Thanks for looking and explaining Christian, I will have to re read your posts when I get the time this weekend but appreciate your time and explainations!
Good point Christian. My room is untreated and quite small, but no door in the back and just a thick drape. I also have shelves and stuff in the back or the room and it is not really rectangular. For its size it is surprisingly good though, but to really evalulate the plates I prefer to make measurements outside. Not a good time for that in Sweden now though 
I do my indoor measurements to evaluate panels at around 30cm distance, which is a bit close perhaps to get the complete picture, but reduces the effect of the room.
For actually tuning speakers for use in the room I instead use RTA and white noise and move mic around my typical listening position.
Here are my RT60 params for a plate I just built and is testing now:
Or the mdat if it is anything else you want to look at:
I do my indoor measurements to evaluate panels at around 30cm distance, which is a bit close perhaps to get the complete picture, but reduces the effect of the room.
For actually tuning speakers for use in the room I instead use RTA and white noise and move mic around my typical listening position.
Here are my RT60 params for a plate I just built and is testing now:
Or the mdat if it is anything else you want to look at:
And for comparison here is the same plate but recorded at 1m (green) compared to recorded at 30 cm (red)
I suspect the dips around 200-250Hz for 1m measurement are room modes, but that the close measurement is a bit more jagged because it is picking up some modes more than others when miked so close.
Attached mdat for 1m measurement.
I suspect the dips around 200-250Hz for 1m measurement are room modes, but that the close measurement is a bit more jagged because it is picking up some modes more than others when miked so close.
Attached mdat for 1m measurement.
Yeah the LF is where the close mic is the most deceiving, though in some cases I found the midrange to look much more controlled up close than it actually is at listening distances. I did so many tests last year with a close mic showing really nice response 80hz - 300hz and it took me a while to realize that for every one of those if I backed up a meter or 2 it looked a lot different. I usually look at 1/24 smoothing, while also checking 1/6 and none. I never took any panel outside for a test though one of these days I'll try it. I think the LF looking good up close and poor at a distance is likely caused by the panel bending so that the local pressure is highest by the driver and lowest at the edges, with the net total pressure (as measured at a distance) generally being lower or more volatile than desired.
One could argue that when measuring LF the timing window has to be so long that the room plays a huge role, however I found when testing different panels in the same position in the room some performed well where others performed poorly. I can't say this with certainty, but when we're looking at 5, 10, 15, 20 dB peaks and dips in the FR which do not go away from minor panel placement changes but do go away when the panel is switched to a different type, it seems pretty safe to say those peaky extremes are produced by the panel and not the room. Often when changing the dimensions of the panel the "problem areas" would shift up or down in frequency. This is why I'm pretty comfortable comparing panel tests that were measured at 2 meters.
One could argue that when measuring LF the timing window has to be so long that the room plays a huge role, however I found when testing different panels in the same position in the room some performed well where others performed poorly. I can't say this with certainty, but when we're looking at 5, 10, 15, 20 dB peaks and dips in the FR which do not go away from minor panel placement changes but do go away when the panel is switched to a different type, it seems pretty safe to say those peaky extremes are produced by the panel and not the room. Often when changing the dimensions of the panel the "problem areas" would shift up or down in frequency. This is why I'm pretty comfortable comparing panel tests that were measured at 2 meters.
@SapphireSloth
The room modes affect the LF most as well, so I don't think it so clear cut. What you can try is measuring close but at different positions and see how much it differs compared to measuring further away.
At least in my case, moving the mic at different locations 30cm away does not change the response as much as moving the mic further away. And I do know my room and have observed those dips in the low mid in many different measurements.
Ideal distance to use of course really depends on your room, mic and speaker positions. For me 30cm seems to be the best if I want to get a decent idea of the FR of the plate. If you are actually tuning the plates to the room, 30cm is too short, then you want the rooms effect to be as strong in the signal as your typical listening position, and then 2m might make sense depending or your room as well as speaker and listening positions. But for a measurement of the plate itself, unless you have a really large or well treated room, you would get way too much room information.
The room modes affect the LF most as well, so I don't think it so clear cut. What you can try is measuring close but at different positions and see how much it differs compared to measuring further away.
At least in my case, moving the mic at different locations 30cm away does not change the response as much as moving the mic further away. And I do know my room and have observed those dips in the low mid in many different measurements.
Ideal distance to use of course really depends on your room, mic and speaker positions. For me 30cm seems to be the best if I want to get a decent idea of the FR of the plate. If you are actually tuning the plates to the room, 30cm is too short, then you want the rooms effect to be as strong in the signal as your typical listening position, and then 2m might make sense depending or your room as well as speaker and listening positions. But for a measurement of the plate itself, unless you have a really large or well treated room, you would get way too much room information.
"Room information" I understand and believe, but for me the question is how much does it skew the panel's FR? 5dB? 10? 20? There's been lots of discussion in this thread about how DMLs are less affected by the room than cone speakers and after doings tons of different tests inside my room that idea seems to be true. Or maybe other speakers aren't affected as much as I thought. I don't have much experience measuring systems aside from these DML experiments and one set of older Definitive Technology tower speakers I've got.
One day I will haul my gear outside and test it to find out how much the room affects my measurements. Lots of rain in the forecast though.
One day I will haul my gear outside and test it to find out how much the room affects my measurements. Lots of rain in the forecast though.
Hi there folks
I have finally finished my HYBRID DML speaker project > DML + open-baffle woofer + tweeter.
I will start with the wood-work. ( I had to build with no workshop and only hand-held tools )
The first challenge was to not only make a frame to hang the panels, but also provide a quite large baffle
for the woofer + a mounting position for the tweeter.
Using some imagination, I arrived at a rather unusual & unique design.
My starting point was to provide a strong & sturdy floor base-plate > 620mm x 540mm x 25mm = 5.5Kg.
( because these speakers go strongly down to 30Hz, I didn't want them 'dancing about' )
From this base I built upwards > apart from this base-plate and a 'hanging brace', there are only 3 pieces of wood.
Here are some pictures:
I have finally finished my HYBRID DML speaker project > DML + open-baffle woofer + tweeter.
I will start with the wood-work. ( I had to build with no workshop and only hand-held tools )
The first challenge was to not only make a frame to hang the panels, but also provide a quite large baffle
for the woofer + a mounting position for the tweeter.
Using some imagination, I arrived at a rather unusual & unique design.
My starting point was to provide a strong & sturdy floor base-plate > 620mm x 540mm x 25mm = 5.5Kg.
( because these speakers go strongly down to 30Hz, I didn't want them 'dancing about' )
From this base I built upwards > apart from this base-plate and a 'hanging brace', there are only 3 pieces of wood.
Here are some pictures:
Attachments
CONTINUATION ...
When I first made & hung the 600mm x 500mm DML panels and connected them directly to the amplifier >
I was dreadfully shocked by their horrible sound. They were 'honky & resonant' and tonally wrong.
I thought - boy have I got a challenge on my hands to make these sound good.
The first stage to improve sound and incorporate the woofer and tweeter was to design 'passive EQ' into
the crossover network. This improved sound by 50%.
When I first made & hung the 600mm x 500mm DML panels and connected them directly to the amplifier >
I was dreadfully shocked by their horrible sound. They were 'honky & resonant' and tonally wrong.
I thought - boy have I got a challenge on my hands to make these sound good.
The first stage to improve sound and incorporate the woofer and tweeter was to design 'passive EQ' into
the crossover network. This improved sound by 50%.
Attachments
CONTIUATION ...
The actual sound of the speakers is better than the 'frequency response plot' would indicate.
The front & back radiation creates a multitude of reflections that show up a 'combe filtering'.
I also have some reservations regarding my measurement microphone - it had a drop to a hard-wood floor.
I should state that these speakers are not efficient and need a good 100W per ch. amp for full dynamics.
There is one great thing about these speakers > you don't need to buy or build sub-woofers,
the bass response is really quite stunning
The actual sound of the speakers is better than the 'frequency response plot' would indicate.
The front & back radiation creates a multitude of reflections that show up a 'combe filtering'.
I also have some reservations regarding my measurement microphone - it had a drop to a hard-wood floor.
I should state that these speakers are not efficient and need a good 100W per ch. amp for full dynamics.
There is one great thing about these speakers > you don't need to buy or build sub-woofers,
the bass response is really quite stunning
That looks pretty much like an open baffle sub you just built, why do you not consider that portion a sub?
Yes, contrary to boxed cone speakers, DML panels do seem to sound better than their FR's. Definition and separation of instruments does not seem to be influenced very much by the FR.
I fully understand your question & point ...
It is pretty much the use of terminology. Most (many) consider a sub. to be something in its own box.
Also, open baffle is not generally associated with subwoofers. (many say - open baffle all the way, except for sub.)
I like to think that I've just built an integrated speaker with great extended bass
Hello SapphireSloth
Yes it is what I mentioned by the 1/Window duration. 100Hz is 100ms period so need a window about that. I haven't yet compared a 80ms window to an other type of smoothing.
It is something I also observe with a daex25hfe on EPS even if the bandwidth is much better with a PVA glue exciter as Steve recommend than with the double side tape I used before.
I have currently an EPS test panel. First is was just hung by tape at its top. Terrible how this small tape ribbon was noisy. Tape is no more in my tool box for DML
Jozka's values are very far from what I have. it is why I asked for other values. The reverberation time in my rooms is about 0.5s. A bit less in the living room, A bit more in the kitchen. The C50 is 12dB in the living room at 1m (first value, it is a "fresh" topic) and only 8dB at 0.8m in the kitchen which is more reverberant.
A good additional test, we have example from Veleric, is to check the impedance curve. The peaks show the panel resonance. The first mode might be below and the next higher. An other possibility is a room effect.
Same problem here. We can generate quickly more test cases than we can analyze.
Close mic in DML can give information about what happen locally. Eric shown interesting things with that. Each area of the panel has its own behavior, acting like a small loudspeaker so some observation distance is needed to get the whole picture. In measurement of classical loudspeaker, there is a rule of thumb saying the measurement distance has to be 3 times the largest dimension of the system (ie the cabinet diagonal)... which leads to the mix with the room... and what you do to go by comparison. No magic!
Christian
Thank you for the feedback Leob.
Your room seems also less reverberant than mine and in consequence I guess, the clarity is higher.
Our environment (home) is like it is but it is interesting to have an idea of how reverberant it is, including the place we make measurements.
Just to come on the "methodology" aspect which was mainly the topic in my previous post, The distortion folder of your measurements shows a good noise floor and a peak of distortion at 4,2k which matches with a peak in the FR. Something happens here...
Most of the time, I don't have this look to the distortion for my own measurements!
So I just opened files of a panel I am playing with at the moment to see the distortion curves are awful compare to yours...
In a first one, when the panel was tape suspended, it was clearly audible. With a string suspension, it was improved. I had a doubt if it was fully fixed... I see it is probably not!
Lesson of that. REW is an excellent too. Go to the different views to collect all the information. If something looks weird, it is weird.
I have also done some outdoor measurements. Not easy to do. Need the right weather : even a little wind is a problem, for the mic, against the panel. No noise around.
Those days, I am testing different mics and preamp so I have a look to the noise floor. Even a bird singing outside, all the doors and windows closed changes the curve (not a lot, just to say all of that is very sensitive)
Christian
DML certainly makes issues with the room less obvious, but there is mainly two aspects of room acoustics. The room modes and the reflections.
The modes are the points where you get standing waves at frequencies with a wavelength coinciding with the room dimensions. Those will happen in low frequencies and can result in large shifts i response when moving around only by a few cm as some positions will have modes making a certain frequency stronger, and others anti-modes making it weaker. These are the hardest to treat, and where I think DML will not make that much difference. The higher the frequency, the more modes can fit on a plate, and you get in effect multiple sources for the same frequency, but at lower frequencies it will become more like a single wave and DML becomes more like a regular speaker.
When it comes to reflections, that is the same thing in some ways as the modes, sound source is arriving at multiple different times to the listener. But at higher frequencies we do not experience it as changes to the FR as we move in the room, but we experience it as the sound bouncing around like delay and reverb, and rather than prominent dips at certain frequencies it will have a comb filter effect. This is easier to treat with just some acoustic plates on walls and ceiling, and DML makes a big difference since you have reflections of a bunch of sources (each mode on the plate) arriving at slightly different times, meaning we do not hear the individual reflections. When analysing this is more obvious in decay time measurements like the RT60, or the impulse response, but can be hard to spot in FR.
Also, our ears does a lot of quite advanced processing. We analyse the timing/phase inconsistencies caused by the shape of our ears to locate the source of a sound, which if you think about it is some pretty complicated DSP. We expect sounds to be phase incoherent, since that is how pretty much all sound is generated in nature. And we never have the exakt same sound generated from two sources at the same time, so that is why reflections can be so jarring and disturbing sometimes, it really confuses our hearing and becomes an effort to try to parse the data.
So how a source measures from a mic at a specific point and how we perceive it are two different things, and benefits of DML has very much to do with how our hearing will process that kind of phase incoherent signals arriving at our ears.
Yes, I noticed there was some audible rattle there and I'm gonna investigate what it can be. A bit unusual at that frequency. If something is rattling, typically that happens in lower frequencies.
I think this could be true. There's something about the sound of the DML that is very different from a "normal" loudspeaker.
My brain might be reacting to absence of the usual "something is not completely right" in speaker reproduced sound.
Even when the FR of the DML is quite nonlinear , it still "feels" better to me (within bounds of course).
Last edited: