It's not a shame to not understand digital audio. But I'm not writing about opinions there.
Errors of a floating point calculation are far outside the range of the music signal. That's a fact.
A driver has a non linear frequency response (filter) and with that comes a wild phase response. When you want it more linear you need a filter with the opposite frequency and phase response. That's no opinion.
If you found a real error let me know. I don't mind how everyone likes to listen to music, nobody forces people to or off DSP and digital stuff. It's cool to make it without DSP, it's just harder and some things are not possible to do. But at least get the facts right.
https://aespeakers.com/shop/td/td10m/
The TD10M would be awesome in a 3-way with sub. Using the TD15M and couldn't be happier. 🙂
The TD10M would be awesome in a 3-way with sub. Using the TD15M and couldn't be happier. 🙂
I was talking of course of the adc/dac parts, not the amp design itself.
Now that you've relaxed your requirements for 100Hz to 1000Hz (~2x Fc) for the midwoofer, you'll have plenty of choices.
It will also allow you options to look for drivers of different sizes & sensitivity to match with M74A.
I would focus on the cabinet / diffraction model first, and forget about the woofers for the time being.
It will also allow you options to look for drivers of different sizes & sensitivity to match with M74A.
I would focus on the cabinet / diffraction model first, and forget about the woofers for the time being.
Okay, my 2c-
Since you will not need bass from the TMW, and high sensitivity could be reclaimed...
Acoustic Elegance TD10M
Eminence Kappalite KL3010LF
Eminence Kappa Pro-10LF
Ciare (the one Troels liked for his 4ways)
Wavecor (listed above)
PHL (I second that rec)
Faital Pro 10" (plenty)
B&C 10" (plenty)
Beyma (plenty)
18Sound (plenty)
Since you will not need bass from the TMW, and high sensitivity could be reclaimed...
Acoustic Elegance TD10M
Eminence Kappalite KL3010LF
Eminence Kappa Pro-10LF
Ciare (the one Troels liked for his 4ways)
Wavecor (listed above)
PHL (I second that rec)
Faital Pro 10" (plenty)
B&C 10" (plenty)
Beyma (plenty)
18Sound (plenty)
I did a sim a bit ago in the this thread. It was interesting how little power the PHL needed to make SPL's above the FS.
https://www.hypex.nl/products/dsp-family/dsp3-224
Looks like there will be a new DSP next year but it uses the same DSP chip as the current plate amps. It is a 4way dsp so I wonder if there will be 4way plate amps.
https://www.hypex.nl/products/dsp-family/dsp3-224
But the point is that there's not that many 10" woofers that can go up to 1000Hz without breakup. Surely non of the pro woofers can do that. However there is one 15" woofer that could do that, but we all hate Scan-Speak for it's price.
https://audioxpress.com/article/tes...-8582t00-15-ellipticor-oval-voice-coil-woofer
I honestly think that choice is quite small, and It's between SB, SS and Purifi. If someone likes gambling, and have too much money then there's also Kartesian.
https://www.kartesian-acoustic.com/265-vki
What about just replacing the current 12" SB in the existing woofer enclosure with the NRXL in a first step? @IamJF has shown nice measurements that the M74A works pretty good together with 1-2 NRXL in CB. Is a 4th way just for kickbass really needed?
Yea, as I told you! 😆 🙂
Hypex - I had a conversation about their D/A conversion some time ago, they don't think a higher dynamic range is needed. So we should not count to a bigger improvement in this regard soon. The old D/A was AK4454, seems like they stick to it.
I measured 114/116dB(A) dynamic range with digital input from a FA253 (the 100W amp has less noise) - which is actually really good for a complete system! Nevertheless. the amp could do >10dB better.
S/N with the analog input drops 3dB further.
While these are still impressive figures (don't forget, when you go digital in these modules it's the complete system performance) I just now have the topic that I do an installation of my reference speaker and the customer hears some noise from the midrange dome (Satori 2,5"). So there IS a need for even higher performance.
p.s.: When you do the calculation of S/N with analog HiFi Gear and their mismatched levels you often land in the 90dB S/N area. Just using a Hypex module with digital input is still one of the best "signal chains" you can have.
I measured 114/116dB(A) dynamic range with digital input from a FA253 (the 100W amp has less noise) - which is actually really good for a complete system! Nevertheless. the amp could do >10dB better.
S/N with the analog input drops 3dB further.
While these are still impressive figures (don't forget, when you go digital in these modules it's the complete system performance) I just now have the topic that I do an installation of my reference speaker and the customer hears some noise from the midrange dome (Satori 2,5"). So there IS a need for even higher performance.
p.s.: When you do the calculation of S/N with analog HiFi Gear and their mismatched levels you often land in the 90dB S/N area. Just using a Hypex module with digital input is still one of the best "signal chains" you can have.
Testing from decades ago long before the unreliable simulations of today have shown your chosen faceted baffle is the optimal choice when considering performance, aesthetics and reduced construction costs and efforts………spheres aren’t cheap or easy to make! Lol
If you can facet the two lower corners below the woofer, you’ll wind up with less wiggle in the response in the 100-250hz range and a little less direct floor bounce…..but overall I really like the look. If you can angle that baffle a few degrees, that dome mid’s forward presentation will back off a bit and you won’t have to delay it so much in the XO.
While performing your simulations , there’s always the top down alignment to consider where the dome mid is the top driver and the tweeter below and between the mid and woofer. From an imaging standpoint, the critical midrange sees less baffle and therefore presents itself more coherent within the free air surrounding it……while at the same time the high frequency output of the tweeter gets physical smoothing from the wider baffle. The 1st fwd lobe is then steered upwards which improves height imaging perception which is humans recognize at around 8k.
Hi,
your math doesn't seem to add up. 100Hz has 3.4m long wavelength, so physical features smaller than about 1/10th in size, 34cm, have no effect to response at 100Hz. Baffle is much smaller than this, so even at 250Hz there is likely no effect. If facets at low corners have effect, they have it about 1kHz or so, bandwidth of the mid dome. Also, floor bounce is almost toward listening window with typical listening distance of ~10ft and height of ~3ft so baffle edge has about no effect on floor bounce because the baffle is not in front of the driver, way of the sound that specularly reflects from the floor.
Idea to have slants and narrow baffle around tweeter and wider with mid is to smoothen DI and reduce edge diffraction related secondary sound source, which reduces ripple seen in frequency response, to smoothen response. You advice is opposite and would unsmoothen the response at least in measurements. Height perception is from HRTF and 8kHz sounds plausible without checking it out, but surely crossover main lobe ain't that high in frequency? Also, not sure how it would work if it was, why not just boost 8kHz? Boosting main sibilance range seems counter productive though. If you pointed the speaker toward ceiling specular reflection point this might elevate the image, reducing mid and HF energy from floor and moving it to ceiling moving the "weight" of vertical early reflections + direct sound up.
That option I did consider. There are three disadvantages: (1) The spacing between the woofer cabinet and the main speakers is too large to support a 500 Hz crossover. Simulations and measurements show that 250 Hz is about the limit. (2) The cabinet itself has some resonances in the ~ 400 Hz range (if I recall correctly). If I want to run it higher than 200 Hz, I will need to build a new (better) cabinet. (3) I am planning to upgrade my primary woofer system eventually, and until I do that, I don't want waste time and effort on incremental changes.
Understand, thanks for reply!
Regarding the Hypex DAC: I've read somewhere that the current Fusion DSP module bundles two of the AKM AK4454 DAC channels for the tweeter channel to improve SNR.
Regarding the Hypex DAC: I've read somewhere that the current Fusion DSP module bundles two of the AKM AK4454 DAC channels for the tweeter channel to improve SNR.
Not a bass driver really, so maybe not enough Xmax headroom to figth the non L in the low end ? (if 3 ways I mean) ?
As far i understood, the L version is only better above 200/250 Hz and just profit of less sealed cabinet volume.
I bet the non L to have no advantages in the lows vs two SS 26W Discovery. They also will do better from 200 hz till the Bliesma dome vs the non-L.
I mean hifijim has planed a good dynamic behavior also in the lows (110 dB ?)
What can handle the SB34RNXL with a linkwitz transform below 80/70 hz ?
Actually, if I were to upgrade my primary woofer system today, I would probably use two SB34NRXL75-8's per side. It is a great woofer. A stacked pair of 12's would fit in my space better than a 15.
That conversation is a future thread !
That conversation is a future thread !
This is a baffle design that looks to have some promise: The overall cabinet is 673 H x 286 W x 248 D (mm). The internal modal dimension ratios are 1: 1.3 : 3.33. Midrange-Tweeter center to center spacing is 110 mm, and the frequency where this distance is 1.2* wavelength is 3746 Hz.
The tapered bevel begins just above the woofer and increases linearly to the top. The bevel width at the tweeter is 75 mm wide, and at the midrange it is 42 mm wide. The bevel along the top is a constant 75 mm wide.
I experimented with varying the positions of the tweeter and midrange, and this combination was a good overall balance.
Here is 6 pack simulation for the mid and tweeter. For this baffle, a 3.5k 2nd order works well.
When we add in a woofer response, it gets a bit tricky. A 2nd order filter at 500 Hz results in a rising bump in directivity between 400 - 1200 Hz.
I tried a 3rd order filter, and it seems to work better. I am not sure I like what has happened to the vertical polar response from 500 - 2k, but the overall directivity and PIR response looks nice.
In any case, this is a start, and it looks promising. I am going to continue experimenting to see how optimized I can get this. The goal at this stage is to achieve the best DI curve across the two crossover ranges, basically from 300 Hz to 6 kHz. Above 6k the simulation is not going to reflect the true directivity of any tweeter, and below 300 Hz the room will dominate.
j.
The tapered bevel begins just above the woofer and increases linearly to the top. The bevel width at the tweeter is 75 mm wide, and at the midrange it is 42 mm wide. The bevel along the top is a constant 75 mm wide.
I experimented with varying the positions of the tweeter and midrange, and this combination was a good overall balance.
Here is 6 pack simulation for the mid and tweeter. For this baffle, a 3.5k 2nd order works well.
When we add in a woofer response, it gets a bit tricky. A 2nd order filter at 500 Hz results in a rising bump in directivity between 400 - 1200 Hz.
I tried a 3rd order filter, and it seems to work better. I am not sure I like what has happened to the vertical polar response from 500 - 2k, but the overall directivity and PIR response looks nice.
In any case, this is a start, and it looks promising. I am going to continue experimenting to see how optimized I can get this. The goal at this stage is to achieve the best DI curve across the two crossover ranges, basically from 300 Hz to 6 kHz. Above 6k the simulation is not going to reflect the true directivity of any tweeter, and below 300 Hz the room will dominate.
j.
You could also look at a Duelend style crossover that has low initial slopes but steepens up a little.
https://vcllabs.com/tech-page-01/
Or try elliptical by putting a deep PEQ notch above/below the crossover point, if you want to stay with lower initial slopes.
In a 3 way with a mid that can't be pushed lower it can be hard to stop the drivers having too much out of band energy meeting in the middle without using something like an LR4 slope.
PS: The 2K off axis rising in both looks like trouble to me.
https://vcllabs.com/tech-page-01/
Or try elliptical by putting a deep PEQ notch above/below the crossover point, if you want to stay with lower initial slopes.
In a 3 way with a mid that can't be pushed lower it can be hard to stop the drivers having too much out of band energy meeting in the middle without using something like an LR4 slope.
PS: The 2K off axis rising in both looks like trouble to me.
Thanks for that suggestion, some quick hacking shows this approach has some promise.
Regarding the Duelend, are you aware of a resource or documentation which illustrates how to realize this filter topology using standard IIR or analog building blocks? I realize the answer may be buried in the Laplace (?) polynomial shown in the paper, but I have not done that kind of work since 1989 "feedback and control theory" class.