Blog Forums DIY Speakers and Subwoofers First 3 way design

5 voices
19 replies
  • Author
    Posts
    • #14329
      dameo182
      Participant

      I’ve made a few MTM speakers now and a pair of 2 ways speakers (waiting on parts to finish those) and now my attention is on some descent 3 way speakers. It’s a whole new level for crossover and design so I’m here for (hopefully) some advice. Still being a newbie it’s a bit daunting to start the design of floor standers, I want to have the subwoofer firing down from under the cabinet to stop little fingers from investigating the cone, but I’m not sure how that would affect the sound. I will draw up the design on sketchup when I get time, but I realised I still have basic questions that I dont know the answers to yet. 

      First one being, how to take accurate measurements of the sub to make an accurate crossover design. I think I know that measurements under 200hz aren’t accurate, so wouldn’t that make the method of measuring on tweeter axis at 16″ away in accurate? Also I have no idea how to splice measurements in omnimic yet. 

    • #14331
      ajc9988
      Participant

      @Dameo182 – I would recommend checking out the video on testing subs by Erin’s Audio Corner. You want to use ground measurements. I also noticed you may have a misunderstanding about why below 200Hz isn’t reliable.

      First, here is Erin’s video.
      https://youtu.be/hyaco1vbgXg
      https://www.erinsaudiocorner.com/subwoofer_testing/cea-2010_subwoofer_testing/

      The CEA information is on how to properly test subs. This is doing ground measurements.

      Now, sound obviously can reflect off of objects, walls, ceilings, etc. Nature of sound. So, what reviewers will do is gate off when they do frequency sweeps. Why? Because it excludes all information after a certain time point. What that does is remove all the reflections from the measurement, meaning you only have what the speaker response is. Now, due to the amount of time, the frequency amplitude, etc., you can get an accurate idea above 200Hz, but that is because you gated out the reflections. Below that, due to numerous factors, you cannot get an accurate reading. And to get 200Hz, you still have to select the proper gate to filter out all reflections after that period of time. Longer or shorter gates can cause the resolution to change on how low you can go and still have accurate readings.

      Recently, New Record Day discussed his testing methodology. In it, he discusses at some point gating and the ms used by Danny of GR for the gating. I cannot remember the value, so am posting the video here for a reference if you feel like searching for it.

      Here is the Laurie Fincham interview about resonances. This will matter a bit if you plan to go really deep with adding a subwoofer like I am.
      https://www.youtube.com/watch?v=oRSXhsIy14g

       

      Here is my build: ( https://toidsdiyaudio.com/community/diy-speakers-diy-speakers/dinas-big-brothers/paged/5/#post-7815 for my current build, though life happened, funds go bye bye, and so I will be putting the drivers in an open baffle to use while waiting for building the enclosure, which will take 14 sheets of MDF, give or take, for the 4 speakers (a pair to my parents)).

      One thing to remember, unless building a humongous box to accommodate 3-4″ insulation for the sub, you will never have a perfect box for the sub. And adding 6-8″ in any dimension is huge. There is constrained layer damping and good bracing for the low end. Just be aware of potentials for resonances. As to mids and highs, as Laurie Fincham put it, you can brace to your heart’s content or try to reduce duration of resonance (constrained layer damping (“CLD”)), but nothing takes the place of preventing the resonance in the first place. That is done through insulation. So, for the mids and highs, insulation can truly help. He recommended 1/4″ for every 1″ span. So 8″ wide would need 1″ foam on each wall (1/4*8=2″ then divide by 2 to get the thickness for each wall for insulation). Now, this can vary with the material used for insulation, etc. He was using acoustic foam.

      You might also want to pay attention to sensitivity, as otherwise you will be using resistors to level the speakers and possibly drive up the power needed to run the speakers.

      I hope that helps. I had just learned Sketchup, then trying to do an FEA, I couldn’t get it to export where it played well with other software that can do that, so now learning Fusion 360 and trying to figure out doing an FEA there. I took the extra time to build in dado channels throughout the parts so most all can be done with my dado stack on my saw. There are a couple I cannot do that with, so I’ll still have to use the router for those.

       

      Edit:
      https://www.youtube.com/playlist?list=PLnIxFR_ey0b37Ex4KV2mBz-kYB7QLffR1
      Playlist on understanding measurements.

      https://www.youtube.com/watch?v=Ys7oylGGcgk&t=300s
      Video on time alignment of speakers.

    • #14333
      dameo182
      Participant

      Thank you for that well thought out and informative post, I’ll definately take a look at those videos you posted, hopefully I understand everything. My main issue will most likely be that I have no way to take measurements outside, so far all my measurements have been done indoors using the method that nick teaches, worked well for the mtm designs I’ve done so far. I’ll have to try to come up with a way to measure properly. As it’s late here now I will properly watch the videos tomorrow and see if I learn anything 😁

       

      As for your design, they look like some huge speakers 😂, I’m taking a different route (maybe), dont know how well it will work out but as I’m a fabricator, most of my builds involve sheet brass, copper, aluminium etc so for this build I want to use aluminium sheet, 10mm thick and welded seems, braced where needed and then painted to a high gloss finish. Yeah its more expensive but I like the uniqueness of it

    • #14334
      ajc9988
      Participant

      @dameo182 – That aluminum will have a good thickness. That is like 0.4″. Considering the stiffness of aluminum, that will definitely work. Just have to worry of resonance with aluminum with that being high enough to be in an audible range. Insulation or something to add mass might help. Hard to say as I have never used metal for fabricating a speaker.

    • #14335
      chedwin
      Participant

      @dameo182 one of my favourite sounding PA speaker companies uses aluminium and stainless steel for their speakers and it works really well. not sure about the bigger products but their smallers speaker lines I think are too shallow and/or narrow to have any form of insulation material internally and yet they still sounds amazing

      from the tiny 0.5″ driver speakers for resturant and shop installations

      https://www.k-array.com/en/product/lyzard-kz1/

      up to their flagship 8×8″ woofer 8×4″ mid woofer 4×1.5″ compression driver arena line array modules

      https://www.k-array.com/en/product/firenze-kh8/

      everything is sheet metal. interestingly part of the reason they told me for using metal not wood is to remove cabinet resonance!


      Josh Evans, Professional Live Sound Engineer, High End Commercial AV Install Technician
    • #14340
      dameo182
      Participant
      Posted by: @ajc9988

      @dameo182 – That aluminum will have a good thickness. That is like 0.4″. Considering the stiffness of aluminum, that will definitely work. Just have to worry of resonance with aluminum with that being high enough to be in an audible range. Insulation or something to add mass might help. Hard to say as I have never used metal for fabricating a speaker.

      Yeah I was thinking that the aluminium would give a good solid cabinet and look pretty good too, every speaker I’ve made so far has had some type of metal involved. Usually I build an mdf cabinet that gets clad in either 1.5mm copper or aluminium sheet, I’ve used brass before and aged it to a bronze type colour, but with this build, given the size of the cabinet I’m opting for full thickness metal instead. I just need to figure out the measurement issues before I really start designing them otherwise I could be stuck with half built speakers 😁

       

    • #14341
      dameo182
      Participant
      Posted by: @chedwin

      @dameo182 one of my favourite sounding PA speaker companies uses aluminium and stainless steel for their speakers and it works really well. not sure about the bigger products but their smallers speaker lines I think are too shallow and/or narrow to have any form of insulation material internally and yet they still sounds amazing

      from the tiny 0.5″ driver speakers for resturant and shop installations

      https://www.k-array.com/en/product/lyzard-kz1/

      up to their flagship 8×8″ woofer 8×4″ mid woofer 4×1.5″ compression driver arena line array modules

      https://www.k-array.com/en/product/firenze-kh8/

      everything is sheet metal. interestingly part of the reason they told me for using metal not wood is to remove cabinet resonance!

      I checked out those tiny speakers, theyre amazing, it’s unbelievable that something that small can give good sound. Hopefully my finished speakers will turn out how I imagine and the hard work I’m going to have to put in to learn how to do it will be worth it 

    • #14350
      ajc9988
      Participant
      Posted by: @dameo182
      Posted by: @ajc9988

      @dameo182 – That aluminum will have a good thickness. That is like 0.4″. Considering the stiffness of aluminum, that will definitely work. Just have to worry of resonance with aluminum with that being high enough to be in an audible range. Insulation or something to add mass might help. Hard to say as I have never used metal for fabricating a speaker.

      Yeah I was thinking that the aluminium would give a good solid cabinet and look pretty good too, every speaker I’ve made so far has had some type of metal involved. Usually I build an mdf cabinet that gets clad in either 1.5mm copper or aluminium sheet, I’ve used brass before and aged it to a bronze type colour, but with this build, given the size of the cabinet I’m opting for full thickness metal instead. I just need to figure out the measurement issues before I really start designing them otherwise I could be stuck with half built speakers 😁

       

      And with it being 0.4″, with aluminum having stiffness qualities about 7x that of wood/mdf, while people making towers are using around 0.75″, it will be much stiffer than wood with less flex (hence higher frequency for resonance).

      When I brought up resonance, I didn’t mean it wouldn’t sound good, just wanted to give information on the resonance related to the material used so that you could keep an eye open during the design phase.

      So, I would start, as I always do, with throwing the T/S parameters in whatever program you enjoy using to estimate box volume needed, then work that backward with the face and basket/diameter of the speakers to see how close you want the drivers to each other and the distance overall needed, that way to constrain the height dimension, so you know you cannot go below driver1+ … Driver n + buffer space above + buffer space below, which usually helps me. After I get one dimension mostly settled, I can look at how to apportion the width and depth. I do similar with width, using the diameter of the largest driver + a buffer on either side of that driver. That leaves how deep the speaker has to be. I then, if I don’t like the depth, change the others until I reach a measurement I like.

      Now, if using insulation or bracing, you may want to estimate the volume of those, add those to the needed box volume, and then solve for the dimensions. I do not know your process, but that is roughly what I do with mine. Once you have the rough dimensions, you can do the estimate on how much aluminum you will need.

      I’m pretty sure you know your process, but wanted to mention how I would do it. I apologize if this came off as anything other than sharing my methodology, as I did not mean it in any other way (I realized after typing this it could come off as patronizing, etc., when that was not my intention, but I am too lazy to revise what I typed, so leaving it and adding this to apologize if seen as other than what I meant).

      I do look forward to seeing what you come up with! Do you have measurement equipment like a calibrated microphone and DATS? I would love to see some measurements on this!

    • #14353
      dameo182
      Participant

      @ajc9988 

      Dont worry i realised you were just giving your method of designing and yeah thats pretty much the same method that I use too, I mainly focus on trying to keep the dimensions different from each other to try and avoid standing waves, doesn’t always give the best aesthetically pleasing design though so I adjust to get the best compromise. I’m by no means an expert, but I think my last builds turned out pretty good. 

      I do have dats and omnimic for the measurements I take, had a bad start using them but with help from people on here I got there in the end, my laptop was causing a colouring effect to the graphs I was getting so I put the tracks onto a cd instead, seemed to be ok with that. Using some of the features of omnimic is very confusing though when you dont really have anything to learn from, there doesn’t seem to be much in the way of tutorials out there to see how it’s properly used, I learn best from seeing it done rather than reading about it. 

      I listened to those videos posted above while I was at work today, and though they are informative and good for measurements of subs, I’m not sure how to make it work for measuring a 3 way speaker in the way I was hoping, mainly because I want to take the measurements on the design axis, of the whole speaker, and to get the z offset of all the drivers at the same time. I want to use the tweeter as the design axis, I figure if I measure from any other point I’ll be starting off with slightly lower high frequency which could affect the sound after crossover design as ill be listening on tweeter axis most of the time. I dont know if that’s the right approach tbh but it’s just what seemed to me to be the right method, but it wouldn’t surprise me if I’m completely wrong on that 😂 i just hope I dont have take separate measurements and merge them because I don’t have the first idea how to do that 

    • #14354
      ajc9988
      Participant

      @dameo182 @123Toid – sounds like we have a video request showing how to use the omnimic and software features properly! More content the more better, right?

      I do not have an omnimic, so I cannot give advice on it or the features with it. I have a UMIK-1.

      As to testing, think of it this way, you need the subwoofer ground measurement to ONLY figure out the crossover point of the sub and mid/woofer. So, put the speaker on its side, then offset the directly away from the sub over to the tweeter (what normally is a vertical offset when standing, but is horizontal while the speaker is on its side). Repeat sweeps for the mid-woofer as well. Since you are likely dealing with the 60-150Hz range, below the gated response, you will get roughly what you need, while doing the measurement closer to where you will vertically do the measurement once the speaker is standing up again. You repeat the measurement standing up, this time for the woofer and tweeter. Compare the data for the woofer from the ground measurement and standing, just to see that it will roughly be similar enough that nothing is going to go wonky from doing it in that manner.

      As such, that is 4 measurements, 2 ground measurements for the low end crossover, then the 2 measurements for the mid to highs.

      In other words, even with doing the more accurate ground measurements, you could always measure in both positions, then compare and use your best judgment on when to override the ground measurements with the measurements done when aligned. You could also do the ground measurement straight from each driver, then in the tweeter position while its on its side, then in the standing position from the tweeter.

      How does that sound for an approach?

      Edit:
      https://www.youtube.com/watch?v=-WyW7S6u1FI

      This is his video on why he switched, but I think you are looking for more of a step by step tutorial.

    • #14355
      dameo182
      Participant

      @ajc9988 

      That seems like a good approach, thank you, I think I will start the driver choice and box design then I can model it in sketchup, see what im dealing with and once I’ve got that far ill upload the pics on here for critique. 

      As for the video, that would be awesome, I’m sure many people would benefit from that, the main things for me are the waterfall plots, I worked out how to get the z offset using omnimic, but couldn’t get the Spectral decay plot to show a constant graph, kept changing every other second, maybe its supposed to do that but I’m really not sure. The graph looked clean, no ringing in the drivers once the crossover was wired in, just the changing graph seemed strange. I’m sure theres a lot more to the programme I’m just nervous about trying anything after the teething problems i had with it. That wasn’t the programme though, just my inexperience. 

       

    • #14357
      123toid
      Keymaster

      @ajc9988 

      I probably should do a video on it.  I was just talking to one of my Patreons who just switched and he noticed a difference right away (in a good way). I think the Omnimic software is so easy, it makes it hard. What I mean to say is everything you need is right there for basic measurements, that you think I must be something wrong. But in my opinion it gets more useable easier graphs faster. Of course, I still like the Umik and rew.  If just can take some time getting used to the software.  It is very comprehensive. Maybe I can do a video on the basics of the Omnimic software sometime though. Not sure when I’ll get to it, but in the meantime, feel free anyone that needs it to make a post on here and I’ll answer any questions I can. 


    • #14358
      123toid
      Keymaster

      I also want to add that I think most measurements of subwoofer performance, ground plane or not is pretty moot.  At least the actual response that is.  It is more important to see what it’s anechoic F3 is (which can be done via ground plane). After that, your room is going to dictate the response between the F3 and your crossover point.  Which is why, most people firmly believe in EQ of your subwoofer. In fact, I could care less if the response has a hump, just as long as it doesn’t have a huge dip.  Humps are easy to EQ down.


    • #14359
      dameo182
      Participant

      @123toid 

       

      Thats exactly how I’ve felt when using omnimic, that it’s too easy to just open the programme and take a measurement, and that I must be doing something wrong, I had no confidence in the results I was getting at first, I’m a bit more trusting of the results now though since hearing the speakers. You’ve always done great videos in the past so having one to refer to when using this software would be really helpful so I hope you do manage to make one at some point, it would be much appreciated! 

      Regarding your advice about the subwoofers freq response being moot, did you mean I’m basically overthinking all of this? I had it in my mind that I had to get a perfectly accurate response of the sub to create the crossover to the mid woofer, but after reading your reply I’m now thinking that I could, in theory just use the method you taught me before for measuring the mtm I built. As the crossover will likely be around 500hz (correct me please if that is a bad place to cross) I should see enough of the bass response in my measurement to be able to get the crossover work done, would it be ok to work it like that? 

    • #14361
      123toid
      Keymaster

      @dameo182 

      I’m not sure where you would want to cross your sub over.  That is really more dependent on the subwoofers nearfield response and the distortion profile.  Having said that, you will still want to find your z offset.  You can do this by measuring just the tweeter and sub together in parallel and finding the offset like normal.  I would still take a measurement of all three in parallel.  It should be pretty close when all is said and done. 

      *Don’t be surprised if the below 800hz range is not lining up perfectly.


    • #14364
      dameo182
      Participant

      @123toid 

       

      I looked at the response given on the data sheet of the sub I’m considering and it looks like it starts to break up around 1000hz so i was planning to cross around 500, if that sounds right, then if I’m doing it the way you showed me before then I’ll take measurements of the tweeter, then the woofer then the sub, then one of them all playing together and match the up the responses in xsim, or omnimic, to get the z offsets of each one, and then design the crossover like normal. I hadn’t realised that I could do it that way tbh, it’s the first time I’ve done anything with a sub involved so I wasn’t sure. Does that method seem ok? 

      • I have another question though, when I come to do the measurements, should I keep the distance at 16′ or increase it as the sub has a bigger cone size? 
      • Should I take a near feild measurement of the sub in the cabinet as well to double check the crossover point? 
      • What’s the reason that below 800hz might not match up so good?

      Sorry for all the questions, I like to pick your brain when I can as I learn a lot 😃

    • #14448
      elliottdesigns
      Participant

      @dameo182 Just saw your earlier post about using 10mm aluminium. It sounds cool and all, and it will certainly be stiff, just remember that metal rings, so maybe apply mass damping or something like noico butyl rubber sheets that are usually used for damping sheet metal in cars like car doors. (For the inside of course, unless you don’t care about aesthetics, otherwise both in and out is probably best)

    • #14469
      dameo182
      Participant

      @elliottdesigns 

      I’m hoping I can design the cabinet well enough to not need that, but I know what your saying, it’s going to be trial and error anyway so once it’s built I’ll be able to take measurements and adjust as needed. My main concerns are avoiding standing waves, and creating a speaker that’s ‘shouty’ from the wide baffle that will be needed for the woofer. So to try and combat that I may design it to have a tapered baffle (thinner at the top, wider at the bottom) and also tilt the baffle backwards by about 5°. Thats all just ideas for now, I’ll have a better idea once I get it drawn up , thanks for your input though, every bit of advice is appreciated 👍

    • #14476
      123toid
      Keymaster

      @dameo182 

      16 in would be fine. I would still do a Fairfield measurement as well. And by farfield I mean 3 ft. And that’s after you design the crossover. You can also do one 10 ft back. The reason why you might want to do that is some people like their basd response hot. But if you make that flat at 16 in or even 3 ft, your bass might feel like it’s lacking 10 ft away

       


    • #14477
      dameo182
      Participant

      @123toid 

      Ok thank you for that, for the 10′ measurement would I need to adjust the gate as I’ll be measuring in room? If so whats the best time window to use for that? 

You must be logged in to reply to this topic.

©2022 KLEO Template a premium and multipurpose theme from Seventh Queen

CONTACT US

We're not around right now. But you can send us an email and we'll get back to you, asap.

Sending

Log in with your credentials

or    

Forgot your details?

Create Account