Help Understanding Tekton Tweeter Array Schematic?

I have a pair of Tekton Double Impacts and I am trying to understand the details of the tweeter array. I created the attached schematic from my pair.

Can any of you crossover gurus break this down and explain the functions of circuit?
 

Attachments

  • Double Impact Tweeter Crossover.pdf
    301.6 KB · Views: 930
frugal-phile™
Joined 2001
Paid Member
I am no XO expert but this is what it looks like to me:

750094d1555442878-help-mid-range-driver-tekton-double-impact-txo-png


dave
 

Attachments

  • tekton-double-impact-TXO.png
    tekton-double-impact-TXO.png
    12.7 KB · Views: 5,573
If planet's scema is right it makes sense to me. I see it as an array that grows in size as the frequency goes down.

Ta is the "point", then gradually at a lower freq Tb Tc come into play and at this stage there is rightly a vertical "line array" of 3 units, then eventually at a lower freq still the entire array is active.
 
diyAudio Moderator
Joined 2008
Paid Member
The resistor shelves the capacitor at some frequency *. Around this point and at higher frequencies the capacitive reactance is below 1 ohm and the resistor is effectively shorted. At lower frequencies, 1 ohm is added to what the 15uF capacitor is doing.

The 1uF capacitor almost looks to be unnecessary, but is likely making a subtle but wanted change in the response. Simulating would make this much easier to understand.

* [about 20kHz - 1/(2*pi*1*0.0000082)]
 
Last edited:
Could someone explain the second order high pass and equalization circuit? I cannot find any resources that explain the equalization portion of the circuit layout. Thanks!

I believe that mess of capacitors in the top left of the circuit is largely an attempt to protect the amplifier and to lower the cost of the circuit.

For instance, a very high quality 15uf cap is expensive. But it looks to me, that they've done the following:

1) the 1uf cap is probably an expensive one, and they're using it to bypass the 15uf cap and the 8.2uf cap. (If you put a 1uf cap in parallel with a 8.2uf cap, you get a 9.2uf capacitor array.)

2) The resistor is likely there to raise the overall impedance, to protect amps that can't deal with a 2ohm load.

If the designer didn't care about blowing up amplifiers or saving money on capacitors, he could've made the circuit with ONE capacitor, instead of three caps and a resistor. Once you boil it down that way, you can see that the cap and inductor in the start of the circuit is a simple 2nd order high pass filter.
 
  • Like
Reactions: 1 user
The resistor shelves the capacitor at some frequency *. Around this point and at higher frequencies the capacitive reactance is below 1 ohm and the resistor is effectively shorted. At lower frequencies, 1 ohm is added to what the 15uF capacitor is doing.

The 1uF capacitor almost looks to be unnecessary, but is likely making a subtle but wanted change in the response. Simulating would make this much easier to understand.

* [about 20kHz - 1/(2*pi*1*0.0000082)]

Here's a sim.

10-DI-tweeter-array-767x1024.jpg


First off, here's our drivers : SB29RDNC-C000-4
$58 from Meniscus : SB, SB29RDNC - Meniscus Audio

beta6A-size500.gif

Eminence Beta-6A
$65 from PE : Eminence Beta-6A 6-1/2" High Power Midbass Midrange Woofer Speaker

Side note: I don't know how they make a profit on these speakers. Even if you built these yourself, you're looking at nearly $700 for the midrange and tweeter array alone. The Internet says the speakers retail for $3000 a pair? That's a crazy price.

8uxma8o.png


First off, here's what the circuity would look like with no xover filter whatsoever. This is the basic response you'd expect from a nice soft dome tweeter. 87dB efficiency, a F3 at 1200Hz and output to 20khz.

myazKjf.png


Here's the response with Tekton's seven element array. There's a series of things happening here:

First off, with a center-to-center spacing of about two inches, the array elements will begin to destructively interfere with each other at about 2khz if they're all playing full range. That's a big IF; the whole point of this xover is to "power taper" the elements on the outside, similar to what D.B. Keele does with his CBT array. Only the center element is getting full power.

Instead of conventional power tapering, the Tekton array filters out the highs but not the lows. This means that below 3khz, the seven element array is basically behaving as a single diaphragm that's approximately seven inches in diameter. This is nice because it narrows the directivity, similar to what a waveguide would do.

In the sim, you can see that the center driver (red) gets full power, the top and bottom tweeter (yellow) get approximately 100% as much power (each) and the outside tweeters get 25% as much power (each.)

The graph is a bit deceptive, because the yellow curve represents *one* tweeter, not two. IE, the combination of the two will be louder than the yellow curve suggests.

The most important part about the whole circuit, is that every tweeter is lowpassed except the tweeter in the center.

As if all of this wasn't confusing enough, the low pass filters on the outer elements do one more thing, which is that they CURVE the array. This is because the inductors introduce a delay. So the lowpass filters curve the array, just as if the baffle of the loudspeaker was curved. Curving the array should extend the highs and widen the beamwidth of the array.

All of this is fairly standard stuff if you've read up on the Keele CBT papers. Very similar idea.

In a nutshell, the Tekton speaker uses an array of seven tweeters to raise the on-axis efficiency to about 93dB. The tweeters hand off to an array of high efficiency prosound midranges. The net result should be a speaker that looks (fairly) conventional but has efficiency and directivity control that's similar to a horn. In a lot of ways, it's a two dimensional CBT array.

Here's some stuff that might help explain this speaker:

1) passive loudspeaker delay : Passive Loudspeaker Delay
 
  • Like
  • Thank You
Reactions: 2 users
In post twelve, I made a couple of arguments:

1) The six tweeters in a ring are not playing full range, they are low-passed

2) That mess of capacitors in the front of a circuity isn't necessary. You could do this with a simple 2nd order filter. The mess of capacitors is likely to lower cost by using one expensive cap bypassed with an array of inexpensive caps.

To test this hypothesis, I made another model.

This model has a couple of features:

1) the mess of capacitors is replaced with a single cap

2) The ring of six tweeters is replaced with a ring of six midranges

Tekton-DoubleImpact-top.jpg

Picture this array, but replace the six tweeters in the outer ring with six midranges instead. The center tweeter remains unchanged.

Here's the results:

Ce2EW0M.png


The idea is the same. A single tweeter playing full range in the center, surrounded by six more drivers that are low passed. In this case, the drivers are Tymphany TC6 midranges instead of SB29 tweeters. This drops the cost A LOT. Seven tweeters costs over $400, whereas one tweeter and six midranges costs $106. The use of midranges instead of tweeters makes the rolloff of the array shallower. The overall efficiency is a little bit higher. The impedance is a LOT lower; about half as much. So if you did this, you'd have to tweak the crossover to match it to your amp.

Basically the use of midranges instead of tweeters in the ring lowers the cost a great deal, but the efficiency suffers because the tweeters have an efficiency of 87dB while the midranges have an efficiency of 84dB.



If it were me, I'd use midranges in the ring instead of tweeters. The use of tweeters raises the efficiency a tiny bit, but at an enormous cost, basically quadrupling the cost of the array.
 
Last edited:
Nicely done Patrick!! I looked at the review, and it also seemed as if the slopes were too low. Higher order would give significantly better off axis response.

I'm stuck on a cruise ship so I really went down the rabbit hole of "loudspeaker shading" today.

Possibly the "purest" example of a shaded speaker is something like the Keele CBT. There is no xover, just shading and curving. A lot of the data that's published on it looks spectacular.

One of the interesting things is that there aren't many measurements of an *actual* CBT. A lot of the data is theoretical.

If you're trying to get an array of identical drivers to work together (like the Tekton array) the rule of thumb is that your upper limit is:

Center-to-center spacing / 3

For instance, Monte Kay built a CBT speaker using 1" Aurasound speakers. The center-to-center spacing is about 1.5". That would indicate an upper limit of 3khz. (13500 / 1.5" / 3 = 3000hz)

The CBT curving and shading should extend that upper limit, but how high?[/i]

Ddwtzry.png

Here's Kay's measurement of his CBT. I've highligted in blue the area below 3khz. You can see that it's fairly well behaved below that frequency. The red box shows the beamwdith of the array. It varies with frequency, but it's approximately 90 degrees. Each curve is 10 degrees. The measurement goes from 0 to 70. (mfk-projects CBT center speaker)

Above 3khz, you can see that the polars aren't as epicly smooth as you get with a waveguide.

This isn't bad performance at all, but it is not as consistent as you can get with a waveguide.

At the same time, the CBT is obviously a lot more compact.

If anyone is wondering why I'm talking about CBTs in a thread about the Tekton Impact, it's because they're fairly similar in concept.

zZFGBOW.gif



mVgtA55.gif



Here's a couple of sims I did, which show how an array of three tweeters work when they're mounted vertically, and spaced 2" apart. (Very similar to the Tekton impact array.) In the first sim, all three are playing 100%. In the second sim, the outer tweeters are attenuated by 3dB.

You can see that the shading makes a SMALL difference. But there's still lots and lots of comb filtering above 3khz. Basically the shading helps to a degree, but the pattern is largely determined by spacing.

Hence, the never ending quest for smaller and smaller elements for line arrays.

Maybe I'm just a wet blanket, but these sims seems to indicate that something like the Snell Expanding Array, or the array that Keele did with Horbach in 2007 (http://www.linkwitzlab.com/Horbach-Keele Presentation Part 2 V4.pdf ) is more promising.
 
diyAudio Moderator
Joined 2008
Paid Member
After reading the review I'm maybe a little surprised to see they work as well as they do. This can be seen in the tweeter range (approx 1kHz and up) on the horizontal polar (fig.4). But I might have a few concerns.

I read the comment: ["The closer I sat to the speakers, the less focused the aural images were;"]. This leads me to believe that either the baffle edge contribution, or the vertical polar response (fig.5) is troublesome.

Then I read this: ["Getting the Impact Monitors to focus their best required placing them closer together than I ever could have imagined would work—the centers of their central tweeters were only 5.6' apart"]...

So I was inclined to question the horizontal smoothness. On the polar there is a flawed region between 3k and 6kHz. Could be Patrick is right in suggesting a lobing issue?

Maybe they should have cut the outer tweeters harder at the top. Was it a compromise between achieving adequately narrow directivity, and smoothness in directivity because the sources were too far apart? The review says: ["the music was slightly blurred and indistinct", which just maybe could be due to the source separation, as some people have subjectively noted about discrete arrays that are imperfectly implemented.

So I decided to consider what you said here:
I looked at the review, and it also seemed as if the slopes were too low. Higher order would give significantly better off axis response.
..I'm inclined not to use an overly steep cross unless I'm specifically avoiding issues.. otherwise a minor mis-match between ways can become more pronounced. But I don't know the specifics of the cross. I wonder what leads you to feel this way?

In the measurements I can see baffle diffraction artifacts between 500Hz and 1kHz.

Below this, the horizontal and vertical kick wide suddenly. The response has been compensated in the upper bass region, but I feel this region is so important for weighting the system in. Maybe this speaker is difficult to fit into a room.