Design Check/Input TABAQesque with TB W5-2143

[Deude_Mann]

I would like to get feedback on this design from those who are willing. TIA.
Goals:

• More compact than the TABAQ for 5" shown here TABAQ design (S1 - S3 is much smaller than usual Sd rules of thumb)
• Do not exceed xmax below the unloading frequency (xmax with damping is @ 40 Hz)
• More low extension than a simple BR
• Start with a Tang Band W5-2143 (not great xmax but upper response looks flat) TB W5-2143

Results (so far, not great):

Observations:

• The BR in WinISD has better SPL (+6 dB) at the same xmax limitation as the MLTL.
• The MLTL is a little better on its -3dB point but not much.
• Looking at the MathCAD worksheets for the TABAQ (in the long TABAQ thread) it also has a SPL of about 90 dB at xmax. Its extension is better (due to more volume?) but is not hugely better.
• Peak group delays are similar
• The total volume of the BL is 17 liters (WinISD) and the MLTL is 18 liters (Hornresp).
• I played around with port length and settled on the length that provided the flatter response at the bottom (below 100 Hz or so). It was possible to get more extension with more port length but the width of the 1st bandpass response started to suffer; the roloff after ~100Hz started to suffer.

What am I missing here? I do not see much benefit if any to the MLTL over the BR as I have modeled it. I AM NOT BERATING THE MLTL; it is clearly a proven approach. I am wondering what I may have missed in my simulation that limited the potential in the MLTL approach since I am not a long-time user of Hornresp.

I have read through a lot of the TABAQ thread, and used Hornresp to model several different MLTLs, TLs, and tapered TLs trying to come up with a design that beats the BR but I can't get there for some reason I do not understand. If so, then I could use help with my modeling and/or optimization approach. Again, this is not an attack on the MLTL; I only mention this again since some here seem to get quickly offended if there is even a suggestion of such a thing.

Thanks in advance

 

[Deude_Mann]

Here are the exported Hornresp files and WinISD files for those who might want to dig deeper.

 

[GM]

Short on time, so just a few notes:

Forget using Sd ROTs

Speaker design trade-offs

Per the above, MLTLs should always be bigger than an equivalently tuned BR since you're comparing a Helmholtz resonator to a vented closed cylinder, closed cone TL

 

[Deude_Mann]

Thanks. So if I understand your point, basically you are saying my goal of smaller size (cross-section) while retaining extension for a given MLTL are fundamentally in conflict.

 

[GM]

You're welcome! Sad, but true, can't fool Mother Nature. Better though to think in terms of acoustical loading, i.e. 'bigger is better' (BIB) rules for greater LF gain bandwidth (BW) and ideally need to understand the fundamentals of speaker design if you want to do more than just randomly 'experiment' with software.

Also, please Export the Hornresp (HR) file(s), so we can Import them to see the whole design. Ditto any WinISD file(s) for those members that still rely on it, though best to design in HR and only use other programs for feature(s) it doesn't.

 

[Deude_Mann]

Thanks. The Hornresp file as well as the WinISD file were already exported in the zip file in the original post. They are zipped since the native files did not seem to be allowable here. If that is not what you are looking for let me know.

I did not go into the pros and cons of the fundamentals of speaker design for this context simply because I wanted to keep the post concise. Most people, including me, turn off to 'word walls' it seems. Hornresp is what is new to me.

 

[woodo]

I’ve most likely misunderstood but this paper of Martin’s attempts to “unify” the various enclosure physics on offer pointing out advantages and disadvantages. Might go some way to answer some questions as well as raise many more. http://www.quarter-wave.com/TLs/TL_Alignments.pdf.

 

[GM]

Thanks. The Hornresp file as well as the WinISD file were already exported in the zip file in the original post. They are zipped since the native files did not seem to be allowable here.

You're welcome!

Thanks!, didn't look in the .zip files since we can use HR's Export/ Import feature to attach/DL its .txt file and I quit using WinISD (can't attach IIRC) once HR became good enough to do all the sims I was interested in and at higher accuracy overall.

 

[GM]

Dunno, guess I'l have to read it when time permits, but FWIW his Classic TL alignments were spot on from a T/S theory POV, so not sure what he might have added in this new doc, but it pared all the 'fat' (acoustic gain) out of TL design, so wind up with DM's dilemma.

 

[Deude_Mann]

FWIW here is the same design, changing only the S1-S3 sizes which increases the volume of course. The bass extension is considerably improved though the SPL stays about the same. The loss of ripple above 200 Hz is from increased path distance.

 

[Deude_Mann]

Now it turns out WinISP uses a 2pi steradian assumption, and I had Hornresp set for 4pi, so that explains the gain problem. Major facepalm. I had wondered about this but it took a bit to find the answer. Now the SPLs between WinISP and Hornresp match closely. Also, the MLTL has an edge on the response below 60 Hz in that the rolloff is much slower; at 40 Hz the MLTL is down 6 dB and the BL is down 12 dB. That is huge!

If I could go back and edit the original question I would!

 

[GM]

Yeah, noticed that. Another thing is dialing in driver/vent pathlength differences relative to one's ears that reduces mid band 'ripple', reducing amount of damping required, though need to have it behind the driver, not just partially. Also, if not going full digital, then need to add any series resistance since raises Qts, filling out the bottom end even more.

What got me interested in column/tower/TL alignments early in my 'adventures in audio' was the pipe loading of the vent allowed bigger ones for a given length (Lv) or a length reduction for a given vent size (Av), so even more bottom end and/or could be tuned lower to better blend with the room as well as improve transient response by somewhat protecting the driver from turntable (TT) 'rumble'/audible over excursion.

 

[woodo]

Now it turns out WinISP uses a 2pi steradian assumption, and I had Hornresp set for 4pi, so that explains the gain problem. Major facepalm. I had wondered about this but it took a bit to find the answer. Now the SPLs between WinISP and Hornresp match closely. Also, the MLTL has an edge on the response below 60 Hz in that the rolloff is much slower; at 40 Hz the MLTL is down 6 dB and the BL is down 12 dB. That is huge!

If I could go back and edit the original question I would!

View attachment 1159498

Quite a good result for a 5” speaker! I emailed TB some time back to check if the 2143 was available from a local distributor but sadly not—I would need to import. As GM raised, the path setting will assist with the simulation as discussed in the hornresp thread recently https://www.diyaudio.com/community/threads/hornresp.119854/page-669#post-7266491
Edit: I see you have path dialed in. filling to the recommended density along 3/4 of the line might smooth out the last ripples. Can adjust when the box is built to taste.

 

[Deude_Mann]

Agreed. I played with the damping quite a bit and basically added it until the returns (ripple reduction) began to diminish. Adding beyond that just diminished the low end response which is what one would expect from a TL normally of course. So I think I am into the point of simulation error and will just adjust based on measurements of the finished design.

 

[eusubius]

How does this translate to pipe area, vent size/placement, and driver placement?
After my W6 project I have a friend who wants a W5-2143 in MLTL.

119 cm is the horn length, so the box will have to be around 4' tall, but I assume the driver is lower.

If there's a good intro to HR video on YT, let me know, as I know this stuff should make more sense than it does.

Thanks!

 

[GM]

Hmm, not familiar with Tabaq design theory or even if this is part of it, just know it's a well proven, popular offset, so in a 'nutshell', set driver at 'x' distance to floor, then add (0.5361 * 'x') to it to find its total height (L) for a typical acoustic 3rd harmonic (0.349) driver offset; vent(s) are normally down low in the pipe's 7th harmonic or 'L' * 0.152, i.e. we normally always want them at an odd harmonic. All dims ideally i.d.

 

[eusubius]

Beautiful! I like math.
Found this video:

hopefully I can use the math, and learn from the video so I can model and understand what I'm modelling.

 

[eusubius]

@Deude_Mann have you built this yet?

I'm learning how to man the S and Con parameters... however HR seems to have changed, and I can't find info for line dampening, or figure out what point the port is offset from top of cab.

So far I'm leaning toward building this, or something very similar to. May play with design a little yet.

 

[woodo]

can't find info for line dampening

Opening the help file and searching on "filling" brings up some info. Also when you create an initial design in the "design wizard" (for a transmission line) or finesse it in the "loudspeaker wizard" in the tools menu, there are controls to adjust how much and location of filling. There are likely ways others have used it too.

 

[GM]

Here's a max flat MLTL (Vb = Vas, Fb = Fs) with an approximation of a TABAQ style vent down at the bottom.