Totally Flat MK II

[aliouindie]

Greetings all,

This will be my >$200 entry this year at MWAF. I brought the partially complete speakers to InDIYana and they were well received. Thanks for all the positive feedback. Someone asked me if these were the "Totally Flat" speakers that I had built last year. They look similar, but are completely different; new box, new drivers, new crossover.

Based on feedback, I will be revising the crossover to reduce the woofer level by about 1dB. It may have been the room, but the woofer seemed just a little too overpowering, masking the detail just a tad in the mid and high frequency areas.

Driver selection:

Tweeter: Peerless DA25TX00-08 (8 ohms) (see pic). This is the poor man's high Onlinesbi sudoku aadhar card end tweeter, with sound quality comparable to some of the very best tweeters available, regardless of price.

Midrange: Tang Band W3-1797S 3" aluminum sandwich full range (4 ohms) (see pic). This driver has the midrange clarity and detail of a good planar type driver, such as the B & G Neo-8 or Neo-10. TB also makes a 4" version of this driver, but I decided to go with the 3" because it has 20% less cone mass. Power handling will be a little less than the 4" version, but it should be able to hit my goal of roughly 95dB in the listening chair with both speakers playing.

Woofer: Tang Band W6-1139SIF (4 ohms) (see pic): I wanted to keep baffle and cabinet size to a minimum, yet have a system that could dig deep with authority. This subwoofer can go right down to 25Hz in a very small box (3/4 cubic foot or so). The trade off is efficiency, which should be OK as long as I use an amplifier with sufficient power that can handle the load.

Impedance: The mid and sub are both 4 ohms, which causes the overall system Z to dip to about 3.5 ohms at 600Hz. Phase angle, however, is zero at this point. The amplifier output stage will get very warm but this should not be a major problem.

Efficiency: Probably somewhere in the 83dB/1W/1M area, which is quite low. Again, a big power amp that can handle a 4 ohm load will be necessary.

Progress: I am currently working on re-painting and re-finishing the box. I'm also replacing the particle board xover box with a much improved hardwood unit. And I just put in my parts order for a couple crossover revisions.

Next up: A few historical construction posts. I started building these about 4 months ago and am now approaching the finish line. In the next few weeks, I'll finalize & post the completed xover, cabinet, and FR measurements. Comments and questions are always welcome.

This should help to spread out internal resonances compared to a rectangular shape.

[raiderone]

Don't know how I missed this build as I'm a big fan of the TB 3" flat honeycomb...

The low pass on the W3 looks really smooth, if I may ask what does the C5 .22uF cap do to shape the response and is the L5 .51mH necessary? Also, 1.7kHz crossover to the tweeter seems a bit low, how did you arrive at that crossover point? Thanks.

[Wolf]

That tweeter Bill used in another build with xover at 1kHz, and according to measurements it doesn't complain. My thought would have been to run the mid in about the same range, just shifted higher. Even so, that was purely my initial speculation. Bill did a nice job with these!
Wolf

[Wolf]

To answer your query, the 0.22uF provides a tank notch with the low pass filter on the mid. It can both steepen the roll-off and suppress nasties. The 0.51mH coil makes the low pass a 3rd order electrical, and likely matches the high pass on the tweeter since it is also a 3rd electrical. Being that these drivers have wider bandwidth than they are being used, the acoustic roll-off does not likely come into play and requires the 3rd orders to get better summation.
Later,
Wolf

[raiderone]

Hey Wolf, thanks for the information, especially the comment "Bill did a nice job with these." I wish I could have heard them too. I've always thought to use the W3 as a mid/tweet; the off axis is really good for a 3". This build has given me some motivation to try to run a pair as a mid. Not too many builds using these, though there was/is a pro build (Boenicke) using a pair that seemed to be well received.

[4thtry]

Hey Wolf, thanks for the information, especially the comment "Bill did a nice job with these." I wish I could have heard them too. I've always thought to use the W3 as a mid/tweet; the off axis is really good for a 3". This build has given me some motivation to try to run a pair as a mid. Not too many builds using these, though there was/is a pro build (Boenicke) using a pair that seemed to be well received.

Thanks for the positive feedback on my speakers. I agree with everything that Wolf said and would add the following:

C5 & R6 form a tank filter which helps to pull down the high frequency peaking of the TB mid-range. This driver starts to peak up significantly above 7kHz and this filter pulls down the response by an additional 15dB. I was debating whether or not to delete this filter, because the response was already down by about 40dB in this area. But I decided to leave it in because it helped to provide a better match to my LR 24dB/octave target slope. R6 has no effect on the response. It's purpose is to protect the power amplifier's output stage from going into oscillation.

L5 forms the third element of the 18dB/octave mid-range low pass electrical filter which is, as Wolf pointed out, necessary to match the tweeter's 18dB/octave high pass electrical filter (C1, L1, C6). These two electrical filters, combined with the driver roll offs, produce almost perfect LR 24dB/octave high and low pass acoustical slopes with excellent phase tracking and a sharp reverse null. Combining these two drivers using standard 12dB/octave electrical high and low pass filters results in a 90 degree phase difference between the two drivers. All of my attempts to develop a crossover using 12dB/octave parts failed and I was forced to go to a higher order.

1.7kHz does, indeed, seem a bit low for this driver combo. My initial plan was to cross somewhere in the 2-3kHz range. But when I tried to cross higher, I could not get anything to work out using either 12 or 18dB/octave electrical parts. I was able to achieve a fairly flat on-axis 3kHz crossover model using 12dB/octave electrical parts, but the phase tracking was all messed up (see attached 3kHz XSim model). In addition, the lower xover point helps to improve the vertical polar response pattern.

Hope this helps to provide further clarification,

Bill

Side note: The XSim program does not provide target filters, as I described above. However, you can generate target filter curves using a dlr's WinFilters program and then import the target curves into your XSim model using the "get file" drop down menu.

[raiderone]

1.7kHz does, indeed, seem a bit low for this driver combo. My initial plan was to cross somewhere in the 2-3kHz range. But when I tried to cross higher, I could not get anything to work out using either 12 or 18dB/octave electrical parts. I was able to achieve a fairly flat on-axis 3kHz crossover model using 12dB/octave electrical parts, but the phase tracking was all messed up (see attached 3kHz XSim model). In addition, the lower xover point helps to improve the vertical polar response pattern.

I suppose it does not matter since your speaker design sounds good, but what does "messed up" phase tracking do to the sound?

[4thtry]

I suppose it does not matter since your speaker design sounds good, but what does "messed up" phase tracking do to the sound?

I guess the only reason to keep the phase in alignment is so that the crossover does what you expect it to do. I do not really know if there is a sonic penalty from poor phase tracking. What I need to do is to intentionally "mess up" one of my next crossovers and then use kenrhodes new xover switching devise to quickly switch back and forth between two crossovers to see if I can hear a difference. If the phase tracking differences alter the horizontal or vertical polar response patterns of the two speakers, then it should be audible.

By messed up, I mean that the phase is "non-parallel" or "criss-crossing" through the xover region. My understanding, from reading various threads that talk about this topic, is that you want the phase curves of the two drivers to either be right on top of each other or parallel to each other as you go through the xover region. For a LR 24dB/octave target, the phase curves would be on top of each other, producing a sharp reverse null at the turnover frequency. A target 18dB/octave butterworth would have phase curves 90 degrees apart, but parallel, through the xover region. There would be no sharp reverse null.

Bill

[raiderone]

Bill, I'd really like to try your crossover/speaker. I've had the TB W3 for years and in many ways a special driver, but with a metallic zing/bite/grain at the top end. Even so, I never thought to use a tweeter or take it down to 200Hz(!). Your speaker seems to address this shortcoming, but to me in an unexpected way and it is hard to argue against your approach to let the driver determine the crossover. Right now, I think I have most of the parts to try your crossover as a template, albeit with different tweeter and woofer. Thanks for posting all the information about your design.