advertisement


The LS50 is dead ...

I wasn't saying that the number of crossover parts was the problem. I was saying that narrow notch filters can be problematic. This is because drive unit parameters have quite wide tolerances and they change a lot with temperature and voice coil position/excursion.

I know that temperature can influence low frequency performance, but It's not something I've really looked into because every speaker is affected by temperature and I've never seen it as a problem. A vented voice coil and possibly a large magnet to draw away the heat seem to help, but this is in my very limited experiments. I always measure the thiele/small parameters with a warmed up, run in driver, and I assume that's what other good speaker designers would do. I don't think temperature affects notch filters because they are generally used in the mid to high frequencies. If they do affect notch filters, I can't say I've noticed subjectively.

One of the most lively and enjoyable speakers I've built had this crossover. Ignore below 300hz in the frequency response graph.

Eminence Deltalite 2510 (10" midbass) and 18Sound NSD1095N / XT1086 Horn

YHAH48al.jpg
 
My experience suggests that complex crossovers are more likely to screw up the phase relationships between different parts of the signal.

It's not something I can ever remember seeing on diy speaker building forums, but interestingly it's something I've seen mentioned on hifi forums in the past. Phase shift at the crossover caused by different acoustic (not electrical) slopes is well known, and to achieve the correct slope for good phase alignment, you may need to use a complicated crossover.
 
It's not something I can ever remember seeing on diy speaker building forums, but interestingly it's something I've seen mentioned on hifi forums in the past. Phase shift at the crossover caused by different acoustic (not electrical) slopes is well known, and to achieve the correct slope for good phase alignment, you may need to use a complicated crossover.
How is that measured? Only, if the music signal is phase-shifted by frequency (in the crossover), then even if the drivers correctly reproduce the phase, relative to each other, you broke it before you fixed it, surely? And you only fixed it so it accurately reproduces what you broke.
 
How is that measured? Only, if the music signal is phase-shifted by frequency (in the crossover), then even if the drivers correctly reproduce the phase, relative to each other, you broke it before you fixed it, surely? And you only fixed it so it accurately reproduces what you broke.

I don't know if it's me being thick (won't be the first time) but I keep reading your post over and over again, and I don't understand the question.
In my defense I have got a splitting headache from a hangover.
 
I don't know if it's me being thick (won't be the first time) but I keep reading your post over and over again, and I don't understand the question.
In my defense I have got a splitting headache from a hangover.
I may have misunderstood your post, but what I took from it was that you could 'fix' the acoustic phase issues in the crossover region by a complex crossover. I presume this means that you can ensure the woofer and tweeter work in phase for any given frequency (to acceptable limits). They'll be phase-coherent. I presume this is measured by way of test tones at various frequencies? So, what you can say is that, for a given frequency, woofer and tweeter both go in and out in phase with each other. What you don't see is whether woofer and tweeter go in at frequency X, but out at frequency Y, say. In which case, any music consisting of X and Y will no longer be reproduced in phase.

So, if the crossover messes with the phase in the signal in the first place, then the woofer and tweeter will deliver a phase-coherent rendering of that signal. The problem is that the signal is not itself phase coherent, any more, because the complex crossover doesn't preserve the phase relationships between low frequencies and high frequencies. So the drivers may accurately reproduce a broken signal.
 
I may have misunderstood your post, but what I took from it was that you could 'fix' the acoustic phase issues in the crossover region by a complex crossover. I presume this means that you can ensure the woofer and tweeter work in phase for any given frequency (to acceptable limits). They'll be phase-coherent. I presume this is measured by way of test tones at various frequencies? So, what you can say is that, for a given frequency, woofer and tweeter both go in and out in phase with each other. What you don't see is whether woofer and tweeter go in at frequency X, but out at frequency Y, say. In which case, any music consisting of X and Y will no longer be reproduced in phase.

So, if the crossover messes with the phase in the signal in the first place, then the woofer and tweeter will deliver a phase-coherent rendering of that signal. The problem is that the signal is not itself phase coherent, any more, because the complex crossover doesn't preserve the phase relationships between low frequencies and high frequencies. So the drivers may accurately reproduce a broken signal.

The thing that I think may be confusing you (because it confuses most people) is the difference between acoustic order and electrical. A 1st order 'electrical' crossover for a tweeter high-pass filter would be a single capacitor - BUT, to achieve an acoustic 1st order slope (The measured (acoustic) frequency response is what really matters), it could take a third order electrical (series cap, shunt inductor, series cap). It all depends on the frequency response and impedance as to how many parts it will take to achieve a certain measured (acoustic) slope.

As for the measurements and phase alignment. To make things simple, imagine the drivers are time-aligned. You take measurements of both woofer and tweeter and import into your speaker design software. You then model a symmetrical acoustic slope (1st, 2nd 3rd order etc) with however many parts are needed, and the software will have a page that shows you how well the phase tracks throughout the crossover, as long as it's within certain limits, your frequency response should be reasonably flat on and off-axis. It would have to be a very bad crossover to have phase problems, and that would manifest itself as a wonky frequency response on and/or off-axis throughout the crossover. This is why it's important to take on and off-axis measurements with the crossover, to make sure your simulations match the actual response. If they don't, then you've done something wrong.
 
Last edited:
I don't think LS50s work particularly well in a big room.

I almost bought a pair and this aspect worried me - well not so much the large room as the performance when playing loud.
In a small 2-way DC driver like that in the 50, the radiating area for bass is very small, even smaller than for a traditional 5" woofer because the central portion of the cone is missing. Your tweeter 'baffle' (the bass driver cone) is also shifting point in space to a far greater degree than with a large DC driver such as a Tannoy.
As that cone is also covering the mids I can see issues.

In the end I went with Elac who produce a similar sized unit but in a 3-way config and squeezing in a dedicated small bass driver. Can play very loud for a small box without obvious character shift or fatigue setting in.

The Meta absorption system sitting behind the tweeter in the new LS50 does look interesting though.
 
Do speaker FR plots take into account phase shifts between drivers, or do you measure with a tone sweep? My experience suggests that complex crossovers are more likely to screw up the phase relationships between different parts of the signal. This probably isn’t audible in the basic frequency response, but tends to manifest as a reduction in realism, or credibility - the recreation of a live-feeling event.

Steve, I think it's also important to have spectral decay information along with the response data and that is often not published.
A lovely flat 'speaker response can hide a multitude of swirling currents beneath the apparently calm waterline. Buried driver resonances which aren't in plain sight on a typical response sweep can be very audible IME.
 
Do speaker FR plots take into account phase shifts between drivers, or do you measure with a tone sweep? My experience suggests that complex crossovers are more likely to screw up the phase relationships between different parts of the signal. This probably isn’t audible in the basic frequency response, but tends to manifest as a reduction in realism, or credibility - the recreation of a live-feeling event.

They do in the sense that relative phase differences between drivers can affect the frequency response. However, the frequency response alone cannot tell you whether a dip or other issue is caused by a phase related issue.
 
I had a pair in a big room but it was in a H/T setup and everything below 100hz went to a big sub. Worked great.
By filtering out deep bass, the HT amplifier removes the main problem, which is large cone movement modulating the tweeter dispersion.
Simply adding a subwoofer in a conventional stereo system does not do this
 
By filtering out deep bass, the HT amplifier removes the main problem, which is large cone movement modulating the tweeter dispersion.
Simply adding a subwoofer in a conventional stereo system does not do this

You can do this if you have the right gear. Some subs have a 'main in / main out' so you can go from your preamp to the sub, then from the sub to the amp. The amp and main speakers will only see the 'above the crossover point' part of the music.

If you want a 2.1 system in a bigger room this is the way to go.
 
Darko didn't seem blown away from sound quality improvements. Indeed, in the comments section he says "slightly better than previous version". Quite the contrast to Guttenberg's hyperbole on the metas. What gives?
 


advertisement


Back
Top