Purifi + Waveguide Project

I agree the mic holder is critical.
I've made a mount to attach a 1 to 2m long aluminium tube onto a lighting stand
I like the 3D printed parts...

My mic stand is substantially less elegant... I threw it together one morning from scraps I had laying around...


1688163590529.png

- - -
1688163618791.png

- - -
1688163641501.png


j.
 
  • Like
Reactions: 9 users
Once again, I have confirmed that for low frequency crossovers, I just do not like filter slopes higher than 2nd order.

I tried a 4th order crossover instead of the baseline 200 Hz LR2 crossover. My thinking was that I might achieve something positive by more thoroughly isolating the bass frequencies from the midrange, and keeping midrange signals out of the woofer. I tried crossover frequencies from as high as 250 Hz down to a low of 120 Hz. I thought that 160 Hz sounded best, or least bad, but it was not as good as the baseline 2nd order.

I immediately noticed a dullness on bass transients with the alternate 160 Hz LR4 crossover, particularly with electric bass and drums. It was as if the leading edge of the notes were removed. There was also an inferior presentation with nearly all bass register instruments, including double bass, bass clarinet, electric piano, baritone sax, piano, all percussion instruments. It was harder to follow the bass line in all music. Male vocals suffered.

I have noticed this before with other systems I have built. I have a theory that what I am hearing is the effects of group delay.

Shown below is the six pack for the two different DSP filters. This is my "pop music" DSP preset which has a slight downward slope.

1688591265877.png

- - -
1688591292550.png



Now shown is a close up of the phase and group delay plots

1688591332979.png

- - - -
1688591359635.png


Overall, there is very little difference between the the two DSP filters in terms of frequency response. The sound power, DI, horizontal polar response, all of these are quite similar. What is dramatically different is group delay below 500 Hz.


1688591424190.png



So that is my theory. Group delay below 500 Hz is an important factor in the quality of bass presentation. Or to be more precise, group delay below 500 Hz is highly correlated to my perception of bass realism and accuracy. I suspect that I am more sensitive to this phenomenon than most.

I know this is not exactly ground breaking research, and the concept that group delay is important to bass quality is an idea that has been percolating about for decades. But the psychoacoustic studies are inconclusive, and I am not aware of any industry guidance on what level of group delay is acceptable. So for me, this is interesting information.

j.
 
  • Like
Reactions: 4 users
But the psychoacoustic studies are inconclusive, and I am not aware of any industry guidance on what level of group delay is acceptable.
There isn't a definitive conclusion with a specific number but many of the studies converge in a broader sense.

GD Audibility Graph.PNG


I think this one is good from the point of view of a worst case scenario, in that the test is difficult and it was done with headphones.
Keeping GD under 1ms from 300Hz to 1KHz, was the threshold of not being able to hear the difference.

GD-Aud.png


You might also want to look at the in room decay that you get and whether it changes with the different crossovers.
There was a strong correlation to lower decay being preferred at certain frequencies, more so than the evenness of the frequency response in this paper and the later follow up

https://www.avsforum.com/attachments/jaes_v60_5_perception_modal_control-pdf.2273992/

https://core.ac.uk/reader/30732229?utm_source=linkout
 
  • Like
Reactions: 3 users
Have you tried pushing the crossover much higher up, like 400hz? [augerpro]

I would try that if I could... but my bass driver and cabinet don't play nicely when pushed up that high. I have minimal accoustical damping (foam/stuffing) within the bass box... and the box bracing/structural damping was designed around a low frequency pass band. Another complication, the distance between the midrange and the woofer is about 670 mm. Thats fine for 200 Hz, problematic for 400+ Hz.

Simulations with a 400 Hz LR4 show that group delay rises from 0.5 ms at 1k to about 2 ms at 300 Hz, and then stays constant down to 100 Hz where it starts to rise again. So maybe this would be fine? Maybe it is the change in group delay in the crucial 100 - 400 Hz range?
 
Kimmo has suggested looking at excess group delay and keeping "Excess GD <2 ms at 100 Hz"
My experience seems to confirm that guidance. Thanks for that tip!
Piggybacking on Fluid's comments, what do the early reflections look like, broken out into horizontal, floor, and ceiling? Might be something interesting there that isn't captured by the gobal DI and PIR.

See if anything here makes sense to you... The horizontal power and DI are so similar between the two filters, I very much doubt they have any bearing.

1688600296197.png


The vertical Power and DI curves do have some differences. I am not smart enough to have an opinion about these vertical curves
1688600409937.png


1688600442973.png
 
  • Like
Reactions: 1 user
You just have to listen to which xo works best in your room. LR2 gives wider dip for ceiling reflection, isn't that a good thing? I have liked LR2 more in all my dsp-speakers, despite LR4 measured better!

Another thing to audition is bass level below 100Hz.
 
Last edited:
  • Like
Reactions: 1 user
David shows low frequency wavefront detection is in the microseconds (less than 10), the gd research milliseconds and at best a bit below 1 millisecond, that is too large a gap to believe either. But i i would bet, i would bet on David Griesinger, given his research on proximity.
 
  • Like
Reactions: 1 user
Member
Joined 2005
Paid Member
Interesting.

In an MT, I’ve always preferred acoustic LR2, where possible, due to a) less components in a passive crossover b) smoother gradual downslope in power response for active/passive crossover.

For WM crossover I haven’t come to my own conclusion whether I prefer acoustic LR2 or LR4. Although I try to aim for middle C (~252Hz), most of the time the XO point tends to to be dictated by what the mid or woofer is capable of doing.

I wonder if you take a measurement at the listening position using the moving mic method to elucidate if there are other (non anechoic) differences.

Of distortion measurements between the two systems.
 
  • Like
Reactions: 1 users
I also think LR2 on a 3-way speaker bass-mid sounded better than LR4, but I might have had misconfiguration of DSP and so on :) I definitely had problems with bass boxes and think that shallower slope would need very good box with no noises and resonances, while steeper slope could possibly allow sloppier box. I think FIR would help get best sound of a 3-way speaker, especially if its with big bass box. FIR makes group delay non issue while allowing some extra adjustment for any box issues and room interaction stuff. Definitely do not overlook box sound, worst case seems to be quite audible, especially with party SPL.
 
  • Like
Reactions: 1 users
David shows low frequency wavefront detection is in the microseconds (less than 10), the gd research milliseconds and at best a bit below 1 millisecond, that is too large a gap to believe either. But i i would bet, i would bet on David Griesinger, given his research on proximity.
I'm not sure which specific part of Griesinger's research you are referencing, but he does mention the Just Noticeable difference of inter aural time delay as 2 microseconds. Binaural hearing thresholds are not the same as group delay thresholds so I still don't see a contradiction.
 
  • Like
Reactions: 1 user
I'm not sure which specific part of Griesinger's research you are referencing, but he does mention the Just Noticeable difference of inter aural time delay as 2 microseconds. Binaural hearing thresholds are not the same as group delay thresholds so I still don't see a contradiction.
I have been searching for hours for his publication covering the time ( something like 7 usec) it takes to detect a wave start in the low frequencies (half-rectified, the haircell firigs give the frequency pitch) and the sound envelope where phase is important to keep the shape of the envelope. Helas, so i emailed Mr Griesinger, i hope he responds.
I need it for an entirely different reason altogether, but that is way out of scope of this thread.

Of course by the time we realise what we hard, a lot of processing in our nerveus system/brain has taken place.
That is way out of my ability to understand.