Phase Coherent Speaker Project?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I was given a pair of Spica TC-50 speakers. I am quite pleased with their sound. I have built a number of speakers (Jim Griffin Jordan/Aurum Cantus Monitors, Metronome with Fostex 127s, ART Arrays), but none have the rock solid imaging of the Spicas.

After reading about them they seem to have a couple of unique qualities (at least in literature and interviews with the designer).
  • Phase Coherent
  • 1st order crossover
  • tight tolerances of components

Has anyone come up with the DIY equivalent of the Spica's? Not necessarily the shape of the enclosure (triangle), but something that has the same design principles and results?
 
The term should be phase perfect or transient perfect here, as they use the 1st order filters, the only filters which are transient perfect. 4th order Linkwitz Riley crossovers, for example, are phase coherent, but they exhibit 360 degree phase shift.

I've not tried any 1st order designs, but I've come accross a few drivers that I've thought could be applied in such a design. Generally a woofer with extended upper range and smooth roll-off and a tweeter loaded with a waveguide could work.
 
Probably they exploit the natural "phase travel" that even a single driver exhibits over it's operating range? If the range of the low frequency driver ends at 200 degrees, probably they made the high range driver at the same shift at the crossover point and then it is controlling the phase?

Is there an easy way of "guessing/predicting" the phase at certain frequency by just using a ruler to measure cone depth and a pocket calculator to calculate the delay?
 
Probably they exploit the natural "phase travel" that even a single driver exhibits over it's operating range? If the range of the low frequency driver ends at 200 degrees, probably they made the high range driver at the same shift at the crossover point and then it is controlling the phase?

Is there an easy way of "guessing/predicting" the phase at certain frequency by just using a ruler to measure cone depth and a pocket calculator to calculate the delay?

200 degrees?

Usually you can relate the geometry with a group delay curve.

That is, the difference in depth to the voice coils of the two units will give an air path delay that corresponds with the two plateaus that you will see in the group delay curve. Its hard to generalize about the phase curve in the transition region, since their are a number of options available when you design the system. With most pairings you can get a couple of combinations of rolloff rates and absolute polarities that work passably well. Each would have a different phase transition unit-to-unit.

Outside of the crossover region phase is strictly determined by the drivers, their Hilbert transforms and their relative depth.

David S.
 
Phase of each band pass is determined by the combined phase of driver and crossover. Both affect phase across the entire band pass. The way Spica designed the speakers was that the woofer response was tailored to a 4th order Bessel low pass response The Bessel response has constant group delay (linear phase) over much of its band pass. The 1st order HP on the tweeter was used because it has very little additional delay, but still, combined with the tweeter's natural response the HP is actually a 3rd order acoustic response. Anyway, Spica then offset the tweeter and staggered the crossover points so that the propagation delay at high frequency was the same delay associated with the Bessel LP section on the woofer. That is why the baffle is so sloped on the Spica. The result is something approximating a transient perfect design but the crossover still has irregularities in the crossover region. Additionally, since the tweeter offset is so great the TP behavior degrades rapidly off axis.
 
Has anyone come up with the DIY equivalent of the Spica's? Not necessarily the shape of the enclosure (triangle), but something that has the same design principles and results?

Specifically on the topic I have made a humble attempt to do something similar without even having heard about the Spica.

It's a suspended OB with adjustable tilt.

Here it is: http://www.diyaudio.com/forums/full-range/131590-mjk-s-jordan-jx92s-ob-goldwood-gw-1858-woofer-h-frame-55.html#post2365991
And here again: http://www.diyaudio.com/forums/full-range/131590-mjk-s-jordan-jx92s-ob-goldwood-gw-1858-woofer-h-frame-51.html#post2197163

No one mentioned or asked about the tilt of the OB... and somehow I didn't mention it...

The two supporting planks are pivoted and allow the OB to be tilted back up to about 20-30 degrees angle.
Depending on the listening position I move the whole OB back and forth and tilt it to different degree or listen to it upright...

And there definitely is difference, until I get it right the scene and imaging are partial. In my case almost always the vocals are centered and the guitars are well positioned. But my biggest problem is cymbals coming in front of my face, get it wrong and cymbals overtake the scene...
The guitars and voices are less of a problem because I use the midrange as an extended range from 200-250 Hz to 7-10 kHz and I rely on their natural roll off and impedance peak after 8-10 kHz without a bandpass crossover - only first order single cap, Zobel, L-pad, high frequency extension circuit, baffle step compenstation and... and that's about it :eek:

And actually adding a high frequency extension messed quite allot the imaging... :eek: it's RC in parallel to each other and in series with the speaker - the R does not modify the phase, but the C does. (it's not shown on the schematic in the referred topic - I experimented recently by bypassing the first R of the L-pad with a C...)

In conclusion: tiltable OB or front baffle can show very good results for obvious reasons - altering the distance of each driver to the listener and thus the relative phase of the sound from each source.

Best Regards!
 
Last edited:
I am at CES, and found that some of the best sounding speakers were phase coherent. (Vandersteen and Dyanudio, to be specific.)

As JGH noted in 'why hifi experts disagree'*, I think that audio is highly subjective, and what one person finds pleasing another person may not. But phase coherency seems to be something that I'm in tune to.

So....

As the original poster asked, 'what are some good DIY phase coherent speakers?' IIRC, the ACI Jaguars were phase coherent. Is there anything else? LR4 seems to rule the DIY world, but maybe 1st order crossovers aren't so bad (at least for some people.)

Also, if there are used speakers that you'd recommend, I'm all ears. I noticed on CL and Audiogon that used Vandersteens can be had for less than $500 each.

* Why Hi-Fi Experts Disagree | Stereophile.com
 
I am at CES, and found that some of the best sounding speakers were phase coherent.

Hi Patrick... What goes on in Vegas, Stays in Vegas!

I enjoy honest audio humor:

"The stereo illusion is just a Jedi mind trick."
"The audio industry should require that all recordings be performed in Symphony halls with just two microphones.
***Dipoles = "They are Here" soundstage.
***Corner controlled directivity monple speakers = "You are There "-- 1st Row Center SoundStage.
***Optimized in-room monopoles = "You are There" -- back in the rear SoundStage.


Thiel Inc. is technically based upon 1st order acoustic crossovers, supported by physically time aligned speakers on slanted baffles. "Back in the day" a 1st order crossover was a single capacitor or inductors, a philosopyh of simplicity=perfection. A REAL 1st order acoustic Xover has dozens of components(fig 1&2).

The Coaxial speakers from Theil and KEF appear to be a "white flag" that old-school phase coherence between tweeter+mid is difficult/expensive.
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
Phase coherence is perhaps the biggest advantage of full range drivers and why so many people in the Full-range forum won't go back to multi driver speakers once they have experienced a full range. There are no cross overs, no tweeters, maybe some folks use a sub. For the most part, we spend time on cabinets that bring out the bass extension. You cannot beat the sound stage, or spatial imaging of a full range and its ability to reproduce vocals as there are no cross overs to deal with in the critical band of the human voice. It has problems with rock or complex symphonic pieces, but for many, the ability to imagine you are right next to the performer in the same room is all worth it. And you cannot beat it for simplicity: direct drive from amp to single driver - just two wires and that's it. I had been listening to multiway speakers for most of my life and when I discovered the full range speaker, it was an epiphany - being able to hear many of my CD's in a way I had never before imagined that they could sound that good. Give it a try. :)
 
Last edited:
Specifically on the topic I have made a humble attempt to do something similar without even having heard about the Spica.

It's a suspended OB with adjustable tilt.

Here it is: http://www.diyaudio.com/forums/full-range/131590-mjk-s-jordan-jx92s-ob-goldwood-gw-1858-woofer-h-frame-55.html#post2365991
And here again: http://www.diyaudio.com/forums/full-range/131590-mjk-s-jordan-jx92s-ob-goldwood-gw-1858-woofer-h-frame-51.html#post2197163

No one mentioned or asked about the tilt of the OB... and somehow I didn't mention it...

The two supporting planks are pivoted and allow the OB to be tilted back up to about 20-30 degrees angle.
Depending on the listening position I move the whole OB back and forth and tilt it to different degree or listen to it upright...

And there definitely is difference, until I get it right the scene and imaging are partial. In my case almost always the vocals are centered and the guitars are well positioned. But my biggest problem is cymbals coming in front of my face, get it wrong and cymbals overtake the scene...
The guitars and voices are less of a problem because I use the midrange as an extended range from 200-250 Hz to 7-10 kHz and I rely on their natural roll off and impedance peak after 8-10 kHz without a bandpass crossover - only first order single cap, Zobel, L-pad, high frequency extension circuit, baffle step compenstation and... and that's about it :eek:

And actually adding a high frequency extension messed quite allot the imaging... :eek: it's RC in parallel to each other and in series with the speaker - the R does not modify the phase, but the C does. (it's not shown on the schematic in the referred topic - I experimented recently by bypassing the first R of the L-pad with a C...)

In conclusion: tiltable OB or front baffle can show very good results for obvious reasons - altering the distance of each driver to the listener and thus the relative phase of the sound from each source.

Best Regards!
 

Attachments

  • 1stOrder_Challenge.JPG
    1stOrder_Challenge.JPG
    92.3 KB · Views: 242
  • 2Way1stOrder.jpg
    2Way1stOrder.jpg
    6.7 KB · Views: 238
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.