Phase correction

[echrisconnor]

I have a pair of speakers I've built which I love the sound from but also love to tweak things. I'm curious about how to alter/correct the phase relationships between the tweeter and woofer. My question revolves around the fact that if I reverse the polarity of the tweeter, I don't hear a huge drop or spike in response. This makes me think that the network is at some odd phase angle. If I want to nudge the phase in one direction or another is there any rule of thumb for how to do so without altering the overall performance of the crossover?

[curt_c]

That’s a tough one without measurement tools.

You probably don’t have so much an odd phase angle but perhaps an odd order crossover. Often the response of ‘real’ drivers will add one to two orders of transfer function to the electrical transfer function of the crossover network.

If you used drivers that you can obtain ‘typical’ response curves from, there is freeware available at the FRC Consortium which will allow you to get a handle on what’s likely happening with your crossover.

-Otherwise, read up on chaos theory… ;^)

C

[unknownuser]

> I have a pair of speakers I've built which I
> love the sound from but also love to tweak
> things. I'm curious about how to
> alter/correct the phase relationships
> between the tweeter and woofer. My question
> revolves around the fact that if I reverse
> the polarity of the tweeter, I don't hear a
> huge drop or spike in response. This makes
> me think that the network is at some odd
> phase angle. If I want to nudge the phase in
> one direction or another is there any rule
> of thumb for how to do so without altering
> the overall performance of the crossover?

Reversing the polarity doesn't necessarily cause huge drop or spike in response. It all depends on the type of crossover. The odd order butterworth crossover, for example, will allways partially out of phase (90 degrees), and reversing will not cause any change in output.

The easiest way to see amplitude/phase relationship between the woofer and the tweeter is thru a measurment software, and measure the actual frequency response. Otherwise, it's just going to be hit or miss deal. Oh yeah, crossover design software helps out a LOT too. BTW, most crossover design software has measurement utility. Just get a good calibrated microphone and preamp.


(Originally posted by: kyle)

[Chris Roemer]

> I have a pair of speakers I've built which I
> love the sound from but also love to tweak
> things. I'm curious about how to
> alter/correct the phase relationships
> between the tweeter and woofer. My question
> revolves around the fact that if I reverse
> the polarity of the tweeter, I don't hear a
> huge drop or spike in response. This makes
> me think that the network is at some odd
> phase angle. If I want to nudge the phase in
> one direction or another is there any rule
> of thumb for how to do so without altering
> the overall performance of the crossover?

In a HP network, (this is being VERY general here - ROT) every series cap and/or shunt coil shifts the phase about 45 degrees to the positive. Conversely, in the LP, each series coil or shunt cap tends to shift the phase a negative 45 degrees.

As was previously mentioned, a great way to get a handle on what's going on here is to get Crossover Simulator from the FRD Consortium. You can plug caps, coils, and resistors into a wide variety of xo topologies and watch the SPL, transfer function, impedance, and phase curves move around. I don't have my own measuring equipment so I'm just using the Freq and Imp .txt files from PE, but you WILL learn a lot. There's also a curve tracing tool (SPL Trace) that lets you generate your own data files from "pictures" of manufacturer's data. VERY cool stuff.

[madrok]

> Reversing the polarity doesn't necessarily
> cause huge drop or spike in response.

And even when it does, some people don't notice it with music. The ear is very forgiving. Try listening to test tones, a sine wave sweep at the xover frequency, or pink noise and adjust the placement of your ear (directly on axis, tilted up and down) and you should be able to hear the difference.

[georgew]

> I have a pair of speakers I've built which I
> love the sound from but also love to tweak
> things. I'm curious about how to
> alter/correct the phase relationships
> between the tweeter and woofer. My question
> revolves around the fact that if I reverse
> the polarity of the tweeter, I don't hear a
> huge drop or spike in response. This makes
> me think that the network is at some odd
> phase angle. If I want to nudge the phase in
> one direction or another is there any rule
> of thumb for how to do so without altering
> the overall performance of the crossover?

Download Carlos Lorenzo Driver offset/crossover spreadsheet.

<A HREF="http://www.diyspeakers.net/projects.php3">http://www.diyspeakers.net/projects.php3</A>

[echrisconnor]

Thanks all. I thought my ears were pretty sensitive and that I'd be able to hear this. I always could with 2nd order xos in the past. This one is a 4th order acoustic Linkwitz-riley.

Interestingly I downloaded the Lorezo Carlos spreadsheet. From whiat this models, it looks like the crossover dip just shifts a little when you reverse polarity. Seems like it may be an indicator of what's happening.

Thanks again!

chris

> Download Carlos Lorenzo Driver
> offset/crossover spreadsheet.

> <A HREF="http://www.diyspeakers.net/projects.php3">http://www.diyspeakers.net/projects.php3</A>

[Dirk]

> Thanks all. I thought my ears were pretty
> sensitive and that I'd be able to hear this.
> I always could with 2nd order xos in the
> past. This one is a 4th order acoustic
> Linkwitz-riley.

> Interestingly I downloaded the Lorezo Carlos
> spreadsheet. From whiat this models, it
> looks like the crossover dip just shifts a
> little when you reverse polarity. Seems like
> it may be an indicator of what's happening.

> Thanks again!

> chris

The other way to alter the RELATIVE phase is to change the physical location of the woofer or tweeter. If the two drivers are surface-mounted, the acoustic center of the woofer is further back into the cabinet than the tweeter. If you were to bring the woofer forward, it would reduce the phase delay....or make it worse by pushing the phase closer to 180 degrees (where you could do a polarity-flip on the tweet). The easiest way to test this out with a typical 2-way "TM" design is to simply tip the speaker back 15-20 degrees or so. This will place the tweeter further away from your ears (relative to the woofer). If you can, give that a try and see if anything sounds different.

I must note that this is also very dependant on your chosen crossover point. Small changes in offset will be more apparent with a 3khz crossover point than a 1.5khz crossover point (because the wavelengths are longer and more distance would be required to substantially alter the phase).

D.

[echrisconnor]

Provided Link: http://home.stx.rr.com/poneal/word_docs/AP130ZO.do (http://home.stx.rr.com/poneal/word_docs/AP130ZO.doc)


The tipping idea is interesting. One thing I do hear is significant change in the sound when I move vertically off-axis. So, when I'm sitting in front of them at ear level they sound one way, and then when I stand up they sound very different. The difference is not a severe when I move laterally off-axis and stay in the same vertical plane. Honestly, I may prefer what I hear when standing since it's open and broader soundstage. Directly on-axis though, the mids extremely detailed.

Here's the link to the speaker design I used. My design is just the same as this only it uses Dynaudio D-28 tweeters instead of the Morels.

> The other way to alter the RELATIVE phase is
> to change the physical location of the
> woofer or tweeter. If the two drivers are
> surface-mounted, the acoustic center of the
> woofer is further back into the cabinet than
> the tweeter. If you were to bring the woofer
> forward, it would reduce the phase
> delay....or make it worse by pushing the
> phase closer to 180 degrees (where you could
> do a polarity-flip on the tweet). The
> easiest way to test this out with a typical
> 2-way "TM" design is to simply tip
> the speaker back 15-20 degrees or so. This
> will place the tweeter further away from
> your ears (relative to the woofer). If you
> can, give that a try and see if anything
> sounds different.

> I must note that this is also very dependant
> on your chosen crossover point. Small
> changes in offset will be more apparent with
> a 3khz crossover point than a 1.5khz
> crossover point (because the wavelengths are
> longer and more distance would be required
> to substantially alter the phase).

> D.

[curt_c]

> The tipping idea is interesting. One thing I
> do hear is significant change in the sound
> when I move vertically off-axis. So, when
> I'm sitting in front of them at ear level
> they sound one way, and then when I stand up
> they sound very different. The difference is
> not a severe when I move laterally off-axis
> and stay in the same vertical plane.

This is a consequence of the driver spacing and the crossover frequency (fc) used. At this fc, and assuming perfect phase tracking between the dirvers at fc, the first null would be approximately 3 degrees vertically from the design axis when the listener was 1 meter away. (Somebody please check my math!) This is because the difference in path lengths between the woofer to the listener, and the tweeter to the listener would be 1/2 wl of fc, or out of phase with respect to each other.

The drivers are aligned along the vertical axis, so the lateral response off-axis will be better, but might be somewhat compromised by the beaming of the woofer at higher frequencies. Both could be improved by lowering the crossover frequency, but I suspect Paul’s intent was to push the crossover above the ‘intelligence’ band. Speaker building is all about choosing the series of compromises that best fit your design objective, and its pretty tough to mitigate them all.

C

[echrisconnor]

Well I'm very happy with the overall design and these things do sound marvelous. I'm just looking for greater understanding and ways to tweak things. I'm not likely to move the drivers back an/or forth, but was looking for simple ways to subtly alter this phase relationship and see if it changed things notably.

c

> This is a consequence of the driver spacing
> and the crossover frequency (fc) used. At
> this fc, and assuming perfect phase tracking
> between the dirvers at fc, the first null
> would be approximately 3 degrees vertically
> from the design axis when the listener was 1
> meter away. (Somebody please check my math!)
> This is because the difference in path
> lengths between the woofer to the listener,
> and the tweeter to the listener would be 1/2
> wl of fc, or out of phase with respect to
> each other.

> The drivers are aligned along the vertical
> axis, so the lateral response off-axis will
> be better, but might be somewhat compromised
> by the beaming of the woofer at higher
> frequencies. Both could be improved by
> lowering the crossover frequency, but I
> suspect Paul’s intent was to push the
> crossover above the ‘intelligence’ band.
> Speaker building is all about choosing the
> series of compromises that best fit your
> design objective, and its pretty tough to
> mitigate them all.

> C

[curt_c]

> Well I'm very happy with the overall design
> and these things do sound marvelous. I'm
> just looking for greater understanding and
> ways to tweak things. I'm not likely to move
> the drivers back an/or forth, but was
> looking for simple ways to subtly alter this
> phase relationship and see if it changed
> things notably.

Don’t get me wrong; I’m not criticizing the design. I’m just suggesting a possible explanation as to why you hear what you do. As Dirk said, tilting the speaker may be advantageous in determining the best overall phase response. Doing more by ear would be difficult. The mind is pretty forgiving of phase issues, and it will only be confused through the crossover region, and any phase anomalies with a 4th order acoustic transfer function would be limited to perhaps an octave either side of fc, . Listen for differences in perceived position of instruments in that band, and adjust the tilt of the speaker to try to minimize them.

C