advertisement


Active Crossover Frequencies

McBride was my inspiration to start tweaking Naim circuits over 20 years ago! Lots of good ideas.

The values in the XO schematics are very close to what I have calculated, too. These are classic 3-pole Bessel filters with excellent transient response. They are my baseline design at the moment.

I just received a miniDSP to experiment with. Amazing variety of curve types and poles available. Plus the room compensation feature that is effectively an equalizer. I bought the UMIK-1 microphone with it for that purpose, too. the thing that strikes me first is how complex the setup is. Seems to me that the vast majority of people will need a dealer to set one up (and reset it every time you change your room or furniture!). this is in comparison to the relatively simple NAXO type of active crossover.

Also just read about Dutch & Dutch speakers which appear to have all the DSP circuitry, the microphone, the xover, and the power amps built in so you don't even need a preamp, xover, separate amps, etc etc. Just plug the speakers into the internet and pick up your remote control! I'm wondering if there will be anything left of analog in the future?
 
Sound is analogue, analogue is going nowhere, sure good digital implementations can do a lot and processing power is cheaper everyday as are very competent codecs offering ADC and DAC at sample rates vastly above audible range. You will still need to wrestle my record deck from my cold dead hands to separate me from it :) I say that as somebody that has done digital filter design for industrial instrumentation.
 
Just did an initial listen to the miniDSP vs a prototype Ryan Sound Lab crossover. This was done on my 2nd system which consists of an RSL 72, Mivera Class D power amps, and DIY transmission line speakers. Here are a few key specs of each xover setup:

miniDSP: 2x4 HD model with best processor (Shark)
-Linkwitz-Riley 8-pole filters (48db/octave slope)
-setup as stereo 2-way crossover @ 3400 Hz. (nominally -6db points for L-R type filter).
-adjusted actual crossover points for constant power output (-3db at 3400Hz): low freq = 3700Hz, high freq = 3100Hz. (this gives flat power through the pass band).
-did not adjust for time delay (might make a difference since the -3db points are no longer the same).
-tweeter output adjusted down -2db - about the same to my ears as turning the tweeters down manually to balance with the woofer-mid.

RSL prototype xover: 2-way
-Bessel 3-pole filters (18db/octave slope - same as NAXO)
-Crossover point is 3400Hz also adjusted for constant power across passband. (-3db points: low freq = 3600Hz, high freq = 3200Hz approximately)
-tweeter turned down for balance by ear.

Setup: same amps, speakers, source, and preamp for both xovers. No change in volume setting on the preamp when switching (both xovers have unity gain).
I won't elaborate here on setting up the miniDSP. Only need to mention that after the initial hassles with downloading software and understanding how to use it, the process of controlling the miniDSP is amazingly easy - selecting curve types, # of filter poles, frequencies, etc. Nicely done package.

Initial listening reaction using CD and LP sources:
- miniDSP is pretty decent but not great. Voices have added coarseness and a slightly hollow, metallic sound. Not terrible but a bit irritating. Same with piano - leaner, meaner - lacking in great tonality. On the plus side, the dynamics seem to be all there, at least on this system which is not my best one.
-RSL xover has all the dynamics plus the smoothness and tonality you expect from an analog system. Not to say it's perfect but not fatiguing like the miniDSP for sure.

There are dozens of possible filter setups for the miniDSP and I should try a couple of others to see what that does to the sound. Maybe these defects can be ameliorated by careful selection of parameters.

Then, there is the whole equalization feature which I didn't even look at (although I have the microphone to try it sometime). Since that was a constant between the two components, at least the initial impressions weren't affected by that.

I'm thinking of moving this evaluation to my primary system (Isobariks) to get more revealing information. Since a single miniDSP can only do one 3-way system (or two 2-way systems) I will try putting it on the woofer and midranges first, while leaving the tweeter on the existing analog xover to start.

More reports to follow!
 
RyanSoundLab said:
Just did an initial listen to the miniDSP vs a prototype Ryan Sound Lab crossover. This was done on my 2nd system which consists of an RSL 72, Mivera Class D power amps, and DIY transmission line speakers. Here are a few key specs of each xover setup:

miniDSP: 2x4 HD model with best processor (Shark)
-Linkwitz-Riley 8-pole filters (48db/octave slope)
-setup as stereo 2-way crossover @ 3400 Hz. (nominally -6db points for L-R type filter).
-adjusted actual crossover points for constant power output (-3db at 3400Hz): low freq = 3700Hz, high freq = 3100Hz. (this gives flat power through the pass band).
-did not adjust for time delay (might make a difference since the -3db points are no longer the same).
-tweeter output adjusted down -2db - about the same to my ears as turning the tweeters down manually to balance with the woofer-mid.

RSL prototype xover: 2-way
-Bessel 3-pole filters (18db/octave slope - same as NAXO)
-Crossover point is 3400Hz also adjusted for constant power across passband. (-3db points: low freq = 3600Hz, high freq = 3200Hz approximately)
-tweeter turned down for balance by ear.

Setup: same amps, speakers, source, and preamp for both xovers. No change in volume setting on the preamp when switching (both xovers have unity gain).
I won't elaborate here on setting up the miniDSP. Only need to mention that after the initial hassles with downloading software and understanding how to use it, the process of controlling the miniDSP is amazingly easy - selecting curve types, # of filter poles, frequencies, etc. Nicely done package.

Initial listening reaction using CD and LP sources:
- miniDSP is pretty decent but not great. Voices have added coarseness and a slightly hollow, metallic sound. Not terrible but a bit irritating. Same with piano - leaner, meaner - lacking in great tonality. On the plus side, the dynamics seem to be all there, at least on this system which is not my best one.
-RSL xover has all the dynamics plus the smoothness and tonality you expect from an analog system. Not to say it's perfect but not fatiguing like the miniDSP for sure.

There are dozens of possible filter setups for the miniDSP and I should try a couple of others to see what that does to the sound. Maybe these defects can be ameliorated by careful selection of parameters.

Then, there is the whole equalization feature which I didn't even look at (although I have the microphone to try it sometime). Since that was a constant between the two components, at least the initial impressions weren't affected by that.

I'm thinking of moving this evaluation to my primary system (Isobariks) to get more revealing information. Since a single miniDSP can only do one 3-way system (or two 2-way systems) I will try putting it on the woofer and midranges first, while leaving the tweeter on the existing analog xover to start.

More reports to follow!
Click to expand...
Hi,
I think i saw that you mentioned that the 4th order had a dip in the middle - i think you were referring to the Linkwitz-Riley filter. This is the optimal filter, since although there is a dip in the middle at the crossover frequency, the sum of the filters responses means that the overall response is flat.

In addition to this, the Linkwitz-Riley ensures correct phase response at the crossover frequency.

https://en.wikipedia.org/wiki/Linkwitz–Riley_filter
http://www.linkwitzlab.com/crossovers.htm

I use a program called Crossover-Pro by Harris Technologies for passive loudspeaker design, which allows for non Linkwitz-Riley crossover approaches with different orders - and lets you see if the magnitude and phase of the system with the drivers used- from the database of drivers.

I was going to take the approach you have in regards to the analogue crossover - but with the orders required using multiple "2nd order single opamp implementations", and the delay filters, the number of opamps to be used was high, and the frequencies are set based on the cutoff frequencies for each filter. You can change the frequencies by changing the resistor values - but it is a pain, given that you may or may not like the change.

My preference would be to progress with a DSP solution - i am surprised the MiniDSP gave the results you have mentioned. Did you use digital in, and analogue out ? This is my aim - to use a digital input, and the Analog Device ADAU1452 device which provides 4 stereo outputs. I will connect these to the Texas Instruments 32bit DAC's for upto a 4-way crossover, and implement the volume control in the ADAU1452, programmed to allow for the generic pre-amplifier remote controls (most remote controls for audio follow a standard such as RC-5). With the DSP solution - the ease and flexibility are much greater than an analogue filter approach.

For the tweeter crossover frequency - i am using the SB Acoustics Satori TW29RN-B - and the power rating in the datasheet can be misleading. They are using an IEC filter cutoff at 2.6kHz, for a specific input power of 80watts. The entire 80watts is NOT applied to the tweeter, but a smaller percentage of this as per the 2.6kHz high pass filters. What i am getting to - is that careful consideration of the cut off frequency for the tweeter (high pass) is required to ensure that the signal applied does not contain too low frequency energy and hence too much high power. The tweeter travels +/- 0.5mm based on the input signal, where as the midrange i am using travels +/- 5.5mm.

How did you set up the drivers for the correct signal level ??

I too will have to purchase a microphone - so i can adjust each driver volume level when setting up the crossover. I was going to use the laptop - with a tone sweep generator - to see what the response is across the band. Probably in the back garden - to stop reflections etc., within the room.

You seem to be much further along than me - so hope it goes ok.

Regards,
Shadders.
 
Very ambitious projects you have lined up there! Lots of hardware and software to work on! Let me respond to your comments:

1. The "dip" at crossover that I refer to occurs even after you add to the two bands together. It's there for any order L-R circuit. Here's the problem: With a nominal design, each output of the crossover is down 6 db (1/2 voltage) at the crossover frequency, so it appears that the total of the two is actually flat. However, what we hear is not voltage but rather power (sound pressure level) which is a function of the square of the voltage. Thus 1/2 * 1/2 = 1/4 power from each speaker giving you only 1/2 power total at the crossover point. That's the dip I'm referring to. This problem can be almost eliminated though by altering the crossover frequencies of the two speakers so they overlap more. For example, for a nominal 2.6kHz crossover, if the tweeter is set for 2.4kHz and the woofer/mid for 2.8kHz the power will be almost flat. The downside is that the phase is no longer exactly correct but may still be tolerable. There's no free lunch with the L-R setup, I'm afraid.

2. To see the theoretical power response for different frequency setups you can use LTSpice modeling. Once you've set up the circuit, you can see the voltage, power, and phase relationships all on one plot.

3. Also, as you say, the tweeter needs to be protected from too much signal so you might just raise the whole nominal crossover point to 3kHz or even higher. One thing that I do now (and it's worked to prevent another tweeter blowout) is to put a large, quality polypropylene capacitor in line with the tweeter. I use about 47uf for this and it sounds just fine. No more blowouts.

4. Setting the driver level is a matter of listening. There is no test gear that can integrate the total speaker response in your particular room into an apparent volume level like your ears. It's much easier to do once you've tried it.

5. My comment on the SQ of the miniDSP is based on using it in an analog in-analog out setup. So all the ADC-DAC circuits in the box are used. This means more conversions with their inherent loss of quality along the way. Still, the sound is very good - just not quite as good as a pure analog setup to my ears. That said, I've left the miniDSP in my 2nd system and the sound is very acceptable.

Haven't yet gotten to the microphone testing. Too many other projects in the way right now!

Best regards.
 
RyanSoundLab said:
1. The "dip" at crossover that I refer to occurs even after you add to the two bands together. It's there for any order L-R circuit. Here's the problem: With a nominal design, each output of the crossover is down 6 db (1/2 voltage) at the crossover frequency, so it appears that the total of the two is actually flat. However, what we hear is not voltage but rather power (sound pressure level) which is a function of the square of the voltage. Thus 1/2 * 1/2 = 1/4 power from each speaker giving you only 1/2 power total at the crossover point. That's the dip I'm referring to. This problem can be almost eliminated though by altering the crossover frequencies of the two speakers so they overlap more. For example, for a nominal 2.6kHz crossover, if the tweeter is set for 2.4kHz and the woofer/mid for 2.8kHz the power will be almost flat. The downside is that the phase is no longer exactly correct but may still be tolerable. There's no free lunch with the L-R setup, I'm afraid.
Click to expand...
Hi,
I understand what you are saying in terms of the power of the signal into a known impedance. For active filters, power is not an aspect that needs to be considered. For an opamp filter implementation, the circuit is generally a voltage operated circuit, as the relatively high impedances means that there is very little power dissipated (low current, low voltage).

Regards,
Shadders.
 
Perhaps my explanation wasn't clear enough. Half the voltage out of the active crossover means half the voltage coming out of the power amps, means half the voltage going to the speakers. That half voltage is half power at all points, including the sound pressure level coming out of the speakers at the crossover point. You still have the dip, unfortunately unless you try some kind of compensation like moving the crossover frequencies as mentioned above.

I'm not saying that you might notice this dip when listening (although I did)- with a very steep filter, such as 48db/octave, the dip won't be very wide. The miniDSP gives you the option of up to 48db/octave.
 
RyanSoundLab said:
Perhaps my explanation wasn't clear enough. Half the voltage out of the active crossover means half the voltage coming out of the power amps, means half the voltage going to the speakers. That half voltage is half power at all points, including the sound pressure level coming out of the speakers at the crossover point. You still have the dip, unfortunately unless you try some kind of compensation like moving the crossover frequencies as mentioned above.

I'm not saying that you might notice this dip when listening (although I did)- with a very steep filter, such as 48db/octave, the dip won't be very wide. The miniDSP gives you the option of up to 48db/octave.
Click to expand...
Hi,
Thanks - i see what you mean - the power dip in the loudspeaker at the cutoff frequencies of the filters. Does seem like a case for using DSP and higher order filters than 4th order.
Regards,
Shadders.
 
Hi,
I had to re-examine the issue of the crossover - it did seem odd that Linkwitz-Riley is established as the optimal crossover for speakers - but is not implemented in a passive design due to too many components causing losses. I did locate this article :

https://www.rane.com/note160.html

States "The acoustic sum of the two driver responses is unity at crossover. (Amplitude response of each is -6 dB at crossover, i.e., there is no peaking in the summed acoustic output.)"

I have not studied the speaker drivers - so, is the SPL a function of voltage, assuming that the amplifier can deliver the current ?

Regards,
Shadders.
 
I THINK this is the answer ...

The air displacements caused by the two drivers at the crossover frequencies, will sum to more or less double at some locations of interest i.e. listening positions. This is wave superposition.

This doesn't imply that acoustic power is doubled for free.

At crossover frequencies there would be other regions in space - e.g. some locations higher or lower than the drivers - where the air displacements caused by the two drivers will sum to almost cancel each other. (And generally, you'll see FR dips when measured from vertically off-axis.) Again, wave superposition. Those regions don't imply acoustic power from the drivers is being destroyed, either.

See the query https://www.researchgate.net/post/Superposition_does_not_conserve_energy.
 
Last edited:
Hi,
I examined Vance Dickasons Loudspeaker Design Cookbook, 7th edition, and as per page 147, if two acoustic sources transmit at 100dB, and they are in phase (correlated) at the point of listening, then the acoustic sum is 106dB. This means that although each speaker driver is 6dB down at the crossover frequency, their acoustic sum is still at the same level as the designed flat response.

Regards,
Shadders.
 
You must log in or register to reply here.


advertisement


Back
Top