OSWG 12" 2-way

I've decided to do something with some flat-pack MDF I had lying around here and started a 12" 2-way.

First up a test baffle, here being given its first finishing pass. The camera pic and lighting makes it look a bit weird.



I decided to go for a nice and large waveguide so that I could try out various options and play around a bit if need be. The baffle is a about 16" wide and 2.5" thick.

Tomorrow I'll cut and glue up the cabinet bracing and test some CD options.

Here a mounting plate for a 1" CD which I'll recess into the rear of the baffle. I hope that I'll find a driver that is up to the job and not too complex to filter.

plate.jpg
 
Last edited:
Here the bracing mocked up. The side panels show a property of MDF when dadoed, release tension on one skin and it will cup slightly. I clamp the one side to the table and the other gets pulled into the bracing so the net result is fairly flat.

IMG_20160328_123819.jpg


Look carefully and you'll spot where I butchered the bracing to fit the bits for this project. Some is just hanging on a tether until the glue up is done before I'll knock it out.

Tomorrow I need to cut up a rear panel and some port roundovers though I'll probably do a hot melt glue up first and do some measurements.

I have a pair of P.Audio WN12s I want to use eventually as they are immaculately behaved up to around 2kHz (crossover will depend on polar measurements I guess)
 
Thank you. Seems like there are some horn enthusiasts here. I like your avatar as I'm a big fan of horns.

I haven't been active here for a long time. I use Mach3 as a machine controller and cambam for the CAM. The machine I built using a Bridgeport Milling machine in a friend's workshop with a few key parts farmed out to engineering shops in the area. The head is a bit flexible and this job here is just about on the limit of the Z axis clearance of the machine.
 
I draw it out in CamBam, good simple 2D CAD inside the package. Has 40 free trial sessions if you want to give it a go. I have a common interlocking bracing scheme that I use that works well for me.

The horn profiles I typically do in a spread sheet and manual G-code.

If I were to count all the tool paths on horns I'd probably end up beyond 30km by now.

A quick measurement with a couple of drivers showed up the expected ~4kHz dip on axis but the 90 degree waveguide came up a bit short on acoustic loading so the filter to make it cross low enough may not end up trivial. I suspect a couple of impedance correcting networks in my future.
 
I quickly glued and taped on some bits to get just the basic info. I think I'll increase the mouth radius a fair bit more, as the radius on the baffle does very little to tame that on axis dip (a known feature of the OS waveguide)

OS12FRs.png


Here a little family photo with the newest member looking bigger as it's closer to the camera. The horns are awaiting a new mounting, but they work very well.

IMG_20160329_202436.jpg


Now for the real question, if I double the radius of the WG Terminus, how much will the 4kHz dip change or does one just live with it?
 
The perspective skews things somewhat. The white system is considerably larger with a ~112Hz horn doing duty on the bottom end. It's just roughly put down for the moment but it works really well. The drivers were one-off made to order for the MB (BL>30), the mid is a modified 3" compression driver turned on a lathe after extensive AkAbak modeling and tests on a home made impedance tube and the HF changes like the weather.

The dip in the on axis improves drastically by even just slightly moving off axis, I'm not even sure whether I should machine yet another profile as it's quite laborious. Some modeling would be in order.
 
A quick synopsis of modeling various round overs....the larger diameter round overs at the terminus seem to do two things: push up the frequency of that on axis dip and widen the frequency range over which it applies...it also reduces the gain at the lower part of the pass band for an equal size baffle forcing one to a narrower coverage angle...methinks i'll leave it be.
 
A larger radius probably won't help with the axial dip much. Regardless of the radius the distance from the throat to the mouth and across the mouth at all angles will be equal thus the dip. Deviating from a round shape will help to eliminate/mitigate the dip, or using a variable roundover radius would help if you want to stay with a round wg. Or you could call it a day and listen off axis like most of us with cdwg systems do 😉

What a larger radius *will* help with is maintaining even coverage throughout the desired bandwidth. As it is now with the smallish radius you have I'd expect the pattern to narrow a good bit between 1khz-2khz. It seems a roundover radius of 1/4wl of the lowest freq of interest works well. My current waveguides have a 1.5"r and they exhibit a good bit of pattern narrowing at 1.5khz. For a 1khz low xo I'd want about a 4"r.
 
I did try a euclidian spiral and for the baffle transition the difference was negligible compared to a radius. Perhaps if it were a loose unit and rolled back.

On to more relevant things, the crossover for this with the time alignment is a bear to get right. With an active setup less so, but the passive will be quite a challenge.

Anyone done this before and care to suggest a topology? 1st order HP (with a few conjugate networks) 2nd order LP doesn't get the phasing right other than for a 1.6 kHz crossover and I am aiming a bit lower. What target transfer functions are worth a try? But3/LR4?
 
@Nate Thanks for the kind words. I used a roughly 2.5" radius, the MDF layers mask the appearance some. If the sub 2kHz rise is due to an increase in DI, I'll have to compensate carefully in the crossover.

@Rewind, this is also a bit of an experiment, I wish to compare CD versus rising directivity horns. It is also a passive setup as I dislike my horn setup when not time-aligned.
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
the crossover for this with the time alignment is a bear to get right. With an active setup less so, but the passive will be quite a challenge.

Anyone done this before and care to suggest a topology? 1st order HP (with a few conjugate networks) 2nd order LP doesn't get the phasing right other than for a 1.6 kHz crossover and I am aiming a bit lower. What target transfer functions are worth a try? But3/LR4?
Since the woofer is closer, you may have more luck with an assymetric crossover. As a starting point I'd suggest looking at the difference in axial distance, then dividing that by quarter wavelengths at the crossover point and using that many orders of filter more on the woofer.

Since you are using HolmImpulse, you can take a useful phase measurement of the two by cutting them in at the same time point. Simulation from there may help.. You might consider taking these at the listening angle.