 2 exciters in series, one being shunted by a capacitor for HF performance in order to avoid any possible comb filter.
A combination of several 19CT might be a good idea in order to get advantage of their HF qualities (low mass, low inductance)... not seen here i think and as you said additional complexity or at least a new path to explore.
Christian
Hello Audiofrenzy,With multiple exciters there is some degree of comb filtering and or intermodulation as Bertagni states. The most exciters I have used on 1 panel is 4 exciters. Even with 4 exciters the spl does not increase much. There are techniques to reduce intermodulation between more then one exciter on the same panel but there is no way to completely eliminate it. The more exciters used on one panel will produce more intermodulation. Intermodulation seems to effect the HF a bit more then the LF.
How to fix this problem. Solution> Instead of using multiple exciters on one panel you use multiple panels with one exciter on each panel. When doing it this way you will notice that the spl will be a lot higher then if you used mulitiple exciters on one panel which once again will prove that there is intermodulation.
You can easily test this theory for yourself the only problem is you will need an amp/receiver with many channels.
The way I do it is by using the dedicated center channel on my receiver with a DML panel with one exciter. Since the center channel is in the middle it is far enough away from the main L/R speakers to prevent any comb filtering.
Of course my way is for audiophile home use and not PA.
I am not sure to follow you in the wording.
Clearly, comb filtering can occur in the situation of multiple sources playing the same signal when the distance between then becomes of the order of magnitude of the signal wavelength (in the air). If I remember previous posts, there is no consensus of that for DML (lack of evidences, I don't say it doesn't exist, just the proof has not been shared here).
I know intermodulation as defined by wikipedia (intermodulation is the amplitude modulation of signals containing two or more different frequencies, caused by nonlinearities...). This occurs in any system (DML also) but not because of the number of sources but because of the non linearity like it was shown in power amplifier, radio amp.
Having one panel per exciter will reduce the intermodulation has several panels having their own exciter should have a lower displacement so lower non linearity than a single one with several exciters.
Even if the role of the area is not demonstrated we should think it is a good think also for efficiency taken the benefit of a larger total area in addition of the multi exciter arrangement.
From a comb filtering point of view, it might not be an improvement as the distance between the sources (mainly true in HF which are supposed not travel widely on the panel) is greater than in a single panel with multiple exciters so the frequency where the comb filtering occurs is lower.
Anyway, interesting.
Christian
Wow! They sure are expensive! Thanks for that link. I had gone to several outlet sites, but what they had was like 25 roll/cases for around $65-70/roll. I think you're right about using epoxy for permanent bonds. However, I attached mine about 15 months ago using the VHB that they came with, and they are still doing well. I think the key is to let them set for about a day before using them, letting the glue set.
I didn't like the price of the 3M tape so have decided to try this, picked as it's rather thin also considering what Dayton put on their exciters
https://www.amazon.co.uk/gp/product/B07C2TGKGT/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1
Will they fall off, Have to wait and see.
https://www.amazon.co.uk/gp/product/B07C2TGKGT/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1
Will they fall off, Have to wait and see.Hi Christian,
I can concur. During testing I have experimented with two exciters separated by 100mm on a single panel. There's no evidence of audible comb filtering. At all. Not vertically, not horizontally, not in the measurements, and no matter what signal sources are used (music, guitar, piano, keyboard, orchestra, white noise, pink noise, swept sines, spot sines etc etc etc.)
(I think I saw some inexperienced DIYer's a while ago referring to the graphic of an RTA mode pattern on a measured panel as "comb-interference." But, even making allowance for their measuring environment, I suppose such expertise should just be ignored. )
Yes indeed, comb filtering is unavoidably manifested in the acoustic medium with a coherent wave-front (in air for a standard pistonic driver,) but I'm sure you agree that DML panels are probably as different to cones as what F1 race cars are to bicycles.
To confirm:
Speed of sound in air is around 343m/s.
Speed of sound in a (polycarb) DML Panel is around 2250m/s.
In air, two pistonic drivers separated by 100mm will start combing at 3,4kHz. Very audible, with beaming and lobing, and leading to a tiny listening sweet-spot only a foot or two wide.
On a DML panel, two exciters separated by 100mm will start "combing" (if that's the correct word) at 22,5kHz if AND ONLY IF the DML panel delivers a coherent, single wave-front. Which it does not. And which is probably why audible combing does not exist in DML sound.
I'm currently testing with 8 x 40W drivers (8-ohms, 320WRMS) on 1200 x 500mm panels. I found the (panel) sweet spots for the drivers by using various combinations of Chladni patterns with spot sine-waves, and real-time RTA FR using white noise. This eventually resulted in the drivers being spread out in a mirrored configuration, and I suspect that a maximum number of panel modes are activated.
Golden-Ears-Jerry was here over the weekend, and we set up two such DML panels out in the garden. No matter where you stand, the musical reproduction was like giant head-phones: Smooth. Full stereo image everywhere. No sweet spot. No combing.
Good article in Wiki, that.
If anybody else is interested...
Single pistonic speakers that attempt to produce a full range response are especially susceptible to IMD because (low-excursion) HF signals in such a cone are directly superimposed on the (large-excursion) movements in the same cone which is required to reproduce LF signals. This means that the HF signals directly undergo a Doppler effect producing signals that are not present in the source signal, and, besides HF beaming—leading to combing, hot-spots and dead spots—these are all unavoidable issues especially in pistonic full-range cone drivers. These are physical principles that cannot be addressed by any amount of engineering or signal processing without adding band-width-limited domes or cones... Enter multiple drivers, cross-overs, phase shifts and complexity. 🤦♂️
Agreed!
A DML panel requires much much less membrane movement to produce LF than what a cone driver does. Therefore IMD is very significantly reduced in principle, and might be reduced even further by careful design and optimisation of vibration modes.
On a slight tangent, I confirmed a while ago that multiple drivers per panel do not necessarily produce higher SPL the more drivers are added. The best configuration seemed to be 4 drivers per panel, carefully placed, wired as a 4-ohm load, and additional drivers need to be split onto independently-damped membranes if SPL is to be increased.
But I need to test this again since I now have a power amp that can drive low impedances with constant voltages. I suspect that the only reason I did not previously get a concomitant SPL increase with multiple drivers was because the amp PSU was sagging under a 4-ohm load.
The surprising thing is that, probably because of the very high speed of sound in a panel, carefully-placed multiple drivers do tend to transform across the panel to make it vibrate in such a way that the DML action is not degraded, and pin-sharp imaging is still manifested across a massive listening sweet-spot whether-or-not multiple drivers are installed across a single panel or even across multiple panels.
Obviously the confusion arises from inexperienced beginners mistakenly thinking about the DML wave-front as a single wave-front, such as produced by pistonic cones, and this confusion leads to more confusions about things like "combing." But DML speakers produce a diffuse wave-front and which appears to be a holographic signal, and which is why it does not manifest many of the unavoidable problems such as found in traditional cone speakers.
Interesting Indeed.
Thanks Christian, it's always instructional to get such solid engineering input from your discussions.
Last edited:
Thanks Christian, this is good.
Is this an Excel spreadsheet by any chance?
Hello André.
For sure there is. Here are the original file (ODS Open Office format) and the XLS export
for what its worth.... from a pdf from PUI Audio
https://oemelectronics.se/images/produkter/ljudindikering/Exciter-PUI.pdf
If multiple exciters are needed for higher output, try to avoid spacing them evenly about a surface (or panel). This introduces comb filtering on the surface, where one exciter starts to cancel out the midrange to high frequency output of another exciter.
I noticed they say "This introduces comb filtering on the surface" and not mentioning if this comb filtering reaches the listener.
For multiple exciters they also recommend 1/5 - 2/5 spacing.
https://oemelectronics.se/images/produkter/ljudindikering/Exciter-PUI.pdf
If multiple exciters are needed for higher output, try to avoid spacing them evenly about a surface (or panel). This introduces comb filtering on the surface, where one exciter starts to cancel out the midrange to high frequency output of another exciter.
I noticed they say "This introduces comb filtering on the surface" and not mentioning if this comb filtering reaches the listener.
For multiple exciters they also recommend 1/5 - 2/5 spacing.
It's a wide field set up problem but I have seen a paper from the far east that suggests using polar measurement to determine if a panel is optimised. Not mentioned if for wide field or not.
More interesting was a video demonstrating a comsol simulation video of an excited plate. First step was to find the first n natural frequencies. They are spaced. The effect of that could be viewed as a comb filtering problem but isn't. Interestingly as the computational load is high it uses symmetry and just models 1/4 of the plate with a small steel area attached at a corner. Materials are selectable. Personally not sure that is valid but it's probably an extremely expensive package.
I have some doubts about the precision of this paper :
- the FR shown in page " are based on a circular plate with no mention of the exciter location neither of the boundary conditions (free edge, clamp...). Circular? What a strange choice.
- it is strange to see that a 4mm plate (190mm diameter) can reach lower frequencies than a 2mm one. Some explanations are missing in my opinion
- it recommend a lightweight and dense material. Means thin?
- in page 4 the role of the membrane characteristics seems oversimplified (density for HF, flexibility in LF)
- the sentence about the exciter placement versus "standing waves" is not precise or could even result from a bad understanding of what are modes in a plate.
The diffuse nature of DML is in the papers describing how they works : see "Peter Mapp, A flat response". Extract below.I've been talking about DML's "DIFFUSED" sound since forever.
Anyone know what comb filtering sounds like in conventional cone drivers? When comb filtering occurs it will have a softer and diffused type of sound, sounding less prominent and or out of phase.
DML's do produce a diffused wave front which is why there is way less comb filtering/intermodulation to begin with then conventional cone speakers but there is still comb filtering going on but at a lesser degree. This is why one exciter per panel sounds the most prominent/coherent, well maybe not to PA ears.
1.This diffused wave reduces comb filtering.
2.This diffused wave is also what reduces wall/room reflections.
3.This diffused wave is also what reduces microphone feed back.
If you understand the pebble in the pond analogy you would know why DML's have a diffused front wave in which it produces a holographic type of sound.
It creates a kind of network of mini sources "pseudo-randomly" placed on membrane with the advantages you describe.
Comb filtering appears when a source is duplicated by delayed path.
The difficulty I have with comb filtering is it occurs in any stereo system in a room with even full range speakers : 2 sources playing partially the same signal, the effect of the floor and the ceiling. Everything is here to create comb filtering.
From here, I am not able to jump to the conclusion of the behavior of multiple exciters on the same panel.
I am wondering which setup could allow anyone, starting by me, to identify a comb filter effect in room condition, not a simulation with headphones. The auditory system in listening conditions (meaning with the reflections from the environment) has to be involved.
Extract :
What would help in this topic would be to compare the modal figures of plate driven by one and more exciters to see how the emitting areas change. The limit is the effect of the possible comb filtering is in mid or high frequency which need many nodes. For now here, the simulation is at the very early stage (I mean in DIY community).
Splitting the membrane in 4 parts for simulation to reduce the computational load works if the configuration is fully symmetric.
The app I mentioned is likely to be able to cope with whole plates and probably more exciters. It's my guess that choosing to simulate a 1/4 plate speeds thing up but if the app didn't account for the rest there would be no point in using it.
If you google opensource comsol alternatives you'll probably come up with a list. I did but applying them is unlikely to be simple.
If you google opensource comsol alternatives you'll probably come up with a list. I did but applying them is unlikely to be simple.