Crazy idea on lowering the frequency of cancellation below tuning?

[lowerFE]

I just had a crazy idea on the possibility of lowering the frequency when the cancellation starts for a vented enclosure's tuning. I need to brush up on the detailed theory of how vented enclosures work, but I won't have time for some time, but I do want to share this and see if you guys think this is even possible. 

 

So the basic principle of vented systems work by delaying the rear wave from the driver by one cycle so it will combine with the active woofer in phase, thereby increasing the output near tuning. Below tuning the driver and the resonator become out of phase, group delay shoots through the roof, and they cancel, causing a 2nd order cancellation roll off starting at the tuning frequency.

 

Now, consider a speaker with 2 active drivers and 2 passive radiators with a divided wall, and we'll call the 2 woofer and PR sets A and B . Let's say the tuning point is 60Hz. Then, I split B's signal into 2 bands, >60Hz and <60Hz for the second woofer. Then I'll add a delay equivalent to 1 cycle at say 40Hz to B's <60Hz and A's signal. Now, the 40Hz output of driver A should be in phase with the 40Hz output from set B. If I had enough taps I can use FIR filters to tailor the phase shift to be exactly inverse of the phase shift caused by drivers playing below tuning to make it even more effective, if this method works. 

 

The numbers are arbitrary BTW, just something less than one octave away from the first tuning frequency.

 

Could something like this counteract the cancellation below tuning for some fractional octave band below the tuning point of a vented system?

[3ll3d00d]

I can't tell which is A and which is B, is A 2 woofers and B 2 PRs or A is a woofer and PR and B is another woofer and PR? 

 

FWIW your description of how a ported system works is wrong. The driver and port are in phase well above Fb (which is irrelevant as the port has so little output at this point), in quadrature (90 degrees out) at Fb, with the port dominating output, and end up with the port is 180 degrees out of phase with the front output of the driver well below Fb. This results in the 4th order (not 2nd) rolloff. Obviously the phase relationship between the two doesn't just abruptly shift but transitions from in phase to out of phase as it moves through the passband.

[lowerFE]

I can't tell which is A and which is B, is A 2 woofers and B 2 PRs or A is a woofer and PR and B is another woofer and PR? 

 

FWIW your description of how a ported system works is wrong. The driver and port are in phase well above Fb (which is irrelevant as the port has so little output at this point), in quadrature (90 degrees out) at Fb, with the port dominating output, and end up with the port is 180 degrees out of phase with the front output of the driver well below Fb. This results in the 4th order (not 2nd) rolloff. Obviously the phase relationship between the two doesn't just abruptly shift but transitions from in phase to out of phase as it moves through the passband.

 

A is one woofer, one PR, B is other woofer, other PR. So think of a subwoofer like

 

WP (set A)

Dividing wall

WP (set B )

 

Your explanation makes sense. I did a quick read up on Wikipedia and the Loudspeaker cookbook (which didn't explain how ported alignments work) and the Wikipedia article didn't explain much. I had a sense of how ported alignments worked, but didn't quite figure out in my mind how it transitions from in phase to out of phase. The smooth transition in phase as it approaches Fb makes sense. I didn't realize at Fb they are 90 degree out. 

 

It's a 4th order roll off, but only 2 of the 4 orders are caused by the out of phase cancellation. The other 2 is just the normal 2nd order driver roll off. 

 

Anyways, I realized I could just test it out myself, and it would take 30 minutes or so. I only did some simplistic testing, and TL;DR, It does seem to work, but most of the gains starts half an octave below tuning. However, this can probably be improved with FIR filters. 

 

Here's a link to an REW file of my results of 25 measurements from 0ms delay to 24ms delay. 

 

https://www.dropbox.com/s/q3sncmb25caadue/Ported%20enclosure%20cancellation%20delay%20experiment.mdat?dl=0

 

I did this with my micro 3 way speaker. It has a Tang Band W4-1720 4'' woofer tuned to 47Hz. I had one speaker play normally while the other speaker having the delay settings. The speakers were placed side by side, and a UMIK-1 was on a stand measuring around 10'' from both speaker. I don't think location matters, as long as the measurement conditions are the same, all we care about is the delta between the measurements. 

 

I placed a LR8 crossover at 47Hz (the tuning frequency), and then delayed the <47Hz band, and then summed the two signals together. However, achieved better results by setting the crossover at 55Hz, probably because of the phase shifts of a LR8 filter. With FIR I won't have this problem, but I don't have enough taps for this experiment. I only tried those 2 frequencies. I'm sure I could find a better crossover frequency, but the results from the 55Hz crossover is enough to make a point. 

 

Here's a picture showing 3 of the measurements. The red is no delay. The purple is with 7ms delay. The green is with 21ms delay. 

 

 

Looking at the 7ms delay, it starts having an output advantage starting around 37Hz, and maintains a 2dB gain from around 20-35Hz, and gradually increasing to a 6dB gain at 15Hz. 

 

Looking at the 21ms delay, the output gain doesn't start until 21Hz, but it quickly increases to a 10dB gain by 15Hz.

 

I'm thinking, with the right phase compensation with FIR, it might be possible to achieve substantial output gain everywhere below tuning to the point it becomes a natural 2nd order roll off.

 

What do you guys think?

[SME]

No, this is not an effective strategy.  Your measurements are likely affected by the locality of the measurement mic.  Try putting it in a far corner of the room instead.  Assuming A and B are identical woofer + PR systems and are precisely co-located, you should see the most output when both are playing signals with the same phase.

 

Even if you could somehow eliminate the 12 dB/octave of roll-off, you still have the problem of the woofer unloading below the tuning frequency, which frequently necessitates an additional 2nd order HPF or even 4th order HPF to protect the drivers.

[lowerFE]

No, this is not an effective strategy.  Your measurements are likely affected by the locality of the measurement mic.  Try putting it in a far corner of the room instead.  Assuming A and B are identical woofer + PR systems and are precisely co-located, you should see the most output when both are playing signals with the same phase.

 

Even if you could somehow eliminate the 12 dB/octave of roll-off, you still have the problem of the woofer unloading below the tuning frequency, which frequently necessitates an additional 2nd order HPF or even 4th order HPF to protect the drivers.

 

They are 2 identical speakers with the same woofer, PR, and driven by the same amp too. 

 

I will try putting them in a corner, but I suspect the lack of headroom will cause a problem in getting a good reading. Each speaker has only a 4'' woofer. The mic was 1 feet away, and the pair struggled to put up just 70dB at 15Hz without the delay. I don't have access to speakers with more capable speakers until I go home around New Years. 

 

What is the "locality of the measurement mic" problem? I've never heard of this. 

[SME]

 

What is the "locality of the measurement mic" problem? I've never heard of this. 

 

I just mean that you have multiple things making sound, and with such a close measurement, the mic is likely significantly closer to one or the other source.  This will not give you a reliable indication of response at larger distances.