DIY digital crossover project

[boxerfan88]

I think I just got more confused

Maybe I'll shall my confusion.

Let say

Driver 1 & 2 are a woofers
Signal is a 50hz, do I think 1 duty cycle is 20ms

So there are 2 signal gen/amps driving these 2 woofers.

If I delay the signal on amp02 by 10ms or half duty cycle, then the 2 signals are out if phase by 180°.
So the 2 signals will kind of cancel each other, macam wiring L/R spkr out of phase.

But if amp02 is delayed by 1 full duty cycle, then both are in phase. But the 'pulses' are not the same, right? One is 20ms slower.

So is there any audible effect on the combined output?

I did a simulation in Excel ...

• signal runs for 4 complete cycles
• woof1 no phase shift (no delay) - blue line
• woof2 180 degree phase shift (delayed by the 180 degree time period) - green line

The result will look like this (red line):




We'll get half a wave of output from woof1.
Then woof2 comes in and begins to cancel.
During cancellation, the combo is theoretically zero.
At the end of the signal, woof1 is silent, but woof2 got the remaining half wave to play.
So the result is that very weird red waveform...

Now for the complicated part - real music is not sine wave - and we got 2 ears - how the woof1 output & woof2 output interact and combine at the left ear vs. right ear plus how our brain interpret the combined signal is quite complex to "model the behavior"

 

[boxerfan88]

Level matching the two DACs for higher SPL.

• Soekris volume is now -6
• SMSL volume is now 86

 

[boxerfan88]

After undergoing boom boom volume at bro Kenny's place, decided to push my setup up another 3dB.
Now can almost touch 100dB SPL at max. digital volume at MLP.

But, I dare not push it any further as it is really really loud ... and I worry about pissing off neighbors.



Level matching the two DACs for even higher SPL.

• Soekris volume is now -3
• SMSL volume is now 93

RT measurement at MLP over 10 minutes playback at max. digital volume:




I do still have headroom coz the Neumann speaker gain setting is at the lowest (94dB), still have 3 more steps (100dB, 108dB, 114dB).

 

[boxerfan88]

Ooh … the 8ch DAC is on the way!





Now to figure out where to fabricate the TRS-XLR cable.

 

[synthesis]

Now to figure out where to fabricate the TRS-XLR cable.

Can go to LHS at Sim Lim Tower 2nd fl

 

[boxerfan88]

Nice toy to replace the physical measuring tape when doing speaker positioning...

 

[boxerfan88]

Boom boom ... tonight hit 98dB peak

Track: "Fading Sun"

https://tidal.com/browse/album/3343902

 

[boxerfan88]

Installed the Topping 8-channel DAC into the system.




Got some TRS-XLR cables made at LHS, decided to splurge on shielded power cords for the speakers.

I also decided to remove the Analog crossover, and simplify the cabling.

Since I was doing lots of rewiring, I decided to reconfigure the Yamaha AVR from the A38-Usher into the Neumann system.

 

[boxerfan88]

One important observation, when connecting the Yamaha AVR RCA output into the Neumann XLR input, the SPL was really a lot lower.

This is because the RCA operating voltage (1V-2V) is much lower than XLR operating voltage (2V-4V).

I had to re-do the Yamaha YPAO calibration so that the correct levels can be set inside the AVR.

After that, I decided to just re-do the PEQ.

All good now, just need to get used to watching shows at AVR volume of -10 instead of -25.

 

[boxerfan88]

Next, I turned my attention to the stereo setup.

First order of business is to level match the DACs.

It was a super fast job! The Topping DAC volume control is dBFS based, just like the Soekris.

All DACs level matched as follows:

• Soekris volume -3
• Topping volume -3
• SMSL volume 93

• red = Soekris+SMSL sweep as reference
• green = Topping first sweep at -6
• blue = Topping second sweep at -3 (matches red)

 

[boxerfan88]

Next step was to redo the PEQ.

I did a "no EQ" sweep of Topping vs. (Soekris+SMSL) from 10Hz-600Hz to compare the bass curves.



Bullseye! 99% match on both L-channel and R-channel, except for the variation around 90Hz which is immaterial to my PEQ algorithm.

So nice to just re-use my existing EQ. No further EQ tweaking needed. Sweet!!

 

[boxerfan88]

Finally, tidying up foobar configuration:

• disconnected & removed analog crossover setup.
• created Topping configuration and re-use the existing DSP/EQ configuration.

Revised system architecture:

 

[boxerfan88]

Clock Drift Compare (between the old SKR+SMSL and the new Topping).

The original SKR+SMSL combo, where the clock drift showing up as phase shift in the bass region.



The new Topping is rock solid, no differences between the 5 sweeps that are 30secs apart.

 

[boxerfan88]

Re-measured the sweeps for reference.

Frequency response: other than the messy 100Hz-200Hz area, pretty flat and even at MLP.




Impulse: nice and tight.



Distortion: below 0.5% from 300Hz, even the bass region is pretty good.

 

[boxerfan88]

After 10 days using the Topping DM7 8-ch DAC ... compared to my R2R DAC ... I find it a tiny little bit brighter (in a good way). It does help create the perception of "more details". My observation seem to align with some of the comments by YT reviewers of ESS9038 based DACs.

Measurement wise, the Topping DM7 measures much better SINAD than my R2R DAC.
My R2R DAC has very low noise floor but has quite a bit more harmonics (kind of typical of R2R DACs).
I wonder if that extra harmonics make the R2R DAC sound more "analog" ?
(tube also add on those extra harmonics)

I wonder if the cleaner SINAD has anything to do with the clarity ...


(ps: I am using the dac1421 measurement as a proxy for my dac1521 ... both designed around the same time just that mine has 4 R2R strings instead of 2)

 

[boxerfan88]

Topping DM7 sounds really good now.

Wrapping up my tweaks for the year.

Final "noSub" PEQ:




And the final Digital Crossover (with Sub) PEQ:

 

[boxerfan88]

Some good subwoofer integration advice/tidbits that I saw elsewhere. My experience with subwoofer integration journey and method aligned very much with his advice. Cross-posting here to share share.

Keep your driver acoustic centers about 1/4 wavelength away or closer at the acoustic crossover frequency.
So @80hrz the wave is about 168" so keep drivers within about 4 feet
So @ 100hrz the wave is 135" so keep the drivers within about 3feet
@ 150hrz the wave is about 90" so keep within about 2feet.

You can fudge this a little and expand the distance as there is so much bass bouncing off boundaries so soon a typical room that it is not even close to an exact science at these large wavelengths. Plus in stereo you already have 2 or more woofers much further apart than the above yet often playing the same frequency at the same time in each L and R channel(but not mixed in stereo).

Also remember electrical slopes are not what matters at all, acoustical slopes and acoustical summing is the key so measure and experiment with slightly different values for high and low pass. You can also try 48db slopes on the sub if you cross really high. I often use 48 for the lowpass and 24 for the high pass when I go really high , like 150-200hrz.

Also I suggest the moving mic method averaged over a period of say 30seconds to a minute for this kind of measurement. Not single sweeps.
Also in the bass for adjustments below say 150hrz ish, use 1/12 or 1/24 octave with high Q filters. Above that 1/6 or even 1/3 is totally fine.
After granular adjustment look at the pyschoacoustic or 1/2 smoothing and see if you lost to much bass tone, if you did shelf the bass up a bit(or down if it high)

Changing the crossover is not the only variable. Each time you adjust you may have to tweak in relationship to other variables. Also there are room modes, SBIR, Subwoofer(s) location, location of mains to walls etc.

I also apply PEQ/room adjustments above about 150hrz to each L & R channel independently. Below that I generally apply globally.

Howdy, I don't agree. I don't find this is the same as matching something like a tweeter and midrange. Blending the sub is not delicate, it is almost a bit of brute force crude endeavor.

The wavelengths are enormous and nearly 100% omni(unless crossing high to say make a diy 3way) and blend together very easily. They do not need to matched like the ideal on paper especially as they absolutely to do not behave that way in rooms.

Then the many, many different room effects are massive and much room correction is used below 150hrz relative to other regions.
I've never had a problem getting superb blending with a little time involved to try a few different things. Just measure with every adjustment and take some time.

 

[boxerfan88]

Some interesting research findings...

Let's say we want to subjectively enhance sound (adjust it to our liking), by the means of various parametric EQ filters.

• Bass : we can typically enhance bass via a low-shelf filter, Fc 100 Hz, Q 0.65
• Male vocal warmth : peak filter, Fc 180 to 200 Hz, Q 2.0
• Female vocal warmth : peak filter, Fc 400 Hz, Q 2.0
• Male vocal clarity : peak filter, Fc 3000 Hz, Q 2.0
• Female vocal clarity : peak filter, Fc 4000 Hz, Q 2.0

Source for warmth and clarity : https://larryjordan.com/articles/eq-warm-a-voice-and-improve-diction/

If we add the previous criteria, we get :

• Bass (Low shelf, about 100 Hz)
• Male vocal warmth (Peak, Fc 180 to 200 Hz, Q 2.0)
• Female vocal warmth (Peak, Fc 400 Hz, Q 2.0)
• Treble (Peak, 1200 Hz)
• Midrange accuracy/Shoutiness (Peak, 1250 Hz)
• Male vocal clarity (Peak, Fc 3000 Hz, Q 2.0)
• Female vocal clarity (Peak, Fc 4000 Hz, Q 2.0)
• Detail/Sibiliance (Peak, 5000 Hz)
• Grain/Clarity (Peak, 7500 Hz)
• Airiness (High shelf, 10000 Hz)

 

[boxerfan88]

Is today’s weather & electricity ultra clean?

Donno why tonight the music is ultra clear and nice.

 

[boxerfan88]

Minor tweaks this weekend.

-> some changes to subwoofer plate amp settings

• subwoofer filter changed from 50/24 to AVR/12 to allow full slope control by digital crossover
  • the above change caused a lot more subwoofer bass energy to be pushed into the room
  • volume reduced by 5 clicks from mid-point to reduce bass power from overloading room

Updated EQ36b.



^ my ears prefer the -1dB bass trim (200Hz low shelf)

Resultant response:





.