New project has come alive. Seas 27TBFC / Peerless 830884 TMM

[fjhuerta]

Hi all,

I have everything I need to start building my new tower speakers. The plans are:

1) TMM, Seas 27TBFC on top, Peerless 830884 woofers on a TMM configuration.
2) Ported alignment, 4" port.
3) 96L enclosure size, tuned to around 30 Hz.
4) 1" chamfer all around.
5) 2" front baffle (MDF), 1" MDF all around.
6) XOver point around 1.4 KHz, 4th order, L-R.
7) Front facing port.

The design is inspired on the JBL L250, although (obviously) it's vastly different - this will be a 2.5 way, with a much smaller woofer. I also found the baffle shape helps quite a bit with tweeter diffraction.

These are the basics. I just delivered my plans to my local furniture builder. I'd love to hear any opinions on the driver configuration and positioning I arrived to. I have no experience with TMM speakers, so any help is greatly appreciated!

This is the front baffle configuration. The base is quite wide - 45 cms. This is because, in order to get 96L out of the enclosure, I'd need a box 40 cms. deep (quite deep for my room, actually).

I may build the columns a bit slimmer (40 cms. wide) and build the box 45 cms. deep. It doesn't really affect baffle diffraction measurements. Any ideas or hints? I'd assume a slimmer baffle would look prettier.



This is the baffle diffraction I'd be expecting:



I think the tweeter diffraction will be easily controllable, and considering the 27TBFC's dropping response below 1.7 KHz or so, it may actually be helpful for the crossover point I'm looking at.

At this point, I'd be very grateful for any opinions regarding driver placement and baffle width, since I'll give the local furniture maker an advance payment on Tuesday, and the design needs to be finalized by then (I don't expect major changes, though).

One thing that makes me think I may have my baffle configuration wrong is Jon's M8a design, where his woofer is a bit far away from the tweeter. Is this because of floor bounce, or any other reason? Am I on the right track?

EDIT : Changed thread name to a more descriptive one

[cjd]

So, tweeter to lower mid-woofer distance is greater than the crossover point would support... A 2.5 way would work great though.

Also, the tweeter being so far off the vertical axis will tilt the image - you'll lose some diffraction but you may not get the imaging that you desire.

Otherwise, looking good. Neither is a show-stopper by any means, just something to be aware of.

C

[fjhuerta]

C, you are right. I forgot to add - it's a 2.5 way design.

I'm playing round with tweeter positioning, so I can center it a bit more without adding too much diffraction. Thanks for your input!

[fjhuerta]

Revised front baffle....

[dlneubec]

Javier,

This second layout looks improved. You might shoot for the Golden Mean ratio for the distances of your tweeter from the 3 adjacent sides as a start. I would guess you are close to that now anyway.

[TacoD]

Why is the tweeter at the edge of the baffle? My experience is that if you have some baffle above the tweeter you got less edge diffraction.

[cjd]

Nah, you just move the diffraction down in frequency. However, I would think that with the box shape what it is, moving the tweeter down and centering above the mid-woofers would lead to good results diffraction-wise, and better results horizontal lobing wise.

C

[fjhuerta]

Dan, I tried the Golden Ratio, but it placed the tweeter too low on the baffle (due to the 1" chamfer I'll be using). Diffraction did look quite a bit better, though

Taco, the tweeter is not at the edge of the baffle. It's at the edge of the chamfer. It's 2.5 cms. below the edge.

C, I followed your suggestions. Although moving the tweeter down made the diffraction effect far more pronounced, centering it basically made it identical to the off-center response. So that's the baffle I'll be using! Thanks for your suggestion.

[fjhuerta]

I'll try to document this project as much as possible. If I'm posting too much information please let me know.

This is the front baffle diagram. I'll try to post the ported bass response later during the day.

[dlr]

That alignment looks better

The highly offset tweeter even looks out of place. Your final one has better aesthetics.

I made a small bookshelf 2-way for my brother years ago that had the tweeter tight into the corner. This was before the BDS. The response was the smoothest I had ever gotten from a tweeter diffraction-wise.

Keep in mind that you'll have diffraction from the mid-woofer below it. That is at times stronger than the edge diffraction on-axis. Off-axis will be another story.

Have you investigated your intended listening axis in the BDS? Remember that even a 5-10 degree off-axis position is common when placing in the room. I seldom listen to a set with the tweeter pointed straight at my ear. Some degree of off-axis listening is typical. The diffraction can be significantly different.

Some judicious use of felt for the remaining diffraction control will help. One reason I'm a big proponent of felt for diffraction control is that it makes the response on all axes much improved so that correction is not needed much in the crossover. This means that if you decide that you need to change the listening angle you'll have a more uniform response at any position. Having the tweeter centered as you now do will help that as well.

[fjhuerta]

Finally... some progress

The cabinets were delivered today. They look really nice. I can't take credit at all for them, though - a friend of mine is great with woodwork.

They are braced and pretty heavy. I'd say around 60~70 pounds. The front baffle is 2" MDF, the rest is 1" MDF.

The grilles are made out of metal rods, separated about 3/4" from the cabinet. I think this may help minimise diffraction, compared to a regular wooden frame. I'll need to take some measurements.

I took some pics of them with my phone camera. They look very fuzzy, but I just couldn't find my "regular" camera...

Here's the image of the speaker that inspired my design...



Here are the finished cabinets, with the tweeter mounted just to see how they'll look:






Hey, I have attachments Not trying to be political here and #10000! Do I get a prize? :B

[TacoD]

Wow very nice!! No you don't get a price for these id's, but you've got my attention .

[ThomasW]

I'll try to document this project as much as possible. If I'm posting too much information please let me know.

There's no such thing as too much documentation when it's related to the design process of an actual project.....

Cabinets look good....:T

The raised grills may cause as much or more diffraction problems than flush mounted ones. The effects should be fairly easy to measure.

[fjhuerta]

Unibox sim

Here are the Unibox plots I was missing.

Basically, it's a 96L box, tuned to 30 Hz. I could go down lower, but I feel I won't be missing much, music-wise.

Port size is 4".

Frequency response:



Step response:



Port air speed:



Impedance:





Peak cone excursion:

[fjhuerta]

The wiring is ready, so with any luck, I may have some frequency response plots in-cabinet this weekend.

Just one question, though. I usually damp the walls of my cabinets with eggcrate foam. I read on Zaph's blog this isn't the best way of doing things - he suggests using carpet padding. Is there really a big difference? If so, I need to get some carpet padding... fast.

[ThomasW]

My recommendation is regular old fiberglass insulation. I know it's itchy and nasty stuff to work with, but it's coefficient of absorption is great....

[Paul H]

My recommendation is regular old fiberglass insulation. I know it's itchy and nasty stuff to work with, but it's coefficient of absorption is great....

Thomas and all,

Next time you need some, try a mineral wool batt insulation like Roxul makes - similar or better absorption, without the itch. It's sold in the big orange box store and other building supply places around here.

[ThomasW]

Paul,

As far as I know it's not sold in the US. I've certainly never seen it anywhere.

[fjhuerta]

Well, I finally lined up the cabinet, yet I used a mix of 3/4 foam, and 1" egg-crate foam. I couldn't find any carpet padding near home, and I was aching to finish the speakers.

I also wired the drivers with 14GA Radio Shack cable and PE terminals. Nothing fancy.

Perhaps tomorrow I'll be able to take the MLS measurements (going back to school is seriously limiting my free time!).

Everything seems fine. The cabinet is extremely heavy, now that it has the drivers. I usually take measurements with the speaker on top of a 50 cms. base, which I place on a 50 cms. high step at home. Hopefully, I'll be able to lift the speaker myself (there's nobody to help me around here).

One thing - the speakers look quite.. hmm... "strange" with the Peerless drivers. The light gray border really stands out. I don't know why, but they remind me of vintage speakers. I'll try to take a decent pic tomorrow.

[fjhuerta]

MLS files and graphs

I had some free time today, and hurried up and took the measurements.

1M, 4ms. The woofers have had their nearfield+farfield measurements merged, and I added the BSC, so they should be ready for modeling.

Lower woofer



Upper woofer



Tweeter

[fjhuerta]

I couldn't upload 6 files, so here goes the one that's missing.

[69Stingray]

Looking Good!

[fjhuerta]

Thanks Stingray!

I think I'll devote one or two weeks to crossover design. As it is right now, I find myself faced with two options.

Option #1: The "Zaph" design (Apologies to John Krutke if I'm wrong, but I got the idea for it after reading his design mantras - don't throw components at a FR curve just to make it extremely flat). As far as I can tell, it'll only need two components for the tweeter and three for the woofers, and it yields a nice flat response. 4th order acoustical, crossed over at 1,500 ~ 1,800 Hz seems to work fine.

I actually measured the 30° off-axis frequency response. So far, the "Zaph" design yields the flattest power response curve.

Option #2: The Jon Marsh / Evil Twin design. I haven't yet begun this design, but it obviously involves very steep slopes (6th ~ 8th L-R), and I may employ a C-E variant. I have experience with the usual parallel crossovers, so C-E presents quite a bit of a challenge.

Given these are well-behaved drivers, I may go with Option #1, although I'll be simulating for quite a while...

[mmoeller]

I don't believe that a C-E design would have to be one of Jon's series/parallel designs like the NatP. It's a matter of adding the extra zero to the transfer function in the correct place, and making sure the "rebound" from that zero doesn't cause trouble. The closer to the crossover point you choose the zero the greater rolloff, but also the greater "rebound". That all shouldn't stop you from adding it in a parallel design. I could be wrong though

[robertwb70]

Paul,

As far as I know it's not sold in the US. I've certainly never seen it anywhere.

They sell both Owens Corning and Roxul HERE and their prices seem pretty reasonable for the Roxul stuff (especially considering I can't find anything locally)

[rmht]

Paul,

As far as I know it's not sold in the US. I've certainly never seen it anywhere.

Thomas

A Denver guy here...long time lurker...

Anyway Plateau Insulation is in Aurora, much better prices than above on rock wool, 703 and 705, will deliver free any amount in metro area. I use them all the time, ask for Lou...3.371.0505

I have fabricated my floor to ceiling traps and all reflective panels using them, also customer's projects from time to time.

[fjhuerta]

OK, I need some help... please...

I'm having some issues with destructive interference starting at around 800 Hz.

After aimlessly trying to design a crossover without first really thinking about how to do the 0.5 woofer, I'm back to square one.

I wired both woofers in parallel on LSPCad. I then added an inductor in series with the 0.5 woofer, and tried to flatten FR with it.

Much to my surprise, I couldn't get a flat response. The 0.5 woofer is cancelling some of the midwoofer's output at 800 Hz and above.



Common sense tells me I should just ignore its effects, since they'll vary according to the listener's distance from the speaker.

Anyway, that's what I get with a 4mH 15GA air core inductor...

Am I on the right track?

[Evil Twin]

OK, I need some help... please...

I'm having some issues with destructive interference starting at around 800 Hz.

After aimlessly trying to design a crossover without first really thinking about how to do the 0.5 woofer, I'm back to square one.

I wired both woofers in parallel on LSPCad. I then added an inductor in series with the 0.5 woofer, and tried to flatten FR with it.

Much to my surprise, I couldn't get a flat response. The 0.5 woofer is cancelling some of the midwoofer's output at 800 Hz and above.



Common sense tells me I should just ignore its effects, since they'll vary according to the listener's distance from the speaker.

Anyway, that's what I get with a 4mH 15GA air core inductor...

Am I on the right track?

Welcome young Padawan to the world of 2.5 way crossover design... your initial efforts with cabinet development and construction have been most impressive, but as others who precede you on your path have found, a 2.5 way is not a simple design to master.

Yes, you are getting cumulative phase shift on your 0.5 woofer which is causing some destructive interference - you may also be getting some based on the "measurement point" in your simulations or measurements- for example, if you stack two 8" or 10" or 12" woofers and measure on axis to one, you will get destructive interference due to arrival time differences and phase shift. This is one reason to be careful about the baffle width and frequency for which you decouple your 0.5 driver from the main.

Now, you could "ignore" or minimize the perception of this effect by measuring/simulating the midwoofers half way between the two- this is the equivalent to your on axis driver response if what you had was a single oval shaped driver defined by the upper and lower boundaries of your midwoofer pair.

Consider that the 884 is a largish driver anyway for a two way, it looks like you have about 8 cm between the 884 frames, this is equivalent to driver ~53 cm tall you're operating up to 800 Hz. (~18" woofer). Do you see any potential problem with that?

One possible approach to compensate for this issue would be to use a lower 0.5 crossover point, employ a small amount of BSC in the "full range" 8, (which you are crossing at what frequency to the tweeter?) - several advantages to this, because the Baffle step response it self induces a phase lead in the 1.0 woofer which contributes to time difference and cancellation.

If given your problem, my first attempt would be to move the range of the BSC down, perhaps close to an octave, so that the arrival time differences would be a much smaller phase shift.

It might also help to google some successful designs on the web and study their transfer function. You've created something of a challenge for yourself with your choice of drivers and spacing.

But perhaps you would do well to ignore these comments- after-all, this aging Sithlord has never done a 2.5 way himself, only something very similar in the Wilson X1 Klone where the low bass module is used with a slow roll off to provide BSC to the upper range module (100Hz & up)

[TacoD]

Welcome young Padawan to the world of 2.5 way crossover design... your initial efforts with cabinet development and construction have been most impressive, but as others who precede you on your path have found, a 2.5 way is not a simple design to master.

I totally agree, a 3-way is easier to get right. I have done several 2.5 ways (the latest with 2x Dayton RS180 and Scanspeak 9600SE).

Most of the time I cross in the 100 - 200 Hz region, and as Jon wrote you need to do some minor BSC for the 1.0 woofer. But be carefull, in some of my designs the second impedance peak of the BR loading (tuned at 45 Hz you get a nasty peak @ 80 - 90 Hz), screwed up the working of the x-over. So you have to apply a notch to get the proper transfer functions. Also woofers with a lot of output used in a small frequency range (if you're doing a 2.5-way with 6" or 7" woofers) are hard to filter, you may need some extra components to get the x-over point.

A nice trick for 2.5-way x-overs is to cascade the 0.5 filter to the filter of the 1.0 woofer. You get the transfer function of the 1.0 woofer and then you don't need much extra parts to get the transfer function.

[fjhuerta]

Evil Twin,

I should have supposed the Dark Side of the Force would be strong on a design with woofers so separated from one another. The way I saw it, many 2.5 designs had the lower woofer closer to the floor, and I thought that both woofers would be better integrated if they were as close together as possible (they are actually 5 cms. apart). Guess I was wrong - they were far away from one another because of the cancellation!

TacoD, my intention is using the extra inductor for the 0.5 woofer roll-off. Unfortunately, it's simply not enough - phase cancellation is happening earlier than I expected. I took a look at the unfiltered summed response from both woofers, and I have a lovely, just lovely, phase cancellation issue happening at around 2 KHz. It's something like -6 or -10 dBs in magnitude (I need to check). There's no way I'll be able to get around this with a single inductor.

I'm thinking about the following things.

1) Ignore the minor interference.

2) Simulate some measurements at 1M, 2M, 3M. Quite probably, the destructive interference will present itself at other frequencies. If this is the case, then I'll always have an issue. If I can at least isolate the affected frequencies, then I can roll-off the 0.5 woofer earlier, like Evil Twin said.

3) Instead of using a single inductor as the 0.5 woofer component, try going to a higher order crossover. Highly doubtful - I expect phase shift issues will not allow me to do such a thing.

Rats. What if there's no real solution to this problem?

[augerpro]

Cascading the filter like Taco said is done in the best designs I've seen around teh web. I think that method is becoming pretty popular.

[fjhuerta]

Cascading the filter like Taco said is done in the best designs I've seen around teh web. I think that method is becoming pretty popular.

Yeah, but unfortunately, the cascade filter is not giving me enough attenuation...

[TacoD]

In my applications I've a pretty steep slope for the 1.0 woofer, so I never encountered this problem. So I would suggest to lower the x-over point, or increase the x-over slopes.

[fjhuerta]

In my applications I've a pretty steep slope for the 1.0 woofer, so I never encountered this problem. So I would suggest to lower the x-over point, or increase the x-over slopes.

Good idea...

I'll try a 1.3 KHz or so 6th order slope. See where that gets me...

Thanks!

[fjhuerta]

OK... 6th Order L-R seems to be the best compromise so far between interference and FR.



Filter topology:



Raising the 0.5 cascaded inductor value to, say, 4mH doesn't give me a flatter response. I think I'll need to raise it to at least 3.5mH, since the lower woofer is still playing at a high level in the midrange.

What do you guys think... am I on the right track? So far, the only changes I'm thinking are raising the inductor value, and raising the -6dB point to 1,400 Hz (I think 1.3KHz is too low for the Seas 27TBFC).

[fjhuerta]

After a couple of tries, I decided it didn't make any sense to use paper drivers with a very high order crossover... I just didn't see any benefit.

So here's my first "production" try. A 4th Order L-R simple crossover. The tweeter part comes from Zaph's MTM, with a few modifications. I tried to make FR as flat as possible while taking into consideration phase cancellation issues.

Here it goes...

Schematic

[TacoD]

I do not understand why you want to cross the 0.5 woofer that high. And I agree, with paper cones you don't need 8th order Cauer .

[fjhuerta]

TacoD,

If I use a higher crossover point for the 0.5 woofer, FR is not as flat as I'd like.

Then again, considering I used the KHF tool to obtain the baffle step response, it may be possible that it isn't as exact as I think it could be. And I'd definitely have the 0.5 woofer playing as far away from the midrange as possible, so I'll follow your suggestion.

I raised the inductor value to 4.7 mH.






This is the crossover I'll build. Considering I'll do a lot of listening off-axis (the speakers will be used at my living room), the extra HF response will actually flatten at around 30° (I took 30° off axis measurements, so I'm sure of this!). A 4 Ohm / 4 uF conjugate network on the tweeter would flatten FR on-axis perfectly.

A thing I wasn't expecting - the crossovers are a bit more expensive than I thought. They were $110 for a pair... the 4.5 mH iron core Erse inductors were $41 a pair. Yikes.

[TacoD]

Good work! A lot of thought got in your project .

You can buy some cheap ferrit core coils to try the x-over first. An x-over is always expenisve, that's why commercial offerings try to minimize the number of components or cross higher so that they can buy smaller coils.

[fjhuerta]

Thanks a lot, TacoD!

I just came back from Canada. I bought a pair of Energy C-50 speakers while I was there (too cheap to pass up). I'm listening to them right now, and I'm being reminded why DIY is so much better... they sound pretty bad to my ears!

Anyway, as for the crossover -the parts arrived today. I need to find out a way to finish them by Friday night... basically, everything is ready, except for the ports (I need to glue them to the exact length). I can't wait to listen to the speakers - it's the first time I've really been able to control tweeter diffraction.

[fjhuerta]

OK, the crossovers were built yesterday, and the speakers are assembled. The only thing left to do is tuning the ports... Probably tomorrow I'll finally know how they sound.

[fjhuerta]

Finally!

Listening impressions...

1) These speakers are very dynamic - quite a bit more than my (tower) Nat P's. They also have a lot more bass response. They are problematic in that respect - they have too much bass, IMHO. I need to play with positioning, but they are so huge, this is going to be difficult.
2) Somehow, they are smooth. Very smooth. Again - compared to the Nat P's, there's something I don't understand. I didn't design any BBC dip into them, yet they aren't as "in your face" as the Natalies. They aren't recessed - they simply have a more distant presentation, which I find quite nice. I'd assume this is because of the drivers, which are paper, instead of aluminum. But it doesn't make any sense - after all, if FR is similar, both speakers should have the same character...
3) The 27TBFC seems to have a bit more "bite" and presence than the Dayton RS28, which sounds (to me) like a cloth tweeter. I prefer the 27TBFC, actually - apparently, not having an upper end roll-off really makes for a noticeable difference.
4) I listened to the speakers for around 6 hours with no noticeable listening fatigue. Even my wife commented on this - she couldn't stand the Nat P's for too long.

At this point, I'm completely confused. Why should these speakers sound so different to the Natalies? If I had to describe what I heard, I'd say the Nats are more detailed and accurate, and these speakers are far more forgiving and relaxed. But FR seems to say both speakers should sound very similar. I just don't get it.

Also, what can I do to tame a bit bass response? I just lined the interior of the boxes with foam. Would filling them up with dacron or something similar help?

[JonMarsh]

Tuning the port/boxer lower will probably help out- as it is, you're only down about 2 dB at 40 - 50 Hz. Moving the tuning lower will put a bit more of a slope in the Fb between 100 Hz and the Fb- I'd try for around 25 Hz.

Also, be sure to pay attention to the ratio's of distance of the speaker with regards to adjacent boundaries- spread the LF reinforcement evenly through using golden ratio staggered distances, like in this example





This can also play a big role in taming the perception of too heavy bottom end, because if two boundary distances stack up, you'll have both an unwanted peak and a dip above.

[TacoD]

Good to see that you're making progress! I also recommend to tune the port lower. Also make some measurements for the system + filter.

[Mazeroth]

Tuning the port/boxer lower will probably help out- as it is, you're only down about 2 dB at 40 - 50 Hz. Moving the tuning lower will put a bit more of a slope in the Fb between 100 Hz and the Fb- I'd try for around 25 Hz.

Also, be sure to pay attention to the ratio's of distance of the speaker with regards to adjacent boundaries- spread the LF reinforcement evenly through using golden ratio staggered distances, like in this example



This can also play a big role in taming the perception of too heavy bottom end, because if two boundary distances stack up, you'll have both an unwanted peak and a dip above.

Jon,

Which program is that?