Active Bi-Amping/Tri-Amping FAQ

[etc6849]

Here are some THD curves.  Parts are lower than before and some are slightly higher.

 

Measurements below are at MLP, no subs, no Dirac and both use the Ashley XR1001. 

Note for the distortion readings top one uses XMC-1 (HDMI) and bottom uses Oppo HA-1 preamp (USB DAC).  This could explain why parts of the curve have lower THD.  Got too tired to hook the Oppo back up, but it is probably the nicer preamp between the two wrt to noise (asynchronous USB DAC, fully balanced, no video circuits, no FM/AM, etc...).

 

IMD is also lower with no passives.  Someday I will hook up the Oppo HA-1, but I have it listed for sale on craigslist so hopefully no one buys it first!  Ok, They will in next post.  Hit the 2Mb limit again.

 

[etc6849]

Measurements below are at MLP, no subs, no Dirac and both use the Ashley XR1001. 

IMD is also lower with no passives.  No passive XO's uses XMC-1 processor via HDMI.  Passive XO's uses the very nice stereo Oppo HA-1.  Even with the noise advantages the Oppo has, IMD has lowered considerably for some of the side-bands with the passive XO's removed.

 

[Khornukopia]

1 hour ago, etc6849 said:

I still have lots of phase shifts

I don't think the vertical lines are phase shifts, but are breaks in the sloping line so it can fit within the space of the graph display. Maybe someone can explain to us how to "unwrap" it when we use REW, and read the phase graph accurately so it can be easier to interpret.

 

Interesting project you are doing, thanks for sharing.

 

 

[etc6849]

Any time you change shift by 180 degrees you have a shift in polarity, but the wrapping was confusing me apparently.  The attached is a much better way to view phase (I clicked controls in REW then unwrap).

 

I assume ideally, you'd want as flat sloping (very linear) curve as possible?  Mine has a drop at the tweeter and changes slope through the mid-range.  Not sure what to do to make mine more ideal as the Ashley XR1001 is an analog active crossover, but Dirac will clean up the phase plots.  Can't wait to do the other channel.

 

Measurements below are at MLP, no subs, no Dirac and both use the Ashley XR1001.

 

26 minutes ago, Khornukopia said:

I don't think the vertical lines are phase shifts, but are breaks in the sloping line so it can fit within the space of the graph display. Maybe someone can explain to us how to "unwrap" it when we use REW, and read the phase graph accurately so it can be easier to interpret.

 

Interesting project you are doing, thanks for sharing.

 

 

 

 

 

[Khornukopia]

7 minutes ago, etc6849 said:

I clicked controls in REW then unwrap

Well, that is going to make my occasional REW activities much more fun to look at.

[Khornukopia]

16 hours ago, DrWho said:

Side note: passive xovers work on the principal of adding series impedance to block frequencies, which is effectively increasing output impedance over certain frequencies - and therefore increasing distortion (even if the xover components are perfect, which they aren't).

I always suspected there was ANOTHER qualified reason my tri-amp Klipschorns sound better than my passive x-over Klipschorns in the same room.

[Chris A]

40 minutes ago, etc6849 said:

I assume ideally, you'd want as flat sloping (very linear) curve as possible?  Mine has a drop at the tweeter and changes slope through the mid-range.  Not sure what to do to make mine more ideal...

 

Look at excess group delay, which is visible after you hit the "Generate Minimum Phase" button under the "Controls" button dialog.  It will show you the delay of each channel in milliseconds.  For instance, a 1979 Cornwall with CT125 tweeter:

 

[Chris A]

2 hours ago, etc6849 said:

But mid and tweeter have a hiss that is audible at the MLP with nothing on but just the amp.

 

I hate to add a wire wound resistor back, but was thinking to put a 4.5ohm in series with each HF and MF driver...So would you live with a hiss or add a resistor?

I'd change the gain structure to reduce hiss first.  My initial guess is that your crossover may be the long pole in the tent in terms of noise, although it could also be the higher power amplifier channels.

 

Failing that, I'd probably add a resistor like that which you've mentioned.  The amplifier will actually like that and it will make the amplifier output feedback less sensitive to back EMF (making the amplifier look more like a current source...which is a good thing). 

 

Chris

[etc6849]

It's definitely the amp (e.g. with only the amp hooked up mid and high has hiss).

 

Ordered a selection of the Vishay/Mills 12W non-inductive wire wound resistors (expensive at over $4 a peice, but I didn't want to use a regular wire wound resistor like what radioshack has).  Vishay resistors:  http://www.vishay.com/docs/31801/mra.pdf

 

The Palladium brochure says this about the midrange:

 

"The inverted dome midrange driver operates from 500 Hz to 3.5 kHz and is a unique design, not found on any previous Klipsch product. The driver sensitivity of 106 dB allows minimal excursion at normal playback  volume. As cone excursion reduces, so does distortion. The 4.5-inch aluminum dome is driven by 3 high temperature N35H neodymium magnets, ensuring linearity under a range of listening volumes..."

 

And the spec sheet says mids are crossed at 3.2kHz?!?

 

However, I get less THD and less dip when using 4kHz. 

 

I am seeing an anomaly on the excess group delay plot at 4.76kHz.  Must be the room?  Otherwise, I think group delay looks decent at the MLP (about 2 meters)?

 

13 inches from high and mid drivers:

 

AT MLP:

 

2 hours ago, Chris A said:

I'd change the gain structure to reduce hiss first.  My initial guess is that your crossover may be the long pole in the tent in terms of noise, although it could also be the higher power amplifier channels.

 

Failing that, I'd probably add a resistor like that which you've mentioned.  The amplifier will actually like that and it will make the amplifier output feedback less sensitive to back EMF (making the amplifier look more like a current source...which is a good thing). 

 

Chris

 

[Chris A]

2 minutes ago, etc6849 said:

I am seeing an anomaly on the excess group delay plot at 4.76kHz.  Any idea what is causing this?

Yes.

 

From the excess group delay plot, you need to delay the tweeter about 1.2 ms and the midrange about 1 ms.  The plot is showing time misalignment due to the crossover filters--90 degrees of phase lag on the lower frequency driver per order of the crossover filter.  So if you're using a second order, that's a half wavelength of delay at the crossover frequency on the woofer, and the same on the midrange-tweeter.

[etc6849]

Not surprising I guess since the drivers are not physically aligned in the speaker?  I was going to let Dirac handle the alignment as my active XO is analog.

 

 

5 minutes ago, Chris A said:

Yes.

 

From the excess group delay plot, you need to delay the tweeter about 1.2 ms and the midrange about 1 ms.  The plot is showing time misalignment due to the crossover filters--90 degrees of phase lag on the lower frequency driver per order of the crossover filter.  So if you're using a second order, that's a half wavelength of delay at the crossover frequency on the woofer, and the same on the midrange-tweeter.

 

[Chris A]

Most of the phase delay is probably from the crossover filters in the Ashly analog crossover.  IIRC the drivers in the Palladium are within a couple of inches of being in line in the vertical plane.  So everything else is due to the crossover filters. 

 

The period of a 4700 Hz acoustic wave is 0.2 ms (i.e., 1/4700).  So that discontinuity that you're seeing in the excess GD curve at 4760 Hz is about a one wavelength time misalignment at the crossover between the midrange and tweeter drivers--leading to that big rise in the group delay at that frequency.

 

Chris

[etc6849]

Thanks again Chris for all your help.  I owe you and your wife dinner next time I'm in Arlington!

 

Make sense.  Could be that with the mic 13" from the midrange, the midrange was dominating and so the excess delay issue did not show up.  I think my system is going to sound fabulous when I'm done (well already sounded great before).  I'm half way finished removing the XO's from the right speaker.

 

1 hour ago, Chris A said:

Most of the phase delay is probably from the crossover filters in the Ashly analog crossover.  IIRC the drivers in the Palladium are within a couple of inches of being in line in the vertical plane.  So everything else is due to the crossover filters. 

 

The period of a 4700 Hz acoustic wave is 0.2 ms (i.e., 1/4700).  So that discontinuity that you're seeing in the excess GD curve at 4760 Hz is about a one wavelength time misalignment at the crossover between the midrange and tweeter drivers--leading to that big rise in the group delay at that frequency.

 

Chris

 

[Chris A]

Drop in when you're next in these parts...

[etc6849]

Got my right speaker all back together and I'm just wowed.  Despite the medium to low amount of hiss you can only hear when nothing is playing, things sound out of this world.

 

Real tri-amping combined with 1-2 years worth of acoustic improvements make a great combination.  I have never heard anything with this level of clarity and smoothness (some tracks before were a little harsh, but sound incredible tri-amped).  On some of the most dynamic tracks I have, strings and plucks are indistinguishable from real instruments where they only sounded kind of real before.  The same holds true for voices which are absolutely jaw dropping.

 

Although they probably have some of the better passive crossovers, I have no idea why Klipsch would spend years developing these speakers with such excellent drivers, then cripple them like they did.  Why can't manufacturers at the $20k price point provide an outboard box with an active crossover in it (bought my pairs used for $8000 and $6400 which is still a huge amount of money)?

 

I am on a slippery slope now as my center (which I don't have room for now) and rear speakers pale in comparison with their passive crossovers installed.  Spending another $2-3k to build more matching shelves, buy three more XR1001's and add more amps is worth it given the improvement that is apparently possible.  Just not sure about the space constraints.

[Dawson's Ridge]

10 hours ago, etc6849 said:

 Despite the medium to low amount of hiss you can only hear when nothing is playing...

 

What do mean by hiss, you can hear a hissing sound from your speakers at your listening position when no source is playing? Does it intensify as the volume control is increased? Is it at the same level as you switch sources or does it become worse on phono?

[Chris A]

12 hours ago, etc6849 said:

Got my right speaker all back together and I'm just wowed...Real tri-amping combined with 1-2 years worth of acoustic improvements make a great combination.  I have never heard anything with this level of clarity and smoothness (some tracks before were a little harsh, but sound incredible tri-amped).  On some of the most dynamic tracks I have, strings and plucks are indistinguishable from real instruments where they only sounded kind of real before.  The same holds true for voices which are absolutely jaw dropping.

That's also been my experience.  I've found that given a reasonably well-proportioned room to start with, the amount of acoustic treatments that are required are visually doable.  For more difficult rooms, more treatments are required.

 

Fully tri-amping (or bi-amping two-way loudspeakers) isn't something that can be easily described, I've found...but the difference is startling. 

 

12 hours ago, etc6849 said:

Although they probably have some of the better passive crossovers, I have no idea why Klipsch would spend years developing these speakers with such excellent drivers, then cripple them like they did.  Why can't manufacturers at the $20k price point provide an outboard box with an active crossover in it (bought my pairs used for $8000 and $6400 which is still a huge amount of money)?

 

At the time when the Palladiums hit the market ~8 years ago, most digital crossovers were still fairly expensive, and good analog crossovers didn't have something like Dirac to do the parametric EQ and time delays upstream.  Now there are many price points and effective product combinations to tri-amping. 

 

Now Klipsch is testing the waters with its brand new Fifteen, but I assume that they'll continue to expand into this domain with future offerings, perhaps with a replacement line(s) for the now-discontinued Palladium series.   It just makes a lot more sense.  It also takes a lot of time out of developing new loudspeaker models (i.e., not having to develop "one-size-fits-all" passive crossovers), and also makes them much more flexible to specific room installations (i.e., near corners, walls, or spaced away from room boundaries) as well as potential bundling with room correction software/firmware.

 

What you're going to find is that getting the frequency response flatter, even as flat as ±1 dB from the room's Schroeder frequency (~100 to 200 Hz) up to 10 kHz, the more dramatic the sound quality improvement will be; that's been my first-hand experience.  More available PEQs/shelf filters (i.e., more usable biquads per channel), the better the resulting sound.

 

Chris

[Chris A]

As digital crossover hardware continues to increase in capability and decrease in cost, I think that you're going to start seeing more FIR filtering used to correct not only the frequency response even better, it can also correct the phase response to yield much better impulse response (which is audible). 

 

Once that happens (i.e., economical FIR filtering for hi-fi), I think that you're going to start seeing more powered loudspeaker offerings (like the Klipsch Fifteen) but with their own integrated room correction software for using the loudspeakers in arrays (2.0, 2.1, 3.0, 3.1, 5.1, etc.). It's going to get cheap enough to provide that capability in integrable "building blocks" that preclude the need for putting everything into an AVP or AVR, and that can be tailored for the capabilities of the individual loudspeakers themselves.  All you're really going to need is a smartphone running an app, or dedicated remote to act as a calibration microphone and controller, and some sort of player/streaming device to feed the loudspeakers...either by wired connection or wirelessly.
 

Chris

[etc6849]

Those fifteens look interesting for non-techies and techies alike.  I bet they'll be very popular.

[Khornukopia]

On ‎1‎/‎21‎/‎2017 at 10:02 PM, etc6849 said:

three more XR1001's

Have you considered getting digital active crossovers sooner, rather than next time?