Layering in a Second Speaker System With Narrow Bandpass?

[Ron Resnick]

Let's say an audiophile likes his stereo system, but he feels there is a dip in frequency response between 5 kHz and 8 kHz that he would like to ameliorate. His speakers are small to moderate-sized, full-range floor-standing speakers. He's always been fascinated by plasma tweeters.

He buys a pair of Lansche floor-standers of similar height to his existing speakers. He positions them on the outside of his existing speakers.

He connects them to the same brand and model of amplifier he is using for his existing speakers. (Let's assume his existing speakers and the Lansche speakers have the same sensitivity.) He puts in attenuators in front of the input to his amplifier so he can adjust the volume of the standalone Lansche system.

His goal is to have the Lansche system reproduce mainly the 5 kHz to 8 kHz frequency range. He buys a Marchand cross-over box. He sets the cross-over to 24dB slopes, plugs in a 5 kHz high pass filter and an 8 kHz low pass filter.

Our intrepid audiophile has now layered into his existing system a completely new loudspeaker system. He plays a bit with the cross-over frequency on the Marchand and adjusts the volume of the Lansche system using the attenuators in front of the amps driving the Lansches.

Sitting in his listening chair he feels he has successfully raised the prominence in his system of the 5 kHz to 8 kHz frequency range.

What could be wrong with this approach?

A sense of incoherent drivers between the original loudspeakers and the Lansches?

Phase anomalies resulting from having two different drivers (the tweeters in his original speakers and the plasma tweeters in the Lansches) reproduce the same exact frequency range?

What else?

 

[fbhifi]

Is there a measured drop in frequency response between 5kHz and 8 kHz in this audiophile’s system,
or is the question purely hypothetical?

 

[Ron Resnick]

The scenario is hypothetical.

Please assume there is a measured drop in frequency response between 5 kHz and 8 kHz of about 5dB.

 

[Robh3606]

The scenario is hypothetical.

Please assume there is a measured drop in frequency response between 5 kHz and 8 kHz of about 5dB.

Why would you do this? Why not just add EQ and add a slight boost in the 5-10K range. There is very little power there so a 5 db rise won't add a significant power increase in the driver.

Your adding a crossover and an extra power amp so what would be the objection to high quality analog EQ?

That way you save a ton of cash and don't have any second source issues.

Rob

 

[Ron Resnick]

High quality analog EQ is my suggestion as well.

 

[sbnx]

IMHO, this would be a disaster. Extremely difficult to pull off even with the utmost care and precision. Most people can't even integrate a subwoofer at 40Hz let alone trying to get a second speaker, standing next to the original, to integrate at 5kHZ. Also, I would not try band pass. Just roll off the speaker at 5kHz and bring in the other one.

I agree that if we are talking about pushing the signal through some type of crossover, digital or analog, then why not just boost the signal at 5kHz using only one speaker. Much higher chance of success. I also do not like running the signal though such a device. There is going to major loss of transparency. It makes no sense to talk about buying such an expensive speaker as Lansche and Marchand in the same sentence.

Before I would do all of that stuff I would look at the on and off-axis measurements of my speaker. Is the 5-8kHz dip built into the speaker? If it is and you don't like it then get a different speaker. If it is not built in then the dip is likely due to speaker positioning. Reposition the speaker and the dip goes away.

Then there is always Bacch SSP if this hypothetical person can't find a way to do it any other way.

 

[hopkins]

If you boost the signal on a given band, don't you have to lower the overall signal to avoid clipping?

 

[Carlos269]

I also do not like running the signal though such a device. There is going to major loss of transparency.

As I have explained to Ron and others here, transparency is a function of frequency and it can be added or taken away. It is not the case that transparency has to suffer because you add a device. If the device is parametric transparency can actually be enhanced.

 

[sbnx]

As I have explained to Ron and others here, transparency is a function of frequency and it can be added or taken away. It is not the case that transparency has to suffer because you add a device. If the device is parametric transparency can actually be enhanced.

I guess we will disagree on this. Transparency has little to do with the actual frequency response.

 

[Carlos269]

I guess we will disagree on this. Transparency has little to do with the actual frequency response.

Interesting that you would say that. What do you think transparency has to do with if it is not frequency related? I looking forward to seeing what you attribute transparency to.

 

[Audiohertz2]

Let's say an audiophile likes his stereo system, but he feels there is a dip in frequency response between 5 kHz and 8 kHz that he would like to ameliorate. His speakers are small to moderate-sized, full-range floor-standing speakers. He's always been fascinated by plasma tweeters.

He buys a pair of Lansche floor-standers of similar height to his existing speakers. He positions them on the outside of his existing speakers.

He connects them to the same brand and model of amplifier he is using for his existing speakers. (Let's assume his existing speakers and the Lansche speakers have the same sensitivity.) He puts in attenuators in front of the input to his amplifier so he can adjust the volume of the standalone Lansche system.

His goal is to have the Lansche system reproduce mainly the 5 kHz to 8 kHz frequency range. He buys a Marchand cross-over box. He sets the cross-over to 24dB slopes, plugs in a 5 kHz high pass filter and an 8 kHz low pass filter.

Our intrepid audiophile has now layered into his existing system a completely new loudspeaker system. He plays a bit with the cross-over frequency on the Marchand and adjusts the volume of the Lansche system using the attenuators in front of the amps driving the Lansches.

Sitting in his listening chair he feels he has successfully raised the prominence in his system of the 5 kHz to 8 kHz frequency range.

What could be wrong with this approach?

A sense of incoherent drivers between the original loudspeakers and the Lansches?

Phase anomalies resulting from having two different drivers (the tweeters in his original speakers and the plasma tweeters in the Lansches) reproduce the same exact frequency range?

What else?

Head hurts at the thought

 

[sbnx]

Transparency is the absence of blur. Blur can be caused by many things. Acoustic misalignment, intermodulation distortion, lack of resonance control, careless handling of the signal (e.g. poorly designed equipment (even if it has dead flat frequency response), etc.

 

[Carlos269]

Transparency is the absence of blur. Blur can be caused by many things. Acoustic misalignment, intermodulation distortion, lack of resonance control, careless handling of the signal (e.g. poorly designed equipment (even if it has dead flat frequency response), etc.

Sound has three components: frequency, amplitude and phase. Perceived transparency is about the spectral content in the 4KHz range. Those items you mentioned are mechanical, not surprising coming from a mechanical engineer, and while they may effect the perceived transparency, they do so by altering the frequency spectral content of the sound reaching our ears.

 

[Robh3606]

If you boost the signal on a given band, don't you have to lower the overall signal to avoid clipping?

Hello hopkins

Well depends how much and where. Music power requirements are not linear so lets say you add +6dB at 50 hz. You would require 4 times the power. Do it on the other end you also add 4x the power but you use less in the upper octaves.

It's always better to use cut only EQ if you can but in this case it doesn't make sense unless you drop the level over all and take the SPL cut.

You should have more than enough headroom designed into your system. But it can get complicated as it is driven by how sensitive your speakers are power available and so on. This gives higher efficiency systems an advantage WRT headroom playing at the same average SPL.

For example my active system is 98dB 1 meter. I listen at about 85 dB average with the power I have available I can hit 115dB peaks clean so 30dB of headroom. I can afford a 6dB narrow band boost.

If my speakers were 88dB it would drop to 20 dB of headroom. Maybe you can maybe you can't?

Not addressing power compression, listening distance or average levels.

System dependent you would have to look at.

Rob

 

[Duke LeJeune]

His goal is to have the Lansche system reproduce mainly the 5 kHz to 8 kHz frequency range. He buys a Marchand cross-over box. He sets the cross-over to 24dB slopes, plugs in a 5 kHz high pass filter and an 8 kHz low pass filter...

That dip region is less than one octave wide, so I'm not sure how much better a Lansche plasma tweeter would sound compared to something more conventional.

Having two slightly different arrival directions (per channel) could blur the imaging, and if your head is not exactly in the right location side-to-side then you also have two different arrival times per channel, which (given the short wavelengths at those frequencies) results in comb-filter effects that will vary every time you move your head. These comb-filter effects will be happening BEFORE the precedence effect kicks in, which is undesirable. I'm thinking that not only could clarity be degraded, but imaging might get fuzzy when you have a lot of energy in that region. You might even hear the two sound sources separately, like a little duet; I had that happen once with a horn midrange and a horn tweeter whose outputs overlapped.

Please assume there is a measured drop in frequency response between 5 kHz and 8 kHz of about 5dB.

In order to bring up that region by 5 dB the secondary sound source will need to be almost as loud as the main speakers, assuming it's adding in-phase.

I would want to diagnose before prescribing... what's causing the dip?

In general I prefer to use an acoustic solution before falling back on EQ, BUT in this case, depending on what the cause is, EQ might make the most sense.

One example of a situation where an acoustic solution might be feasible is this: Suppose there is a crossover between a cone midrange and a ribbon tweeter at about 8 kHz, and the cone midrange is beaming badly at the top end of its range. So the on-axis response is pretty good in this region, but the off-axis response of the cone midrange is dipping badly in its top octave, resulting in a net in-room response with the dippage you describe.

Simple acoustic solution: Sit much closer to the speakers, such that the direct sound dominates over the reflected sound. This won't eliminate the dip in the net in-room response at the listening position, but will reduce it.

Complex acoustic solution (but no more complex that what you've already described!): Leave the speakers where they are and add a rear-firing tweeter adjusted to fill in the off-axis dippage in the 5-8 kHz region. This avoids the issues arising from having two horizontally-spaced sources per channel for the direct sound while adding energy where and when it's needed.

If the issue is that the drivers operating in the 5-8 kHz region have the dip in their on-axis response, then imo EQ is probably the best solution.

(Of course if you hypothetical friend is secretly looking for an excuse to buy a pair of Lansche speakers, don't show him this post!)