Blog Forums DIY Speakers and Subwoofers My first design: questions, pointers etc

last updated by jrand 5 months ago
4 voices
20 replies
  • Author
    Posts
    • #23006
      jrand
      Participant

      Hello everyone!

      I am designing a pair of floor standing speakers, at first I thought of trying to copy Toids <300$ speakers (https://toidsdiyaudio.com/2021/02/23/diyspeaker/) but have since then thought of a slight upgrade.

      I have a few questions, first about Toids speakers, and secondly about general design.

      So, starting with Toids speakers, in the build he says that the drivers used are the RS180S-8, however during the video he shows the specs of the RS180-4, citing the vented volume for a response down to 33 Hz. When in winISD, however, I can only get the F3 of the 8 Ohm version (actually the non-shielded version, RS180-8, which still has a slightly lower frequency response) down to 36.6Hz while I can get down to 34.4 Hz (sounds like splitting hairs though) with the RS180-4. This is for 55L (1.94 ft3) boxes with filters applied to limit the cone excursion.

      I am convinced that indeed he used the RS180S-8 for the build, because I would imagine that that is also important for the crossover design as the impedance would be quite different. But as I started playing around with winISD, I saw the website in the video and used the RS180-4 to a satisfactory result (I would be happy with a 35Hz response) but later realized that the driver was actually the 180S-8 which is has a slightly higher lower-frequency. Not so much of a question but more an indirect hint that I would love to know way more about the design behind those speakers and also their final response at the lower end.

      Regarding the actual design of those speakers, I do have a question though, which is the fact that he paired a driver with a rated sensitivity of 86.9dB/2.83V/m (RS180S-8) with a tweeter that has a sensitivity of 93.5dB/2.83V/m (RST28F-4). However in the final crossover theoretical curve, the two speakers have a flat response at around 85dB. Is the 7.5 Ohm resistor of the crossover used solely for a sensitivity matching , or is it an important part of the actual high-pass? (I still did not delve into crossover design). My intuition tells me that the resistor also ends up raising the impedance of the tweeter side of the circuit, and I could imagine that this is also a way to match the impedance between an 8 Ohm driver and a 4 Ohm driver.

      I thought that one would try to match the sensitivities of the used drivers for a flat response, but in this case the chosen drivers have a 6.5dB difference in sensitivity. So, does this mean that In principle, I could always try to pair drivers with different sensitivities and then make a crossover that brings the curve of the louder driver down?

      Now, questions regarding design in general, because I am thinking in doing a slight upgrade of the build to the RS225-8, as I can get an F3 of 31 Hz with those, and in terms of price, where I am located the jump from the RS180-8 still makes it within my budget.

      Firstly, when I am designing a box in winISD, from following Toids tutorials, I add highpass filters to limit the excursion of the cone at my rated power, however I do not understand where these filters will be in the actual speakers. Toids did mention that most amplifiers have either way a built-in 20 Hz cut-off filter, but how do I set any other necessary filter to protect the driver? Is it with an EQ, in the amplifier, as DSP? As an example, on my current design for the RS225-8 box, I use two 2nd order butterworth highpasses, one with a cut-off of 20 Hz and the second with a cut-off of 34 Hz. Where would I set these up? I think they are called infrasonic filters? (As you may be guessing, I am very far removed from the audio world). This is quite important for me, as while tuning the things in winISD I would like to know if I could add an arbitrary amount of these highpasses, or change their order, cut-off etc, so I need to have a more accurate overview of what I will be able to do when setting up the actual pair of speakers.

      Secondly, after looking a bit around, it seems that if I do indeed would go for a pair of RS225-8s, and keeping it a two-way design, I could keep using the RST28F-4, but modify the crossover to push it down to the minimum of the tweeter at 1.4~1.5kHz. Is this a good idea? Toids speaks about keeping the distortion low, but I guess that a 3rd order shifted lower to allow for a 1.4kHz cross point instead of 1.8kHz would still be quite effective at protecting from distortion, or am I wrong and this might have a tremendous impact on performance? I guess that I am willing to sacrifice a bit of beaming for the lower end response.

      Alternatively, to avoid crossover problems, I could pair the RS225 with the RS100-8, but I guess that to keep the power handling to be consistent between the woofer and the full-range I would need to make some trickery using 2x 8 Ohm in parallel and then match the sensitivity with the crossover. However this approach means sacrificing better high-frequencies and will be quite more expensive if going for 4 full-range drivers instead of 2 tweeters.

      Which leads me to another question that is related to crossovers and the air velocity in the port. Is there a definite amount of dB that the first port resonance needs to be attenuated by (calculating the value from the crossover’s lowpass order and its cut-off)? Or in other words, how far above the crossover should I make the first port resonance lie? Because while attempting to recreated Toids design of the RS180S-8 speakers, it seemed that getting the port resonance above the crossover meant getting very close to or above 20m/s of air velocity in the port, so I am wondering how that part of the design goes.

      Well, this is already getting very very long, and maybe several of you will facepalm due to my thoughts, but I would appreciate some sort of feedback.

      If you read part or all of my post, thank you so much for your time!

      PS: This is my second time trying to post this, I wrote a very long text one week ago and after a few days wondered about it, and saw that it did not exist anymore. I find it really strange, because the day that I posted it, I even came back several minutes later and edited my post.

    • #23049
      123toid
      Keymaster

      I apologize I haven’t been on. It is a crazy busy week with my sons birthday, the Super Bowl and family visiting. I will look at this sometime the beginning of next week if not tomorrow. In the meantime @elliottdesigns or @tvor-ceasar may have some input. Or even @zarbo-audio-projects who is a very talented designer. I have ahd the privlidge of hearing a few of his designs and they never disappoint.


    • #23059
      jrand
      Participant

      No need to apologize!
      I guess I also went overboard and should have made my questions bite sized, and one at a time. In the meanwhile I have been trying to get more acquainted with the design procedures and getting more informed, but I am also not having a lot of free time at the moment!

    • #23061
      tvor-ceasar
      Moderator

      I’ve got a lot going on this week into next and won’t be able to delve into the minutiae until then. Just wanted to let you know that I’m not ignoring anything, just can’t get there at the moment.


    • #23071
      tvor-ceasar
      Moderator

      Ran a quick number on the -8 version and in a 1 CuFt box (referenced from the video), I get a Fb / Fp of 34.02 and an F3 of 33.11 with a -2.3 dB response at that point. The 2″ port is just a rough guess – might need a larger port, but you get the idea.


    • #23072
      jrand
      Participant

      Hi @tvor-ceasar ! Thanks for the input, that gives me a really nice confirmation that I am going towards the right direction and also that indeed Toids used the 8 ohm version. (I actually saw yesterday that in the actual youtube video he commented that it was a mistake and that indeed he used the RS180S-8, so that is cleared as well!).

      I have a question though, so while playing with winISD, I can have nice boxes with really low responses, but after I add for example a target power of 85% of the rated RMS power handling, I have to add highpass filters to limit the cone excursion and the response is then raised several Hz up. That is why I also ask how does that work then in real life, and not just theoretically.

    • #23090
      elliottdesigns
      Participant

      I’ll definitely have a read through on the weekend and give you any thoughts I may have, busy until then though unfortunately! I had a super quick skim, seems very interesting!

    • #23098
      tvor-ceasar
      Moderator

      The way the filter works is that there is what they call the Knee Point – that’s the frequency you design for. This is the Pont where you are starting to actually look down (or up, depending on direction) your design slope. This knee point is not a razor sharp point, but rather more of a Point of Intersetcion (PI) like you would have in a curve as drawn when you took geometry. The PI is the intersection of the 2 tangent lines and sits somewhere outside the arc of the curve. The slope of the filter determines how close to the beginning and end of the curve it is located. For example, a 6dB filter is a much more gentle (larger radius) curve before you get to the actual slope than a 24dB filter (steeper slope, smaller radius). Both are still outside the curve, but the 24dB is closer to the ends than the 6dB. So, when you add in the high pass at resonant frequency (33Hz?) It is actually attenuating the resonant frequency somewhat. Try moving the knee lower and using a sharper slope, and see if it still keeps the xmax under control.


    • #23100
      elliottdesigns
      Participant

      Yep, what Charlie says about the knee point of highpasses is totally correct, but it becomes a lot sharper (and moves closer to the frequency the slope is designed for) the higher order of a filter you apply (steepness of slope, the Q (quality) of the slope also changes this). When doing it passive, it’s just like a high-pass that you’d have in the crossover, except of course much lower in frequency which typically means much larger capacitors. 6db slope would be 1st order and has 1 capacitor in series with the woofer. 12db is 2nd order and uses a cap in series and inductor in parallel, etc, etc. By the time you get to something fairly steep like 24db you have A LOT of components. The other alternative is an active slope which can be done really easily depending on what your audio source is, so I’d need to know that first before offering any advice on it.

    • #23101
      elliottdesigns
      Participant

      Also, as for it being how it works in real life. Yes and no, if you want to properly limit distortion, you don’t want to go past Xmax, so you have two choices, cut off at a higher frequency and lose bass, or cut down how much power you are feeding it. As with everything (especially engineering, which this is) there are always compromises.

      I said yes and no, because technically you can still go beyond Xmax, you just start getting some more noticeable distortion, some people are ok with this and might not even notice it. As long as you don’t reach Xmech, this is where the cone actually reaches its limit and will massively distort and possibly damage the driver.

    • #23105
      jrand
      Participant

      Thanks @elliottdesigns and @tvor-ceasar, yeah, filter theory is something I thought I was quite familiar with (I mean, butterworth filters, different orders, active or passive), I just had the notion that the cut-off frequency (maybe what you are calling the knee point) is defined as being always at -3dB, but it seems that that is only applicable for butterworth filters. It seems the definition of the cut-off is the intersection of the stopband (the downward slope) and the passband (“horizontal” no attenuation), which then is valid for filters which aren’t smooth like butterworth filters, but also for filters with oscillations in the pass- and stopbands, things that I had never considered.

      So, from the designing theory I had understood that you add those highpasses already in the design to be sure that at fairly loud volumes you do not induce distortion due to the cone excursion, as you know it will be protected by the said highpasses. From your post I get that, if you want to add those highpasses in real life and make it a permanent addition to your boxes, you should build one and add it to the connections of the speaker, is that right? But like you said, a capacitor that will be able to withstand 70W and has a low cut-off, it will be huge and expensive.
      But I guess that some DSP could also be set to include such a filter, no? And what about infrasonic filters? Do all amplifiers have them?

      I worry a lot about this topic because when designing the box in winISD, it seems that to push the low-end as much as possible with the size and port tuning, that the driver will exceed xmax at really low powers, which to me feels like a waste, if a rated 80W RMS driver can be driven only at 5W max until distortion kicks in (with no filters) albeit reaching down to say 30 Hz. I rather have a speaker that I could drive all the way to say 70W (although I can imagine that would be a very very rare use case, that instead for everyday listening I would be closer to the 5W level), that maybe only reaches 34 Hz. I don’t know, it feels more like I did what I really could to push as much performance as possible from the driver.

    • #23112
      tvor-ceasar
      Moderator

      You would put that first in line on your bass driver. It’s not necessary on mids or tweeters as their crossovers take care of that already.

      As to the ratings, 70 watts at 8 ohms is around 24 volts RMS, and if you were to actually go with the 4 ohm driver, just double that. Then when you settle on a uF value, make sure you get a voltage value that is greater, say 50V for 8 ohm and 100V for 4 ohm. Physical size is actually up to the manufacturer. Yes, there will be a minimum size for Value/voltage ratings, but many “hi-end” tend to make their caps on the overly large size. (It’s a psychology thing).


    • #23115
      123toid
      Keymaster

      I do apologize, you are correct I used the shielded version. Interestingly enough, the only reason I used the shielded version was because of the lower extension.


    • #23116
      123toid
      Keymaster

      For a high pass, if using a receiver of some kind, they typically have an electronic high pass you can add to it. Like you said, it protects the drivers from over excursion, typically below tuning frequency. In addition, this will also typically give you better power handling.

      Typically when viewing those wattages in WinISD, there are things you can do. For example, create a smaller box, This will increase your power handling. However, this will also affect your tuning frequency. In addition it’ll also affect your first port resonance. These are the trade-offs you have to decide upon when designing a box. What is more important to you. Is it low frequency response? Or is it higher power handling?

      Another thing to consider is will you be using a high pass. The Cinema 10 I designed was designed around them being crossed over no later than 80hz. I know that at that, you should have no issues with port noise or power handling if you use them as they are designed. Every time you design a speaker you have to decide what your design goals are and go from there.


    • #23202
      elliottdesigns
      Participant

      Just to add on to what Nick and Charlie said, nearly all amplifiers have a subsonic filter at 20hz, but this isn’t always true, also as Nick says some processors and 2in1 amps (pro and amp) have some sort of optional high pass filtering, in case you don’t have that, there’s always miniDSPs, I’ve also had a breakthrough with my raspberry pi 4 DSP project that I’m hoping to post sometime next month. If you go the DSP route for your highpasses, you also now have the option to do a fully active crossover network.

    • #23555
      jrand
      Participant

      Hello everyone once again.

      I did not have the opportunity to thank you all for your responses, but I do so now. I had been extremely occupied and if I have some free time, I spent it instead on winISD or crossover designing.

      At first I did not fully understand what @123toid was mentioning, about handling of the power etc, and I was still not understanding exactly where I could place the highpass filters, but all of that is in a way solved after playing a lot around with winISD and by just slowly internalizing the concepts and reading your answers multiple times.

      I also found this build which kinda corroborates some of my decisions: http://speakerdesignworks.com/Finalists_1.html

      So, here is where I currently am and I have now a few more specific questions which I think will be easier to address. I will also start a thread just devoted to the build itself, justifying choices, showing the progress, etc.

      My driver selection:
      -1x RS225-8
      -1x RST28F-4
      -1x RS100-8 (question about opinion on this choice).

      Cabinet size:
      48L (1.695 ft³)
      Port tuned to 30.25 Hz. First port resonance tuned to 900 Hz

      The above choice gives a very flat response down to 40Hz, an F3 of 30.5Hz and allows the driver to be used up to 52W if we consider x-max, however this also means that the port has air speeds up to 22m/s. But this should be plenty loud, my listening distance is about 3m and the estimated SPL is 93.5dBSPL, which seems a lot when I search online for loudness levels.

      Here is a preliminary crossover design with the following things in mind for the overall curve:
      -Baffle step compensation of about 3dB (400~500Hz baffle step)
      -Decrease of overall sensitivity between 1kHz and 5kHz of max 2dB

      Now, regarding my choice of the RS100-8, it was just by chance as I have them, so I decided to use them, however they are rated for way less power than the other guys in the build and have a considerably lower sensitivity. What are your guys thoughts on this?
      Should I think of going for 2x RS100-4 in series instead?
      Because currently I am juggling things a bit with the crossover to have the first port resonance 15dB below the target of ~90dB while trying to give most of the power both to the RS225 and the RST28F, instead of having a more clear low-mid-high separation which makes me believe that there will be more problems regarding phase. It would also make it easier to handle the first crossover and deal with first port resonance, also allowing me to shift it a bit lower and so avoid chuffing.
      What are your thoughts on the preliminary crossover? Looking at both the RS100-8 and RS100-4, what would be your choices for the mids, also considering my comments? Single driver, dual?

      (I must add that I ordered both a reference mic and a DATs to analyze the speakers, cabinet and the z-offset, and of course end measurements.)

    • #23559
      123toid
      Keymaster

      So far. I think you are on the right track with the build. I think you are fine with just one RS100. The only thing I would recommend is to attenuate the 5Khz-20Khz about 2db. I think it will be bright on the high end. Having said that, some people like it bright, so that is really a judgement call And honestly it is really easy to attenuate that once you have everything set-up. I always say, order a few 1-3ohm resistors so that it is easier to fine tune later. Or even buy some a little lower value than you think you will be using. It is often surprising how chaning that resistor value will make a difference.


    • #23603
      jrand
      Participant

      Thank you for the feedback, its nice to know that what I am doing is making sense.

      Yes, I was wondering about that as well (the 2dB attenuation at the higher end). I play a lot around with the resistors after the tweeter high pass and I guess I will do as you say. I do prefer the sound to not be so bright as well, so I will see the range for fine tuning and also get those going.

      Will start working on the design of the boxes soon!

    • #23716
      jrand
      Participant

      Hello again, here I am with questions once again.

      So, first, would you recommend that I send one or both of my RST28F-4 back and get new ones?

      Attached you can see the free-air impedance measurements and one of them has a completely different curve than the other. One of them has 20% lower impedance at Fs and a prominent peak at 2.8kHz, which I guess would affect a lot the crossover design, and I am not very inclined in having each speaker have its own crossover. Also, they both do not pass the rub and buzz test although the software somehow is not calculating Fs correctly (it just takes the first slight bump of the curve instead of the main peak).

      Seconly, after measuring my RS225-8, I got thiele parameters differing a bit from the listed ones, specially Fs which I got for both drivers that I bought 31.8 Hz instead of 28 Hz. And I wonder one thing: I always heard that speakers need to be broken in. Do the Thiele small parameters change a lot after several hours of use (for example Fs, as the suspension (surround?) becomes more ductile I guess Fs would decrease). If yes, does this mean that the data provided by dayton audio pertains to broken in drivers or new? How does one takes this into account in their design?

      I ask this because the parameters I measured change quite a bit the best box and the response’s F3 quite a bit.

      Lastly, if I would like to measure the z-offset, I should mount them in a test box. I gather that at least the x,y positions of the drivers should be the same as in my final design, but I wonder if I need to make the test with a box, or could I for example just mount them on the front baffle of my final design and measure their response without a box. Because I could in principle build a box like the final one, but then I would need to figure out how to mount the crossover inside (I guess I could use the hole of the RS225, but that sounds annoying). Its not simple for me at the moment to build three boxes, and I would like to see if I could leave it at building only 2 final boxes.

      I am sorry for being so annoying with all this questions.

    • #23721
      123toid
      Keymaster

      A couple of things you need to keep in mind. T/S parameters are not constant. They do change based on the temperature among other things. Interestingly enough, even if you test the T/S and they differ, you can try to simulate it, and more often than not (with a good company) the results in winisd will be similar. I did a video a while ago about Do I need to break in a sub? It is found here:

      https://youtu.be/CccMMFL0_Nc

      For the most part, companies like Dayton, I have found their t/s parameters to match their winISD simulation. For the tweeters, they are relatively similar. They aren’t going to be identical, as they aren’t matched pairs. How the the response look? My guess is the minor differences you are seeing will not affect the response. But if it does, absolutely return them.

      As for the Z offset, you could mount them on the baffle. Just know that you will not be able to create the crossover from that. You will want the drivers in the box to properly account for baffle step.


    • #23722
      jrand
      Participant

      Wow! Breaking in makes indeed a huge difference!! I am impressed that I nailed the concept of breaking in and Thiele parameters without having any background on it besides having heard that people only rate speakers after “breaking them in”.

      So, after your input regarding Dayton’s reputable listed specs, I think I will keep my original box design as at the moment I do not have a good way of breaking them in (still in the process of finding an amp).

      And thanks for the tips for the z-offset. I guess I will build the boxes and then figure out a way to mount the crossover inside using the hole of the RS225 to access the back board of the test box, while the other box I will leave open until the crossover is mounted.

      I have yet another question:
      In the uglies you add this foam to the walls to dampen standing waves, as you say it. I am thinking in doing something similar, my main motivation is to try to dampen the coupling with the MDF (my own thoughts) while also dampening standing waves, but I do not know how to implement that in winISD. Or, actually, I know I can model stuffing inside by changing the Qa in the advanced box parameters, but I wonder if just lining the walls with some absorbing material warrants changing that value, or if I should just keep it as is (I actually see an improvement with a value of 80~90 of Qa).

      I cannot thank you enough Nick (if I may) for your patience answering all my newbie questions.

      As soon as I have finalized a CAD drawing of the boxes, I will start the new thread.

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