goudey,
I am sorry, I forgot to include the link where I copied these graphs from. And, you are right that is electrical phase (not acoustical phase). Can you give an idea of how acoustical phase will look like for that.
These graphs are for M3Ti - I am using these as my main speakers. I picked these up a couple of years back after listening to several popular budget speakers (Paradigm, NHT, PSB, etc.), many of them at home (had to buy some of them). At that time I had no idea of speaker design (at all). Got into DIY a couple of months back and opened these up - nothing - no stuffing, no bracing (though the box is wedge shaped), almost no crossover (just a resistor/cap on the tweeter), and cheap MCM-like drivers. I used the graphs for this speaker since I can easily correlate measurements with real-life. Getting back to original question:
So what would the phase plot (electrical/acoustic) look like for this design? Also, is there a commercial design with similar goals, that I can audition?
I am sorry, I forgot to include the link where I copied these graphs from. And, you are right that is electrical phase (not acoustical phase). Can you give an idea of how acoustical phase will look like for that.
These graphs are for M3Ti - I am using these as my main speakers. I picked these up a couple of years back after listening to several popular budget speakers (Paradigm, NHT, PSB, etc.), many of them at home (had to buy some of them). At that time I had no idea of speaker design (at all). Got into DIY a couple of months back and opened these up - nothing - no stuffing, no bracing (though the box is wedge shaped), almost no crossover (just a resistor/cap on the tweeter), and cheap MCM-like drivers. I used the graphs for this speaker since I can easily correlate measurements with real-life. Getting back to original question:
goudey said:
So what would the phase plot (electrical/acoustic) look like for this design? Also, is there a commercial design with similar goals, that I can audition?
to jag:
> So what would the phase plot (electrical/acoustic) look like for this design?
That would take some work (model fitting), and I can't spare the time. Since there are exceptions to phase distortion from crossovers (first order crossovers, phase corrrected crossovers) and other assumptions would need to be made, I am not sure how useful it would be for any particular system design.
My guess about the phase response is that phase distortion in the crossover region is quite large and significant because typical tweeters need significant attenuation at 500hz so they won't be destroyed.
The comment about the low frequency phase was easy to make, and from that I inferred that it was probably the phase of the impedance (that plus the shape of the phase at high frequencies; it looks like the impedance through the crossover region was a design consideration as it should be). I do perform model fitting for individual drivers (and driver/enclosure combinations), since this is necessary to understand what is required for crossover/equalization design. For this I usually make the measurements myself, so phase is available from the measurements. I sometimes use amplitude information from other sources to get a head start, and in these cases phase is often not available.
> Also, is there a commercial design with similar goals, that I can audition?
Most commercial designs do not strive for low phase distortion from 80hz-4khz. Such designs will not show well in specifications, so they are not so easily marketed. Marketability is far more important than long term listening enjoyment when it comes to market success.
One commercial design that comes close was the Martin Logan CLS electrostatic system. I have a pair (no longer functional) and I am striving to create a reasonable replacement that meets the phase distortion requirement. Once I experienced the good attributes of a nearly full range electrostatic I find it hard to accept anything less. Others have similar experiences with nearly full range drivers (hence the interest in this thread).
Some manufacturers claim to have designs that use first order crossovers. This can be done, though I don't believe everything I read. Other design considerations are also important to me, such as a clean "waterfall plot" within the high frequency and ultrasonic regions. Altogether the combination of requirements is somewhat hard to achieve (an interesting design challenge).
> So what would the phase plot (electrical/acoustic) look like for this design?
That would take some work (model fitting), and I can't spare the time. Since there are exceptions to phase distortion from crossovers (first order crossovers, phase corrrected crossovers) and other assumptions would need to be made, I am not sure how useful it would be for any particular system design.
My guess about the phase response is that phase distortion in the crossover region is quite large and significant because typical tweeters need significant attenuation at 500hz so they won't be destroyed.
The comment about the low frequency phase was easy to make, and from that I inferred that it was probably the phase of the impedance (that plus the shape of the phase at high frequencies; it looks like the impedance through the crossover region was a design consideration as it should be). I do perform model fitting for individual drivers (and driver/enclosure combinations), since this is necessary to understand what is required for crossover/equalization design. For this I usually make the measurements myself, so phase is available from the measurements. I sometimes use amplitude information from other sources to get a head start, and in these cases phase is often not available.
> Also, is there a commercial design with similar goals, that I can audition?
Most commercial designs do not strive for low phase distortion from 80hz-4khz. Such designs will not show well in specifications, so they are not so easily marketed. Marketability is far more important than long term listening enjoyment when it comes to market success.
One commercial design that comes close was the Martin Logan CLS electrostatic system. I have a pair (no longer functional) and I am striving to create a reasonable replacement that meets the phase distortion requirement. Once I experienced the good attributes of a nearly full range electrostatic I find it hard to accept anything less. Others have similar experiences with nearly full range drivers (hence the interest in this thread).
Some manufacturers claim to have designs that use first order crossovers. This can be done, though I don't believe everything I read. Other design considerations are also important to me, such as a clean "waterfall plot" within the high frequency and ultrasonic regions. Altogether the combination of requirements is somewhat hard to achieve (an interesting design challenge).
goudey said:
Oh no, I am not asking for an accurate plot. All I want to know (approximately) is where I am right now, and where I should be (for phase Nirvana)?
Put another way, what does a typical 2-way design phase plot approximately looks like? It would be great if you can give me that approximation for M3Ti (so that I can easily correlate - may not be possible from the information provided). And, how would that look like in best case (using wide range driver), within practical limits. I do not want you to bother about specific drivers and designs. OR, just let me know what you have seen in your experience: typical 2-way and best case. Thanks!
generic examples
to jag:
Well, it worked (previous post), just didn't show in the preview.
In the previous post I show a generic example using a 4th order L-R crossover at 1.4 khz. The plot shows amplitude response in dB for low pass, high pass, and coherent sum. Using the same scale the phase is also plotted. This phase plot tells us absolutely nothing about audibility, not even a hint.
The plot also shows 'phase curvature'. This is a more useful metric that indicates something about the relative group delay distortion. My experiments tell me that I PERSONALLY do not like this distortion to be significant between 1kha and 4 khz.
In the plot I show in this post the example is for a 4th order L-R crossover at 6khz. This will sound much better to me. Many people, I am sure, when presented with both, would not hear a difference (different people hear differently).
Most commercial designs will, for me, have significant phase distortion.
Hope this helps.
to jag:
Well, it worked (previous post), just didn't show in the preview.
In the previous post I show a generic example using a 4th order L-R crossover at 1.4 khz. The plot shows amplitude response in dB for low pass, high pass, and coherent sum. Using the same scale the phase is also plotted. This phase plot tells us absolutely nothing about audibility, not even a hint.
The plot also shows 'phase curvature'. This is a more useful metric that indicates something about the relative group delay distortion. My experiments tell me that I PERSONALLY do not like this distortion to be significant between 1kha and 4 khz.
In the plot I show in this post the example is for a 4th order L-R crossover at 6khz. This will sound much better to me. Many people, I am sure, when presented with both, would not hear a difference (different people hear differently).
Most commercial designs will, for me, have significant phase distortion.
Hope this helps.
Attachments
another generic example
While I still remember how I did it, here is another example.
This is for a 4th order L-R crossover at 350hz. The region where the "phase curvature" metric is significant (to me) extends up into the 1-4khz region.
Note: Its best not to fixate on a single performance metric. Within our budget limits we need to make the "smart compromise". The phase distortion becomes an important consideration only after everything else is "almost perfect". In such cases reduced phase distortion can make a considerable difference, but not otherwise.
While I still remember how I did it, here is another example.
This is for a 4th order L-R crossover at 350hz. The region where the "phase curvature" metric is significant (to me) extends up into the 1-4khz region.
Note: Its best not to fixate on a single performance metric. Within our budget limits we need to make the "smart compromise". The phase distortion becomes an important consideration only after everything else is "almost perfect". In such cases reduced phase distortion can make a considerable difference, but not otherwise.
Attachments
Re: generic phase distortion examples
A gif looks a lot better than jpeg for graphs (or any other line art).
I guess I know what is going on now: We want to avoid crossover phase distortion by taking crossover out of the critical range. And we want to avoid woofer low frequency rolloff phase distortion by choosing a wide range driver that goes significantly lower than critical range (even better to put a crossover on lower end of the wide range driver outside the critical range to avoid more serious phase distortions introduced by natural rolloff). Finally we want to carefully select a wide range driver so that the driver itself does not introduce phase distorion in the critical range. And, as mentioned before, a driver with good SPL response should have a good phase response as well.
I guess, I got some reading to do - but now I know what to look for. Thanks.
Very well said!
goudey said:
A gif looks a lot better than jpeg for graphs (or any other line art).
I guess I know what is going on now: We want to avoid crossover phase distortion by taking crossover out of the critical range. And we want to avoid woofer low frequency rolloff phase distortion by choosing a wide range driver that goes significantly lower than critical range (even better to put a crossover on lower end of the wide range driver outside the critical range to avoid more serious phase distortions introduced by natural rolloff). Finally we want to carefully select a wide range driver so that the driver itself does not introduce phase distorion in the critical range. And, as mentioned before, a driver with good SPL response should have a good phase response as well.
I guess, I got some reading to do - but now I know what to look for. Thanks.
goudey said:
Very well said!
Another driver contender
This ATD driver is from Italy so I 'll bet it is available in a lot of places. Seems to fit the band in discussion, 93dB isn't super high, but much better than average!
http://www.e-speakers.com/products/atd-wide-range.html
This ATD driver is from Italy so I 'll bet it is available in a lot of places. Seems to fit the band in discussion, 93dB isn't super high, but much better than average!
http://www.e-speakers.com/products/atd-wide-range.html
Re: Another driver contender
It has a nice Q for putting in a TL as well. You might have trouble reaching 80 Hz thou.
dave
Variac said:
It has a nice Q for putting in a TL as well. You might have trouble reaching 80 Hz thou.
dave
Good question, lawriebuck. I'm coming rather late to this thread as I'm a relatively new member. Are you still with us, Surf, Sun & Sound and has your speaker been built yet?
I've used the Bandor 50mm units extensively and think that they are superb. I used 4 of them in the original 7th Veil "System IV" and will use 4 again in my new speaker system (to be unveiled at the Frankfurt High End Show, end of May). In the new system they will be crossing over to a bass unit at 100Hz.
I don't use any other tweeters as the Bandors go all the way and I think that any gains by putting in a tweeter would be offset by the probelems caused by the extra crossover and by having different 'flavours' of units as they go up the spectrum. It's a judgement call and I wouldn't criticise either way. However, if you're using the Bandor 50mms, I wouldn't crossover at anything less than 7kHz and probably higher.
One thing about using the Bandors for high frequencies, and particularly if you're using more than one in a linear array ...
Using 4 units works perfectly (if they're mounted above each other without gaps) for the sweet spot. Horizontal dispersion is also no problem and they image beautifully. But, stand up and the top end disappears. With 1 or 2 units this wouldn't be a problem. Of course if you're using a tweeter, again no problem.
I hope that this is helpful and not too late.
Good luck
Steve
I've used the Bandor 50mm units extensively and think that they are superb. I used 4 of them in the original 7th Veil "System IV" and will use 4 again in my new speaker system (to be unveiled at the Frankfurt High End Show, end of May). In the new system they will be crossing over to a bass unit at 100Hz.
I don't use any other tweeters as the Bandors go all the way and I think that any gains by putting in a tweeter would be offset by the probelems caused by the extra crossover and by having different 'flavours' of units as they go up the spectrum. It's a judgement call and I wouldn't criticise either way. However, if you're using the Bandor 50mms, I wouldn't crossover at anything less than 7kHz and probably higher.
One thing about using the Bandors for high frequencies, and particularly if you're using more than one in a linear array ...
Using 4 units works perfectly (if they're mounted above each other without gaps) for the sweet spot. Horizontal dispersion is also no problem and they image beautifully. But, stand up and the top end disappears. With 1 or 2 units this wouldn't be a problem. Of course if you're using a tweeter, again no problem.
I hope that this is helpful and not too late.
Good luck
Steve
Bandor drivers are closely related to Jordan's. This isn't suprising since both share a common heritage. Bandor's owner, Doreen Bance is the ex-wife of Ted Jordan and was awarded equal access to their joint intellectual property rights in their divorce settlement. While Ted Jordan's strength is design, Doreen Bance's strength is production. As of this writing, Bandor's drivers are more available than Jordan's.
Although, like Jordan, Bandor makes drivers using curvilinear profile aluminum foil cones in a variety of sizes, ..
http://www.snippets.org/ldsg/sect-5.php3
Quite remarkable..
Although, like Jordan, Bandor makes drivers using curvilinear profile aluminum foil cones in a variety of sizes, ..
http://www.snippets.org/ldsg/sect-5.php3
Quite remarkable..
i thought that Ted had resurected himslef when a few years ago i saw another company in Germany sporting his name. I wonder if ALR-Jordan is still active.
now what would one suggest to top of this driver (JX92)?
seems to me a system using fullrange drivers like the JX92, FE103, TB W4-xxx and W3-xxx, etc. and a woofer below 100Hz seems a good design.
my wife wants me to attempt the WAF of B&W's VM-1 and FPM 5.
seems to me a system using fullrange drivers like the JX92, FE103, TB W4-xxx and W3-xxx, etc. and a woofer below 100Hz seems a good design.
my wife wants me to attempt the WAF of B&W's VM-1 and FPM 5.
I've been considering the possibilities of something like that too. Wouldn't it be easier, or more prudent, to cross higher than 100Hz. Maybe even 900-1000?
I would have thought that the JX92 is well capable down to 100Hz.Timn8ter said:
I would take it full-range if you're using separate subs, or down to between 150Hz and 450Hz with a bass driver in the same cabinet. Crossing over at 450Hz would allow the system to exhibit a little more slam than at 100Hz (depending on the bass drivers and configuration used).
The main point of using a 'full-range' driver like the Jordan is to avoid a crossover having a negative effect on the mid-band.
Similarly, if you're using a separate tweeter, it might be better to cross over at 5kHz or above.
Just some thoughts.
Low crossover points are easier to manage with electronic crossovers (duh) but I get concerned trying to make passive crossovers that low. Getting it out of the mid-range is a good point. I'm still trying to dial-in my BV2 and am now using a slightly overdamped 1st order series XO at ~4000.
7V said:
my interest is in making the system as discrete as possible. so i thought 2 4" wide range drivers (jordan, fostex, TB, etc) in a small 10 liter box with a good tweeter could cover from 100Hz on up. since the system will be wall mounted (like B&W's FPS and VM1) there might be little need for BSC.
the system would then be supported by atleast 2 subs (possibly even 1 more for the rear).
now the tough part....the system should be capable of audophile level sound qualtiy. While I was impressed with the WAF of the B&W speakers mentioned earlier I was not impressed with their sound qualtiy. dynamic range was very restriced. the spekaers had more shrillness than "air". to be honest i do prefer a more laid back sound that most and while this characteristic might nt make for impressive HT/AV effects lisetner fatigue in audio mode is limited.
with the advent of plasma and LCD projetion screens i expect every major manufacturer will offer a speaker or 2 that is similar to the VM1 or FPS.
Sounds good to me. Presumably you've done the sums on the 10 litre box for two drivers.navin said:
Of course, using two drivers makes a tweeter essential because of the beaming and cancellation at high frequencies (for the benefit of any newbies thinking of using two full-range drivers).