advertisement


DiscoCAT Speaker Design

S-Man

S-Man

StrivingON
After my success at (dramatically) improving my IPL M4 kitchen speakers, I decided to attempt a small 2-way speaker design that would beat my Murphy CAOW1s - not a trivial target!

https://pinkfishmedia.net/forum/threads/improving-ipl-m4s.273191/

I've had the CAOW1s for about 10 years and for the last 8 have been running them as sealed boxes crossed over to my "bass extenders". I have tried several modern and classic mini-monitors but couldn't find anything quite as good as the Murphys. I'm left wondering what on earth is going on with commercial designs that with their huge resources and custom parts they can't make a better speaker than a talented amateur using off-the-shelf drivers!

And why might I think I might be able to do even better with my first from-scratch design???

Well, there are two main reasons I was hopeful:
1) The improved M4s gave a glipse of some sort of clarity in the midrange that the CAOW1s seem to cover with broad brush strokes (although the CAOW1s were much better overall)
2) I'm not constrained by trying to get a quart of bass out of a pint po(r)t, like 99% of what's out there.

My design objectives:
* 2-way with cabinet width of 200mm. -6dB somewhere between 80 and 120Hz. Sealed box = 2nd order rolloff.
* Designed from scratch to work with the grilles in place
* Easy to drive with medium efficiency
* Accurate tonal balance (certainly not "hi-end" hyper detail and sod everything else!)

Driver choices:
I looked at a lot of bass/mid drivers and the vast majority are aimed firmly at extracting a quart of bass, usually compromising other areas! However, there are few e.g. some Monacors and some older and lower tech units that seem to do what I want. The ScanSpeak Discovery 15W8434G00 was selected as the unit with a very usable frequency response and reasonable distortion. It's also very reasonably priced.
I decided a 1" tweeter would be a better choice than 3/4" - I can't heard much above 12KHz nowadays so it makes sense to go for the biggest frequency overlap possible to make the crossover design easier. Not really sure why I chose the Morel, it seems to have a reputation for punching above its weight and the cost was reasonable. Anyway the CAT308 was selected, which has enabled me to come up with the slightly ridiculous name of "DiscoCAT" for the speaker :).

Enclosure choice:
I decided to attempt a 2nd order acoustic crossover and I thought this would be easier if I time aligned the drive units by sloping the front baffle back. Some calcs indicated that 15 deg seemed about right.
Previous experience (loads of amps that never got boxed) made me decide to build a "proper" box straight off. It was very tempting to do a prototype box.
I had a lot of surplus 18mm and 9mm ply and underlay for LVT. I didn't need much internal volume, so the design morphed into a CLD construction with a recessed back for the crossover to inhabit.
The ply was ordinary stuff with the odd void, not the Baltic Birch stuff. My theory is that the voids help break up resonances ;).


TBC...
 
I like big tweeters as well (oh, err, miss). I particularly the 34mm Audax that was used in few BBC designs, and many other places - it has a magnet that looks right for a woofer. Some of the designs that I heard that use it sounded really dynamic. Fairly sensitive, good power handling, reasonably low resonance. I am sure there are more modern parts that do the same thing better, but it was used in some of my favourite boxes.
 
Being a thicko on construction, I looked up CLD and I'm pretty sure that your construction technique has very little to do with Chronic Liver Disease. :)
 
Or perhaps it means Curiously Lopsided Design? :D

IMG-4150.jpg


IMG-4151.jpg


IMG-4155.jpg
 
quote: ...'so the design morphed into a CLD construction with a recessed back for the crossover to inhabit.'...

If by cld you are suggesting 'constrained layer damping', you are mistaken. What you have is a glulam (contraction of glued laminate), known for at least 100 years. Ply has very little damping, mdf even less. So you should get a load of off-axis vibrations.
 
Helen Bach said:
quote: ...'so the design morphed into a CLD construction with a recessed back for the crossover to inhabit.'...

If by cld you are suggesting 'constrained layer damping', you are mistaken. What you have is a glulam (contraction of glued laminate), known for at least 100 years. Ply has very little damping, mdf even less. So you should get a load of off-axis vibrations.
Click to expand...

Apologies if I used the wrong term.

I thought I had a CLD construction because from inside to outside I used: 2 layers of underlay- 9mm ply - 2 layers of underlay - 18mm ply.

i don’t actually know the properties of the underlay, but it’s certainly dampens sound – which is pretty obvious when you put it under LVT. I think/hope it has some viscoelastic properties.
 
Underlays like that are about dealing with impact noise transmission to the floor below; mostly about adding compliance , much more than 'damping' but yes you'll get some of that for free.

Looking forward to how this whole project comes out as a first draft : D


Oh, and plus-1 for tweeters of a decent diameter and large magnets for dynamics and precision - over ultimate extension.
 
martin clark said:
Looking forward to how this whole project comes out as a first draft : D
Click to expand...

Actually the project is at a reasoably advanced stage but I decided to "serialise" it. Post #1 was done up to around March '23.

By May I had got to this stage:

IMG-4528.jpg


IMG-4529.jpg


I had to fabricate another pair of baffles after the dreaded paint reaction thing struck :(.
It turned out OK because the originals didn't include the 15 deg slant. It was a challenge to route the roundovers at the top and bottom but they definitely look better with the CLD (Curiously Lopsided 15 Degrees).
 
Raw drivers response in the box. Microphone is 1m in front of and level with the tweeter - so 15 deg off axis. Measurements are approx 9dB down from 2.83V.

Raw-drivers-in-box.jpg


Effect of grille frame and grille material (with some felt diffraction control efforts) on the 15W:
Green = no grille
Effect-of-FFG-on-15-W.jpg


Effect of grille frame and grille material (with some felt diffraction control efforts) on the CAT308:
Blue - no grille
Effect-of-FFG-on-CAT.jpg


So my decision to design for grille-on adds extra challenges.
 
Nearfield driver measurements:

Nearfield.jpg


This shows -6dB at ~83Hz. 80Hz would be perfect to match my electronic crossover. I don't think I will bother to adjust for the 5% error.
SIMs showed -6dB at ~100Hz, so the 15W goes lower in my sealed box than expected. I was planning to change my electronic XO to 100Hz, but no need.
 
a.palfreyman said:
What X/O are you using? LR2, LR4, active?
Click to expand...

The target for the passive crossover in the DiscoCAT is a 2nd order acoustic, somewhere between 2K and 3KHz.
I use an active XO for the integration between the DiscoCAT and the bass extenders: low pass is an active LR2 at 120Hz cascaded with another active LR2 at 80Hz. High pass is an active LR2 at 120Hz which combines with the 80Hz (actually 83Hz) acoustic rolloff of the DiscoCATs. Although this might seem odd, the phase alignment is correct and the amplitude sums to within about 1dB IIRC from when I did the sims. It also means I only need 1 buffer in the mid/high pass section and I found it challenging to get audibly transparent filters.

The bass extenders go down to 10Hz (-3dB) with a 1st order rolloff characteristic. The point of this is not to get stunt bass - quite the opposite - it's to get bass with no overhang.

a.palfreyman said:
Be really interested to see how the sloped baffle works WRT time-alignment.
Click to expand...

DiscoCAT:
It's purely to align the acoustic centres of the 15W and the CAT. The acoustic centre of the 15W is ~18mm behind that of the tweeter, the slope of the baffle hopefully compensates for this.

Bass Extenders:
The BEs comprise 3 X 7" drivers in a 200mm wide box that doubles as the stand for the DiscoCATs. Hence the need for them to also be 200mm wide. They are therefore time aligned because they are directly below the 15Ws (of course the BE drivers are slightly further away because they are below the listening axis, but this error is mimimal at the wavelengths (>2.86m) involved below 120Hz).
 
This was the sim of the 1st crossover I built:

B1-Crossover.jpg


It sounded very detailed but a bit lean and bright.
(Note that <250Hz is not valid due to the windowing used).

In an attempt to reduce the lean/bright effect I modded the crossover to this:

B2-Crossover.jpg


The result was indeed an overall darkening of the sound. However the result was a rather lacklustre sound. Phase tracking was not great, maybe this contributed to the mediocre result?

However it seems to me that the bass series inductor sets the overall tone of the speaker by tilting the response in the 100 - 1500Hz region.
Up to this point I had been trying to choose the crossover frequency by varying the inductor. I decided to try a 1.8mH inductor to get the right lower/mid balance and then adjust other components to get a suitable crossover frequency and decent phase tracking. This might sound simple, but for anyone who has tried, it's like trying to solve a multidimensional puzzle - especially as this is all just for one measurement point!
 
My 3rd crossover built (I designed many more on the simulator):

B3-Crossover.jpg


This sounds much more balanced than the first two. Seems like 1.8mH is about the right value for the overall balance.

XO point is around 3Khz which might result in less ideal off-axis response as the 15W starts to beam.
Phase tracking is pretty good. Adding 43uF in the C4 position reduces the rate of tweeter roll off and fits the 2nd order target curve down to below 100Hz, which helps phase tracking at lower frequencies. I haven't tried this for a number of reasons: increased lf energy into the tweeter, the phase response is only "correct" at the microphone position, phase tracking is decent for >1 octave below the XO point with C4 shorted.

C3/L3/R5 is a notch filter centred around 11KHz that takes the level down about 3dB. This has a very subtle effect on the sound - it seems to make the treble just a touch sweeter and more airy sounding.

I'm wondering if the peak at 5.6KHz needs addressing? I thiink it's caused by the reflection off my grille frame and I'm not sure how much of an issue this is subjectively. I probably need to measure off axis to see if it goes away.

Compared to the CAOW1s, the DiscoCAT is a bit clearer and resolves more detail. With a good recording the DCs are probably better. The CAOW1s do have a wonderful way of handling any recording and type of music, they are possibly just a touch euphonic. I think I'm actually getting somewhere with these!
 
Thank you so much for sharing the steps of the process and illustrating the difficulties involved in speaker design!

I really enjoy reading on how you work yourself towards your goal, even though speaker design is outside of my personal diy-scope!

Best, Magnus
 
5.6kHz is 60mm for reinforcement at that freq. Foam square diffuser on the tweeter faceplate that stands flush with front of the grille frame? Either that of make the cutout 'jagged' like on the AR22 Tribute speakers on here a few years ago.
tFNT6F
 
a.palfreyman said:
5.6kHz is 60mm for reinforcement at that freq. Foam square diffuser on the tweeter faceplate that stands flush with front of the grille frame? Either that of make the cutout 'jagged' like on the AR22 Tribute speakers on here a few years ago.
tFNT6F
Click to expand...

Good suggestions! I must post up my experiments with felt and foam diffusers... not what I was expecting!
I have considered chamfered grille frames (indeed the CAOW1s have these). I have left them square until I complete the diffuser experiments. Hadn't thought of jagged though - are there any pics around?
 
You must log in or register to reply here.


advertisement


Back
Top