This means that you can not do anything in the bass really. It's just too little. My Mac Mini from 2012 run 160k taps as we speak - with 17% load. What was your plan for FIR?
S/N ratio ≤112dB - this means that SNR = 56dB is within spec ;-)
Nice subs!!!
//
I'm aware that 960 taps is not a lot, and like I said not sure if I will have much use for it. It seems like a lot of DSP units supporting FIR have quite limited number of taps. I guess a lot of the DSP power is already allocated to the other processes and FIR gets what is left over.
It should be able to handle doing x-overs at least, but will probably stick with regular HPF and LPF unless I have issues with phase alignment, but especially when using DML tops I doubt that will be needed.
I don't use FIR filters a lot during production since usually the pre ringing is more of an issue than phase, but a bit curious if it might be useful for system tuning. I was hoping to be able to try convolution of the reversed response offset for the tops, but maybe that is hoping for too much from the unit.
It should be able to handle doing x-overs at least, but will probably stick with regular HPF and LPF unless I have issues with phase alignment, but especially when using DML tops I doubt that will be needed.
I don't use FIR filters a lot during production since usually the pre ringing is more of an issue than phase, but a bit curious if it might be useful for system tuning. I was hoping to be able to try convolution of the reversed response offset for the tops, but maybe that is hoping for too much from the unit.
Got the DSP processor and amps connected up. No subs yet so just tested a bit with a couple of plates.
Going from RME Digiface to AES input works flawlessly and got quadraphonic panning and fx set up in my DJ setup using Envelop for Max4Live.
Of course the K420 is insanely overpowered for driving just a couple of plates, but turning down output from DSP by -40dB I get more than a degree of adjustment on the amp between silent and super loud
A bit temporary setup with plates standing on the floor leaning against wall, but my impression is that sound is excellent, and best sounding amps I tested on the plates so far. If I turn up amp to max without anything playing I cannot hear any noise, so both DSP and amps are dead quiet!
Very happy with the DSP. Thought maybe it would be crappy in some way with those IO capabilities and those features for that price, but extremely little noise, feels really solid and excellent interface in both hardware and software.
Haven't had a chance to play around with FIR yet, but noticed they now have a FIR section on the AutoEQ page, which I hope means that it will do the kind of inversed magnitude difference filtering I wanted to try out.
Now I should have some weeks to touch up the plates and see what the DSP can do to them before the subs are ready!
Going from RME Digiface to AES input works flawlessly and got quadraphonic panning and fx set up in my DJ setup using Envelop for Max4Live.
Of course the K420 is insanely overpowered for driving just a couple of plates, but turning down output from DSP by -40dB I get more than a degree of adjustment on the amp between silent and super loud
A bit temporary setup with plates standing on the floor leaning against wall, but my impression is that sound is excellent, and best sounding amps I tested on the plates so far. If I turn up amp to max without anything playing I cannot hear any noise, so both DSP and amps are dead quiet!
Very happy with the DSP. Thought maybe it would be crappy in some way with those IO capabilities and those features for that price, but extremely little noise, feels really solid and excellent interface in both hardware and software.
Haven't had a chance to play around with FIR yet, but noticed they now have a FIR section on the AutoEQ page, which I hope means that it will do the kind of inversed magnitude difference filtering I wanted to try out.
Now I should have some weeks to touch up the plates and see what the DSP can do to them before the subs are ready!
Hi Leon, I await your tests and also the videos as soon as you can, I had to suspend for several months but I had come to the conclusion that perhaps the frame, such as thicknesses, etc. was important but it is not necessary to have large thicknesses (at least I hope) the problem is engineering the structure system so that it is easy to assemble everything. As for the electronics I used a professional power amplifier from the Powersoft KIT DigiMod 500 old model but it works well, unfortunately it does not have a DSP
Hello DML enthousiasts,
I got inspired by this PA panel built, and am going to experiment. Thanks Leob!
I bought 4 Xcite drivers, and have started experimenting with foam,paper sandwiched foam, and also ordered material to make a carbon/nomex 3.5 mm panel.
FYI, Tectonic seem to have stopped the DML production and selling, and now only do BMR things. Also, on their twitter page they give some extra clues about damping, plate support placement, and driver placement. A bit like they want others to experiment further? https://twitter.com/tectoniclabspro
One of their latest plates uses unidirection carbon to make it stiffer in one direction - funnily xps also has this property because of the extrusion proces.
I'll post some results when anything meaningful happens.
Regards, Hans van der Zijpp, Amsterdam
I got inspired by this PA panel built, and am going to experiment. Thanks Leob!
I bought 4 Xcite drivers, and have started experimenting with foam,paper sandwiched foam, and also ordered material to make a carbon/nomex 3.5 mm panel.
FYI, Tectonic seem to have stopped the DML production and selling, and now only do BMR things. Also, on their twitter page they give some extra clues about damping, plate support placement, and driver placement. A bit like they want others to experiment further? https://twitter.com/tectoniclabspro
One of their latest plates uses unidirection carbon to make it stiffer in one direction - funnily xps also has this property because of the extrusion proces.
I'll post some results when anything meaningful happens.
Regards, Hans van der Zijpp, Amsterdam
A 4th order LR lowpass at 100Hz is possible at 48kHz and 1024 taps with a tiny bit of ringing.
A 2nd order highpass at 100Hz is also JUST about possible with 1024 taps.
At 96kHz, you need double the taps to get the same.
Or 4 times for 192kHz.
I would consider that the absolute minimum (if you want to include the 100Hz highpass filter, otherwise you can go lower)
For just some param EQ with a relatively low Q and boost/cut it's probable also fine for anything above 200Hz.
I think in the order of about 16384 taps for 48khz or 65536Hz for 192kHz would give you enough wiggle room.
Just put the response in rePhase and just see what fits!
Small update on my dml experiment. I have built a 3mm nomex/carbon panel (60x42 cm). The making of the panel is easier than I had thought. I hand laminated the panel. It is of course a bit heavier than a vacuum formed panel (1140g/m2).
The panel sounds a lot less agressive and distorted than my xps foam panel, but also a lot quiter. I Begin to think that for PA applications the foam might be more practical - it is about half the weight of the carbon panel. Or maybe a lighter/bendier sandwich panel? Any tips/hints where to search for the best compromise? Greetings from Amsterdam, Hans
The panel sounds a lot less agressive and distorted than my xps foam panel, but also a lot quiter. I Begin to think that for PA applications the foam might be more practical - it is about half the weight of the carbon panel. Or maybe a lighter/bendier sandwich panel? Any tips/hints where to search for the best compromise? Greetings from Amsterdam, Hans
Hi Hans,
I've used EPS in live gig panels. My prototype panels were skinned with Kevlar, and another pair with Dyneema. Both types of panel were extremely efficient but VERY mid-heavy. The higher frequencies don't like propagating through polystyrene. Its too soft. I believe Leob uses graphite-impregnated XPS for his gig panels. I'm not familiar with the properties of that material, but it sounds like it's worth chasing.
Solid, hard cores (plywood, solid wood, solid acrylic or polycarb, glass, aluminium) are generally too dense, leading to low efficiency, which is why lightweight honeycomb cores are ideal. Nomex is obviously the best, but to bond it to a carbon fibre skin is a PIA. The amount of resin required usually doubles the weight of the panel if done by hand. Aluminium core is also good, but again, bonding it to a CF skin is a b!tch. CF and aluminium are both conductive, and if an alu honeycomb core is skinned with CF, then galvanic erosion becomes a problem if they make contact.
There are so many other options...
I think one of the best is XPS skinned with Kraft paper using cold glue as the adhesive. I'd cut the XPS down to 5mm thickness if possible.
Normal Twinwall polycarb as-is, with careful edge damping, and a few coats of spray-paint, is one of my favourite solutions. I use 5mm thickness. It is anisotropic, but this could be an advantage when building very tall or very wide panels.
I believe there's a polycarbonate honeycomb available from a manufacturer in Belgium.
https://econcore.com/en/technology/technology-for-licensing/polycarbonate-panels
Now THIS is what I would consider first-prize, taking into consideration sound, ease of manufacture, longevity and price.
Have fun
I've used EPS in live gig panels. My prototype panels were skinned with Kevlar, and another pair with Dyneema. Both types of panel were extremely efficient but VERY mid-heavy. The higher frequencies don't like propagating through polystyrene. Its too soft. I believe Leob uses graphite-impregnated XPS for his gig panels. I'm not familiar with the properties of that material, but it sounds like it's worth chasing.
Solid, hard cores (plywood, solid wood, solid acrylic or polycarb, glass, aluminium) are generally too dense, leading to low efficiency, which is why lightweight honeycomb cores are ideal. Nomex is obviously the best, but to bond it to a carbon fibre skin is a PIA. The amount of resin required usually doubles the weight of the panel if done by hand. Aluminium core is also good, but again, bonding it to a CF skin is a b!tch. CF and aluminium are both conductive, and if an alu honeycomb core is skinned with CF, then galvanic erosion becomes a problem if they make contact.
There are so many other options...
I think one of the best is XPS skinned with Kraft paper using cold glue as the adhesive. I'd cut the XPS down to 5mm thickness if possible.
Normal Twinwall polycarb as-is, with careful edge damping, and a few coats of spray-paint, is one of my favourite solutions. I use 5mm thickness. It is anisotropic, but this could be an advantage when building very tall or very wide panels.
I believe there's a polycarbonate honeycomb available from a manufacturer in Belgium.
https://econcore.com/en/technology/technology-for-licensing/polycarbonate-panels
Now THIS is what I would consider first-prize, taking into consideration sound, ease of manufacture, longevity and price.
Have fun
Hello Andre,
Thanks for the reply! The nomex panel I made is heavier than optimal, but not as much as I feared. Optimally It should have been aprox. 950 g/m2 (2X200g carbon, 2x225g resin + some resin/glue/fumed silica for the nomex bonding and the nomex itself. So my 1140g is not that far off. I bonded the nomex with some extra fumed silica in the resin, a tip from easycomposites.eu . We'll see if it holds/doesn't delaminate, but for now it looks and feels better than expected.
Before I ordered the Carbon and nomex I also skinned a 2cm xps panel with paper and it was (of course! duh!) much too stiff. Nicer working with wood glue and paper though. The sound was mega mid/trebbly/screaming as you also mentioned I might try to skin a much thinner panel, but have to make a hot wire construction first.
Purely unscientific I have the feeling that the foam resonates in kind of raspy way-as if the fibres/beads in the foam rub each other. I know this isn't the case, but that's the way it sounds to me. Kind of interesting and lively, but too distorted for my intended purpose. Does the thinner plate help against this "distortion/plasticy ringin" or is it only for better transfer between frant and back skin of the panel?
I'll look into the Belgian Econcore, but don't know if it's available for "genral public".
Thanks again, Hans
Thanks for the reply! The nomex panel I made is heavier than optimal, but not as much as I feared. Optimally It should have been aprox. 950 g/m2 (2X200g carbon, 2x225g resin + some resin/glue/fumed silica for the nomex bonding and the nomex itself. So my 1140g is not that far off. I bonded the nomex with some extra fumed silica in the resin, a tip from easycomposites.eu . We'll see if it holds/doesn't delaminate, but for now it looks and feels better than expected.
Before I ordered the Carbon and nomex I also skinned a 2cm xps panel with paper and it was (of course! duh!) much too stiff. Nicer working with wood glue and paper though. The sound was mega mid/trebbly/screaming as you also mentioned I might try to skin a much thinner panel, but have to make a hot wire construction first.
Purely unscientific I have the feeling that the foam resonates in kind of raspy way-as if the fibres/beads in the foam rub each other. I know this isn't the case, but that's the way it sounds to me. Kind of interesting and lively, but too distorted for my intended purpose. Does the thinner plate help against this "distortion/plasticy ringin" or is it only for better transfer between frant and back skin of the panel?
I'll look into the Belgian Econcore, but don't know if it's available for "genral public".
Thanks again, Hans
Hello Hans
Yes 1140g/m² is a good realization performance but unfortunately seems heavy from target point of view.
Reading all the posts about DML is difficult. Let me remind the figures I have in mind :
- the efficiency is linked to D/µ³ (bending stiffness, areal density) which is homogeneous to E/rho³ (Young modulus, density)
- the thin plate behavior is such there is a trade off between efficiency and high frequency (higher efficiency, lower bandwidth to the HF)
- for room application (meaning not PA), the poplar plywood has a correct efficiency, similar to a low efficiency cone speaker. Poplar plywood density is in the range 400 to 530kg/m³ depending on the source so roughly 1.5kg/m² in 3mm thickness.
- there is also a relation based on the same parameters that gives the coincidence frequency which is the frequency at which the waves in the plate have same speed than the sound in the air. Setting it in the upper range of the desired bandwidth lead to search for material around 400g/m² in PA (probably too challenging), 600g/m² in room applications.
- the EPS which is the lightest material used here is generally in its 25kg/m³ density version (see Leob and André and others) so in a standard 20mm thickness already 500g/m² (oups... correct? I didn't remember that!). Seems correct because I have in my notes 300g/m² for the light EPS available in DIY store.
- The "cost" of additional layers of watered glue on EPS is about 50g/m²
- In the core materials tested (see Veleric or Sandasnickaren posts), there is the balsa (2mm thick seems ok). A material listed in papers but not tested here as far as I know is Rohacell.
Christian
@sand
Thank you for the link. It was the opportunity to go through it. All the papers from B Zenker about DML are very informative:
- The introduction gives a very good overview of the topic of DML, the target to get a flat response linked to the work from F Toole
- It is funny to see, they seem to have build a low performance panel in order to get get some dips that might be less pronounced by some other construction solutions. It is not said but I understand it is a clamped panel (no peripheral suspension) driven at its center. If we take the negative, it seems, this solution is not recommended...
- The proposed solution is quite simple in the end : a wall at about 10mm of the membrane with a about 25cm diameter hole centered on the exciter.
What is missing in the paper is the influence of such a solution on how the panel is appreciated while playing music in a room. The rear wave which is an element of the behavior of the panel in a room is here changed. The FR is flatter which is good but what about the overall performance.
This paper is like most not to say all the other one... like if the authors don't listen their panels.
Christian
Indeed, it`s a simple solution to try. The audition is reserved for us although… concealed from the papper(s) maybe will be translated to Zenkers brand, namely in the diy kit:
https://www.xcite-audio.com/products/
https://www.xcite-audio.com/products/
