Jump to content

Active Bi-Amping/Tri-Amping FAQ


Chris A

Recommended Posts

Ellery,

 

I haven't tried this yet but have you tried using a Harsch crossover configuration (4th order Butterworth LF, 2nd order Bessel HF with 1/2 crossover wavelength delay)?  (The presentation in French can be found here.) The impulse and phase response are apparently improved over Linkwitz-Riley. 

 

If you're looking for crisper impulse response, this is an easy and cheap way to do it, assuming that you're already running on a digital crossover.

  • Like 1
Link to comment
Share on other sites

Very neat.  Thanks!

 

So, getting phase slope to match closely seems to have resulted in a very audible improvement to me.  Strings, voices, string brushes sound very real now.  Some of this could be due to the Benchmark AHB2 amps I added, in addition to better phase tracking.

 

A harsch crossover configuration sounds like it would flatten the slope of my phase response (a lot less overall phase shift due to the crossovers).  What audible improvement would be expected?  I know many say we cannot hear differences in phase, but I know my bass sounds far more accurate when the phase slope is linear/consistent and without bumps for the subs/LF crossover.

 

I'm thinking this could be combined with an FIR xo for the MF/TW?

 

On 2/17/2017 at 11:10 AM, Chris A said:

Ellery,

 

I haven't tried this yet but have you tried using a Harsch crossover configuration (4th order Butterworth LF, 2nd order Bessel HF with 1/2 crossover wavelength delay)?  (The presentation in French can be found here.) The impulse and phase response are apparently improved over Linkwitz-Riley. 

 

If you're looking for crisper impulse response, this is an easy and cheap way to do it, assuming that you're already running on a digital crossover.

 

Link to comment
Share on other sites

2 hours ago, etc6849 said:

What audible improvement would be expected?...I'm thinking this could be combined with an FIR xo for the MF/TW?

It would improve the impulse response of the system.  Crisper transients would be the audible result.

 

Steady state, you can't hear the improvement in phase response, but music impulses--particularly percussion hits--reveal phase alignment issues.

 

If you're going to use FIR filters, then you don't need the Harsch approach, since those filter types used by Harsch are IIR filters.  You would simply measure the impulse response of your system using REW, export to a WAV file, then import into a convolver (like RePhase) to generate the FIR filter taps to align both amplitude and phase vs. frequency.

Link to comment
Share on other sites

I wish the Xilica was that flexible.  I can only tell it how many taps to use and the xo frequency.  Taps are limited to 1500 (for all channels combined); this means the FIR xo's are not very useful for lower frequencies like MF-LF.

 

 

44 minutes ago, Chris A said:

It would improve the impulse response of the system.  Crisper transients would be the audible result.

 

Steady state, you can't hear the improvement in phase response, but music impulses--particularly percussion hits--reveal phase alignment issues.

 

If you're going to use FIR filters, then you don't need the Harsch approach, since those filter types used by Harsch are IIR filters.  You would simply measure the impulse response of your system using REW, export to a WAV file, then import into a convolver (like RePhase) to generate the FIR filter taps to align both amplitude and phase vs. frequency.

 

Link to comment
Share on other sites

Chris!

 

I JUST stumbled upon your post here.

 

It was you that suggested active digital crossovers for a 2-way Cornscala project I had just begun to entertain.

 

The seed was planted and NOW I find your post.

 

Very comprehensive, even for a relative newby, with excellent references to illustrate the points.

 

Anyone reading this should be clear of one caveat in life: 

 

"PAY FOR QUALITY ONCE, OR PAY FOR CHEAP FOREVER"

 

There is nothing "wrong" with passive crossovers, "IF" you can't afford active crossovers. They are a less complex, instant gratification answer with given results. Fast food drive through comes to mind... Mmmmm.... cheeseburger please, drag it through the garden!

 

I have just arrived at the point where I cannot afford NOT to set aside the resources to pursue an ACTIVE DIGITAL CROSSOVER for my system.

 

The passives will no doubt find their way into another pair of 2-way Cornscala's down the road, nothing else I have can compare to them, YET!

 

Thanks for your efforts here Chris, any suggestions on an exact model or short list I should concentrate on finding?

 

 

 

Link to comment
Share on other sites

Right now the digital crossover market seems to have a threshold price for good quality units--both used and new, that is.  The minimum used market price is about $300-$500(US), and with slightly higher prices than this garnering disproportionate increases in quality (i.e., $500-$800US).  The best units that I've found (through either direct or indirect experience) are Yamaha SP2060, ElectroVoice Dx38 and DC-One, Ashly Protea, and Xilica (if you can find them used). 

 

New prices have been pushed down by one company and model series--Xilica XP--and they now represent a sort of gold standard that's pushing down the prices of high quality units. 

 

The other market is the low price end represented by miniDSP.  Various reports are that the miniDSP "HD" units are just capable of doing the job from a noise floor, dynamic range, and reliability standpoint.  I've not had an HD crossover to play with, but it would indeed be a big step forward if these units were really up to the task.  My guess is however that you'll have to wait for the next round in product lifecycle updates to really be the equivalent of a Xilica or used EV.  In other words...the risk of losing your investment to something that doesn't really do what you want it to do is still too high in my estimation, until someone shows that it works with high efficiency Klipsch loudspeakers.  I'd recommend adding a little more money and getting something that is known to work and to be reliable over a reasonable operating life.  If I get my hands on a miniDSP HD unit, I'll certainly let you know how it performs.

 

When you look at these prices and capabilities relative to custom passive crossovers that are being sold by third parties, quality digital crossovers now are less money than high quality passives.  If you're only looking at lowest, rock bottom prices, then passives are still the economic alternative of choice but note that we're now talking about a maximum of $300-$500 per pair of passive crossovers before the digital units become the overwhelming best choice economically.  That's a fairly low price threshold and one that most real audio enthusiasts can easily hurdle by saving a little of their spare change for a short period. 

 

Also note that digital units are not relegated to a single loudspeaker model and type, but can be used for any loudspeaker type/model that the owner has on hand.  A lot of people I think miss this important point. Once you own a digital crossover, chances are that you'll keep it for a very long time to use on future projects--just like amplifiers and preamps. 

 

Chris

  • Like 1
Link to comment
Share on other sites

Also note that the price of using REW, a calibration microphone, and a cable to connect your computer to your preamp is now at such a low dollar amount that it now seems likely that instead of audio enthusiasts learning about soldering, they are now learning how to use REW to dial everything in using a digital crossover or some sort of A/V preamp of some quality.  This is a much more useful and powerful skill for the average audiophile to master, IMHO.

 

Another related audiophile skill that I envision by a new class of audiophiles is the ability to demaster their own music using freeware to remove the artifacts of "mass market appeal" clipping and EQ present on almost all available recordings.

 

Once you are enabled with these two new skills, I believe a new class of audiophiles will actually begin to rise...like that seen in the late 1940s-early 1960s with the original rise of hi-fi.  The economics of these two areas now set the stage for change away from current audio nostalgia into higher performance hi-fi at much lower cost.

 

Chris

  • Like 1
Link to comment
Share on other sites

@Chris A

 

I was thinking last week about your goal of having no phase wraps.  I was able to put a large well padded La-Z-Boy Pembroke love seat along my back wall, and it almost fixed the phase wrap I had at 238Hz!  Excess group delay is very smooth at this frequency now, but there is still the amplitude dip.

 

I ordered two ottomans and a chair to match the love seat (at a very good price saving about $800 versus my local la-z-boy dealer).  I believe these last few pieces of furniture will fix the room related phase shifts. 

https://www.wayfair.com/La-Z-Boy-Pembroke-Premier-Stationary-Club-Chair-LZ1408.html

and

https://www.wayfair.com/La-Z-Boy-Pembroke-Premier-Ottoman-LZ1409.html

 

Then with the Xilica XD, I am thinking about doing FIR crossovers at an acceptable slope for: 500Hz, 3.8kHz.  For the 80Hz crossover for the subs I am planning to move up to 110Hz and use a first order LR filter (to eliminate most of the phase shift from the crossovers).  Hopefully doing this would eliminate most of the phase wrapping in the previous REW file I posted for my quad amped setup.  I guess an alternative would be LR 2nd order and inverting the sub polarity or doing the Harsch XO for sub/woofers?

 

I am thinking the shallower slope for the first order linkwitz-riley will mean both my subs and woofers will be working at the same time for the most part, so decay times below 110Hz would have to improve due to better room coupling (more drivers as always led to better decay times for me provided they are time aligned).  IMD would be worse off with a shallow sloped crossover, but I doubt this will be noticeable with tri-amped mains ;) I haven't seen this line of reasoning written anywhere, but it would seem it will give much better in room performance versus the 48dB/octave BW crossover I have now at 80Hz.  Very interested to see if anyone has used this approach to improve decay times (and at the same time eliminate phase shifts that higher order crossovers will cause).  I don't think most speakers by themselves have an in room -3dB point close to 20Hz, so probably not recommended for everyone...

 

Will be a while before my oversized chair and ottomans come.  I'll post the results here next month, but wanted to get your thoughts on using the first order filter for the sub/LF transition.  To ensure time alignment, I would do a band limited sweep in REW and match up the resulting impulse plots for both the woofers and subs (maybe a 40-140 Hz sweep).  I'm thinking I would temporarily change the Xilica's crossover setting to do this measurement so both woofers and subs would give the same amplitude for the 40-140Hz band.  After I get the proper delay, I would time align the woofers and subs in the Xilica and put the crossover settings back where they should be.  I would then sweep drivers individually and do a overlay plot of the phase, and might alter timing in .1ms increments in REW until there is a very smooth phase in the crossover region.  Any thoughts on this approach?

Link to comment
Share on other sites

Ok, I decided to redo my active crossover settings, but wanted to keep phase shift from the crossover used to a minimum so I selected 12dB/octave bessels for woofer and sub.


I also had to invert the sub to get the phase tracking to match between the subs and woofers around the crossover point (110Hz).  Phase looks pretty decent (but I'm not a pro), and the bass is the best it's been; using a shallow slope for the subs and low frequency drivers gives even better bass (was very good before, but has a tad more impact now).  Too late to run Dirac tonight, but I'll post an REW file in the next week or so comparing the presets on the XD4080.58b280c67e450_rightnew.thumb.jpg.34d810bde4493d09de44bc726c216a84.jpg

58b282eb12750_rightnewunwrappedphase.thumb.jpg.a19fc86450bc82bf6b10574926ebb417.jpg

New bessel and FIR XOs LF.png

New bessel and FIR XOs HF.png

New bessel and FIR XOs MF.png

New bessel and FIR XOs subs.png

New bessel and FIR XOs.png

Link to comment
Share on other sites

On 2/25/2017 at 5:06 PM, etc6849 said:

...but wanted to get your thoughts on using the first order filter for the sub/LF transition...

Well, this is somewhat complicated at such low frequencies.  You're into the deep (i.e., can't localize the sound location) sparse-mode region of the room where the placement of the drivers to fill those room modes up with LF energy is an "orthogonal process".  In other words, if you're not injecting LF at a corner of the room and at a mid-point of each dimension of the boundary walls/ceiling/floor, the energy doesn't get into those room acoustic modes. 

 

So the real issue is phase and required excursion of the Palladium front woofers, which are also covering higher frequencies, so your getting increased THD (harmonic) and TMD (modulation distortion).  The ear's masking also plays a role in hiding higher frequency distortion products--so this situation usually comes down to making a few measurements and listening, then making your choice on what to do.  I've found on the K-402-MEH--that having that LF injection into the room at the midpoint on the front wall is worth more than having lower THD and TMD at that point.  YMMV.

 

The biggest distortion issue is modulation distortion, which piles up at higher frequencies that are being produced simultaneously.  This makes the overall sound opaque due to those sidebands, even if masking is occurring at higher frequencies, mainly due to the lower sidebands and increased sideband modulation that is occurring off of the lower frequency sidebands. So higher frequency clarity and presence suffer when you make the front and center loudspeakers try to reproduce very low frequencies.

 

On 2/25/2017 at 5:06 PM, etc6849 said:

To ensure time alignment, I would do a band limited sweep in REW and match up the resulting impulse plots for both the woofers and subs (maybe a 40-140 Hz sweep).

Time alignment of woofers-to-subwoofers is generally not audible if you're crossing below 80 Hz or so (...the lower you go, the less audible the time misalignments).  You can have a great deal of time misalignment below 80 Hz and not hear it.  The only issue is phase cancellations between the woofers and the subwoofers due to the different (non-flat) phase curves of the woofers and subwoofers vs. frequency in those frequency bands where the two drivers are both audible and additive--as in able to cancel each other.

 

Chris

  • Like 1
Link to comment
Share on other sites

On 2/26/2017 at 1:25 AM, etc6849 said:

Ok, I decided to redo my active crossover settings, but wanted to keep phase shift from the crossover used to a minimum so I selected 12dB/octave Bessels for woofer and sub...I also had to invert the sub to get the phase tracking to match between the subs and woofers around the crossover point (110Hz).  

 

Phase looks pretty decent (but I'm not a pro), and the bass is the best it's been; using a shallow slope for the subs and low frequency drivers gives even better bass (was very good before, but has a tad more impact now)...

 

This is as good a phase response at a listening position that I've seen (but my experience base isn't very large).  I didn't realize that you were crossing so high (110 Hz).  Since your subs and your fronts are basically touching one another at the front of the room, apparently you can get away with this crossover frequency/wavelength. That's as high a frequency/shortest wavelength as I've seen anyone do successfully.  Having minimum phase growth due to the crossover filter used is a pretty big deal  (I'd think) in getting both the woofers in the fronts and the subs successfully playing together with the width of the interference band that you've chosen.

 

I am doing the same thing here. I'm fiddling with using REW measured response in-room to fashion FIR filters using rePhase to generate the FIR phase correction filters for the Jubs (i.e., above the subwoofer-Jubilee bass bin crossover frequency), and foobar2000 (using its convolver) to play that linear phase response from the player in my laptop. The crossover frequency to my woofers is 40 Hz currently.  The Jub bass bins are good to about 30-32 Hz, and the TH subs are good to 14 Hz, so having such a low crossover point works, if I ignore phase at these low frequencies--which is reasonable I think.

 

But it would be nice if the digital crossover itself could be running those FIR filters, which would make it a full-time solution--not just playing music on foobar2000.

 

Chris

  • Like 1
Link to comment
Share on other sites

17 hours ago, Chris A said:

Time alignment of woofers-to-subwoofers is generally not audible if you're crossing below 80 Hz or so (...the lower you go, the less audible the time misalignments).  You can have a great deal of time misalignment below 80 Hz and not hear it.  The only issue is phase cancellations between the woofers and the subwoofers due to the different (non-flat) phase curves of the woofers and subwoofers vs. frequency in those frequency bands where the two drivers are both audible and additive--as in able to cancel each other.

 

Chris

Chris,

 

Thanks for answering this non practical question of mine.  I had edited my original post as I quickly realized my sub would need to output frequencies well above 1000Hz to handle a 6db/octave crossover (which from my research was desirable for minimum phase)!  My subs aren't designed to do this so I didn't attempt it.  I may give it a try still, but the 2nd order bessel works very well for low frequencies.  The Palladium woofers would handle the first order slope well as I picked 110Hz to allow for a gradual high pass slope.  However I agree that modulation distortion would have to be worse off.

 

This Rane note is what led me to going with a second order bessel which kept group delay very low: http://www.rane.com/note147.html

 

"Note that for the second-order phase-match design, low-pass and high-pass group delays are exactly the same."

 

This has resulted in EGD that is very well matched all the way down to 20Hz.  I will post the REW file that has Dirac and non-Dirac comparisons of using high order BW filters versus these new active xo settings.

 

I would agree with your earlier post that eliminating phase shifts gives better imaging.  For low frequencies, I think I can hear and feel a difference, but have no way to do proper testing of this.

 

I can see that the sides of my waterfall plots pre-dirac have improved slightly using a 2nd order filter versus an 8th order BW.  Also, EGD improved drastically from before.

 

 

Link to comment
Share on other sites

3 hours ago, Chris A said:

This is as good a phase response at a listening position that I've seen (but my experience base isn't very large).  I didn't realize that you were crossing so high (110 Hz). 

Thanks.  Wait until you see the post Dirac result ;). Could not have got it without your Xilica recommendation and other advice in this thread which led to selecting a somewhat affordable active xo that did FIR filters.

 

I would think there would be a Linux or virtual soundcard solution that can use HDMI out as an active 8 channel xo.  I saw one for Windows, but forgot the name.

 

As you know, regardless I don't think you can setup a PC solution for multi-channel.  Perhaps you could do 5.1 and use the remaining 2 PCM channels over HDMI for active bi-amping the fronts along with mixing 80hz and below and passing it to the sub channel?  If this is all done as a virtual sound card, you would just use KODI for playback of all movies/music/TV on the virtual soundcard, and any processor with HDMI in and analog preouts should work (you'd just lable channels 7 and 8 as left and right WF out, etc...).

 

I was going to attempt this, but it will not work for tri-amped front left and right and 5.1 content as HDMI can only do 8 PCM channels.

 

I also don't know if the windows software I found is this versatile...

Link to comment
Share on other sites

  • 4 weeks later...
On 2/24/2017 at 7:12 AM, Chris A said:

Right now the digital crossover market seems to have a threshold price for good quality units--both used and new, that is.  The minimum used market price is about $300-$500(US), and with slightly higher prices than this garnering disproportionate increases in quality (i.e., $500-$800US).  The best units that I've found (through either direct or indirect experience) are Yamaha SP2060, ElectroVoice Dx38 and DC-One, Ashly Protea, and Xilica (if you can find them used). 

 

New prices have been pushed down by one company and model series--Xilica XP--and they now represent a sort of gold standard that's pushing down the prices of high quality units. 

 

The other market is the low price end represented by miniDSP.  Various reports are that the miniDSP "HD" units are just capable of doing the job from a noise floor, dynamic range, and reliability standpoint.  I've not had an HD crossover to play with, but it would indeed be a big step forward if these units were really up to the task.  My guess is however that you'll have to wait for the next round in product lifecycle updates to really be the equivalent of a Xilica or used EV.  In other words...the risk of losing your investment to something that doesn't really do what you want it to do is still too high in my estimation, until someone shows that it works with high efficiency Klipsch loudspeakers.  I'd recommend adding a little more money and getting something that is known to work and to be reliable over a reasonable operating life.  If I get my hands on a miniDSP HD unit, I'll certainly let you know how it performs.

 

When you look at these prices and capabilities relative to custom passive crossovers that are being sold by third parties, quality digital crossovers now are less money than high quality passives.  If you're only looking at lowest, rock bottom prices, then passives are still the economic alternative of choice but note that we're now talking about a maximum of $300-$500 per pair of passive crossovers before the digital units become the overwhelming best choice economically.  That's a fairly low price threshold and one that most real audio enthusiasts can easily hurdle by saving a little of their spare change for a short period. 

 

Also note that digital units are not relegated to a single loudspeaker model and type, but can be used for any loudspeaker type/model that the owner has on hand.  A lot of people I think miss this important point. Once you own a digital crossover, chances are that you'll keep it for a very long time to use on future projects--just like amplifiers and preamps. 

 

Chris

I'm looking at the Ashly Protea actives. I like them because they are made in the U.S. , but there are so many models. Can you recommend one that would work well with Jubilee's ? The Yamaha, I can't find at all.  Thanks... 

Link to comment
Share on other sites

The Protea 3.6SP and 4.8SP are the two Ashly units that I'd recommend.  If you're only biamping Jubs, then the 3.6SP is more than sufficient and will allow you to also add a couple of subwoofers full time in the future when you're ready.

 

Chris

Link to comment
Share on other sites

  • 2 weeks later...

One feature I find myself really liking is the AES input on the Xilica XD4080 (thinking it is made overseas though, but sounds fantastic).  Added one of these XMOS digital interfaces to my PC for $169 and it sounds fantastic too:  https://www.amazon.com/gp/product/B00PU4IXA0/ref=oh_aui_detailpage_o04_s00?ie=UTF8&psc=1

 

It seems worth while to ditch a preamp altogether if you are adding a digital XO (with digital input) and use a PC with software like Foobar2000 with a well implemented 32 bit volume control (and ASIO soundcard drivers if using Windows) and the interface I linked to.  Seriously, bypass the Windows audio stack (or use Linux), ran across these measurements the other day:  http://archimago.blogspot.com/2015/11/measurements-windows-10-audio-stack.html

 

On ‎3‎/‎25‎/‎2017 at 11:42 AM, Speed said:

I'm looking at the Ashly Protea actives. I like them because they are made in the U.S. , but there are so many models. Can you recommend one that would work well with Jubilee's ? The Yamaha, I can't find at all.  Thanks... 

 

Link to comment
Share on other sites

Interesting.  It appears that you are converting USB out from your computer (using something like foobar2000) to AES/EBU into the Xilica XD crossover, bypassing conversion to analog and using the 32-bit digital volume control in your converter as your volume control.  I assume that you're talking two channel only operation.  By far greatest volume of music available to us is two-channel, so I could see the advantages of switching between this converter and a typical AVP for multi-channel operation.

 

Chris

Link to comment
Share on other sites

The 32 bit volume control is foobar2000's, so it's done in software.  I wish there was a 5.1 solution that worked like this, but then I'd be tempted to tri-amp my other 3 channels, and the setup is already way too expensive (but it does perform and sound much more expensive).

 

Basically, there is a Music and a Watch TV button on my remote control.  I use input channels 3 and 4 on the Xilica XD4080 for my Emotiva XMC-1 when I want multi-channel, and input channels 1 and 2 for stereo only are tied to the AES digital input channel 1 in the Xilica's setup menu.

 

I hooked a serial cable up to the Xilica, and have a script on my home automation (HA) server that mutes channels 1 and 2 and unmutes 3 and 4 when I want to watch tv/movies.  Had to use a GC100-06 to turn my front amps on and off too as part of the macro: https://www.globalcache.com/products/gc-100/models1/
 

Part of the macro also starts and controls two windows programs for music or movies: Foobar2000 (music) and KODI (TV and movies).  I use MCE controller to start these programs, ensure they are in the foreground, etc: https://mcec.codeplex.com/

 

MCE  controller receives commands from my HA server over a TCP socket connection and is 100% customize-able.

 

My signal chain for stereo:

Foobar2000 (controls volume)->Dirac via ASIO4ALL driver->Dirac then uses the Xmos Xcore ASIO driver->outputs USB->XD4080 AES input 1->XD4080 output to subs and amps.  One could use the Xilica's input volume control to raise/lower volume.

 

6 hours ago, Chris A said:

Interesting.  It appears that you are converting USB out from your computer (using something like foobar2000) to AES/EBU into the Xilica XD crossover, bypassing conversion to analog and using the 32-bit digital volume control in your converter as your volume control.  I assume that you're talking two channel only operation.  By far greatest volume of music available to us is two-channel, so I could see the advantages of switching between this converter and a typical AVP for multi-channel operation.

 

Chris

 

  • Like 1
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...