[DaveC113]

What about if the filtering is in the low level signal?  For example, my system has a Marchand active xover that uses 24db slopes with each output going to its own power amp.

It's problematic as passive line level xos (PLLXO) component values depend on the input impedance of the following component. So, put it before your amp and you can't change amps without changing the xo unless they have the same input impedance. So line level xos need to be active to avoid this issue, introducing it's own issues...

Passive speaker level xos are still the best way to do it, in general... but the issue of executing a seamless crossover in the ~(600Hz-5 kHz) or so range is difficult for even very expensive speakers to do well, obviously one big advantage of the Omega 1.5 way design. There are some big advantages to using DSP and class D amps for the woofers, including not having to build a passive high pass filter, which can be an issue as component values can be very high, so they are large and expensive. For mids and tweeters it's nice to use 1st order passive xos though, you end up with a speaker with active bass basically.

What's best really depends on your preferences though, this is in the context of preferring single driver speakers and appreciating their advantages. For someone with different preferences the best way to execute xos would be completely different and some aren't bothered by xos in the range I indicated either. So we have a large selection of speakers to choose from.

[FullRangeMan]

Actually 4th order are super-common and they work well with woofers, hard-coned drivers and more.

Almost all subwoofer amps have 24 dB xos built in, the Crown XLS has only 24 dB xos. Having a woofer, especially a subwoofer, playing audibly into the lower midrange is generally undesirable.

I forget HT subs, I was refer to hifi speakers. Use 4th order to subs make all sense since their fratulent nature.

[pstrisik]

Actually 4th order are super-common and they work well with woofers, hard-coned drivers and more.

Almost all subwoofer amps have 24 dB xos built in, the Crown XLS has only 24 dB xos. Having a woofer, especially a subwoofer, playing audibly into the lower midrange is generally undesirable.

That's what I have between the SAMs and Rythmik 8" units and between the Rythmik 8" and F12 subs.  The first is with the Marchand active xover with a variable freq xover point, the second manually set freq on the two plate amps.  Both Phil Marchand and Brian Ding have sung the praises of the 24db rolloff. 

With the Marchand, he describes it as a Linkwitz-Riley configuration:

It's problematic as passive line level xos (PLLXO) component values depend on the input impedance of the following component. So, put it before your amp and you can't change amps without changing the xo unless they have the same input impedance. So line level xos need to be active to avoid this issue, introducing it's own issues...

Passive speaker level xos are still the best way to do it, in general... but the issue of executing a seamless crossover in the ~(600Hz-5 kHz) or so range is difficult for even very expensive speakers to do well, obviously one big advantage of the Omega 1.5 way design. There are some big advantages to using DSP and class D amps for the woofers, including not having to build a passive high pass filter, which can be an issue as component values can be very high, so they are large and expensive. For mids and tweeters it's nice to use 1st order passive xos though, you end up with a speaker with active bass basically.

What's best really depends on your preferences though, this is in the context of preferring single driver speakers and appreciating their advantages. For someone with different preferences the best way to execute xos would be completely different and some aren't bothered by xos in the range I indicated either. So we have a large selection of speakers to choose from.

I guess I am minimizing the damage since the xover point between the SAMs and the 8"ers is below 100.

My upwardly pointing supertweeters just have a 2nd order line level high pass filter (Harrison).  Given the diffusion pointed upward and that these are rolling off from 15kHz, I don't imaging the filter is that negatively impactful in this case.

It is so atypical with the Monitors handling ~ 60Hz on up and the other three drivers working in very limited and specialized ranges.

.

........Peter

[pstrisik]

Opinions?

If I were to have a high pass filter made for the tweeters, would you advise 1st or 2nd order (4th may be to steep for this purpose since it overlaps with the monitors on the high end)?  These are the tweeters that will have a nominal xover frequency of ~ 12-15kHz and pointing upward.

Thanks for thoughts,  ........Peter

.

[roscoe65]

Opinions?

If I were to have a high pass filter made for the tweeters, would you advise 1st or 2nd order (4th may be to steep for this purpose since it overlaps with the monitors on the high end)?  These are the tweeters that will have a nominal xover frequency of ~ 12-15kHz and pointing upward.

Thanks for thoughts,  ........Peter

.

I would use a first order.  I use a simple cap and attenuator in the HF driver of one of my setups (8khz first order) and it works perfectly.  At the frequency you are proposing you are complementing the natural rolloff of the main driver, which presumably falls off at 6dB/octave or so.  You are also firing the tweeter 90 degrees off axis of the main driver as well as at a different distance from the listener.  Combine that with the 180 out of phase behavior of the 12 dB Harris filter and it would be pure accident if the tweeter output arrived in phase with the main driver.

If it were me, I would bring it in with a first order filter off the amplifier.  You may get a bit more coherence with the sound of the main driver and the amp is barely going to notice the load.  You can also experiment with bringing the tweeter in lower if you wish.  IIRC, the NS1000 crossed your tweeter over at 5250 hz with 12dB filter.  I'm sure you could bring it in anywhere from 10khz on up, but you may need an l-pad or autoformer.

JLM might chime in on this, since I believe he uses a pair of up-firing tweeters with his Fostex speakers in LCS configuration.  I also recall a number of discussions from a few years ago about add-on tweeters for Omega's.

[pstrisik]

I would use a first order.  I use a simple cap and attenuator in the HF driver of one of my setups (8khz first order) and it works perfectly.  At the frequency you are proposing you are complementing the natural rolloff of the main driver, which presumably falls off at 6dB/octave or so.  You are also firing the tweeter 90 degrees off axis of the main driver as well as at a different distance from the listener.  Combine that with the 180 out of phase behavior of the 12 dB Harris filter and it would be pure accident if the tweeter output arrived in phase with the main driver.

If it were me, I would bring it in with a first order filter off the amplifier.  You may get a bit more coherence with the sound of the main driver and the amp is barely going to notice the load.  You can also experiment with bringing the tweeter in lower if you wish.  IIRC, the NS1000 crossed your tweeter over at 5250 hz with 12dB filter.  I'm sure you could bring it in anywhere from 10khz on up, but you may need an l-pad or autoformer.

JLM might chime in on this, since I believe he uses a pair of up-firing tweeters with his Fostex speakers in LCS configuration.  I also recall a number of discussions from a few years ago about add-on tweeters for Omega's.

I power the tweeters with their own amp, Bantam Gold.  The Harrison fmod is signal level.  You are suggesting taking the signal from my main amp with a speaker level cap instead (plus attenuator, which I have - a Schiit Sys)?  You don't see this as compromising the SQ of the main amp/speakers compared to taking the signal from before the main amp?

I suppose I could use a single cap at the tweeter instead of the Harrison filter and still use the separate amp.  I could then try different values. 

[roscoe65]

I'm suggesting using the simplest of first order crossover, such as I use in my Altecs or in the Dynaco A25.  In both cases the main driver is run without a crossover and the HF driver is simply "added on" with a cap and attenuation circuit.  The main driver is unaffected by the tweeter, and the amp barely notices that the tweeter is there.  At these frequencies (>3,000hz) there is not a lot of musical energy.  Additionally, you'll find that the impedance of your Alnico driver rises with frequency and is probably 20 ohms at 10khz.  Adding a tweeter up high will still add very little load to the amp.

It is easy enough to try it both ways.

[pstrisik]

I'm suggesting using the simplest of first order crossover, such as I use in my Altecs or in the Dynaco A25.  In both cases the main driver is run without a crossover and the HF driver is simply "added on" with a cap and attenuation circuit.  The main driver is unaffected by the tweeter, and the amp barely notices that the tweeter is there.  At these frequencies (>3,000hz) there is not a lot of musical energy.  Additionally, you'll find that the impedance of your Alnico driver rises with frequency and is probably 20 ohms at 10khz.  Adding a tweeter up high will still add very little load to the amp.

It is easy enough to try it both ways.

Since I already have both an amp with attenuator built in and a free standing attenuator, I can try it both ways using the same cap arrangement at the tweeters.  Thanks for the brainstorming.

[DaveC113]

I bring my tweeters in up high as well, maybe 15 kHz or so but it overlaps below that as well.

I use a single cap, and rather than using attenuation you can simple reduce the value of the cap, which raises the xo frequency and thus attenuates the tweeter's output. At those frequencies it doesn't matter that the tweeter's output is tilted 6 dB/oct up and it's a benefit to avoid the use of resistors.

[pstrisik]

I bring my tweeters in up high as well, maybe 15 kHz or so but it overlaps below that as well.

I use a single cap, and rather than using attenuation you can simple reduce the value of the cap, which raises the xo frequency and thus attenuates the tweeter's output. At those frequencies it doesn't matter that the tweeter's output is tilted 6 dB/oct up and it's a benefit to avoid the use of resistors.

Thanks.  My sense is that the tweeters provide that sense of air rather than provide coverage at particular frequencies.  So not sure about the phase issue and how much it affects things in this scenario.

Makes sense to use a cap value that takes care of it all.  Coming from 2nd order 15kHz, I don't know where that would be with 1st order.  I think the attenuator will help me get a sense before ordering a bunch of caps at different values!  I'm starting with 1.5, which yields a bit over 16kHz.  Since I've been running 15kHz with 12db roll off, I figure this is a good place to start. 

[DaveC113]

If you're looking to fill in way up high the phase won't matter. Even if you calculated and adjusted your head would have to be in a vice for it to matter.

Cap values when used to attenuate depend on sensitivity differences, but consider raising the xo point one octave attenuates the output 6 dB. A simple xo calculator would be fine to use. I'm currently using .57 uF on 106 dB super tweeters and filling in >15 kHz. It makes a huge difference imo, but psychoacoustically if you don't have full bass extension it would sound way too bright. The better your bass is the more you notice the highs, it's interesting...

[pstrisik]

I've noticed that bass effect.  I can tweak my sub levels up just a hair and the midrange sounds clearer!  Must have something to do with over tones or under tones, if there is such a thing!

........Peter

[FullRangeMan]

Undertones are a bit rare to be listen is a speaker due the xover interference.