In need of a builder...

[BKr0n]

Starting to take shape.

 

[ThatSoundsGood]

Interesting thread here. I have done some builds with Purifi 6.5" drivers and one build with the Bliesma T34B. If you're looking for low distortion, it's hard to beat either of them. But I saw you mention a 1st order crossover for the Bliesma. That's possible but if you're going to use the Hypex Plate amp I highly recommend using FIR filters in the crossover. It's a game changeover. You can get really steep with your slopes and keep the tweeter out of any distortion range. I'm happy to assist with the crossover whenever you get there.

 

[BKr0n]

But I saw you mention a 1st order crossover for the Bliesma. That's possible but if you're going to use the Hypex Plate amp I highly recommend using FIR filters in the crossover. It's a game changeover. You can get really steep with your slopes and keep the tweeter out of any distortion range.

Thanks for the heads up Not sure if a 1st order will be the magic bullet, but in theory it makes sense. If the transition point is more shallow than a higher order filter, it should be a smoother transition with no beaming.

I'm happy to assist with the crossover whenever you get there.

I'll definitely take you up on that! Believe it or not, this is actually my first speaker build lol. While I would like to think I have learned quite a bit, I definitely don't know everything. In some cases, I don't even know the question to ask. Now I need to read up on FIR filters

 

[ThatSoundsGood]

Thanks for the heads up Not sure if a 1st order will be the magic bullet, but in theory it makes sense. If the transition point is more shallow than a higher order filter, it should be a smoother transition with no beaming.

The problem with a 1st order filter on a tweeter is that the tweeter will still have a lot of energy below the crossover point and therefore there is a lot of driver overlap with certain frequencies and the tweeter will still have considerable energy below the crossover point that can easily lead to distortion (even with the amazing Bliesma T34B). So I recommend much steeper filters (8th order and/or FIR), especially with an active crossover. The beaming is more problematic with woofers, but since you can also cross them with steep filters, you can keep them within a frequency range before they start beaming. My best sounding build utilizes a Hypex FA253 with FIR filters.

I'll definitely take you up on that! Believe it or not, this is actually my first speaker build lol. While I would like to think I have learned quite a bit, I definitely don't know everything. In some cases, I don't even know the question to ask. Now I need to read up on FIR filters

This is an ambitious build for a first time design. I dig it. FIR filters or not, I think high order filters are the key in an active system.

 

[BKr0n]

The problem with a 1st order filter on a tweeter is that the tweeter will still have a lot of energy below the crossover point and therefore there is a lot of driver overlap with certain frequencies and the tweeter will still have considerable energy below the crossover point that can easily lead to distortion (even with the amazing Bliesma T34B). So I recommend much steeper filters (8th order and/or FIR), especially with an active crossover. The beaming is more problematic with woofers, but since you can also cross them with steep filters, you can keep them within a frequency range before they start beaming. My best sounding build utilizes a Hypex FA253 with FIR filters.

Sure that makes sense. Didn't think of that. I was actually thinking about what @terryforsythe recommended and doing a 3rd or even 4th order.

This is an ambitious build for a first time design. I dig it.

Thanks! Life's too short not to ya know?

FIR filters or not, I think high order filters are the key in an active system.

While I agree high orders do have their benefits (especially the fact that usually in a digital implementation they are far more stable), I still think it shouldn't be a "all or nothing" concept. Remember: a speaker, at the end of the day, no matter how much DSP or filtering occurs, it is still an analog transducer that creates something in the analog domain. I would imagine high order filters don't sound very organic due to the brick wall nature of the curve. Tbf, that may be preference, but I would say its a "case by case basis" sort of thing.

 

[kemmler3D]

I would imagine high order filters don't sound very organic due to the brick wall nature of the curve. Tbf, that may be preference, but I would say its a "case by case basis" sort of thing.

With the Hypex amps I think you owe it to yourself to experiment with steep FIR or other steep filters... after all, what's necessarily "organic" about sound coming from two places at once? And at the same time, you can't really avoid beaming with a 1st order filter, it's all about when the mid's passband stops. If a steep filter doesn't sound good you can always try something else.

 

[terryforsythe]

Sure that makes sense. Didn't think of that. I was actually thinking about what @terryforsythe recommended and doing a 3rd or even 4th order.

One of the significant benefits of DSP is that you can easily experiment with different settings and find what sounds best to you. I modeled my speakers to arrive at initial settings, then experimented a lot with different crossover slopes and frequencies and settled on what provided good in-room measurements and sounded best to me.

Here is how I went about it:

1. With speakers assembled, I took a frequency response/THD measurements of each of my drivers at 0 deg., 30 deg. and 60 deg. (Adding 15 deg. and 45 deg. would have been even better, but I was measuring in my backyard and had to hurry - I had a small time window between when the golf course lawnmowers finished and my next door neighbor's noisy pool pump turns on every day).

2. I also took an impedance measurement of each driver with the speakers assembled.

3. I imported all of that data into VituixCAD and chose initial crossover values based on reviewing the data, as well as driver compression data I got from Erin's review of the speaker (Erin's Audio Corner). Then I experimented with different crossover frequencies until I achieved what I considered to be an optimum response in the model. Knowing I was going to E.Q., for my midrange/tweeter crossover frequency I put greatest emphasis on the directivity index calculated by VituixCAD. The driving factor for my woofer/midrange crossover frequency was the THD and compression data. In the model both 4th order and 8th order crossovers worked well.

4. I setup the DSP according to the crossover settings I determined in VituixCAD with 4th order crossovers, and started listening. I continued tweaking those settings until I found what sounded best to me. I would change a setting, then listen for a while, then repeat. This went on over a few months. I ended up with 8th order crossover slopes. My final midrange/tweeter crossover frequency ended back at the same frequency that I determined in VituixCAD. My Final woofer/midrange crossover frequency ended up being in the same range I started with, maybe 50Hz lower if I remember correctly.

 

[voodooless]

Just remember that steep filters are not always the best solution to lowering distortion or protecting drivers, especially with tweeters. Because of the steeper slope, it also makes them play lower at a higher volume than a less steep slope (given the same crossover frequency). If you're already close to the lowest possible crossover point, this may actually give you more distortion, not less. So always measure what you do to make sure it's still okay. Consequently, if you think that you can lower your crossover point with a steeper filter, you are probably wrong. see:

Crossover Slopes and Protection for Speakers: Does a Steeper Filter Provide More Protection? - Audiofrog

Someone asked recently about choosing a crossover slope and asked, specifically, if using a 48dB/octave crossover would provide better protection than a 24dB/octave slope. I usually recommend 24dB/octave Linkwitz Riley […]

www.audiofrog.com

 

[ThatSoundsGood]

Just remember that steep filters are not always the best solution to lowering distortion or protecting drivers, especially with tweeters. Because of the steeper slope, it also makes them play lower at a higher volume than a less steep slope (given the same crossover frequency). If you're already close to the lowest possible crossover point, this may actually give you more distortion, not less. So always measure what you do to make sure it's still okay. Consequently, if you think that you can lower your crossover point with a steeper filter, you are probably wrong. see:

View attachment 394719

Crossover Slopes and Protection for Speakers: Does a Steeper Filter Provide More Protection? - Audiofrog

Someone asked recently about choosing a crossover slope and asked, specifically, if using a 48dB/octave crossover would provide better protection than a 24dB/octave slope. I usually recommend 24dB/octave Linkwitz Riley […]

www.audiofrog.com

Yes, that is true and I could have been more specific here. It's important to do measurements and keep the drivers within their distortion and excursion limits. I guess I could more accurately say that using steep filters can help accomplish this in some situations but it depends on the initial design, where we are trying to get the drivers to play and how close the crossover is to their limits.

 

[ThatSoundsGood]

Sure that makes sense. Didn't think of that. I was actually thinking about what @terryforsythe recommended and doing a 3rd or even 4th order.

Thanks! Life's too short not to ya know?

While I agree high orders do have their benefits (especially the fact that usually in a digital implementation they are far more stable), I still think it shouldn't be a "all or nothing" concept. Remember: a speaker, at the end of the day, no matter how much DSP or filtering occurs, it is still an analog transducer that creates something in the analog domain. I would imagine high order filters don't sound very organic due to the brick wall nature of the curve. Tbf, that may be preference, but I would say its a "case by case basis" sort of thing.

It's totally a case by case basis and depends heavily on the initial design. I do think that high order filters and FIR filters sound more accurate to me. I don't really know what "organic" means in this case, but "accurate" is a term I can work with. But as someone else pointed out here, a higher order filter isn't always better at keeping drivers within their distortion limits due to the knee of the filter driving more power into the passband area. So, as with everything in speaker design, it is certainly a case by case situation. The crossover ends up being what the speaker needs.

 

[terryforsythe]

Just remember that steep filters are not always the best solution to lowering distortion or protecting drivers, especially with tweeters. Because of the steeper slope, it also makes them play lower at a higher volume than a less steep slope (given the same crossover frequency). If you're already close to the lowest possible crossover point, this may actually give you more distortion, not less. So always measure what you do to make sure it's still okay. Consequently, if you think that you can lower your crossover point with a steeper filter, you are probably wrong. see:

Those are good points to consider. But, what also should be considered are out of band resonances, etc. Steeper crossover slopes can help to better attenuate those. They also may have an impact on the bandwidth of the vertical off-axis dip between the midrange and tweeter when separate drivers are used, but not by much.

In the end, every driver/cabinet combination is different. My suggestion is to make the final determinations based on testing and listening to the finished speaker. Test - listen - adjust - repeat.

In my case I started testing with 4th order crossovers. 8th order for the woofer/midrange resulted in a tad bit less harmonic distortion in the crossover region as measured in the frequency response sweeps. For the midrange/tweeter, it resulted in what I perceived to be sharper images in the sound stage, but the difference was not significant, and may even have been my brain playing tricks on me and telling me what I expected to hear (placebo effect). After much tuning, testing and listening, I settled on what I perceived to sound the best for my setup. That, to me, is what is most important.

 

[Salt]

Is the decision for transducers and enclosure finalised?
If yes, then we can talk about x-over with measurements taken.

 

[BKr0n]

what's necessarily "organic" about sound coming from two places at once?

This is what we call a design tradeoff

In my case I started testing with 4th order crossovers. 8th order for the woofer/midrange resulted in a tad bit less harmonic distortion in the crossover region as measured in the frequency response sweeps.

I think my only concern with really high order crossover is what happens with any other signal that hits a hard stop: an unwanted resonance.

Is the decision for transducers and enclosure finalised?

I believe so Box will be 12x21.5x19.5" with the MTM in their own enclosure (still working on that one in CAD) that is ~8x4x15.5" (need to double check volume since the back half is rounded). Enclosure will be made of HDF and the MTM Enclosure will be 3d printed (still thinking about materials). I figure since it's not a huge enclosure I can deal with the weight. Enclosure will be lined with resonix products (mega CLD and guardian) adhered to the chassis and each other with decidamp. The rest of the void space filled with polyfil. Still haven't decided on finish. Probably something either latex or enamel based. I think that covers everything...

Edit: oh yeah transducers

https://ptt.purifi-audio.com/web/content/product.template/1726/x_studio_data_sheet?download=1

http://www.bliesma.de/Datasheet%20T34B-4.pdf

https://sbacoustics.com/wp-content/uploads/2021/09/10in-SW26DBAC76-4.pdf

 

[terryforsythe]

I think my only concern with really high order crossover is what happens with any other signal that hits a hard stop: an unwanted resonance.

No.

Every complex audio signal is made up of different frequencies (if interested, read about the Fourier transform, which converts signals between time domain and the frequency domain). The active crossover will send the different frequencies to the appropriate drivers based on your crossover frequencies and slopes.

Besides, using DSP, it is easy to change the slopes. As I stated, tune, measure, and listen. After making a change, and ensuring the measurement looks good, listen to determine whether you think it sounds better or worse. Steeper crossover slopes may or may not sound better with your speakers. As I previously noted, I have built MTM configurations using 3rd order passive crossovers and it worked well. I have read that 4th order also works well, but have not tried that. I have not read about higher order slopes being used, but with DSP it certainly is much more practical to implement as compared to using passive crossovers, and easy to test.

At this stage of your project, I wouldn't worry about the crossover slopes. The time to start concerning yourself with crossover slopes is when you start modeling in VituixCAD using actual measurements of your drivers in the enclosure, with damping material, etc. It is from that data that you will want to select your initial crossover frequencies and slopes to start your model. Then, make adjustments in the model based on the predictions being generated by the model.

Once your model is optimized, use that configuration as your initial starting point in the actual crossover. Then measure it. If it does not measure well, then immediately make adjustments. Once the measurements look good, start listening. Live with it for a week or so. Things may start to stand out. If you think you hear something off, look at your data, and make some adjustments. Then re-measure, listen, and repeat until you are fully satisfied with the sound.

 

[BKr0n]

Every complex audio signal is made up of different frequencies (if interested, read about the Fourier transform, which converts signals between time domain and the frequency domain). The active crossover will send the different frequencies to the appropriate drivers based on your crossover frequencies and slopes.

What I meant was impulse response. Typically higher order filters in the analog domain have a chance to have a spike in them. A great example of this is the elliptical/cauer filter. They have terrible stability and lots of ringing.

 

[terryforsythe]

What I meant was impulse response. Typically higher order filters in the analog domain have a chance to have a spike in them. A great example of this is the elliptical/cauer filter. They have terrible stability and lots of ringing.

I tend to use Linkwitz-Riley for audio. Butterworth sometimes works well. A 4th order Linkwitz-Riley is a cascade of two 2nd order Butterworth filters. An 8th order Linkwitz-Riley is a cascade of two 4th order Butterworth filters.

I would not use an elliptic filter for HiFi. It has ripples in the pass band. See the attached images to see the difference between a Butterworth and an elliptic. The Butterworth is nice and smooth. The elliptic has a spike at the end of the pass band.

EDIT: The elliptic filter has a highly nonlinear phase response near its cut-off frequency, resulting in extra delay there which can be perceived as "ringing'" at that frequency. This is not the case for Linkwitz-Riley nor Butterworth filters.

 

[terryforsythe]

Here is a comparison of Linkwitz-Riley vs Butterworth. Butterworth sums to add +3dB at the crossover frequency. Linkwitz-Riley sums to zero (flat).

Of course, the actual results may be a bit different with actual drivers when using passive filters due to their complex impedance. In the digital domain, though, there is no such issue, though phase alignment between the drivers can be an issue. Use time delay to correct for that. All-pass filters also can be used to correct for phase issues, but they are more complicated to implement.

 

[BKr0n]

So here's how I think I'm going to do the MTM enclosure. Because of the wave guide on the front, it's going to have to be mounted from the inside. The front piece and the back piece will be bonded together and sealed via those pegs. Once together, it can be mounted with standard hardware from the back. I'm fairly certain this will need to be 3d printed. Doing this with wood would be more than laborious. With that said, I also had to move the speakers closer together in order to accommodate the guides. Real estate on that baffle is going to be tight.

 

[kemmler3D]

Closer is better! And 3D printing is great, something that size is probably doable one one plate, (maybe in 2 pieces), if you had your own printer it would cost you $10 and take a day at most to print.

 

[BKr0n]

something that size is probably doable one one plate, (maybe in 2 pieces)

Yup that's the idea. That way I can line the enclosure with whatever (more than likely some CLD and some fluff... no room for foam) and have a place to mount the drivers to as well.

if you had your own printer it would cost you $10 and take a day at most to print.

I do not! lol. Additionally, depending on what material I want to use it will not be cheap PEEK or PC may be my best bet. I did find a place near me that has a markforge though so that opens up some options. There really aren't a whole lot of printers out there that can do continuous fiber.