Nice! Glad to hear the coating worked out for you as well.
Making sure the schellack is thin enough is very important, and it is amazing how so little can make such a big difference.
since shellac is so easy to thin and it dries so quickly (goes on cured) you could apply very light coats and listen in short order and repeat if necessary until you are happy with the sound. If your last coat seems too much then you only have one light thin coat to sand off.
You only need a single coat for sure. The question is how much you dilute and how you apply it. I use 4:1 spirits to shellack and apply with a lacquer brush.
Not sure how similar it is to your typical transparent all purpose glue, but usually they seems like some kind of plastic?
Probably a bit similar to epoxy, but doesn't become as hard. And same problems with diluting.
I'm a bit sceptical. Even if you manage to dilute it will become more like a plastic outer layer than a tight skin like with PVA or hide glue.
me too so I have requested technical data sheet/ We will see what they send and I will post then. I have watched some videos and it looked to be a much stronger bond than Titebond ll pva adhesive and is water resistant. Sounds like a modified pva we will see.
Hi Christian,
I hadn't posted about these ones before so no need to search.
-Eps is VH grade 28kg/m3 20mm thick 500/320mm. I got a sheet from a local packaging/building supply plastics place.
-It was the same panel that only played above 300Hz after too thick a coating of shellac, I gave it a quick sand with the orbital both sides and then it played lower.
- suspension 2 each long side approx 100mm in from corner, 1 middle of short side. I made a frame 19w/40dmm from ply, I left it 19/6mm approx at the front face then cut out the back so it is like an L. The panel is spaced out from the front by 6x 20/8mm pieces of closed cell foam, then a strip around the back edge is screwed on to allow more closed cell blocks to sandwich the panel between. Hope this makes sense, I'll take a photo. Leob's mounting as you can move the blocks/suspension to tune.
- The exciter is glued on with epoxy.
Attachments
Sorry I am new to REW and haven't figured out how to scale easily, I'll have a look.
Andre, here are the zoomed FR graphs, slightly less impressive but mostly 5db swings. The orange is the panel I posted and red is the the second panel I made, difference is probably the amount of coating I sanded off. Also the the Red exciter is brand new so maybe needs to break in a little to loosen the suspension?
Attachments
Hi JozkaNZ,
Nice results. Is one of these the one that you said only played above 300 Hz?
Eric
@JoskaNZ very nice results, thank you for sharing. Your distortion values are very low compared to what mine have been and the FR is pretty well controlled compared to many of my tests. I'd love to see some pics of your frame and mounts.
You said you sanded the panel and that extended LF. The sanding could well be the cause, but I've also found with all my testing that remounting a panel can make a pretty big change too. These panels are very sensitive to the way they are mounted, and the driver too if it is braced, and remounting it can make small changes to how pressure is applied and this can have a large effect. One of my latest panel experiments is extremely rigid fiberglassed EPS and it gives the deepest LF of all my tests. I think rigidity is required to control the power at LF.
Your peak around 18khz is interesting. I suspect it might come from your driver, and the driver could be limiting the LF a bit too. I've only got the DAEX32EP-4 and I believe it is known to give stronger LF but weaker HF. I don't know much about voice coil engineering but I've been wondering if there's a better way to make a full range exciter since it seems like there's a lot of compromises made here. I think the Xcite driver is better than these Dayton exciters, based on results others have shared.
I had an opportunity to quickly do a loud test today with one of my panels and after just 1 minute of run time at ~95dB at 100hz my driver started smelling and feeling quite warm
I also had a lot of noise from my duct tape surround flapping...
You said you sanded the panel and that extended LF. The sanding could well be the cause, but I've also found with all my testing that remounting a panel can make a pretty big change too. These panels are very sensitive to the way they are mounted, and the driver too if it is braced, and remounting it can make small changes to how pressure is applied and this can have a large effect. One of my latest panel experiments is extremely rigid fiberglassed EPS and it gives the deepest LF of all my tests. I think rigidity is required to control the power at LF.
Your peak around 18khz is interesting. I suspect it might come from your driver, and the driver could be limiting the LF a bit too. I've only got the DAEX32EP-4 and I believe it is known to give stronger LF but weaker HF. I don't know much about voice coil engineering but I've been wondering if there's a better way to make a full range exciter since it seems like there's a lot of compromises made here. I think the Xcite driver is better than these Dayton exciters, based on results others have shared.
I had an opportunity to quickly do a loud test today with one of my panels and after just 1 minute of run time at ~95dB at 100hz my driver started smelling and feeling quite warm
I also had a lot of noise from my duct tape surround flapping...One thing I've been trying to accomplish with panel experiments is to reduce the deviation from average of not-smoothed frequency response. I want the FR to look less peaky. When comparing say 1/48 smoothed SPL of a test on top of a 1/6 smoothed graph I want them to look as similar as possible. I believe (but am not certain) that the more similar the 1/48 looks to the 1/6 the better the FR will sound. Generally the fewer phase rotations there are the more even the FR will be. Distortion is very important too but I think driver and panel mounts play the biggest role here, as long as the panel material itself isn't noisy.
Impulse Response, Group Delay and RT60 are also important, though more challenging to measure because of room response. I stopped measuring panels very close (which is needed to reduce the impact of the room) because that doesn't accurately represent the listening experience of a DML panel. I got some really good clean smooth results a couple inches away from some panels which don't sound nearly that good at 6ft away. I don't even do 3ft measurements anymore because that's too close for DML imo. A nice smooth reading at 2 inches becomes a lot more peaky and humpy at 6ft because, I think, the sum of the whole panel responds very differently than the few inches of it that dominates the close measurement.
I've done a lot of tests with a very low budget in a poor testing environment and basically just as a hobby. An obsession, really, just exploring different things on a whim as I feel like it without any serious controls or notes with my experiments. What I think I've found though, for whatever it's worth, is that thinner panels tend to produce more stable frequency response (meaning 1/48 more closely matched 1/6) and more rigidity is needed to flatten the overall FR curve and handle power. Lighter weight is needed for better transient response, and then damping is needed to prevent ringing. Wider panels tend to have more 'uncontrolled' FR going on while taller ones give the needed LR.
So I'm liking the taller aspect ratio right now with thin rigid panels, but I'm still working on the mounts and damping. Epoxy on foam certainly does ring. Even bare XPS and EPS ring. One of these days I'll look more closely at acrylic panels because of Valeric's damping measurements.
Impulse Response, Group Delay and RT60 are also important, though more challenging to measure because of room response. I stopped measuring panels very close (which is needed to reduce the impact of the room) because that doesn't accurately represent the listening experience of a DML panel. I got some really good clean smooth results a couple inches away from some panels which don't sound nearly that good at 6ft away. I don't even do 3ft measurements anymore because that's too close for DML imo. A nice smooth reading at 2 inches becomes a lot more peaky and humpy at 6ft because, I think, the sum of the whole panel responds very differently than the few inches of it that dominates the close measurement.
I've done a lot of tests with a very low budget in a poor testing environment and basically just as a hobby. An obsession, really, just exploring different things on a whim as I feel like it without any serious controls or notes with my experiments. What I think I've found though, for whatever it's worth, is that thinner panels tend to produce more stable frequency response (meaning 1/48 more closely matched 1/6) and more rigidity is needed to flatten the overall FR curve and handle power. Lighter weight is needed for better transient response, and then damping is needed to prevent ringing. Wider panels tend to have more 'uncontrolled' FR going on while taller ones give the needed LR.
So I'm liking the taller aspect ratio right now with thin rigid panels, but I'm still working on the mounts and damping. Epoxy on foam certainly does ring. Even bare XPS and EPS ring. One of these days I'll look more closely at acrylic panels because of Valeric's damping measurements.
Hello JoskaNZ
Thank you for sharing the file.
It is for me opportunity to check/improve to my own measurement technique.
Basically REW records a sweep, extract from that the impulse response (response over the time of the loudspeaker AND of the room) and them makes a time to frequency transform and shows several results.
Among the views, it is interesting to keep an eye on the IR (impulse response).
On this one, we see the so call "long tail" IR of a DML (the low decay of the IR) and a strong reflection at 4ms which might come from the floor or the ceiling. Is the mic at 1m from the DML and DML and mic about 1m above the floor? Some absorbent on the floor might reduces it?
I also openend the IR windows by which the window which defines roughly the duration of the IR used in the time to frequency transformation. Here it is the values by default so the FR reflects what happens in the 500ms. That means the reflections are in the FR. The response you get is closed of the energy response or what you obtain sending a permanent pink noise in the loudspeaker and measuring with a RTA.
In the second record, the reflections seem different. Are they surface closer to this panel than the first one? It could also explain some differences in the FR
Your post arrived just at the time I am considering in discussions with Steve the role of the room in recordings or measurement of panels.
There are 2 (at least) interesting information in REW about the conditions of the measurements defined by the room.
You can go in the RT60 and clarity folders. All is summarized in the RT60 folder is you check the "show data panel" box in the controls
First information is the Topt which as far as I understand that the reverberation time. The time for the energy to decay in the room.
Second information is the clarity C50 which is the ratio of the energy in the first 50ms to what happens after.
In comparison, what I just got in my rooms is a reverberation time of about 0.5s and a C50 at 1m in mean value from 250 to 4000Hz of 12dB.
I am curious to know more about the room you are... Which dimensions? furniture or treatment? distance of the panels to the walls? Don't read I could give some advises from that... It is more to build my own understanding.
@SapphireSloth : you seem to have not as smooth FR as Joska (as I have); what about the same characteristics in your room?
@Veleric, @Leob, @Andre Bellwood : as I think you are REW users, same question; What about the Topt, C50 in your measurements? It is my topic of the day!
So maybe last (sorry to be long), there is the possibility to have an idea of what happens in the early ms so to get an idea of the "on axis response" by :
setting the window end before the 1st reflection (quasi anechoic response)
This view in my opinion is a bit confusing because it lets think the FR is fully detailed above about 250Hz which is my understanding not true. The property of the FFT is to get information for each n/T, T being the duration of the window. So if the window is 4ms, you have information each 250Hz. I hope not being wrong on that! No magic to get rid off the reflections.
An other possibilities is to set a frequency dependent window (go in Tools/IR windows)
For those both, you can remove the 1/n oct smoothing as by nature, those operations smooth the FR. You see in the above box that window is set to 4 cycles (something I read recently used in an other forum which is closed of 1/6 of octave, the line below shows 1/5.5)
An other possibility, the one I prefer, is to make a wavelet spectrogram which gives the frequency and the time information
In the colorized graph, horizontal is the frequency, vertical the time. The color gives the level. Under is the a 2D slice at a given time, here 12µ (not possible to set at 0!)
I like this view because by its principle it is close of the filter bank which is in the ear (unfortunately not the brain treatment which is after). The vertical areas in the spectrogram show modes, either panel or room. The horizontal one, the reflections
The interesting property of the DML as a wide range loudspeaker, almost all the band is time aligned.
I stop here. This post is already too long! I just exposed here some possibilities of REW I use.
Christian
PS : I see I have forgotten the "controls" box of the spectrogram for easily getting the same view
Last edited:
Hi again Joska
The post from Sapphire let me to see I didn't have a look to the distortion folder. I don't go often in this view but I recently read it is a good view to have an idea of for example some unwanted resonances or let say some unexpected mechanical problems.
2 remarks :
- In the first and third record, you have a noise to FR difference of about 50dB at the best point (somewhere between 1 and 2kHz) which is also what I have in my measurements. in the second measurement, the noise floor is much higher as if something was recorded in addition of the sweep. It is an other possible explanation to the difference between the panel in addition to the coating
- In the "free panel" measurement, the distortion is much more important than in the other tests. What is the difference? Do you have after a spine (some frame to keep stop the exciter body moving)? Or might it be the effect of the panel weight on the voice coil in the "free" conditions causing a misalignment?
Christian
It's fun to slide the gating limiter on the Impulse tab to see how that unsmoothed FR graph changes with the slide. Ultimately you do need around 80ms to see what 100hz is really measuring, and the total smoothed FR with 80ms isn't usually too much different from the default 500ms, so I don't bother messing with it most of the time.
The room has a big impact on a lot of measurements and my room sucks. I'm taking the approach of having a baseline and comparing differences between tests from that baseline. I keep thinking about moving some tests outside but that's a lot of work and I can't do it for every test so I haven't tried yet. My neighborhood isn't that quiet either.
I wouldn't say that the FR I shared last is less 'smooth' than Joska's, but my FR slopes downwards more. Here's a different test I did recently in an overlay with Joska's, just because this one looks pretty decent and I have it handy:
This is 1/48 smoothing of both with Joska's in green and mine in red. My mic is ~7ft away from panel and the duct tape surround is awful. My driver also has some noise-generating problems and my room is completely untreated. My distortion readings are far higher, RT60 and Clarity also much worse. If I fix the room and panel surround and replace driver then those figures should be closer to Joska's, though his do look really good to me. Despite the current state of my setup we can still see a pretty good representation of FR. I still don't know what is causing that 75hz hole in my test though.
I don't usually do many off-axis tests because I run through so many iterations of testing that I don't have to time or bandwidth to take and analyze them all, but in the past I've seen that any off-axis problem I see with a DML will already be apparent on-axis as long as the mix is far enough away. If the mic is too close then when on-axis it is closer to the driver that measures more cleanly, but when off-axis the mic is closer to panel edges which are more dirty.
I am never going to listen to a DML 3ft (1m) from my head so I am much more interested in the 6 - 10 ft sound where the whole panel is heard more equally. Obviously I can't measure just the panel itself without the room interacting strongly at that distance so that's why I think "relative to baseline" and try to deduce the effects that changes to panels makes between tests using this data. There were a couple weeks where I did every measurements very close, like 3 inches from panels, and I got lots of great results but later when I moved far away I realized they weren't so great after all. Not because of the room, but because of the other parts of the panel that are not right around the driver where my mic was. Moving the mic to different near-field points on the panel and running tests can show wildly different results sometimes.
The room has a big impact on a lot of measurements and my room sucks. I'm taking the approach of having a baseline and comparing differences between tests from that baseline. I keep thinking about moving some tests outside but that's a lot of work and I can't do it for every test so I haven't tried yet. My neighborhood isn't that quiet either.
I wouldn't say that the FR I shared last is less 'smooth' than Joska's, but my FR slopes downwards more. Here's a different test I did recently in an overlay with Joska's, just because this one looks pretty decent and I have it handy:
This is 1/48 smoothing of both with Joska's in green and mine in red. My mic is ~7ft away from panel and the duct tape surround is awful. My driver also has some noise-generating problems and my room is completely untreated. My distortion readings are far higher, RT60 and Clarity also much worse. If I fix the room and panel surround and replace driver then those figures should be closer to Joska's, though his do look really good to me. Despite the current state of my setup we can still see a pretty good representation of FR. I still don't know what is causing that 75hz hole in my test though.
I don't usually do many off-axis tests because I run through so many iterations of testing that I don't have to time or bandwidth to take and analyze them all, but in the past I've seen that any off-axis problem I see with a DML will already be apparent on-axis as long as the mix is far enough away. If the mic is too close then when on-axis it is closer to the driver that measures more cleanly, but when off-axis the mic is closer to panel edges which are more dirty.
I am never going to listen to a DML 3ft (1m) from my head so I am much more interested in the 6 - 10 ft sound where the whole panel is heard more equally. Obviously I can't measure just the panel itself without the room interacting strongly at that distance so that's why I think "relative to baseline" and try to deduce the effects that changes to panels makes between tests using this data. There were a couple weeks where I did every measurements very close, like 3 inches from panels, and I got lots of great results but later when I moved far away I realized they weren't so great after all. Not because of the room, but because of the other parts of the panel that are not right around the driver where my mic was. Moving the mic to different near-field points on the panel and running tests can show wildly different results sometimes.