Introducing the SSA-WG

[sdl2112]

This is my first build of my own design. I have been mentally designing it for years. There were always more questions than answers but now I'm ready to start the build.


Name:
The SSA-WG is a 3-way Scan-Speak Accuton WaveGuide project.

Drivers:
Woofers: 2x Scan-Speak 22w/8857T00 (parallel)
Midrange: Accuton C173-6-090
Tweeter: Scan-Speak D3004/6640 mounted to a Jantzen Waveguide

Crossover:
Duelund with approximate crossover frequencies of 300Hz and 2150Hz

Bass alignment:
Will be capable of sealed or ported. I can reach desired targets for both alignments with the same volume enclosure (details to come).

Cabinet:
46-47"H x 11.7"W x 17"D

approx 68L net bass
approx 13L net mid

1" Bamboo ply sides and back
2" Bamboo ply baffle
18mm Baltic Birch internal bracing
1/16" lead mass loading mid cabinet + some bass lining

cabinet sketch

[5th element]

Awesome! This should be an excellent design both subjectively and objectively. You could reduce the C2C distance by routing into the waveguide but you might not want to do that.

Troels does this himself I believe as do I in one of my projects.

I would be tempted to mount one of the bass drivers as close to the floor as possible and have the other one mounted as close to the mid as possible.

This should help smooth floor bounce, give you increased bass sensitivity by having one woofer floor loaded and help improve mid to bass integration by having one in very close proximity to the mid.

The other option would be to have one bass driver mounted as close to the floor as possible on the front of the cabinet and then have the second one mounted in a similar position but on the rear of the cabinet, but have it mounted magnet facing out. If crossed over right this will eliminate floor bounce in your usual listening position, give you force cancellation from the opposed mounted woofers and give you good cancellation of even order distortion products down low from the push/pull mounting of the bass drivers.

If you're going passive do make sure to keep the sensitivities in check. You may need to use two series connected 4 ohm bass drivers, especially if you floor mount, as floor mouting effectively eliminates eliminates the sensitivity losses associated with bafflestep.

The tweeter will only be 91dB sensitive at the top of its passband regardless of waveguide loading and two paralleled 8 ohm bass units on the floor will go higher than that I think.

[sdl2112]

Thanks for the quick and detailed reply...

As far as the MT C2C...I considered overlap but overall I nixed the idea. I may want to try the "new future tweeter" and trying to duplicate the WG cut was a no go. + I'm trying to keep the x-over low so s/b ok.

Actually..the baffles are already cut so no woofer movement...though initial testing showed no clear winner.

Interesting enough once you BSC down the woofers you are at maybe 87db...the mid is way more efficient plus the mid gain...the tweeter has great boost in the low end but actually sacrifices in the 10khz range.

[5th element]

Well with floor mounted, push/pull you will lower distortion, remove the need for baffle step correction on the woofers, (lowering distortion even more), eliminate floor bounce and gain force cancellation. It would be a clear winner just on technical specs alone.

The scan woofers are very well designed though so the improvements for push/pull loading wont be as high.

The most important thing would be sensitivity matching and if you floor mounted the two drivers you'd be looking at 92dB end sensitivity. This would outshine the tweeter, granted by only 1 dB and then inductor losses on the woofers + the need to perhaps tilt down the response may mitigate that, but it'd be close.

Given the driver layout I'd expect you to need about 4dB bsc. You will see some reduction in the amount required by the close proximity of the bottom woofer to the floor and I've usually found that the full 6dB is a little too much unless your speakers are over a meter away from all nearby boundaries.

I hear you about re-routing another wave-guide should you wish to try another tweeter, this would be extremely hard.

[5th element]

You can always try reverse mounting the lower of the two scan speak drivers though, even with the current cabinet configuration. It shouldn't change any requirements for the crossover and is also rather easy to try, you would need to lead the wires out for it through the port. I'd give it a go, every time I've done this it's resulted in cleaner bass.

[dar47]

Nice concept and great drivers.:T

I know you have stated you may want to try other tweeters although I'm not sure what other tweet will beat the 6640 but cutting the waveguide profile in the baffle would give you the option to move the mid and tweet closer to each other and allow for higher mounting on the baffle. This of course would limit the finishing of the baffle to bamboo or painting black. If you normally run with grills then I would stick with the plastic waveguide as it makes tweeter changing possible. One other option for baffle material which is just as good as bamboo is epoxied maple strips and the look can be matched to many veneers.

[JonMarsh]

I'd suggest following some of these different ideas by literally building some prototype building blocks and experimenting with the concepts and measured results on one prototype system before going full monty on the final build cab investment.

Regardless, you'll get plenty of cheering and interest from the side lines here- best of luck, and don't forget it's about having fun, if you run into some issues, we're all here to help.

~Jon

[sdl2112]

Thanks everyone for the suggestions and offer of support. I will likely need it later in the build

I re-purposed some test cabinets and performed various tests. The results look very promising so it's full steam ahead :T

I tried WWMT and WWTM. I made different wing profiles...I even tried an 8" and a 10" in their own air space but in the end elected to stick with two 8 inchers.

Here's a few pics of the test set-up;

-Scott

WWMT and WWTM



Taking measurements



Room view



Trying sealed, ported and aperiodic loading

[Face]

Not much to add here except it's nice to see another reference build here.

As for the woofers being too efficient, if you stick with an average gauge air core(15-16ga) on the woofers, and don't mind a gentle downward slope on the top of HFs, you should be fine.

[sdl2112]

Sealed and Ported options

The design will have sealed and ported options with the same enclosure. What got me started down this path is first, I just want to hear the difference myself. Second, when I first tested them the T/S parameters were way off from the data sheet. I investigated and concluded it was a motor strength issue. This appears common to Scan-Speak drivers. Their BL is lower and Qe higher than specified.

This lead me to see what bucking magnets could do. I tried 90mm ceramic and 100mm Neodymium magnets. You place the magnet with opposing flux to speaker magnet which forces more flux through the gap....I believe this is correct. Anyway it works and if you reverse the magnet the Q goes way up.

A word of CAUTION:!: neo magnets are incredibly strong. They come packed with 3/4" plywood blocks between them and even with the plywood I pinched my palm and drew bloodh:. Like I said the flux opposes but once it gets close to the speaker back plate it attracts and once its on you have to get a clamp to nudge it off.

I'll start off sealed, the porting is not to see how far I can extend the response but to keep initial roll-off similar to sealed, get some extension and limit cone excursion. Also with the added magnets we get about 1.1db greater sensitivity.

It will be interesting to see the differences.

Below are the drivers with and w/o the added magnet.


Next are the T/S comparison data


Here is the response standard sealed and ported with added magnet, 68L cabinet volume, 50W input


Excursion comparison


Step response comparison

[sdl2112]

Waveguide details

I posted this info a while back in the Ardent thread. I'm finally putting them to use:T

These are constructed with the SS D3004/6640 tweeter and Troels/Jantzen waveguide. I placed a felt ring between the tweeter and the WG due to a gap from the concave faceplate.

I made the rear clamshell to better isolate the tweeter from the mid pressure.


Other side


Detail


Here's the standard tweeter response on an IEC baffle at 0-15-30-45-60deg


Here it is with the WG on an IEC baffle at 0-15-30-45-60deg


Otherwise the design is coming along great. I have test data, I'm starting the x-over design...should have more details coming soon.

I must say having it all start to come together is very rewarding.

[augerpro]

Wow, great match to the waveguide!

[TEK]

I posted this info a while back in the Ardent thread. I'm finally putting them to use:T

These are constructed with the SS D3004/6640 tweeter and Troels/Jantzen waveguide. I placed a felt ring between the tweeter and the WG due to a gap from the concave faceplate.

I made the rear clamshell to better isolate the tweeter from the mid pressure.


Other side


Detail


Here's the standard tweeter response on an IEC baffle at 0-15-30-45-60deg


Here it is with the WG on an IEC baffle at 0-15-30-45-60deg


Otherwise the design is coming along great. I have test data, I'm starting the x-over design...should have more details coming soon.

I must say having it all start to come together is very rewarding.

From the graphs it looks like it falls of a bit earlier in the 3-5k area.
Could you - for a nooob like me - explain the pros and cons you are considering for the wavequide?

[sdl2112]

Wow, great match to the waveguide!

Yes...I'm very happy. I suspect the felt ring may help with irregularities often seen in the upper frequencies with waveguides.

[---k---]

This looks like the beginning of a winning design. :T

[sdl2112]

From the graphs it looks like it falls of a bit earlier in the 3-5k area.
Could you - for a nooob like me - explain the pros and cons you are considering for the wavequide?

Sure...I'll explain what I know.

- It provides more narrow directivity thus working more efficiently in the 1-6kHz range. It also matches the directivity of similar sized midranges providing a better power response (flat response off horizontal axis).

-The increase efficiency means the tweeter dosen't work as hard in the lower range thus lower distortion.

-It move the acoustic center back to a similar sized midrange thus making x-over implementation easier...especially for lower order like LR2.

-The response rise may look ugly but typically a singe cap (of smaller value than w/o a WG) typically brings the response flat with a near second order roll-off.

Cons...? it's harder to implement from a construction standpoint. It's more difficult to keep the spacing between the mid and tweeter close.

[sdl2112]

This looks like the beginning of a winning design. :T

Ryan, Thanks for your vote of confidence. I think it will be too

[Face]

From the graphs it looks like it falls of a bit earlier in the 3-5k area.

Below those frequencies is where the waveguide's boost falls off, leaving the tweeter's native response. A larger waveguide would provide boost to lower frequencies, but then center to center spacing is increased.

-The response rise may look ugly but typically a singe cap (of smaller value than w/o a WG) typically brings the response flat with a near second order roll-off.

A contour filter plus single cap can yield a nice LR4 slope.

Another WG to consider is the SEOS-8: http://www.diysoundgroup.com/plastic...8-tweeter.html

I haven't tried it with my Scanspeak Air Circ due to the small gap left(need to try felt as you mentioned), but have had great results with another tweeter. Plus, they're available in white to match your mid.

[sdl2112]

I haven't tried it with my Scanspeak Air Circ due to the small gap left(need to try felt as you mentioned), but have had great results with another tweeter. Plus, they're available in white to match your mid.

I used this Felt ring from Madisound and it has the perfect ID already.

[Face]

I used this Felt ring from Madisound and it has the perfect ID already.

Thank you, just bought a few to play with.

[5th element]

That felt ring worked surprisingly well, from the looks of things I was expecting it to create a few issues but it's quite the opposite!

[sdl2112]

Yes, it's hard to anticipate the outcome of a tweeter/WG combo. I tried one where the interface from the tweeter to WG "looked" perfect but the response was horrible. I'm lucky these work so well.

[Face]

So it can be avoided, which tweeter would that be? Thanks.

[sdl2112]

I should have said the mechanical interface looked perfect (no gap, smooth transition) between the tweeter and WG, but the frequency response was horrible so I don't know of a sure combo.

[Face]

I look for tweeter with flat faceplate, smooth response and a little lift after 10K. IME, this one works well: https://www.madisoundspeakerstore.co...22tff-tweeter/



I want to give this one a try next: https://www.madisoundspeakerstore.com/ring-radiator-tweeters/scanspeak-illuminator-r3004/6020-10-1-tweeter-soft-dome-ring-radiator/

[jim1961]

Hey Scott. Glad to see you at work.

One thought though. Not to say you havent already thought this through, but I was drawn to the tweeter http://www.scan-speak.dk/datasheet/pdf/d3004-664000.pdf Xover point of 2K with a LR2. That tweeter has the same resonant FR as mine http://www.troelsgravesen.dk/Jenzen-...s/T29CF002.pdf. and similar power capability. But mine X at 2.8K. I also noticed the power rating was for a 2.5K LR2, so your a little under that. If its rugged enough, you might be able to unleash that 250w amp of yours on it

I guess it works out that you are -24db (more or less) at resonance (500hz). That maybe enough.

Just thinking out loud.

[sdl2112]

With the boost the WG gives, to bring the response down you are effectively crossing it over higher than the 2k would indicate. I am keeping my eye on it though. I'm also looking into a Duelund crossover. Both the woofer and tweeter have steeper roll-offs at their extreme than a LR2...details to follow...Oh and it's 400W at 4 ohm

[jim1961]

Looks like those tweets are around 3 ohms in the 1.7k-6k area :bluezoned:

[Jonasz]

Looks like those tweets are around 3 ohms in the 1.7k-6k area :bluezoned:

...wich is perfectly normal for a 4 ohm tweeter.

[Face]

Looks like those tweets are around 3 ohms in the 1.7k-6k area :bluezoned:

Any decent amp should be able to handle that. If were much further down in frequency, then I would be concerned. You would be surprised how many "8ohm nominal" speakers drop down to 4ohms, and "4ohm nominal" drop down to 2ohms.

...which is perfectly normal for a 4 ohm tweeter.

+1...

[jim1961]

...wich is perfectly normal for a 4 ohm tweeter.

Any decent amp should be able to handle that. If were much further down in frequency, then I would be concerned. You would be surprised how many "8ohm nominal" speakers drop down to 4ohms, and "4ohm nominal" drop down to 2ohms.

I didn't mean it was a problem or unusual.

I was kinda making a joke that he will have (maybe) even more power available at 3 ohms (500w ?). But also that the required XO components to make that 2K XO point need to be designed with 3 ohms in mind, not 4.

I am really looking forward to seeing how this project turns out. It has been in the planning stage for 2 years or more.

[sdl2112]

Test box measurements

Here are a few test box measurements.

I'm also investigating a Duelund crossover. It's similar to a LR2 but provides a net total 360 degree phase rotation and higher roll-off rates for the woofer and tweeter at the extremes. Dennis H provided an excel spreadsheet a while back that I imported in Soundeasy...very nice...Thanks Dennis! My initial target values are a center frequency of 800Hz and an aleph of 3. This yields x-over frequencies of about 300Hz and 2135Hz

The first is all drivers. The red is both 22Ws, the green is the C173 and the blue is the 6640. The scale should be pretty accurate. This is at 1M and 10ms window.


Here is one 22W merged far-field and near-field response


The C173


The 6640


Duelund target responses

[sdl2112]

Now for some woodworking...

I want to highlight the end grain of the bamboo. The front and back will be a butt joint with a round-over. For the top I will stack 1 inch strips. I initially wanted a cabinet width of 11.5 inches...12 stacked strips came out to be about 11.9 inches. Once I squared it up, it came out to 11.7 inches.

Here's the clamp photo...


Top cut to size

[Face]

Nice setup!

FYI, I've been bothering Woodpeck to make another run of those bar clamps, but haven't been successful yet.

[sdl2112]

Yes...the bar clamps are very sturdy. Without them I doubt the top would have come out flat and square.

[Steve Manning]

Yes...the bar clamps are very sturdy. Without them I doubt the top would have come out flat and square.
[ATTACH=CONFIG]24639[/ATTACH]

Scott ........ build is looking good so far ....... is this a home made jig or a store bought?

[sdl2112]

Thanks Steve.

These are flat bar clamps from Woodpeckers. Looks like they did just a short run of them.

[Steve Manning]

Woodpeckers seems to make some good stuff ..... too bad this was a limited run. Though they look like they would not be too hard to make your own ....... might not be as pretty, but doable.

[sdl2112]

Crossover development

After much work I have my first prototype crossover on paper. Learning SoundEasy is no easy task (at least for me:E). Little by little the process made more sense...and that really is the key.

I tried many topologies, notch filters, optimization procedures, reducing component count...etc. This all was really the fun part. Some questions are...what do I really want the response to look like? How much BSC is needed?, is a tilted down response what I want. Well...I'll find out with this build.

What I have now is the simulated anechoic response from my test box testing. I do believe I will need less than the 6db BSC. If we say 4db I expect the responses I show below to be a couple dB higher in the low end.

I'll test again when I complete the real box. This will help check for errors and refine the response...this test data gives me something to practice on:T

So here's Proto1 SSA-WG. It has Duelund response targets with a center frequency of 800Hz and a damping coefficient of 3 giving x-over frequencies of around 300Hz and 2135Hz.


Next is the target and driver responses at 1M centered between the mid and tweeter



Here is the optimized system response of all drivers at 2M, 15 deg off-axis, centered between mid and tweeter


Now the Midrange is in phase showing the out of phase notches


The simulated vertical polar responses from 1250Hz to 3500Hz in 250Hz steps

[sdl2112]

Here's a few more. The SPL scale should be pretty accurate but remember at the 2 Meter distance the amplitude is reduced by 6dB. I expect in the upper frequencies it will be about 87-88dB...low end 89-90dB at 1M...we will see.

I have a notch in the low end where I can adjust R33...Here I show 1000(removed), 30 and 15ohms. This will likely come into play more with the ported version with the second higher frequency impedance hump.


Midrange - Adjusting R13 from 4, 5, 6 ohm


Tweeter - Adjusting R16 from 0, 0.5 and 1 ohm


Impedance fattening circuit, R29


This is to illustrate flexibility in the response hopefully with a few resistive components.

[5th element]

Nice work there! The flexibility that you've baked into the xover should be great for adjusting the overall tonal balance.

[sdl2112]

Thanks Matt! Anxious times...really looking forward to seeing the results of the real build and how it compares.

-Scott

[jim1961]

Tweeter - Adjusting R16 from 0, 0.5 and 1 ohm
[ATTACH=CONFIG]24660[/ATTACH]

Having shared some listening lately, I am thinking you might like the green plot best. A little extra treble for those of us with aging ears

[sdl2112]

Hope I don't need more than that because that's wide open...h:

[jim1961]

Here is my FR L/R at the LP. I am about 2db (by design) down in the treble region compared to the mids. If your are flat to out there, I think it will be plenty bright enough.