Announcement

Collapse
No announcement yet.

DIY Flat Panel Speaker Love

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • muxelles
    replied
    I finally figured out the CNC and here's a first draft. The internal surface is a golden rectangle at 32x20cm and milled to 12.5mm thick. A 25mm border with rounded edges remains. I also have a plain rect of the same size, both treated with a 50/50 PVA glue coating. Compared to the plain rect, the milled version definitely sounds clearer and brighter. But I want to be able to quanitfy these results as I continue experiments on larger panels and different shapes.

    What is the best practice for creating freq response charts for speakers? I have a Dayton iMM-6 callibrated microphone (https://www.parts-express.com/Dayton...ophone-390-810) and an iPhone with StudioSix AudioTools/FFT analyzer. But I'm not sure what settings to use to get results that are easily comparable with what I see on the web.

    Here's what I have:
    1/3 octave graph smoothing
    Full Range FFT
    Flat Plot (vs A-wtd/C-wtd)
    Hamming window (vs none/Hann/Blackman)

    Mic distance=1 meter
    Generator=white noise (vs Sine/Square/Pink)
    Does anyone have better suggestions?

    Leave a comment:


  • Unbiasedsound
    replied
    Also that picture I posted above of the Bertagni sm100 speaker is a 2 way as it uses a 16mh capacitor for the exciter in the upper corner, while the exciter in the middle is used for the lower frequencies. There is no inductor for the low frequency exciter, instead Bertagni uses some type of glue/damping adhesive to tune it for low frequencies. 16mh capacitor is around 1200hz so I am assuming he tuned the low frequency exciter to play around 1200hz.

    Those small brown dots are basically weights and act to reduce waves from reaching the high frequency exciter.

    Also the Bertagni panel has various thicknesses throughout the panel with grooves and channels to control and tune it.

    Bertagni's panels are a step up (maybe 2 steps up) from the standard DIY DML panels.

    Bertagni was ahead of his time in bending wave technology and one of the pioneers that I base some of my techniques on.

    Right now I am working on my new Flagship design and my new design will blow Bertagni speakers out of the water. LOL This new design is even better then the two previous videos I posted. If I have time I might post a video of my new flagship design.

    Leave a comment:


  • Unbiasedsound
    replied
    Technically the outer shape of the panel is a rectangle. Rectangles have corners so to isolate the corners a channel was made to (prevent) lessen the vibrations from reaching the four corners of the panel.

    For EPS and XPS 12mm (1/2inch.) is a good choice. Thinnest I have used for EPS is around 6mm (1/4inch).

    Leave a comment:


  • muxelles
    replied
    Originally posted by Unbiasedsound View Post
    What gave you the idea for that particular shell shape?
    For the shell shape, I was following a golden rectangle tutorial on fusion360 and then extruded the outline to fit the XPS foam. I suppose the intuition is that the exciter has a different resonant length for each polar angle--maybe you get a smoother freq response? Of course, you're right about the sharp angle at the boundary. I didn't know what to do with that.

    I like your explanation about the intuition for the spline and the free-edge radiation. but what is the mindset for the various shapes milled into the examples you have shown? and what is the intuition for the ultimate thickness of the milled section of the XPS? I chose 12mm randomly...

    Leave a comment:


  • Unbiasedsound
    replied
    What gave you the idea for that particular shell shape?

    I have mounted the foam to the frame on the inside edge before as this is how its done in one of Bertagnis designs. This will allow more vibration waves to travel to the outer edge giving it a more holographic image as sound will be radiating from the front/back as well as the (outer edge) sides almost like a ominipolar. Mounting the foam to the frame from the outer edge will have less holographic image because sound will be radiating mostly from the fron and back with less vibration reaching the outer edges. This is one reason if mounting on the outer edges of the panel I suggest using 1 inch strips of foam with 1-2 inch gaps inbetween so some sound can come through the gaps, again personal preference.

    A spine is to hold the exciters magnet in place so that it does not sag and to give it something to push off of to transmit more energy to the panel the same way a conventional cone driver uses a basket to hold the magnet in place. Without a spline at high excursion levels the magnet will rattle because there is nothing holding the magnet in place. Bertagni knew that if the magnet was not held in place the voice coil would sag under the weight of the magnet as well as rattle because there is nothing holding it in place to push off of which is similar to a BMR drivers design that is built more like a conventional cone driver. Adding a spine will change the sound a tad bit making it sound more prominent then diffused again its personal preference.

    The spine should be mounted to a frame. Look at photos of Bertagnis speakers on the web and you will see what I mean.

    Click image for larger version

Name:	image.png
Views:	212
Size:	821.9 KB
ID:	1491584

    I believe Yamaha used the shape of bertagnis panel design to create there elephant ear shaped speakers. If you notice the "EAR" shape in that pic it has no sharp corners but more of a rounded shape. This is the reason for ROUNDING the corners. Sharp corners will add sharp peaks in the high frequencies. This is the reason most conventional cone drivers are shaped like a circle as there are no sharp corners which can cause peaks at certain frequencies especially the mid to high frequencies.

    Leave a comment:


  • muxelles
    replied
    Originally posted by Eipromb View Post
    After doing a ton of reading and lots and lots of trial and error, I managed to get my version of a DML working.

    Here are pics, but please ignore all the dust everywhere! It’s a mess in the den.
    I'm inspired by your pics! I've got some 1in XPS boards and a new membership to a fab lab. I'm looking to learn how to CNC and this is a great opportunity.

    I've drawn up a couple of prototypes on fusion360 that are based on the golden ratio. Both have a 75mm circular mount area with markings for a 34mm exciter. The resonant surface is set to 12mm thick.

    Click image for larger version

Name:	DML rect.2.png
Views:	199
Size:	47.2 KB
ID:	1491577 Click image for larger version

Name:	dml shell.png
Views:	202
Size:	35.0 KB
ID:	1491578



    For all the DML vets and experts, please feel free to toss out any suggestions/comments that may improve the design. I've got plenty of questions, for example:
    • straight or tapered edges down to the resonant surface?
    • why is the exciter mount area extruded, for extra strength?
    • what is the optimal thickness for the resonant surface. why 12mm/0.5"?
      • should the resonant surface be flat?
      • sculpted in 3D with additional bracing?
    • would it be better to mount directly on the 12mm resonant surface? or cut it down to something in-between?
    • what is the optimal size of the exciter mount area? I gave it a 20mm border for no particular reason
    • for the rectangular cut:
      • the exciter is centered at 61.8% of the X/Y dim of the resonant surface, this is the golden mean ratio.
      • the border is 1/8 of the short length of the resonant surface
      • the corner radius is 1/4 of the short length of the resonant surface
    • for the shell cut:
      • the shell curve is also based on the golden ratio/spiral
      • I have no idea how to manage the boundary conditions between the beginning/ending points of the spiral. a straight line was drawn, but I'm open to suggestions.
    • has anyone considered mounting the "frame + foam" against the inside edge?
    • what is the purpose of the "spline"?
      • To anchor the exciter housing such that more energy is transmitted to the resonant surface?
      • what should the spline be anchored to? the surface anchoring the foam strips?
    Toss out your theories and I'll try to incorporate them before I cut. My XPS boards are 405mm wide, so that will limit the size of the prototypes.
    Attached Files

    Leave a comment:


  • Unbiasedsound
    replied
    Thanks. Yup, the prices for most exciters have almost doubled in which I cant afford it anymore so my future development of DML projects are put on hold until the prices come down or some how I become rich. lol

    Leave a comment:


  • AcidWash
    replied
    Originally posted by Unbiasedsound View Post
    These are my higher end DML panels playing. DML - YouTube
    Nice job. To me, the prices on the exciters have reached a level where the cost/benefits don't make sense. Too bad.

    Leave a comment:


  • Unbiasedsound
    replied
    These are my higher end DML panels playing. DML - YouTube

    Leave a comment:


  • Unbiasedsound
    replied
    No, but I have used something similar from Lowes called coroplast. I didnt care for it. If you do decide to try that carbon core I would suggest the 6mm thickness.

    Leave a comment:


  • robstivers
    replied
    Ok... I've been going down the rabbit hole here and on diyaudio trying to glean all the info I can about how to pull off a successful build. I was going to start with 1/2 inch formular xps panels, but then I stumbled across another material that's reasonable in cost and looks like it might be promising. Does anyone have any experience with this material? https://www.carbon-core.com/product/...eycomb-sheets/

    Leave a comment:


  • Unbiasedsound
    replied
    What DML material are you referring to, EPS, XPS, Wood, Canvas, Card board, Paper, Plexi glass, honey comb composite, other? Even conventional cone driver materials are not the same, some have heavier/thicker diaphragms and some have very thin/light diaphragms some have more damping properties and some have less while some are aluminum, paper, ceramic , Kevlar, ribbon, poly etc...but they all basically work the same just like DML/BMRs.

    A DML is just a exciter attached to a diaphragm while a BMR is like a exciter but has a basket holding the magnet in place while the diaphragm is supported by some type of surround material just like a conventional cone driver. My DML design mimics a BMR driver as it has a spine to hold the exciters magnet in place while the foam surround supports the diaphragm that is attached to the frame.

    This technology has been around since the 50's or earlier, its not new. To many people are over analyzing and or thinking that these DML/BMR drivers work differently then conventional cone drivers but in fact they are more similar then not.

    Leave a comment:


  • zx82net
    replied
    Originally posted by Unbiasedsound View Post
    Like I said before a good sounding DML panel is all design dependent as the better ones design the better the sound. If one knows how DML's actually work and function its easy to figure out how to tackle transient response. Its pretty similar to conventional cone full range drivers like Fostex, tang band, mark audio etc. Most of those drivers size are 8inchs and under because its all about physics. Even DML's have to abide by those laws if you want them to sound right. Most people who build DML start off with a 2ft. (pink) panel then go to 4ft. and then max 8ft. They dont even make 4ft. or 8ft conventional cone drivers and for good reason especially if using it as full range. This is why I stress smaller panel size as DML's are very similar to conventional cone drivers.

    Also did you listen to my sound clip, and if so what are your thoughts?
    Somehow, I'd missed the post with the clip before, browser didn't refresh or something. I've listened to it now, thank you for posting it. TBH, I don't put much faith in cell phone recordings. The treble sounds perhaps a bit smoothed over, but I'd rather attribute that to the recording, rather than blame the speaker.

    As a physicist, I find it a bit frustrating when people say "it's all about the physics" without citing any physics. Fact is, the DML work I have seen documented on the internet has been methodical engineering trial and error, which is a perfectly valid mode of design, but it's not really science. There are similarities in operation between a speaker cone operating in heavy breakup and a DML, but the material is different, the radiator size is an order of magnitude different, and the radiator thickness is closer to 2 orders different. So, I don't think you think you can assume the performance is similar just because they both exhibit resonant modes. I'm not knocking DML's, I'm genuinely interested to see some deeper analysis of how they work.


    Leave a comment:


  • Unbiasedsound
    replied
    Like I said before a good sounding DML panel is all design dependent as the better ones design the better the sound. If one knows how DML's actually work and function its easy to figure out how to tackle transient response. Its pretty similar to conventional cone full range drivers like Fostex, tang band, mark audio etc. Most of those drivers size are 8inchs and under because its all about physics. Even DML's have to abide by those laws if you want them to sound right. Most people who build DML start off with a 2ft. (pink) panel then go to 4ft. and then max 8ft. They dont even make 4ft. or 8ft conventional cone drivers and for good reason especially if using it as full range. This is why I stress smaller panel size as DML's are very similar to conventional cone drivers.

    Also did you listen to my sound clip, and if so what are your thoughts?

    Leave a comment:


  • zx82net
    replied
    Originally posted by Unbiasedsound View Post
    No, I have not.

    Blurred?
    I've not heard good ones or seen transient response measurements, so I have no authority here. People make amazing claims for them, but transient response seems like it must be a challenge: The radiating area is very high, leading to low displacement (a very good thing), the total moving mass is enormous compared to a normal tweeter (a bad thing), and you have to deal with propagation delay across the surface of the panel (a bad thing). I wonder how these things bear out in practice. Since material properties of the favored panel materials are undocumented and variable, I doubt modeling helps much.

    Leave a comment:

Working...
X