A Classical Three-Way Design

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This speaker has been work in progress for quite a while. I set myself the task of producing a classical three way with as simple a crossover as practical, without compromising sound quality. Over time there have been numerous versions and tweaks but finally, it's where I want it to be, and so the final result is seen here.

The cabinet: I'm not much of a carpenter, so I canibalised an old pair of Musical Fidelity MC4s. I replaced the rear panel with something more modern and made a new front panel out of pine. Eventually I will have a go at creating a cabinet of my own, but for now this will suffice. I've retained the original filling which appears to be some kind of BAF. The cabinet is sealed, which gives the correct bass alignment to integrate with my active sub.

D3H_5330_zpso1pk2u06.jpg


The bass unit: For bass I selected the SEAS L22RNX/P H1252-8. This aluminium coned unit has an excellent response throught the required range with the ability to resolve fine detail in the upper bass/lower mid on a par with much more expensive units. It does have the typical 'severe' metal cone breakup spike, though a notch filter takes care of this quite well.

The mid unit: Midrange is usually the make or break of any system. It has to have a smooth response both inside and outside the pass band and have as few 'nasties' as possible. Poor vocal reproduction has me reaching for the off switch within seconds. My choice was the Morel EM1308. It has a very natural sound, with an easy to work with response.

The treble unit: The SEAS 22TAF/G has all the advantages of a metal dome with none of the disadvantages. It has the ability to disappear, which is how it should be. My only criticism is that it can sound a bit too polite with rock music.

The crossover: Much tweaked, I finally have the compromise between as few components as possible and no sacrifice in sound quality. There are a few extras: the bass has a notch filter - also a small resistor on the bass cap to adjust phase; the mid has a zobel to ease the upper response and to improve phase. The tweeter does have a rise in response which could be cured with a zobel, however, the rise is at such a high frequency it just isn't audible (to me). Using a dome mid with dome tweeter obviates the need for time alignment. All components are glued to a base of 9mm mdf, which in turn is glued to the cabinet walls; tag strips are used for all main solder connections. The crossovers are placed in the cabinet to be as far apart as possible. Decent quality OFC cable is used throughout.

While experimenting, I broke the wire off the 16mfd Supersound mid cap. I had some Z-Caps to the same value so I used those. I noticed an immediate loss of depth, detail and ambience... The XO now has all new components, including Supersound caps, and all is restored. Lesson learned.

During design it became obvious that there is a phase issue between the L22 and EM1308. This has lead to the EM1308 high pass being lower and shallower than is required. In order to get good phase response and a decent reverse nul, this is how it has to be. Having the EM1308 going so low reduces power handling, but I've not yet found any issues with sound quality. On the contrary, midrange is excellent. Listening to the guitars of Rodrigo y Gabriela is enough to put a lump in your throat... Sometime in the future I may replace the L22s with something a little more phase friendly, but for now I'm happy with what I have. Crossover points are 650hz and 4Khz.

A few pics in and around the cabinet:

D3H_5322_zpsy2aavfhw.jpg


D3H_5323_zps8a2qljvy.jpg


D3H_5321_zpsvkmyohtf.jpg


D3H_5328_zps5edrgffe.jpg


D3H_5327_zpsy4xepp8c.jpg


D3H_5329_zps0faleaf2.jpg


And finally, some technical stuff...

Frequency response on axis:
FR_zpsg9qvau05.jpg


Frequency response - reverse nul:
FRRev_zpssq7xtwwx.jpg


Impedance:
Imp_zpsl7zp3f9h.jpg


Crossover:

XO_zpslouhz8kn.jpg
 
Last edited:
Nice design, I've done a similar design ; with very similar drivers and similar crossover. W21EX002, MDM55, 27TBC. I use a 3.9mH for the woofer. My crossover point are 800Hz and 3kHz.
It sounds great, I have a boxy sound but it is a characteristic of the design in a box. The box is a 30L aperiodic enclosure.
dsc_1112.jpg


I agree, the morel driver is fantastic with guitare ;)
 
Nice simming. How does it measure and sound?
Apart from the usual bass peaks and dips from the room, the in room response is fairly flat.

Describing the sound is always difficult as many adjectives mean different things to different people. The whole speaker is neutral sounding, with depth and ambience at all frequencies.
Lower notes on piano have great authority and resonance without sounding warm.
Vocals are clear and without any kind of emphasis or edge; sibilants are acurately portrayed - you can hear every little nuance and roll of the tongue Enya has to offer.
Orchestral pieces have fine separation with no muddying of detail. Violins, massed or solo, are clear, resonant and sweet.
Acoustic guitar is crisp sounding with tremendous body.
Treble is clear and free of artifacts such as fizz, tizz, edginess and any other popular names. It could do with a bit more vigour though.
Jazz sax is about the best I've heard on any system. Last night I had Ike Quebec in my room...

Overall they just allow me to listen to music without any distractions. On that basis I'd call them a success.
 
Hi
nice project! I always find larger mid domes appealing for classic 3ways. how much baffle step is in there, I cant see it on the sims?
I noticed the mid/ is in phase with the woofer , usually the mid is reversed as well as the tweeter in most 3-ways.
interesting your findings about that 16uf cap on the mid , I wouldn't think differences between 2 poly cap brands, could be an issue esp. since the loading and source impedance is so mild. on the other hand as you suggested might of had / have some issues due to low XO point and loudness, so there's that..
 
how much baffle step is in there, I cant see it on the sims?

The 3.5mH inductor provides BSC. The response without BSC is in the original frd file and isn't shown on the sim. Looking at the curve now there is a 2db rise between 200hz and 500 hz, then a 5db rise between 500hz and 700hz.

I noticed the mid/ is in phase with the woofer , usually the mid is reversed as well as the tweeter in most 3-ways

That would only be the case if the bass/mid xo was purely LR2. The actual acoustic slopes are much higher.

I wouldn't think differences between 2 poly cap brands, could be an issue

With suitable high end partnering electronics, there are noticable differences between both brands of caps and other components, such as internal cabling. Sorry to put a dampner on people who believe it's woo-woo ;) I should add clean soldering also makes a difference.
 
I used an offset of 0.8". Measurements are from 1M positioned between the mid and treble. The listening room is very small and so the listening position is quite close and precise - you either sit in the listening position or you don't...

After listening for a long time, I've now gone back to a second order (electrical) high pass on the mid. This gives a measured x/o of 680hz @ LR4. The response isn't as flat and the reverse null is broad and shallow, but it just sounds better.

The 16µF mid cap is glued in place, so I added a 3µ3 cap to it; there was just enough space on the board to fit a 1.5mH inductor. The resistor has been changed from 4R7 to 5R6.

Technical specs are fine, but in the end, the intended use for these speakers is listening to music; right now they are doing that rather well :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.