Announcement

Collapse
No announcement yet.

On-Wall MTM Design: Advice Greatly Appreciated

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

  • billfitzmaurice
    commented on 's reply
    I can see your point as well, but no matter how many times one tries to reinvent the wheel it always ends up round.

  • mobius
    replied
    Point taken.

    I don't know for other people, but for myself I often don't completely get an idea until I've worked it out for myself.

    Leave a comment:


  • billfitzmaurice
    replied
    Regarding the potential for an Allison Effect sourced response dip, go back and read Post #7.

    Leave a comment:


  • mobius
    replied
    Upon a little more consideration, I think I need to amend my suggestion for this particular situation of a wide baffle speaker only slightly raised off the wall. Thanks to Bill's post, I started thinking a little more about the transition from half-space to full-space and I've realized that the accuracy of the rear wall boundary simulation probably depends on the relationship between the baffle width and the distance to the rear wall which the software doesn't account for.

    More specifically, we know that on any given flat baffle, the higher frequencies propagate forward into half-space and the lower frequencies propagate both forward and backwards into full-space and that the transition frequency between the 2 depends on the baffle width. Or to put it another way, we can say that in order for the HF's to start going in the backwards direction, the wavelength has to be greater than the width of the baffle. So in the case of a 15" baffle, the corresponding frequency is about 900Hz. So in other words, all the frequencies above about 900Hz are only going in the forward direction and are therefore not going to create any reflections off the back wall.

    Now if we look at the case of the rear wall boundary gain again when the baffle is only 3" from the rear wall, we can see that the null that is created (and any ripple beyond that) sits at about 1600Hz (see below). But when the baffle is 15" wide, only the frequencies below 900Hz are going backwards. Therefore that rear wall null at 1600Hz and any other ripple can't actually be created because those frequencies are not moving backwards and are not reflected off the wall and therefore cannot create the kind of out of phase interference necessary to produce the null and ripple in the 1st place.

    At least that's what my logic tells me today.

    So in this case, I don't think adding the rear wall gain to the baffle diffraction effects is going to produce a simulation that isn't flawed. I do still think though that if the cabinet is sitting forward from the wall, the baffle edge diffraction effects but without the 6dB of baffle step loss still need to be added to the infinite baffle measurement for more accuracy. That'll take some manipulation of the baffle sim in something like Response Modeler perhaps but I'm not sure if there isn't just a bit of guess work involved at this point. Adding large edge round-overs or chamfers to the cabinet will diminish those effects of course.

    Or again, flush mount it if you can. That is if you can't take the in-wall/on-wall measurements.

    Leave a comment:


  • mobius
    replied
    Yes, you could choose to just use the driver's infinite baffle response for the sims but the moment the box ceases to be flush with the wall, then you introduce changes to the frequency response due to the diffraction that occurs at the baffle edges and due to the interference that results from the reflection of sound waves off the back wall in particular the suck-out that gets created. In effect, there is indeed no rear wall gain and no baffle step loss because they cancel each other out but the sims model more than just these transitions between half-space and full-space, they give you the effects of the edge diffraction and the rear wall interference. When just relying on sims as opposed to measurements, I would want to include those effects in this situation (though minor they may be) to try to improve the accuracy in the sims.

    For the OP, maybe the easiest thing to do since it sounds like you are able to mount the speakers into the wall, is to choose a cabinet depth that will allow flush mounting to the wall, so basically just an in-wall speaker instead of an on-wall speaker. That way this discussion is mute and you can ignore any potential baffle and rear boundary effects. If you can that is. The in-walls could still be hidden with any kind of acoustically transparent art work if you or your significant other want to.

    Leave a comment:


  • billfitzmaurice
    replied
    Originally posted by mobius View Post
    I'm going to again suggest that in the special case of an on-wall speaker (or a slightly proud sitting in-wall speaker as the case may be), you need to include both the baffle diffraction effects and the rear wall boundary gain effects in your sims
    There is no rear wall gain. Sims model in half-space, and that's what you have with a flush mounted or nearly flush mounted speaker. You will muck up the works if you employ BSC, because there's no baffle step either. This is a good thing, as that's two hoops you don't have to jump through to get an accurately modeled result.

    Leave a comment:


  • mobius
    replied
    I'm going to again suggest that in the special case of an on-wall speaker (or a slightly proud sitting in-wall speaker as the case may be), you need to include both the baffle diffraction effects and the rear wall boundary gain effects in your sims in order to approach some measure of accuracy in your sims.

    So for this I'm using Jeff B's Diffraction and Boundary Simulator. I've chosen a 4.5" cone diameter on a 15"x30" baffle with a 1.5" radius with the driver placed closer to the bottom of the cabinet at 7.5"x6" and with the cabinet sticking out only 3" from the wall.

    The 1st pic shows the baffle diffraction effects. The 2nd shows the rear wall gain and the 3rd shows the effects of both combined. Last is the combined effects added to the FR of RS150P-8 (black trace).

    So note that the 6dB of baffle step loss is cancelled out by the 6dB of gain from the proximity to the rear wall. Also note that because the driver is so close to the wall, that rear wall gain continues right up almost to 1000Hz so that the combined effect you are getting is close to a 6dB increase in the FR between about 100Hz and 1200Hz which you have not taken into account in your simulations. Additionally, the effect can be manipulated by changing baffle dimensions, driver placement and/or distance to the rear wall. Of course changing the driver may alter things as well. For its intended on-wall application, there is a reason that those slim Peerless drivers have a FR that looks different from most other drivers. Not that the Dayton looks unworkable - I haven't made the attempt but it looks like a very broad notch filter aimed at the midrange hump might do the trick to flatten out the FR.

    You should also note that in VituixCAD (as well as XSim), when you use 2 drivers in parallel, the program displays just 1 of the driver responses below the summed response which means that it is actually 6dB down from the real combined response of the 2 drivers. Therefore, the actual xo point is not going to be where the (single) woofer FR crosses the tweeter FR but instead be at a point 6dB higher on the tweeter FR. So for example, in your 2nd sim in post 28, it looks like the xo is at about 1900Hz where the 2 FR's cross at about 86dB, but in reality the xo is 6dB higher than that at about 92dB which puts the actual xo point closer to something like 2600Hz.

    Leave a comment:


  • rpb
    commented on 's reply
    Tweeter looks a little hot. Maybe 1.5 ohms in front of the tweeter xo will help both.

  • Sebreaux
    commented on 's reply
    I noticed that too. I attempted the change you suggested and couldn't get the response to flatten out. I'll keep working on it.

  • rpb
    replied
    Originally posted by Sebreaux View Post
    rpb Would something like this be close?
    The impedance is getting pretty low. Try a bigger coil, and smaller cap on the woofer portion, and see if that helps.

    Leave a comment:


  • Sebreaux
    replied
    rpb Would something like this be close?

    Leave a comment:


  • Chris Roemer
    replied
    Your XO looks "okay" (for starters). Should get you into the "tweaking" range. I see about +1 to +2 dB of baffle-step, which should be fine.

    Leave a comment:


  • rpb
    replied
    The lower you cross, the less critical the height will be. If you built a tm instead of an mtm, the tweeter could be at the bottom of the baffle with the woofer above it. From 16', it may be just fine even if it's a little bit high. I cross many of my speakers below 2k. Not all tweeters sound good that low. If you play the system very loud, the xo point may need to be a little higher than if you don't.

    Based on the specs, I'd maybe try to sim the x-over at 1.8k 4th order.

    Take a look at just the woofer summation at 1m on axis, and then alter the listening position to 3m or so, and 10" below the tweeter axis. If you are 10" lower than the tweeter center line, the woofers will have some issues summing as the frequency goes higher. I suspect that at 2k, all will be fine even if your ears are 10" lower. The sim should help you see the potential issue. It may not be an issue at all at that distance. At 8' it might be more of an issue. It's just a geometry thing. The distance to your ears will be longer to the top woofer than to the bottom woofer..

    Leave a comment:


  • Sebreaux
    replied
    Cross is at around 3100 I think. It is just where it fell as I was playing with the components. Would there be a better target to shoot for? Would a lower crossover point help with off axis response?

    I think that the lowest I can manage getting the tweeters is 6-8" above ear level. Main listening position is about 16 feet from the face of the speakers if that makes any difference.

    Leave a comment:


  • rpb
    replied
    Originally posted by Sebreaux View Post
    OK. MANY thanks to those who have helped so far. I have played around with a bunch of different driver combinations in the interim. I have also "built" existing designs in Vituix in an effort to try replicate the designers outcomes and see what effect changing components has each design. I am attaching my current iteration of a RS150P and RST28F MTM. Maybe not the most exciting driver choices, but it seems to be working with a low component count. I would greatly appreciate if anyone could quickly review this for glaring flaws. I have made sure all component values are available in real life and added the actual impedance numbers in for the inductors. The dip at 4600Hz is bugging me, but I have not yet figured out how to get rid of it.

    I am currently stuck at adding the driver layout. I always thought that these were real world physical distances, but now I am reading that I need to run the drivers through Response Modeler to come up with the correct numbers. Although maybe I am conflating driver layout and driver alignment? Seems like more research is necessary.

    Thanks again.
    The dip at 4.6k is in the tweeters response. There's no reason to attempt to correct it. Looks like you are crossing well above 3k. I'd suggest a lower xo point. Did you use target slopes?

    Will the tweeters be at ear level? Location can be important.

    Leave a comment:

Working...
X