Thanks for the reply - that is kind of the point. I don't know. I have used - even made, plenty of x-overs in the past. I assume they were all passive. Are all LR x-overs active? Some of them look like pre-amps or equalisers - that's not what I am looking for. Something - tuneable OK - but fitted inside the cab. I like to get things set up so I don't need to fiddle with tone controls.
The terms "L-R" and "Butterworth" are used to describe the mathematical formula behind the crossover filter. A Butterworth filter can be thought of as sharper, and a L-R filter as shallower.
A passive crossover can be either type, but is it is usually a customized blend which works with the specific drivers in hand. An active crossover is more flexible, and can be adjusted to work with many different drivers. I imagine there are some pro-sound crossovers which allow you to select either Butterworth or L-R.
So give us some more information about what you have, and where you want to go...?
A passive crossover can be either type, but is it is usually a customized blend which works with the specific drivers in hand. An active crossover is more flexible, and can be adjusted to work with many different drivers. I imagine there are some pro-sound crossovers which allow you to select either Butterworth or L-R.
So give us some more information about what you have, and where you want to go...?
The drivers I have are Tannoy dual concentric speakers - so a bass/mid cone with tweeter in the centre of the cone. The LF and HF are separately connected via a x-over, just like separate drivers. The Decca horns fire into the corner - not forward. They were designed in 1949, based on Paul Voight's originals, to provide a domestic version of Paul's large horn loaded speakers, which were mostly used in cinemas and theatres. I am considering the best way of crossing over. I was interested by some reading which suggested the LR option dispensed with a 3dB peak at the x-over point. I think I understand the principles of passive (eg Butterworth) x-overs. But LR x-overs are new to me. I was hoping someone could explain in simple terms the differences and potential advantages or pitfalls.
Lets assume you have a perfect woofer and tweeter (they measure flat). A 2nd order BW crossover will cut each driver by 3dB at the crossover point, and then continue to cut at 12 dB per octave. A 2nd order Linkwitz-Riley (LR) crossover will cut each driver by 6 dB at the crossover point, and then 12 dB per octave beyond that. So which is better?
Years ago (meaning like 40+ years ago), nearly everyone used 2nd order BW filters. It was thought that the woofer and tweeter would sum together at the crossover frequency as uncorrelated signals... so since each was reduced by 3 dB, the result would be a flat response. I think they thought that the total power response in the room was what mattered most, and after all the reflections and diffractions, the two signals would be uncorrelated.
Then came Linkwitz and his buddy Riley, and they taught us all that the first arrival sounds mattered much more than the reflected sounds, so the woofer and the tweeter were correlated signals at the crossover frequency.... in order for them to sum to a flat response, they each needed to be cut down by 6 dB.
Does this make sense?
When I designed the original crossovers for my 3 way speaker system back in 2006 I used Butterworth 2nd order (12dB/octave) transfer function and they summed about 3dB high in the crossover region which never sounded right, and in near-field response measurements I could see the response anomaly. Decades ago they often offset crossover frequencies slightly to compensate for this, but I didn't understand nor did I have all the kit required for the measurements at the time. I replaced them about a year later with a new design based on LR2 which worked much better and remained in use until 2 years ago when I went fully active with line level LR4 crossovers that I designed specifically for this system. (I also have a DSP processor running Dirac)
I would probably go with an LR2 if passive and LR4 if active.
I'm far from an expert and perhaps others will have more useful information to share or perhaps even a design suitable for these drivers in these enclosures.
I would probably go with an LR2 if passive and LR4 if active.
I'm far from an expert and perhaps others will have more useful information to share or perhaps even a design suitable for these drivers in these enclosures.
Thank you gentlemen! I am beginning to get the picture. So - the x-overs in Tannoy's own cabs, with the 3 position switch would the passive LR2 type, and the usual, inaccessible, non adjustable ones - probably BW?
Jim's point about room reflections raises an interesting issue. The Decca horns are designed to reflect the sound. How do you think that should influence my deliberations?
Jim's point about room reflections raises an interesting issue. The Decca horns are designed to reflect the sound. How do you think that should influence my deliberations?
I'm not certain you're getting this.
Linkwitz-Riley, Butterworth, Bessel, etc are all just the shape of the filter. Whether that filter happens in the digital domain, analogue line-level or speaker-level is irrelevant.
I could have a 3-way system that's bi-amped and use DSP to implement a LR crossover for low-to-mid, and then the mid-to-high crossover could be a passive Butterworth design.
It's important to recognise the limited scope of these crossover terms. To give you an idea, the acoustic slope (ie, the shape of the rolloff when that crossover is connected to that driver) is very rarely the same as the electrical slope.
The only time acoustic slope = electrical slope is when you've got a driver that has a flat frequency response for a few octaves either side of the crossover frequency. Doesn't happen particularly often, though I can think of a couple of examples involving wide/full-range drivers.
This is further confounded by the effects of a non-linear driver impedance. Passive (speaker-level) crossovers work by having a rising impedance towards the crossover frequency. The potential divider thus formed will put more voltage across the crossover components and less across the driver.
Result = driver gets quieter past the crossover frequency.
If the driver's impedance also rises at/near the crossover frequency, it'll start to "win" against the potential divider, and the rolloff will be slower.
Tweeters operated near resonance, then, should have an LCR network to flatten the impedance hump at Fs, and LF drivers should have a Zobel network fitted to counteract the rising impedance due to voicecoil inductance.
The outcome of all this is that if I apply a textbook Linkwitz-Riley passive crossover to a real speaker, the shape of the frequency response will almost certainly NOT be Linkwitz-Riley-shaped.
Even if we add a Zobel filter and an LCR network to the passive crossover, the results still might not be as expected: we have things like BSC to deal with, plus the non-linearities of the drivers themselves.
Moving the crossover to before the power amp(s) would mean we don't need to worry about impedance variations (assuming a voltage source - valve amps usually have non-zero output impedance and we're back to the potential divider), but it doesn't fix everything - there are still the non-linearities of the drivers themselves to contend with.
In my experience, the only way make a crossover work properly (as in, integrate two drivers with a flat frequency and smooth phase response) is by using a measurement mic (you can get set up for less than £100) and some simulation software. I like XSim, but others are available.
Chris
What Chris says ^^^ above. I was going to suggest that you need to be able to do basic acoustical measurements to get this right, but he beat me to it.
I recommend something like the MiniDSP UMIK-1 which will provide you with a calibrated USB microphone here: Acoustic Measurement Tools : UMIK-1
When you get the microphone you can download the calibration files from MiniDSP using the mic's serial # as a reference.
For measurement software there is REW, HolmImpulse and Arta amongst others. (In addition to xsim mentioned above)
All will be useful in understanding how to develop a decent passive speaker level XO.
There is another option, and that is to use a MiniDSP based solution with DSP to implement both the crossover and speaker/room correction. You would need two sets of stereo power amplifiers to implement.
I use a MiniDSP SHD with analog I/O to improve the performance of my speaker system in the room. The results are far better than I could achieve with the analog BSS Varicurves I used before.
If budget stretches the SHD will do the crossover and room EQ and is very flexible and quite transparent. Otherwise there are many other good solutions they can offer.
I recommend something like the MiniDSP UMIK-1 which will provide you with a calibrated USB microphone here: Acoustic Measurement Tools : UMIK-1
When you get the microphone you can download the calibration files from MiniDSP using the mic's serial # as a reference.
For measurement software there is REW, HolmImpulse and Arta amongst others. (In addition to xsim mentioned above)
All will be useful in understanding how to develop a decent passive speaker level XO.
There is another option, and that is to use a MiniDSP based solution with DSP to implement both the crossover and speaker/room correction. You would need two sets of stereo power amplifiers to implement.
I use a MiniDSP SHD with analog I/O to improve the performance of my speaker system in the room. The results are far better than I could achieve with the analog BSS Varicurves I used before.
If budget stretches the SHD will do the crossover and room EQ and is very flexible and quite transparent. Otherwise there are many other good solutions they can offer.
I expect you are right - but I am not sure I want to go into it that deeply! I wouldn't be surprised to learn that the unhinged boffins at Decca buzzed around with microphones and test gear 70 years ago - but that is not my ambition.
What I like to do is listen to the kinds of innovations that surprised and pleased listeners when they were new. These cabinets were made in 2 stages - the first in the early 50s when Decca released the plans for DIY builders. The second in the 60s to benefit from stereo. At that point they were fitted with Kef B200/T27s. With the Tannoys I shall be able to restore the original geometry. So I am just looking for the kind of small benefits I might get from more recent developments in x-over design.
I understand from a previous enquiry that the optimum x-over frequency for the Tannoys is 1.8 KHz. But that, presumably, is based upon a forward firing orientation. So I thought it would be nice to have the ability to tinker either side of that.
I believe the reason for the statement was that you seem to see LR filters as a active only option (using extra equipment) when in reality it is just a variation of a passive Butterworth filter.
I'm so pleased that you found my post useful.
FWIW, you can get set up with a measurement mic for well under £100, and the results will be streets ahead of anything you could do by ear.
Would you measure a voltage using a multi-meter, or by putting your tongue across the connection and having a guess?
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.