Custom waveguide results for BMS 4540ND

[mazurek]

I have been looking for quite a while for a solution to the problems involved in finding a tweeter for a three way. I'll go briefly into the design goals, you can skip these if you want to just see the results so far.

I want some good deep bass without a standalone sub, so I am using 2 sealed Dayton 10HF per cabinet (I note that a real woofer opens up more options by going higher, but will not reach as low unless it has extreme surface area). I'm crossing that at about 100 Hz, to a 7" midbass, after baffle step and all, I think a 7" will do much better than a 5". I chose SS 4531G after seeing MarkK's and Zaph's distortion tests. I have been trying to find a tweeter that can match the dynamic range of the rest of the system, and match the power response. My compromise tweeter will be the Peerless HDS, but I anticipate that the BMS4540ND may do better.

Attached is a diagram of the waveguide I built. I have a quarter inch radius at the mouth entry transitioning the driver to the waveguide, I took off 1/32" off so that the radius would be tangent to the horn (see picture). For that, I used a roundover bit on the router, and then a 3/4" end mill. The rest of the horn is straight, I cut two circles in 3/4" thick wood progressively wider, and filled everything with spackle filler. The mouth transition to baffle has some 1/4" high by 1/2" wide felt around it. The measuring baffle is about 20" by 25" wide, with the horn centered in it.

Anyone have any recommendations for filler I can apply thickly that won't crack, the spackle sucks (I realize I'm supposed to apply a little at a time, but I'm impatient). Alternately, if horn lathing offers are still open, I would take one of those too.

Attached are some measurements I took. These are hasty, I stacked things on a ladder in the middle of my room, horn baffle is held over the driver by gravity because the driver has a screw adapter (I'm working something that will fasten it better, plate mount adapters from PE don't fit well for this driver). I held the microphone manually and used TrueRTA because I could take measurement quickly, I need to fasten things down better so I can use my mike stand, and get some more time to deal with properly gated SpeakerWorkshop measurements or my Matlab script. Measurements are from about 70 cm, angle is eyed. Everything is plotted in Matlab, unsmoothed 1/24 octave resolution.

Results are attached, I feel they are promising. Off axis response at 20Khz is good. I feel the equalization will not be too bad, and hopefully some of the abberations go away with a better measurement setup, maybe a wider felt band.

Unfortunately, I think I will need to put a resistor in series with it to quiet it down, its too sensitive. On my speakers the noise floor of my Adcom pre, and homemade amps is just fine, but with the compression tweeter, it is audible. Its too bad because it is a real waste of sensitivity.

[mazurek]

Also, I would like some suggestions on crossover, how do I turn this response into an elliptic filter at about 1900 Hz, I could try to change the profile a bit to move the response if anyone has any suggestions.

Forgot to mention, this is for an active application, so I am trying to figure out what I want to do with the transfer function, and will eventually build it out of active filter sections.

[JonP]

Very interesting... This is very much like something I started and haven't done anything with. I also played around with creating a waveguide by using roundover and 45 deg chamfer bits. I was using a 5/8" piece rounded, to the two .75" MDF 45 deg sections. Would be even nicer if we could do a shallower angle than 45... but seems like this works pretty well.

Images not available

Man, I need to get back into that project. (8" 2 way) Have had other things and projects take up all my time, but so it goes..

Keep us posted!

[JonMarsh]

The 440 has a lot of potential. One thing to watch out for with any waveguide or horn are the impedance peaks you get even in the pass band- these require careful zobel design- see the Isisirs thread for more info.

Here's the measured performance in the DDS ENG1 waveguide, at 0 degreees, 15 degrees, 30 degrees, and 45 degrees off axis.



Here's the impedance curve:



Here's the necessary zobel network




Here's the result of the impedance correction:



The impedance curve will vary both as a function of the driver and waveguide.

I think you're making some good progress, could you consider a gentler transition in the throat region (greater radius of curvature) as well as at the termination to the baffle? This will reduce any tendency to HOM (high order modes).

[mazurek]

JonP, I thought about going in that direction for a while, but was having some problems envisioning the dome tweeter flange / horn guide interface. Also, I played with using a chamfer bit in my router, but purchased one with terrible balance and had to abandon that strategy. I'm hoping that the really wide angle of the horn helps to eliminate some novice design mistakes, the audioheritage wood horn thread suggested problems are lessened at a wide angle, and a shallow matches the woofer time delay better. My feeling is that if the BMS4540 is designed for pro-levels, it can stand some reduced horn loading in a home environment.

JonMarsh: I was inspired by some of the work in the Isiris thread to start on this project. About the zobel, is this a problem with an active setup, one Hypex UCD channel per driver? I haven't ever done passive before because I love building electronics. Would a large resistor in series work (I need to reduce the sensitivity to help work on the noise floor), or do I have to do a zobel network. I don't have an impedence test connector, but could make one. Any suggestion on the radius, or as big as possible. The barrier would be that too big and the depth and diameter of the waveguide are enlarged too much, and the cost associated with a huge roundover bit. I guess at <25 dollars a piece, that can be my fun purchase for the month.

[Paul W]

For a filler, try Wood Putty. It cures quickly and shrinks little if at all. It is fairly hard so, if you want to sand easily for prototyping, mix the powder with plaster of paris before adding water.

A large radius roundover at the mouth will help smooth response.

[JonMarsh]

JonP, I thought about going in that direction for a while, but was having some problems envisioning the dome tweeter flange / horn guide interface. Also, I played with using a chamfer bit in my router, but purchased one with terrible balance and had to abandon that strategy. I'm hoping that the really wide angle of the horn helps to eliminate some novice design mistakes, the audioheritage wood horn thread suggested problems are lessened at a wide angle, and a shallow matches the woofer time delay better. My feeling is that if the BMS4540 is designed for pro-levels, it can stand some reduced horn loading in a home environment.

JonMarsh: I was inspired by some of the work in the Isiris thread to start on this project. About the zobel, is this a problem with an active setup, one Hypex UCD channel per driver? I haven't ever done passive before because I love building electronics. Would a large resistor in series work (I need to reduce the sensitivity to help work on the noise floor), or do I have to do a zobel network. I don't have an impedence test connector, but could make one. Any suggestion on the radius, or as big as possible. The barrier would be that too big and the depth and diameter of the waveguide are enlarged too much, and the cost associated with a huge roundover bit. I guess at <25 dollars a piece, that can be my fun purchase for the month.

If you put a resistor in series, it will form a variable divider with the impedance of the driver (see 2nd plot) and you'll have ripples in the frequency response corresponding to the driver impedance peaks.

I didn't pick up that you're going all active (a little early in the morning when I read this- I'm at the airport now waiting for my flight).

Using an inline attenuation to your power amp might give you the noise relief you want without messing up the driver frequency response- probably 10-15 dB attenuation should do it. Just a small L-Pad, you can even use your ampifier input impedance as the shunt leg, in all likelihood.

For some hints on curvature profile, go over some of the discussion on waveguides- for the throat radius, I'd consider something like a 1" radius of curvature to blend to the straight side, and something like 2" if you can at the waveguide mouth to the baffle. It's all about lessening the discontinuities and impedance mismatch.

Try downloading Horn Response and playing with it- a bit of a learning curve, I'm afraid (boy, is THAT an understatement), but pretty useful if you're serious about designing your own.

Gedlee's softare and book has a lot of good info also.

~jon

[Dennis H]

Spackle is only meant for filling small holes. 'All Purpose Tape and Joint Compound', the stuff drywallers use on sheetrock, is better for large areas. If you need it more than 1/8" thick or so, apply several coats. It's not as impact-resistant as Bondo but it's much easier to sand and I wouldn't think the inside of a horn would need to take much of a pounding.

[Biff]

You can use a laminate trimmer, most of them will tilt and lock to let you reduce the angle, but most only take smaller shank bits, 1/2 need not apply. I have occasionally used shims taped to the baseplate of the router, but an extra 15 degrees or so is about the limit there before the stack got, well, a little less solid.

[Paul W]

For a filler, try Wood Putty. It cures quickly and shrinks little if at all. It is fairly hard so, if you want to sand easily for prototyping, mix the powder with plaster of paris before adding water.

OOPS! I goofed...I meant to say Water Putty, not Wood Putty!

[mazurek]

Thanks for the construction tips. This type of project straddles the line of feasibility. Having a continuous d(radius)/d(theta) and large roundovers is next to impossible without a lathe. I see that 5/8" roundover is the threshold in terms of being really expensive, maybe I can use a combination of a roundover and felt to control the horn response.

I tried downloading and using the program HornResp and found it extremely difficult to use. I've got to work out a solution that I can finish in my limited free time. I guess my horn standbys would be XT19 (I don't think it could go low enough) or the BMS coaxials, which would force me to go from subwoofers to real woofers.

[Paul W]

Have you seen this article? Construction seems fairly straightforward and I believe this throat is more appropriate for Hi-Fi/dome type drivers. (Not compression drivers though.)

waveguides

[JonP]

JonP, I thought about going in that direction for a while, but was having some problems envisioning the dome tweeter flange / horn guide interface. Also, I played with using a chamfer bit in my router, but purchased one with terrible balance and had to abandon that strategy.

Well, if it is out of balance and you can't use it, I'd call it defective and have them give you another one! I got a big one (big enough to chamfer a >3/4" board) from Rockler, comes in handy for opening the back of baffle cutouts for woofers... It was over $40, so I'd look for a sale or 2 for one deal someplace. Rockler has them occasionally.

There's a white paper for one of the big pro manufacturers for a "quadratic horn" it has some interesting ideas that I was following. The main one was the idea of a matching curve in the throat, that keeps the walls at 90 deg from the emitting sound wave. For that, in the sizes I was working with, I found that a 3/4" roundover bit on a 5/8" thick panel gave a 90 deg tangent surface flaring out into the flat 45 deg angle cone of my 45 deg chamfer bit created surfaces of 3/4" MDF. (hope that was easy to follow) So that's how I ended up using the 5/8 BB, then the 3/4 MDF.

I found the hole size is rather critical, so you match well and don't have a step. Though a small step isn't anything a certain amount of sandpaper could cure, but you'ld want your two examples of these to be pretty much the same.
Creating a set of templates for exact holesize would be would be a must. Probably need to assemble as you cut and stack the various plates, it would be very hard not to get these adding your errors as you go.

I'll have to finish making a 5/8" - 3/4" - 3/4" sandwich, and see how a 27TDFC comes out in it..

[mazurek]

JonP, what I did to get the initial throat was first route a one inch hole in my homemade disposable router table. Then I made a 3/4" hole in the first wood layer. I then took my ball bearing round over and traced around that to get a nice, well size hole. Then I shaved off the top layer to eliminate the discontinuity with the top layer. I think that I will either try the 3/4" roundover bit, it is not incredibly expensive, or I may generate a more constant dr/dt profile in cad and make it with paper stacking as in Rod Elliots page.

Maybe if I'm lucky, I can get hornresp to work to stop designing blindly.

[JonP]

Yeah, I fooled around with Hornresp, it isn't a "friendly" program. Seemed to get some results, but haven't tested them to cross check.

Get the roundover bit. You'll likely put it to extra good use for front baffle roundovers, which are the bigger the better.. and other handy uses. There's a Whiteside 3/4" on sale now thru Woodcraft.

I used a pair of calipers (carefully, the magnet wants to pull them around!) and measured the diameter at the top of the plate of my TDFC. It has a bit of a roundover itself, and I eyeballed the edge of it and shot for where it stops rounding. That measurment was for the throat hole. See Zaph's tweeter experiment pics, if you don't have one. Then the 3/4" radius on my 5/8" BB ply. Then eyeballing the edge (or end of rounding) of the 5/8" hole, for the starting hole of the 3/4" MDF 45 deg chamfer cut.

As you see, the down side with my method is the inaccuracy of the eyeballing something not easy to measure, and needs to be pretty tight tolerance. I guess one thing to do would be that you could accurately measure your thicknesses, enter it all in some CAD program, and use that to give you the hole size measurments for them to be perfect. Then make templates and cut against them.

[mazurek]

It seems that HornResp only simulates straight walled and exponential horn sections if I am not mistaken. After messing with it for a while, it is not too bad.

I am going to see if I can get a profile with generous curves (like those suggested by JonMarsh) built for me on a lathe. The idea would be that measurements on a best case (semi oblate spheroid conical) horn for this application will show whether or not this strategy of building a compression horn to match the center-center spacing and directivity of a 7" midbass is a worthy pursuit.

[Paul W]

The more recent versions of HS will also sim OS, Spherical, Tractrix, etc. The help file tells you how to toggle flare types.
(and if you learn how to add an Expo section onto an OS mouth, please let me know )

[rc white]

waveguides

I am new to this forum and yesterday I posted material in another thread, (I have forgotten its exact title, I think it is, "new waveguide studies"), that probably should have been in this one.
This is to tell you of my waveguide article at ESP, mentioned by John Marsh earlier in this thread.
rcw

[rc white]

My previous reference to John Marsh should have been to Paul W.
rcw

[mazurek]

I saw that article at Rod Elliot's site, and have read it. I've shelved this for a while. Seas DXT waveguides offer an exciting new option to avoid the iterations of R&D and hassles of fabrication.