Jump to content

Am I doing this right?


rplace

Recommended Posts

Ah, Wouldn't the H frames be considered a direct radiator in this instance? If so, I would move the horns back and delay the woofers to match.

 

But, I am just learning this stuff. Claude and Chris have done this and can do it in their sleep.

Link to comment
Share on other sites

On 12/27/2018 at 7:46 PM, rplace said:

I know there is a way in REW to have it make the PEQ suggestions for you, but I can't seem to get this to work.

Realizing that doing things manually can be an issue when a trying to set up PEQs, the REW EQ facility makes the process a lot more semi-automatic.  But that process is predicated on the assumption that you need to set up many PEQs all at once to flatten the overall loudspeaker response, not just one or two equalization filters. 

 

The non-automatic part of the process is setting the relative channel gains first--to start the PEQ definitions so that you're using mostly attenuating PEQs instead of boosting PEQs--and then setting the overall gain level for the PEQs when turning on the PEQ optimization routine.  Perhaps I'll be able to say a few words on that process after the following post is digested.

 

Chris

Link to comment
Share on other sites

15 hours ago, rplace said:

What I'd really like to have is a recipe on how to arrive at the correct settings, something like:

  • Set XO points and type
  • Next get PEQs right
  • Then set delay
  • Now House Curve....what ever that is. My preference I think

Additionally I'd like to know why I'm doing it and what I am actually doing.

 

Sorry for the delay.  Your list of things to do is pretty much the order of things.  I'll take them one at a time:

 

0) Remember, take each section one-at-a-time and if you run into problems somewhere, just loop back to the step that you think is where you made a poor choice then repeat the steps, but this time using better informed decisions.

 

1) Set crossover points and type:

 

In order to take a full REW upsweep of your loudspeaker, you have to decide on and set the crossover points first.  In some cases, you'll get information from other places, such as

 

a) the crossover points used in the passive crossover version of your loudspeaker (if any exists),

b) manufacturer recommendations (usually not a very good source, I've found),

c) comments from others that have done the settings using the driver(s), horns and bass bin boxes that you're using

d) looking at the frequency response and phase of each driver used alone without crossover or equalization/delay (including listening to the sweeps when you do them on the individual drivers)

 

As you might have guessed, the last two sources of information are usually the best.  Until a longer discussion of how to do the last two methods together, you can discuss the initial crossover frequencies with someone (including myself) just to get you started.  The good thing is...you can change the crossover points later (at the drop of a hat) if you find that you've perhaps chosen poorly or see the possibility to gain something from moving your chosen crossover points.

 

In the case of Klipsch Heritage loudspeakers, the crossover between the K-77 tweeter and its crossover midrange horn/driver (usually a K-55), this frequency is basically set at 4.9 kHz.  For the bass bin to midrange (assuming a 3-way loudspeaker), this is usually between 400-800 Hz--about an octave's worth of choices.  In the case of two-way, this can be all over the map.  If using full-range drivers in a 2-way loudspeaker, usually this crossover frequency is in the 400-1500 Hz range.  For crossover to subwoofer, that's usually pretty straightforward--wherever the woofer/bass bin starts to roll off in your listening room location(s).  Generally, I find that you want to set it as low as possible to spread the low frequency load out on your subwoofers and woofers (and perhaps associated horns).  I'd recommend always keeping your subwoofer within 3-5 feet of your woofers in order to minimize the possibility of hearing the the separation of subwoofer from woofers.

 

After you set the crossover frequencies, then you need to select the order of the crossover filters.  Sometimes you will have a good idea of what you want to use from someone else or just your own intuition. 

 

When I started doing DSP crossover tweaking, I was using the highest order filters available to me (usually 24-48 dB octave) in order to make the interference band as small as possible between the drivers/horns.  Then I started to look at the resulting step and phase response of the loudspeaker vs. frequency.  I didn't like the messed up phase plots that resulted from higher order crossover filters, so I started to use filters that had the least amount of phase shift (Bessel) and lower order filters.  Finally, I've moved entirely to using first order filters (6 dB/octave) because my ears tell me that's the best setting that I can use. 

 

But you have to choose carefully because you need to make sure that the drivers can handle such a gentle slope crossover arrangement without damage (for instance, playing your loudspeakers at extremely high SPL) and still sound good (the ends of the frequency spectrum associated with each driver usually begin to sound less and less nice the further you go into their roll-off regions--high and low frequencies).  Generally, it's good to talk to someone about the order of the crossover filters, too.  You can gain many insights from these discussions, because there are more than one or two issues that arise when you choose poorly.  Keep an open mind and don't fasten to one setting if you get asked about the settings you chose--would be my advice.

 

2) Get the relative channel gains and then PEQs

 

Once you have set the crossover frequencies and filter types (Bessel, Butterworth, Linkwitz-Riley, Chebyshev, etc.) and their order (6, 12, 18, 24, 36, 48 dB/octave slopes), then you're ready to take a full upsweep using REW.  Generally, I use 10-20, 000 Hz sweeps initially to see the relative gain of the drivers channels.  Generally, there is nothing that I say say generally about this process, other than it is usually in your best interest not to use boosting PEQs when you can set the channel gain a bit high, then use attenuating PEQs to pull everything down instead of boosting.  This is another place where a little judgment is usually in order. Talk to someone if you're not sure, but try your hand at doing it and then listen to the results.  You can only learn by doing.

 

Usually, the tweeter channels need to be set to a lower relative gains than the midrange relative to the woofer channels.  The woofer channels are usually set at zero relative channel gain by convention. You can wind up with almost any combination of relative channel gains depending on the drivers/horns and the total PEQs used on each channel.  Always remember that you can reset the individual channel gains higher or lower to also accomplish many of your PEQ tasks.  I've found that when you can use a higher or lower channel gain to solve very difficult problems when trying to use PEQs to flatten or extend response at the frequency extremes of a driver channel.

 

Once you get the channel gains set about right, then you can start the PEQ flattening process, discussed in more detail here:

If you find that you're running out of PEQs and the frequency response is still not flat, rethink your relative channel gains, set them differently, then try flattening using PEQs again.  Don't be afraid to start over again.  I've found that the best settings that I've found usually occurred after some time playing with PEQs and channel gains, and perhaps several tries, then suddenly you find a magic combination that works and uses fewer PEQs overall.

 

 

3) Set the relative channel delays

 

(You can also do this in step 2, preceding the flattening of response using PEQs)

 

At the point where you have a fairly flat frequency response, then you start the using the channel delays to get time alignment.  The first rule of using channel delays: usually you never delay the woofer (bass) channel, only the midrange and tweeter channels.  (This rule is broken with the Cornwall and Heresy that have long midrange horns and direct radiating woofers--where the midrange sound energy comes in last to the listener's ears.)

 

Generally, I look at the frequency response flatness at the crossover region (there will be a dip in response due to time delay mismatches), the phase curve will have a jump or at least a steeper rise in relative phase, the group delay curve (there will be a lag in the channel overall group delay that you can read like a graph on the frequency vs. group delay curve), and especially the spectrogram plot (the driver channels leading the other channels will show up visibly on the spike of the impulse response spectrogram--just like the group delay curve). More on this later, perhaps with graphics.

 

If you're using a 3-way loudspeaker, first set the lower frequency drivers (midrange) relative to the woofer, then set the tweeter channel delay also to the same delay value as the midrange channel--until you are ready to adjust the tweeter delay relative to the midrange channel.

 

During the process of setting delays, you will need to go back and update your PEQ settings in step 2, and perhaps even flop the polarity of one channel relative to the other channels in order to get smooth response through the crossover interference bands.  This is usually a bit of an iterative process.  I just run a sweep, update the settings, then run another sweep, then compare back to the prior sweep frequency response, group delay, and impulse response plots.  I keep iterating until I arrive at a happy medium setting for delays.

 

4) House curves

 

Since I don't use house curves, instead I use flat response for all my loudspeakers. But I correct the problem at its source: I demaster my recordings themselves to take out the high frequency boosts and the low frequency roll-offs found in the recordings due to mastering EQ used. (IOW: I don't have a lot to say about this subject.)  In general, if you're going to put a house curve on your frequency response, do it after you've achieved flat response in each of your loudspeakers. 

 

Chris

  • Like 1
  • Thanks 2
Link to comment
Share on other sites

In general when you do this dialing-in work, especially coming back to re-do settings to your loudspeakers at a later date, the results should be an immediately apparent improvement to your ears while playing your best quality recordings of acoustic performances.  If they don't sound better, then you've got something more to do, perhaps with the room layout (loudspeaker and/or listening positions) and/or room treatments.  Perhaps your issue is with the loudspeaker's drivers and/or horns, or even the recordings that you're choosing to listen to. Generally, the better the loudspeaker/room performance, the more revealing they are of the recordings that you play.  Be prepared to acknowledge that you might just have imprinted on lousy recordings, too.  You'll know what to do at that point...(⏏️⏏️⏏️ your chosen music in favor of better recordings).

 

Chris

  • Like 1
Link to comment
Share on other sites

On ‎12‎/‎29‎/‎2018 at 9:09 AM, Chris A said:

When I started doing DSP crossover tweaking, I was using the highest order filters available to me (usually 24-48 dB octave) in order to make the interference band as small as possible between the drivers/horns.  Then I started to look at the resulting step and phase response of the loudspeaker vs. frequency.  I didn't like the messed up phase plots that resulted from higher order crossover filters, so I started to use filters that had the least amount of phase shift (Bessel) and lower order filters.  Finally, I've moved entirely to using first order filters (6 dB/octave) because my ears tell me that's the best setting that I can use. 

 

As usual thanks so much Chris. I'd be lost without you. I have in fact talked to at least 3 different owners of my HF set up and have a range of about 100-150Hz of where to cross them. They all seem to like them in a similar range. I think right now I'm at the point of the quoted text above. I've always just owned speakers with passive crossover and they were what they were. When I got new XOs for my Khorns I didn't understand the science behind them...and I still don't. My caps were old and everyone with Khorns at the time on this forum were digging the new ones they were getting so I got some.

 

I see on the Xilica PC Application it can be 6-48. Looking at how it visually changes on the computers interface and coupled with terms I've heard thrown around like "Slope", "Extreme", "Steep" and "Shallow" I gather that this number and the changing graphical representations I see near the cross over point have something to do with how much of the frequencies near the XO point are shared, or blended or given to both drivers (I have a 2-way set up)  is that correct?  If so what does this mean in layman's terms? I would think the less that is shared the better, but I'm sure there is not hard, fast rule. Like so many audio things I'm sure the answer is "it depends". Until recently I though the crossover point was just that, a point. If you were crossing at 400 then everything above 400 went to one and every thing below to the other. If we are in fact sharing over a given range what do those numbers 6-48 mean/represent?

 

I've been playing with my set up typically at 24 because that seemed to be the default. When I was making these changes on the first page I changed them during actual music play back from 6 to 48 and could really hear a difference but could not really say why one was better or worse but I liked 6 much better than 48. What am I doing when I change them and what am I hearing as a result of this. The one thing I did not do was make measurements keeping all other things the same and just changing those numbers below the XO point. Finally If you are messing with say the HP side of things and I've got Bessel, 175Hz and 6dB/Oct can the other side that says None, 1000Hz, 24db/Oct have any effect? I would not think so otherwise it would be a band pass not a HP or LP, right? Just want to make sure I'm getting the full picture here. Band pass would be for the case of a 3-way right? You want some stuff in the middle to go to it and cut out the high side at a given point and the low side at a given point, correct?

 

Am I over thinking things? Nothing really sounds terrible. At times not really better or worse but different. Should I be like @Coytee and take some given settings and just live with it? No matter if it is Bicycles and their parts (I have 6 different bikes for different types of riding), or Beer Brewing (I make my own) I like to make changes, try different things but I also like to know the why and what behind those changes. I have not bought a complete bicycle from a shop or brewed a batch of beer from a kit in over 20 years. I'm starting to really like the process of making my own bass bins and playing with the Xilica, but it is a lot to figure out all at one time.

 

Sorry for the long winded reply but it seems like others on the forum have similar interests in knowing how it all works. Perhaps this will help us all get to where we are going.

 

Rereading your posts above and thinking about what I've done this Christmas break I think I'm going to rethink my approach. I picked the 175 XO point just because I had used 200-300 in the past. No real reason behind it. Then I picked the Bessel because you had given me ones like that before, then I started measuring and making PEQ to raise/lower things to flatten it out. I think I'll go back and remove all PEQs and measure various XO points between 150-300 and then 6-48db/Oct just to see what those sweeps look like and start with the flattest of those before I ever do any PEQs. Actually gain after XO values then PEQs. Again, many thanks!

 

The really cool thing about the xilica, and I'd bet others, is you can save them all. I've already got 25 of the 30 saved on the Xilica itself. Overkill probably but fun none the less.

  • Like 1
Link to comment
Share on other sites

Trying to school myself on things I found this. I realize generalizations are not always the best but it seemed to help me understand a bit of what I was asking above. Would any of you XO-Xperts ;) generally agree with this?

 

Quote

the steeper the slope, the more complex the circuit will be to arrive at such a filter. Also, the phase angle of the signal throughout the crossover range becomes more severe as the slope of the filter becomes steeper. On the flip side, the more gentle the slope, the simplier the circuit for such a filter and the more uniform your phase angle vs. frequency will be.

Generally, the gentler slopes force the speakers to play outside their comfortable reproduction range. Depending on the make/model, some speakers can handle this better than others without distortion. If your model of speaker has terrible response outside its recommended band, then you might want to consider steeper sloped crossover filters. However, as the phase angle becomes more severe, the more current will be demanded from your amplifier and if your amp cannot handle the current, it will overheat. The effect on the phase angle depends on the type of filter design you choose among other things...

I'm not a EE and I do not have a very complete understanding of circuit design, but when it comes to crossovers, it is rarely something you can just "pick and it works". There are many design considerations to think about.

 

 

My take away form that would be all other things being equal if you could use a gentler slop you should. If my HF section is good from 150Hz and above and my LF section is good from 1000Hz and below I should be able to cross between 200-300 say and use as gentle of a slope (lower number like 6 not 48, right) as is pleasing to me, no?

 

If I am right here than why all the fuss in years gone by about ALKs Extreme Slope Networks? I'm not trying to cast any stones here, just trying to understand things. Seems to me like if your horns/drivers would allow you would be after the exact opposite of an extreme slop network. Again, I'm just asking and trying to learn not saying Extreme Slope Networks are bad in any way.

Link to comment
Share on other sites

22 hours ago, rplace said:

I gather that this number and the changing graphical representations I see near the cross over point have something to do with how much of the frequencies near the XO point are shared, or blended or given to both drivers (I have a 2-way set up)  is that correct? 

Yes.  the steeper the slope of the filters (i.e., "higher order"), the less wide is the interference band between the crossing drivers.  First order is 6 dB/octave, second order is 12 dB/octave, etc. (See figure below.)

n160fig5.png

 

22 hours ago, rplace said:

If so what does this mean in layman's terms? I would think the less that is shared the better, but I'm sure there is not hard, fast rule. Like so many audio things I'm sure the answer is "it depends". Until recently I though the crossover point was just that, a point. If you were crossing at 400 then everything above 400 went to one and every thing below to the other. If we are in fact sharing over a given range what do those numbers 6-48 mean/represent?

The crossover point is the midpoint between the two sets of electrical filters (connected to the acoustic drivers).  It represents the center of a distribution of acoustic energy that's blended between crossing sets of drivers.  The narrower the interference band, the higher order filters used.  Nature abhors singularities, so there is always some amount of overlap between drivers, even if using 96 db/octave filters. 

 

With the kind of analog and digital filters that everyone talks about here (Infinite Impulse Response, or "IIR"), the steeper the filters used, the more the phase is messed up between the drivers in the crossover region and beyond.  This results in increased polar "lobing" in the vertical plane vs. frequency (see figure below).  I personally can hear the effect of first order filters in-room, I've found.

 

n160fig2.png

 

 

Only first-order filters (6 dB/octave) have no almost imperceptible phase shifts/lags on the lower frequency drivers, for IIR filters.  For FIR filters, you can have your cake and eat it, too at the same time, but the hardware costs more and the dialing in takes one more step than using IIR filters only.

 

Only linear phase filters (called "Finite Impulse Response, or "FIR" filters) of the digital type can simultaneously support steep slopes and good phase characteristics. 

 

Freq_intro.gif

 

22 hours ago, rplace said:

What am I doing when I change them and what am I hearing as a result of this.

You can see the difference in phase growth using linear phase (FIR) filters below in the red trace vs. typical IIR filters in the blue trace.  The blue trace is showing the phase growth for a fourth order filter (-360 degrees).  This results in the vertical polars which change vs. frequency, as shown in the polar lobing figure above.  This is most probably what you're hearing.

 

phase-shift-lr4-crossover-fir-vs-iir.png

 

The only down side with linear phase filters being that the more processing means more time delay for both drivers.  This is okay to use for home hi-fi, not so good when trying to synchronize your television to the music if the audio delays get to be too long, and a disaster for live sound reinforcement where audio delays can pull off the performers' performance significantly. 

 

The extra processing required to do linear phase filtering leads to more expensive DSP crossover processors and more required memory--thus the DSP crossovers that can do it usually cost more.  (One exception is miniDSP crossovers like the 2x4 HD and the OpenDRC-DA8 which actually cost less, but have other issues relative to the Xilica XP series to deal with.)

 

22 hours ago, rplace said:

Finally If you are messing with say the HP side of things and I've got Bessel, 175Hz and 6dB/Oct can the other side that says None, 1000Hz, 24db/Oct have any effect?

It means that the lower frequency drivers are trying to play full-range, i.e., they're not being crossed over at a chosen frequency.  This isn't what I'd recommend for anything but perhaps low passing subwoofers and woofers in-room.

 

22 hours ago, rplace said:

Band pass would be for the case of a 3-way right?

Yes.

 

22 hours ago, rplace said:

Should I be like @Coytee and take some given settings and just live with it?

I don't and wouldn't.  The reason for DSP crossover is that you dial things in using little to no time relative to passive crossovers, with full control of what each driver is doing.  With linear phase filtering, you also get to control phase at the same time.

 

22 hours ago, rplace said:

I'm starting to really like the process of making my own bass bins and playing with the Xilica, but it is a lot to figure out all at one time.

Just like performance bicycles and beer making, it take a little knowledge to do well and the more you know, the better it usually gets.  No mysteries there.  (Natural talent of course helps, but I've found with DSP crossovers, it just takes a little understanding of what is occurring, both mental models of what you're tweaking and using your ears to interpret what your fingers are doing using the DSP crossover controls.) Just take one subject at a time and read on it.  You'll pick it up quicker than you think.

 

22 hours ago, rplace said:

I picked the 175 XO point just because I had used 200-300 in the past. No real reason behind it.

Horns usually have directivity down to their low frequency roll-off point, while direct radiating drivers typically don't have directivity...until you get to high frequencies relative to the driver's diameter.  This is a big difference that you'll hear if you position your horns well relative to the room's boundaries and use proper amounts of diffusion and absorption to kill the long reverberation times.

 

21 hours ago, rplace said:

My take away form that would be all other things being equal if you could use a gentler slop you should.

There are pros and cons of each approach.  I've become much more sensitive to the audible effects of phase and group delay (the first derivative of the phase curve, i.e., the rate of change of phase).  Others are more interested in minimizing the interference band between drivers.  Pick your poison.

 

21 hours ago, rplace said:

If I am right here than why all the fuss in years gone by about ALKs Extreme Slope Networks? I'm not trying to cast any stones here, just trying to understand things.

I believe that I addressed that subject just above. Some people will demand steep slopes in their crossovers (especially if using drivers at the extreme ends of their frequency capabilities--like the K-55 midrange driver crossing over to a K-77 tweeter...there just isn't much overlap there that sounds very good, IME...you really want to cross at 4.9 kHz, I've found).  Others want more frequency overlap capabilities between their crossing drivers and flatter phase (I'm more in this camp presently).  Some people seem to be almost insensitive to any of these effects.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

Here's a very simple way of looking at things.  I'm going to attempt this by merely describing images you'll have to create in your mind.

 

Picture the side view of a sine wave on a graph.  You've seen them a million times...  It's at the frequency you'll be using for you're xover point.  Now look down at it from the top.  All you see is a straight line, the hills and dales are coming toward and going away from you. Now switch back to the side view where you started and step around so you're looking down the end of the graph as it's coming toward you.  All you'll see is a vertical line extending equally above and below the axis of the graph, which you now see only as a point.

 

If you run that sine wave through an inductor it will rotate that vertical line 45 degrees to the left, while a capacitor in series will rotate it instead to the right.  One element each way like this represents a 6dB/octave filter, and you see from the end (where you're still "standing") that the waveforms are now 90 degrees out of phase with each other. 

 

Add another element to each "way" (a capacitor in parallel after the series inductor and a parallel inductor after the series capacitor), and you'll get a further rotation of 45 degrees each way at that frequency.  What you'll be looking at is now a horizontal line with the two (what were originally the) "tops" at opposite ends from each other, thus completely out of phase with each other.  If you combined the two signals they'd cancel each other out and collapse to just just the point in the middle (the graph axis you're now "looking down the barrel of").

 

(This is why you usually see the "+" and "-" of one of the two drivers swapped in a "conventional" 12dB/octave crossover.  So that the drivers will be restored to being "in phase" at the crossover frequency. But more on this later.)

 

For each added element (another 6 dB/octave), the "end view rotation" twists another 45 degrees in their respective directions.  So at 24 dB/octave the waveforms have ended up back in phase again, but this twisting involves a time element, so the "tops" and "bottoms" (of the traditional "side view") are no longer the exact same "tops" and "bottoms" when they "line back up in phase"; they're instead, while in phase, out of time.

 

I confess that I have little knowledge of how these different filters are synthesized in the digital domain.  At first blush I would assume the filtering could be digitally performed without all the "analogesque" phase shenanigans.  But perhaps that's what's required to effect the filtering...

 

So, when you're adjusting delays in the filters, there are actually two items under consideration.  One is the amount of time it takes the sound to travel from each driver and the other is the amount of time between the now-separated waveforms they're reproducing.

 

Hope this does more good than harm to the discussion ;)

 

Please don't ask me any questions about actual implementation as I have zero experience in that respect.  I could only offer opinion after perusing applicable equipment manuals and I'm not particularly inclined at this time.  My wonderful wife just recently allowed the purchase of a pair of Forte IIIs.  She doesn't even begin to understand the allure of all this, much less how I might could (already!) be contemplating bi-amped Jubilees (or 402MEHs) with all the attendant electronics, and my further involvement here would, well, let's just leave it at my lack of inclination...

  • Thanks 1
Link to comment
Share on other sites

I'd like to show you an impulse spectrogram of a typical Cornwall (1979 model) with its passive crossovers.  (Prepare yourself...):

 

Cornwa II Impulse Spectrogram.jpg

 

Now your ears can hear the high end of the Cornwall that is playing with two sets of drivers above 4.9 kHz--the tweeter and the midrange driver on their separate horns.  The tweeter is leading the midrange by about 750 microseconds (the midrange output is albeit at a lower SPL relative to the tweeter above 4.9 kHz), and the woofer is leading the midrange by about 500 microseconds at the crossover frequency.  This is exactly what a DSP crossover could clean up in mere minutes of set up, taking a measurement, adjusting the DSP crossover delays and crossover filters, then rechecking, until the response looks a lot more like this (in this case, replacing the K-55/K-600 midrange and K-77 tweeter with a single ESS AMT-1):

 

Cornwall bass bin + AMT-1 6dB per oct spectrogram on-axis.jpg

 

Chris

Link to comment
Share on other sites

I just read through my previous post and see that I forgot a mental image item.  The end-view "twisting" is effected much the same as if the waveforms were being "screwed" along a screw thread that has the same thread pitch as the frequency.  One direction along that screw "the zero axis you'd be looking down" makes the wave "recede" from you while the other direction "proceed" toward you.  Thus, in the case of "90 degrees each way" (12 dB/octave), it does not exactly cause a complete cancellation.  There's a half-a-wavelength time difference between the equal-but-opposite polarity values you're seeing from the end view.

 

Maybe I should've just sat this one out?  :)

Link to comment
Share on other sites

On ‎12‎/‎30‎/‎2018 at 9:01 AM, rplace said:

My take away from that would be all other things being equal if you could use a gentler slop you should.

 

21 hours ago, Chris A said:

There are pros and cons of each approach.  I've become much more sensitive to the audible effects of phase and group delay (the first derivative of the phase curve, i.e., the rate of change of phase).  Others are more interested in minimizing the interference band between drivers.  Pick your poison.

 

Chris A is the expert, and shows the measured results of his tests. I also think he can hear the differences better than I can, so I recommend following his advice.

  • Thanks 1
Link to comment
Share on other sites

Try them both ways.  Let your ears be the judge.  When you crank in steeper filters, you have to add a delay to the higher frequency channel(s) equal to the period of a 1/4 wavelength at the crossover frequency for each order of the filters used. 

 

So for a 2nd order filter at 2000 Hz, add the following delay to the HF channel:

 

     1/2000 Hz * (1/4) * 2 [for second order] = 250 microseconds. 

 

For 800 Hz and a fourth order filter, the added delay on the HF channel is:

 

     1/800 * (1/4) * 4 = 1/800 = 1.25 milliseconds.

 

Chris

  • Thanks 1
Link to comment
Share on other sites

Very interesting and educational. I thought the delay was all about the difference in drivers based on my K horn and La Scala understanding with the large difference between the Tweeter and mid-range horn length. 

 

So interesting to find out that the network / filter / crossover is what affects the delay. I guess when you stop learning your dead.

 

Also when you set up AVRs and home theater processors they typically ask for the length frim seating position. I figured that was to set various delays so all the sounds hit you at the right time. Probably is because they would have no way of knowing what your speaker crossovers are

Link to comment
Share on other sites

Like I'd said, there are two distinct delays to factor into the crossover.  Chris since provided the ballpark math for the one; the other involves the speed of sound under particular atmospheric conditions versus the relative distances to the drivers.  The two would "simply" be added or subtracted as needed for the final single value to be entered into the equipment.

 

As you'd figured, the AVR delay is a completely separate (third) consideration, and it would be addressed last.  I rather doubt any such implementations are concerned with whether or what crossovers are in play.

  • Thanks 1
Link to comment
Share on other sites

Horn length difference is also added to the total delay from the electrical filters as mentioned above, as well as the insertion delays of the DSP crossovers themselves.

 

I use an AVP (preamp/processor) to perform the surround channel delays using the little Audyssey microphone, because the delays used in the Xilica vary from the fronts (bi-amped Jubs), center (tri-amped MEH), and surrounds (bi-amped Cornwall + AMT-1s) the last of which are using a Dx38 crossover with different  insertion delays.  After getting the delays for each driver in a loudspeaker time aligned, it's nice to not have to set the time delays from all 6 channels (5.1) because every channel has DSPs--even the subwoofers.

 

Chris

Link to comment
Share on other sites

Yeah, in that case it'd certainly be most expedient to use the AVP for final speaker-to-speaker alignment.

 

Whenever I'm with the missus in a place displaying a multitude of televisions and I point out the stragglers as either doing more signal processing or having a slower CPU, she says "only you would notice something like that."  I just chuckle...

 

I'd started this post yesterday; just got back from enduring fair P.A. equipment being very poorly used by a mediocre small-town band, as a semi-designated driver no less.  Can't say how badly I wanted to commandeer the equipment.  What a way to ring in the new year!

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...