Christian,
Yes, I do see it there too. That article references this one below, but I can't find a (free) source of that paper. It might explain the reason for the 2.5x "rule".
Can anybody find a copy of this paper?
Angus JA. Distributed mode loudspeaker radiation mechanisms. In: 108th Convention of the audio engineering society, no. 5164, Paris, France; September 2000
Hello Sapphire Slot,
I downloaded your measurements and found one think that should not help in understanding/improving those panels : in the IR view, the 1st reflection occurs at 2ms when for about 1m distance, it is expected at about 4ms (see here Floor/Ceiling Reflection Calculator )
This reflection enters dips and peaks in in the FR at frequencies higher than expected. In which quantity, I don't really know...
Is there some wall 50cm from the sound path or is the panel close to the foolr?
Hello Eric,
Seems there is no free source for this paper.
From the same author in the 109th convention : DISTRIBUTED MODE LOUDSPEAKER RESONANCE STRUCTURES. Sometimes the content are similar... Maybe you have already it.
For the 2.5 "rule", I don't remember an explanation. The origin might be in Azima's paper.
The frame when testing has panel 20" (50cm) above floor and 4ft (120cm) in front of TV. I do have this fence to keep the toddler out right behind panel which has been present for every recent test. The mic sits on the corner of the couch cushion seen in the pic, but centered when testing and 3ft (1m) from panel. For close-mic I hold the mic with one hand about 5mm or so from panel with two fingers against frame to steady my hand.
It explains the 2ms. It would be better to have the panel and the mic about 1m above the floor to avoid confusion between the panel behavior and the reflection from the floor... but this might be not easy in your environment.
Thanks for posting Steve, do you recall how thick your low density EPs panel is ? Would that panel be the typical 1 pound density foam? Thanks.
He'll moray James.
The eps panel I was referring to, was the one in this video on the left, which was 1cm thick.
Maybe I should do a video of me actually measuring the panel, to make things clearer.
A good dml panel should be able to reach 10k without problems.
Some types of panels have a higher output above 10k than others.
Sometimes too high.
These are usually harder more rigid panels, which are heavier.
Steve.
The eps panel I was referring to, was the one in this video on the left, which was 1cm thick.
Maybe I should do a video of me actually measuring the panel, to make things clearer.
A good dml panel should be able to reach 10k without problems.
Some types of panels have a higher output above 10k than others.
Sometimes too high.
These are usually harder more rigid panels, which are heavier.
Steve.
Tbh I'm not sure it is a noticeable improvement over the naked basswood ply, I'm still experimenting. When starting out it seemed like the ideal material for a panel would be graphite, due to the stiffness/weight factor. Iirc Loeb found the graphite enhanced eps more favorable (don't want to use foam though). As graphite sheets are not common, I thought the next best option would be to make an approximation by building up layers of graphite — & other seemingly ideal materials like acrylic paints (titanium white) — over a thin substrate. My thinking was also that maybe multiple types of materials in small amounts would help fill out the FR as they each have their own unique sonic characteristics. However, I have seen this basswood ply being sold to luthiers, & it sounds great already, so my treatment may be unnecessary. I will say that the panels with copper powder added in seem to have more energy in the 'presence' region. Of course I wonder if the copper interferes with the voice coil. ???
The finishing also serves a visual purpose, as I don't want to see the untreated wood.
But the stiffening factor is not nothing, the panels seem more robust & don't want to 'flap' around, so to speak. Especially as I go larger.
And yea, this is more of an art project for me, a way to bring some aura to my virtual piano compositions.
-B
Advice please anybody who's got a USB mic.
I recently acquired the Dayton UMM6. I've loaded the calibration files and the measurements are good compared to the previous mic I was using.
But how do you set up loopback for timing references? I've tried everything!
I recently acquired the Dayton UMM6. I've loaded the calibration files and the measurements are good compared to the previous mic I was using.
But how do you set up loopback for timing references? I've tried everything!
Thanks SS, but that doesn't give an accurate enough time delay calculation for the mic/speaker distance.
The problem is that the USB mic hogs both the right AND left inputs. Previously, using a phantom-powered measuring mic, I could use a USB convertor, I could feed the mic into one channel, and then loop the other channel input to the left output, and get a solid timing reference.
The problem is that the USB mic hogs both the right AND left inputs. Previously, using a phantom-powered measuring mic, I could use a USB convertor, I could feed the mic into one channel, and then loop the other channel input to the left output, and get a solid timing reference.
Eucy,
I'm not familiar with Midas but it looks like it has good capabilities. I'd love to compare results some time. Christian, Pway and I made several comparisons and got very similar results. It was pretty informative for all of us. Maybe we can figure out if/why your results were weird. Tell me the properties you used, dimensions, etc and I'd be more than happy to run a quick model on LISA for comparison.
Eric
Andre
Forgive my ignorance, but what makes the timing reference so important? I’m not suggesting it isn’t, just wondering when and why it is? When I was using close mic measurements to show how wave speed varied with frequency, I tried a bunch of different reference choices and the results were indistinguishable. So I stopped worrying about it.
Eric
Forgive my ignorance, but what makes the timing reference so important? I’m not suggesting it isn’t, just wondering when and why it is? When I was using close mic measurements to show how wave speed varied with frequency, I tried a bunch of different reference choices and the results were indistinguishable. So I stopped worrying about it.
Eric
Thanks Eric.
I use it as record for measuring mic/speaker distance. It's just so much easier to duplicate the exact setup when re-doing old measurements.
I suppose it's more to do with measuring room impulse reflections, but I'm not too concerned about that yet. I'm just surprised that the USB mic doesn't allow for loopback.
I use it as record for measuring mic/speaker distance. It's just so much easier to duplicate the exact setup when re-doing old measurements.
I suppose it's more to do with measuring room impulse reflections, but I'm not too concerned about that yet. I'm just surprised that the USB mic doesn't allow for loopback.
The loopback thing has always been a 'just-so' issue for me ever since using the original LEAP/MeLLiSSA/CLIO and CALSODs pre-Windows on DOS in the 1990's. I've always used loopbacks by default—never even considered doing measurement of any kind without a fixed timing reference. Also, because of the various delays through the sound cards and converters and things, it's impossible to line up phase measurements correctly without it.
Last edited:
Update:
It's not possible to do loop-backs with USB mics.
Ok, so as long as I'm not designing cross-overs, I won't need to time-align various drivers to get aligned phase information. So it should be ok for cross-over-less DML measurements.
Funny, it's something I've taken for granted and I've never even considered working without loopbacks. I feel like I'm sky-diving without a parachute.
It's not possible to do loop-backs with USB mics.
Ok, so as long as I'm not designing cross-overs, I won't need to time-align various drivers to get aligned phase information. So it should be ok for cross-over-less DML measurements.
Funny, it's something I've taken for granted and I've never even considered working without loopbacks. I feel like I'm sky-diving without a parachute.
Yeah, USB mics are quite limiting. Also doesn't seem to work with longer cables, so really prefer a regular mic to be able to move around and do measuring at a location as well.
I never felt the need for loopback for doing measurements in REW, but it is necessary to able to do transfer functions in realtime with Open Sound Meter or Smaart. I find REW more useful for both speaker design and room/system tuning. Realtime transfer functions are quicker to work with though.
I did find out that I can plug in my USB measuring mic directly to my phone for doing decent recordings though.
I never felt the need for loopback for doing measurements in REW, but it is necessary to able to do transfer functions in realtime with Open Sound Meter or Smaart. I find REW more useful for both speaker design and room/system tuning. Realtime transfer functions are quicker to work with though.
I did find out that I can plug in my USB measuring mic directly to my phone for doing decent recordings though.
I did some more A/B testing between my tower speakers and DML in stereo and I think I've decided to not use DML for my left and right speakers. The clarity of the center image is so much better with the towers and nothing I tried could get the DMLs close. Damping behind, baffles, toe & position, all brought improvements but still not good enough. Maybe a really large waveguide could solve the problem but I don't have the space for that. When listening to a single DML vs tower the DML easily wins but in stereo it's a different story.
Maybe I'll use a DML for center channel later, seems like that's a good use for it. I'm looking into baffleless / no-baffle now for left and right, to retain the open and no-box sound, though I am concerned about the directivity being too wide and encountering the same issue as DML: unfocused center image. At least if I buy drivers for no-baffle and I don't like it I can just put them in a baffle or box 😛
Maybe I'll use a DML for center channel later, seems like that's a good use for it. I'm looking into baffleless / no-baffle now for left and right, to retain the open and no-box sound, though I am concerned about the directivity being too wide and encountering the same issue as DML: unfocused center image. At least if I buy drivers for no-baffle and I don't like it I can just put them in a baffle or box 😛
Imaging can only be influenced by an imbalance of the left/right signal to the ears.
If your speakers, no matter how cheap they are, if they present exactly the same Fs, phasing and group delay, and reflections to the ears, then your image should be solid.
Keep in mind that, in some ways, DML panels might reflect off the rear wall a bit more than a boxed speaker. But I'm not sure if this is a good thing or not.
If your speakers, no matter how cheap they are, if they present exactly the same Fs, phasing and group delay, and reflections to the ears, then your image should be solid.
Keep in mind that, in some ways, DML panels might reflect off the rear wall a bit more than a boxed speaker. But I'm not sure if this is a good thing or not.
Yeah, I think the issue is too much reflection from everywhere, and DML have really wide directivity which makes that problem even more apparent. If I put lots of absorption on the walls that might resolve the problem, as could having a much wider room, but I can't do either right now. I'll move in a year though and maybe then I can revisit this.
I just realized that the inside vs outside test showed this problem pretty clearly on the IR graph, as does the close-mic vs 1m-mic:
(green is outside, red is inside, both 6ft (2m) mic distance)
When measuring a DML in the test position from 3ft (1m) and comparing to a 3" planar there is a big difference in the room effect on the impulse graph:
(red is planar in a box, green is planar no-baffle, purple is DML)
When I changed the DML freq sweep to 1kHz-20kHz this impulse response got a bit worse, so it's not LF causing this.
I just realized that the inside vs outside test showed this problem pretty clearly on the IR graph, as does the close-mic vs 1m-mic:
(green is outside, red is inside, both 6ft (2m) mic distance)
When measuring a DML in the test position from 3ft (1m) and comparing to a 3" planar there is a big difference in the room effect on the impulse graph:
(red is planar in a box, green is planar no-baffle, purple is DML)
When I changed the DML freq sweep to 1kHz-20kHz this impulse response got a bit worse, so it's not LF causing this.