Jump to content

Active Bi-Amping/Tri-Amping FAQ


Chris A

Recommended Posts

Thanks Chris - indeed, I plan to leave the tweeter and squawker connected to the passive crossover, so I'll just be bi-amping (for now). 

So, I'll go into the cabinet, disconnect the woofer from the crossover and also disconnect the crossover from the low-frequency binding post, then run a new cable from the LF binding post directly to the woofer. Was hoping to avoid doing that, but so be it.

And thanks for the welcome.

Nick

Link to comment
Share on other sites

  • 3 weeks later...

Changes made: 2-way stereo electronic crossover now in place, feeding my Marantz 8b (High and Mid range) and Linn LK85 (Low range). Crossover frequency is approx 800Hz using a 24dB/octave slope and Linkwitz-Riley style response.

 

What a spectacular difference to the output from my Heresy IIIs. Bass response is well improved (which I always felt was my systems greatest weakness when just running 'the 8b with the Heresy's and the primary driver behind making the change to bi-amping with solid-state on the LF), but the most surprising and even more pleasing difference is the improvement in separation of individual instruments, giving wonderful clarity right across the the entire frequency spectrum. I was expecting some improvement, but I thought my old set-up sounded pretty good and I am really quite amazed at what a difference this change has made.

 

The only downside is that my Ashly XR-1000 is getting a little long in the tooth and whereas, in the past, my system was completely silent with no active source, I now have a very feint hiss - imperceptible unless your ear is against the speaker, but it makes me wonder if its caps might need replacement and if so, what additional listening bliss might lie beyond that modification (or a switch to a new Marchand, or similar).

 

Anyone thinking about bi-amping Heresy IIIs would receive my absolute recommendation to go for it. And if you weren't thinking about it, maybe it's time to start!

 

Nick

  • Like 1
Link to comment
Share on other sites

  • 1 month later...

Great thread. 

 

After less than two days of running some mono block amps, I crossed my mains at 80 Hz.  From what I'm hearing there has to be lower IMD as things just sound very clear with great transients too.

 

I was seriously considering selling the mono block amps as my 7 channel amp sounded better in clarity on certain songs, but the mains were previously crossed at 60 Hz.  I realized this wasn't a fair comparison today when I finally remembered what Chris had said about IMD.  I guess I really did get used to hearing less IMD!

 

I want to bi-amp in the future, but crossing at 80Hz to dual subs has additional benefits too such as improving room response.

Link to comment
Share on other sites

I guess I really did get used to hearing less IMD!

 

It's insidious how Klipsch does that to you.

 

 

 

I want to bi-amp in the future, but crossing at 80Hz to dual subs has additional benefits too such as improving room response.

 

More subs scattered about helps LF coverage.  If I had the room, I'd add two more TH subs, placed at the anti-node positions along a wall.

 

I've found that PWK was a real pioneer on the subject of bass bin design to significantly reduce IMD--calling specific attention to the subject when it wasn't fashionable to do so.  All the sidebands of IM distortion are audible for the bass because those sidebands all show up at audible frequencies, especially up to 10th harmonic sidebands, which are not masked by human hearing. 

 

It's best to limit the passband of the lowest frequencies by device type in order to limit the pile up of IMD sidebands at the highest frequencies that each bass bin produces, but also it's important to realize that the source of those IMD components is the magnitude of cone motion. 

 

If you minimize the woofer motion in your fronts by crossing at a slightly higher frequency, you may be significantly reducing IMD that's audible at even higher frequencies in the fronts which are producing using their woofers.

 

Chris

Edited by Chris A
  • Like 1
Link to comment
Share on other sites

  • 2 months later...
On 12/13/2015 at 10:13 AM, Chris A said:

It's best to limit the passband of the lowest frequencies by device type in order to limit the pile up of IMD sidebands at the highest frequencies that each bass bin produces, but also it's important to realize that the source of those IMD components is the magnitude of cone motion. 

 

If you minimize the woofer motion in your fronts by crossing at a slightly higher frequency, you may be significantly reducing IMD that's audible at even higher frequencies in the fronts which are producing using their woofers.

 

Chris

 

Chris,

 

Haven't heard from you in a while.

 

As an experiment, I setup my P-39f's in a quasi tri-amp fashion where I limit the frequencies going to 3 separate amps that then feed my front two P-39f's, but left the passive XO's in place for now as the amps I'm using are plenty big enough even accounting for losses the passive XO's have.  Stereo immediately sounded a little better, especially very loud highs and the mid-range.

 

I then reran Dirac on my PC (using my PC via Dirac Live to do all DSP functions for level, timing and Room EQ and passing everything to my XMC-1 via HDMI.  This takes a lot of DSP load off my processor, which is a weak point compared to what my PC can do.  Thinking Dirac on my PC is doing 32 bit conversions, but hopefully they release a 64 bit version.  For the measurements I used a Steinberg UR22 Mk II and an Earthworks M30 mic.

 

At any rate, the results show about a 50% reduction around the crossover frequencies of the P-39f (specs say 500Hz and 3.2kHz).  I wanted to verify this without unhooking one of my front speakers, so I compared the front left to my rear left.  Both speakers are P-39f's, crossed at 80Hz to my subs using the XMC-1.  In fact, the rear channels are fed from one channel of an Emotiva XPR-2 amp, as is the mid-range of the front MF and HF drivers for each front P-39f.  If anything the rear channel should have the advantage as it is closer to the mic.

 

The difference is the front left has an Ashly XR1001 analog active crossover sending LF, MF and HF to the respective amps as shown below:

front.jpg

 

Even at 100dB and having the rear speaker only 5 feet from the mic and the front being 7 feet, there is a large decrease in distortion over all:

front left distortion.jpg

rear left distortion.jpg

 

I also tried to dig for an old measurement of the front left and found an old measurement for the front right with a similar setup using Dirac Live on my PC.  I'm not sure if SPL was calibrated as is with the new measurements (so not a fair distortion comparison), but both the old and new are near 89dB (much more realistic listening level):

new front right distortion.jpg

 

orig front right distortion.jpg

 

Some other plots of the front right plot from above showing off what Dirac Live software ($690 :( ) and lots of bass traps and absorption can do:

front right waterfall.jpg

front right impulse.jpg

front right FR and phase.jpg

 

PS: this is what distortion looks like below 76dB.  About .13% for the mid-range frequencies.  Dirac and tri-amping work well it seems:

Front right crossed to F212's subs and tri-amped, around 75dB for the mid-range.jpg

 

PPS:

Running a center phantom channel as an experiment didn't go well.  While Dirac Live fixed most of the comb filtering, there is still a large dip at 8kHz and increased distortion.  This was a neat experiment since using ASIO4ALL driver and an HDMI output to send the center channel signal to the processor (mine is physically configured in 4.1 and the subs share a mono output) and measure the response of a phantom center.  Going to put my center back when I switch to a 4k TV this year.  Needless to say the lower distortion gains in tri-amping does not offset the comb filtering and the increased THD of a phantom center speaker (presumably due to comb filtering).

Link to comment
Share on other sites

Hi Ellery!  It's nice to see you here.

 

I have to say that I've never seen a CSD waterfall plot as smooth as yours...that's amazing.

 

It's also refreshing to see measurements.  Talking about "how it sounds" without some sort of measurement...well...gets a little old after a while on the forums.  I appreciate the breath of fresh air.  Your measurements look like you've spent a while working on the setup.  I bet it sounds really good in there.

 

Once again, I think that Floyd Toole is spot on with his discussion of the "fundamental flaw of stereo" on page 151 of his book.  The 1.8kHz notch due to the distance between the two ears (psychoacoustics) adds to the 8 kHz notch that you describe, but you can't see that notch until you put stereo microphones into a dummy head and take measurements (just as Toole describes). 

 

In your text that you quoted from one of my posts where I was referring to modulation distortion.  Your measurements above are harmonic distortion. However, if you want to see modulation distortion, you can go to the "Spectrogram" view and play a dual-tone at something like 60 Hz and 210 Hz (a non-harmonic frequency on top of a fundamental woofer frequency, at pretty high SPL both above your subwoofer crossover frequency), then look at the sidebands.  Then change the crossover frequency to something higher than the lower frequency.  The strongest sidebands should pretty much disappear.

 

Chris

Link to comment
Share on other sites

Thanks Chris.  To get bass response and fast decay times took about 14 large bass traps along with 4" panels on most walls, and then on top of that Dirac is still needed to get it that perfectly flat (only +/-8dB without it).  The setup sounds very very good without Dirac though.  Bass that decays that fast is incredible to experience and hear.  Drums and bass strings are very very clean sounding as you'd expect with THD less than 1% at 100dB at the MLP.

 

Here's the REW file from before with a plots of the left speaker without crossing to the sub and without Dirac:  https://drive.google.com/open?id=0B1J0a4OV_WGLd1ROSjB6anhqZW8

 

It looks like I forgot to ask my question in my long post: why did THD reduce?  REW is doing a frequency sweep, e.g. one frequency at a time, so you'd think that THD would not change as much as it is?  Maybe the crossover points before were not 24dB/octave before like the active crossover has and now there is less frequency leakage into the adjacent drivers so THD for a given frequency is reduced?

 

I'd be interested to see how much less distortion (THD) your TAD drivers are.  I bet they measure incredible.  I agree IMD must have improved the most.  I promise to post those measurements this weekend.  I had forgotten about the spectrogram view -thanks!

 

On 1/15/2017 at 7:20 PM, Chris A said:

Hi Ellery!  It's nice to see you here.

 

I have to say that I've never seen a CSD waterfall plot as smooth as yours...that's amazing.

 

It's also refreshing to see measurements.  Talking about "how it sounds" without some sort of measurement...well...gets a little old after a while on the forums.  I appreciate the breath of fresh air.  Your measurements look like you've spent a while working on the setup.  I bet it sounds really good in there.

 

Once again, I think that Floyd Toole is spot on with his discussion of the "fundamental flaw of stereo" on page 151 of his book.  The 1.8kHz notch due to the distance between the two ears (psychoacoustics) adds to the 8 kHz notch that you describe, but you can't see that notch until you put stereo microphones into a dummy head and take measurements (just as Toole describes). 

 

In your text that you quoted from one of my posts where I was referring to modulation distortion.  Your measurements above are harmonic distortion. However, if you want to see modulation distortion, you can go to the "Spectrogram" view and play a dual-tone at something like 60 Hz and 210 Hz (a non-harmonic frequency on top of a fundamental woofer frequency, at pretty high SPL both above your subwoofer crossover frequency), then look at the sidebands.  Then change the crossover frequency to something higher than the lower frequency.  The strongest sidebands should pretty much disappear.

 

Chris

 

 

 

Link to comment
Share on other sites

9 hours ago, etc6849 said:

It looks like I forgot to ask my question in my long post: why did THD reduce?  REW is doing a frequency sweep, e.g. one frequency at a time, so you'd think that THD would not change as much as it is?

 

When I don't really know, I start enumerating the possible sources that I can think of:

 

1) measurement/microphone differences (probably not in this case)

2) room acoustics differences--assuming that the measurement gates are a little long, leading to more reflections in the measurements (likely)

3) breakup of back-EMF into separate channels from the respective passive crossover sections (highly likely)

4) reduction in complex reactance from the passives presenting a combined load to the output terminals of your amplifier(s), leading to a better amplifier load, including an effectively higher nonlinear damping ratio (not sure)

5) drivers within the passives-voice coils-cone structure forming harmonics with the impedance bounces (not sure)

 

I see no likely smoking guns.  What might be more instructive is if you take the passives out of the loop entirely, use REW's EQ facility to re-flatten the frequency response (i.e., with direct-coupled amplifiers to drivers) and measure again.  If the THD gets even lower, then that would point to passive network-driver-amplifier reactance interactions. 

 

You can also re-hookup the passives in the fronts and measure again without tri-amping, not moving the microphone.  That might eliminate (1) and (2) above as possible sources if nothing really changes.

 

Perhaps Kerry or Roy might give us clues as to the source, also. 

 

Chris

Link to comment
Share on other sites

9 hours ago, etc6849 said:

I'd be interested to see how much less distortion (THD) your TAD drivers are.

TAD TD-4002 on K-402 Horn @1 m 93 dB SPL phase.png

 

TAD TD-4002 on K-402 Horn @1 m 92 dB.png

 

Note that harmonic distortion isn't as strong of an indicator of listening fidelity as dual-tone side band measurements, i.e., higher order sidebands that are audible.  But these modulation distortion side bands correlate only with higher order harmonics.  These side band amplitudes are much lower in horn-loaded drivers than in direct radiating drivers (usually ~20-25 dB lower). 

 

Most direct-radiating driver enthusiasts miss that point entirely: that huge reduction in modulation distortion side band amplitudes with horn loading over direct radiating.

 

Chris

  • Like 1
Link to comment
Share on other sites

Here are some measurements.  They all show small improvements when comparing P-39f front left (tri-amped, Emotiva XPR-1 and XPR-2) to P-39f rear left (not tri-amped, Emotiva XPR-2 where one channel is shared with rear right).  I had to use the preset on the XMC-1, so no Dirac as my Corning Optical USB cable quit and I keep my PC in another room (where Dirac resides).

 

Overall, there are improvements in IMD, although not drastic.  I'm guessing the Emotiva XPR-2 amps are pretty decent wrt to IMD.  What is interesting is I left the rear and front left both crossed to the sub at 80Hz, but there is a still an improvement with tri-amping; could just be the XPR-1 (front left LF) is a better amp than the XPR-2 serving both rear speakers. 

 

I meant to do one not crossed to the subs, but forgot :(, but I have no doubt there is an improvement there too.  Seems like tri-amping is worth it to me; I can definitely hear an improvement.

 

60 and 200Hz FL.png60 and 200Hz RL.png

 

240 and 800Hz FL.png240 and 800Hz RL.png

2.4 and 8kHz FL.png2.4 and 8kHz RL.png

 

REW file is here:  https://drive.google.com/open?id=0B1J0a4OV_WGLLWdIYVVoN2hrZDQ

 

 

  • Like 1
Link to comment
Share on other sites

Could you put together a diagram showing the difference between A and B?

 

Is the improvement from "fool's biamping" with an active xover operating outside the passive xover?

(don't be offended by the term...at least I think it's an accepted term for using multiple amps with passive xovers instead of active xovers)

Link to comment
Share on other sites

Ellery, that's really interesting.  I'm captivated by what I see in those plots.  That's got to be FM distortion that's being induced through connection of the three ways of the passive networks driven by the single amplifier channel.  It's also much bigger than I'd ever have believed. 

 

I could believe AM distortion at low frequencies (due to driver compliance and magnetic motor/limited voice coil length nonlinearities), but that level of FM distortion at higher frequencies is much larger than I'd ever expect.  They're showing up as harmonics, too.

 

Wow.

 

Chris

Link to comment
Share on other sites

I prefer the term lazy tri-amping ;) but this is really quad-amping as I'm still crossing to my dual subs using my processor.

 

If you click on the link at the bottom of my post, the REW file is there.  The diagrams are in order with tri-amping on top, immediately following the normal single amp measurement.  All measurements were crossed to my subs at 80Hz, so maybe I should call this lazy quad-amping?

 

I took all measurements from the main listening position, and just measured my rear speakers (which are also P-39f's I bought from ebay).  I am using REW to generate two test tones, then using REW to capture a single spectrogram of the acoustic output at the MLP.

 

For the first two, we see side bands have decreased by about 15dB when using lazy tri-amping + sub for a dual band 60Hz + 200Hz signal.

 

For the next two measurements at 240Hz and 800Hz, only one side band is decreased about 12dB.

 

Lastly for the last two 2400Hz and 8000Hz dual band measurements, side bands near 5kHz decrease significantly when tri-amping.  In fact, one side band drops 20dB and the other about 18dB.

 

Not sure if this is fools tri-amping as I'm not sending the full signal to any amp, but I agree it is not full tri-amping either.  I'm using two Ashley XR1001 active crossovers, plus the crossover in my XMC-1 processor for the subs to limit frequencies to the respective amps.  I let Dirac fine tune everything on my PC as the software is doing 32 bit math and also has filters to match all sampling frequencies my HDMI output can do (time aligns everything too as shown a few posts up).

 

I am game for trying to remove my crossovers next, but in your professional opinion, would I see a benefit besides improving the damping factor by wiring amps directly to the drivers?  Have to be honest it is scary to rip open speakers I paid thousands of dollars for and risk over powering the drivers if I'm not going to see a medium amount of difference, but if it's going to make it sound better...

 

1 hour ago, DrWho said:

Could you put together a diagram showing the difference between A and B?

 

Is the improvement from "fool's biamping" with an active xover operating outside the passive xover?

(don't be offended by the term...at least I think it's an accepted term for using multiple amps with passive xovers instead of active xovers)

 

  • Like 1
Link to comment
Share on other sites

I'm not so sure that is a good idea breaking into the crossovers to bypass them entirely.  You'd have to re-balance the gains and add the EQ that the passives are providing presently, as well as provide crossover filters.  That will take some time to rebalance. 

 

But from what I'm seeing in your plots, I'd be willing to bet that you'd see even lower harmonics and modulation sidebands if the passive reactances were not in the loop, but probably not a lot less. 

 

Your guess is as good as mine, however, because I'm really not believing what I'm currently seeing above.  Those plots are saying that the mass effect back-EMF is circulating within the local passive network and drivers, leading to much higher FM and harmonic distortion, if I understand them correctly.

 

Has anyone else seen this sort of thing before?

 

Chris

Link to comment
Share on other sites

Hopefully @DrWho responds, but I'm guessing to see if it's back EMF causing this, I'd need to do true fool's bi-amping where each amp gets the entire signal, while keeping the three passive crossovers isolated.  If this plot shows the same side band improvement, I would have a decent verification that it was indeed from back-EMF from the drivers? 

 

I have to believe the large amps I am using have an adequate damping factor for the each driver + passive network, but maybe not it seems?

 

11 minutes ago, Chris A said:

I'm not so sure that is a good idea breaking into the crossovers to bypass them entirely.  You'd have to re-balance the gains and add the EQ that the passives are providing presently, as well as provide crossover filters.  That will take some time to rebalance. 

 

But from what I'm seeing in your plots, I'd be willing to bet that you'd see even lower harmonics and modulation sidebands if the passive reactances were not in the loop, but probably not a lot less. 

 

Your guess is as good as mine, however, because I'm really not believing what I'm currently seeing above.  Those plots are saying that the mass effect back-EMF is circulating within the local passive network and drivers, leading to much higher FM and harmonic distortion, if I understand them correctly.

 

Has anyone else seen this sort of thing before?

 

Chris

 

Link to comment
Share on other sites

16 minutes ago, etc6849 said:

I'm guessing to see if it's back EMF causing this, I'd need to do true fool's bi-amping where each amp gets the entire signal, while keeping the three passive crossovers isolated.

I think that you'd need to bypass the passives entirely and direct couple the amplifier channels--a.k.a. true tri-amping.  If you're using amplifiers with low output impedance/high damping factor, (as you clearly are) you're probably going to minimize back-EMF and room-loudspeaker  microphonics issues.  If I understand what you've done, you've separated the passive crossover sections by driver presently. 

 

16 minutes ago, etc6849 said:

If this plot shows the same side band improvement, I would have a decent verification that it was indeed from back-EMF from the drivers? 

Yes.

 

If you were, say, to find an amplifier that had an output impedance of something like 40 ohms, then the amplifier would simply ignore all the back-EMF generated by the loudspeaker+passives, but you'd be increasing the levels of "loudspeaker microphonics" on the decays that you'd probably hear.  It would probably sound like a "super SET" configuration with lots of apparent depth of the soundstage image (i.e., added room-based microphonics reverberation on decays). 

 

But there aren't too many amplifiers like that out there.  The only ones that I'm aware of are the F1, F1J, F2, and F2J from First Watt (Nelson Pass):  http://www.firstwatt.com/prod.html

 

It's also possible to build an "improved Howland current pump" network of resistors with a amplifier of sufficient reserve power capabilities to approximate a very high output impedance amplifier.

Link to comment
Share on other sites

You'd also lose the damping on your woofers if they were coupled to a very high output impedance amplifier.  That wouldn't sound very good.  The high output-z amplifiers would need to be coupled only to the midrange and tweeter drivers/networks.

Link to comment
Share on other sites

So I swapped in an Oppo HA-1 headphone amplifier/USB DAC.  No subwoofer outputs now so I am only quasi tri-amping, but there are still the side bands (used a digital SPL meter to re-calibrate SPL in REW).  I notice that decreasing the volume just 12dB, almost gets rid of the side bands (results in about a 26dB decrease in side bands).

 

Note 87dB peak overlayed onto the same dual band signal, but with a lower preamp volume.  This makes me think something is going on with the speaker itself (as you are saying) at greater than 75dB acoustic output at about 7 feet (e.g. MLP).

 

test 2.4k and 8k.png

 

test 1.pngtest 2.png

 

1 hour ago, Chris A said:

I think that you'd need to bypass the passives entirely and direct couple the amplifier channels--a.k.a. true tri-amping.  If you're using amplifiers with low output impedance/high damping factor, (as you clearly are) you're probably going to minimize back-EMF and room-loudspeaker  microphonics issues.  If I understand what you've done, you've separated the passive crossover sections by driver presently. 

 

Yes.

 

If you were, say, to find an amplifier that had an output impedance of something like 40 ohms, then the amplifier would simply ignore all the back-EMF generated by the loudspeaker+passives, but you'd be increasing the levels of "loudspeaker microphonics" on the decays that you'd probably hear.  It would probably sound like a "super SET" configuration with lots of apparent depth of the soundstage image (i.e., added room-based microphonics reverberation on decays). 

 

But there aren't too many amplifiers like that out there.  The only ones that I'm aware of are the F1, F1J, F2, and F2J from First Watt (Nelson Pass):  http://www.firstwatt.com/prod.html

 

It's also possible to build an "improved Howland current pump" network of resistors with a amplifier of sufficient reserve power capabilities to approximate a very high output impedance amplifier.

 

Link to comment
Share on other sites

Same test as above (Oppo HA-1 feeding tri-amped P-39f and no subs),  but replaced the Ashley active crossover with a XLR Wye adapter.  Now this would be fool's tri-amping ;)  Appears to be a lot more side bands (and at higher amplitude) versus having the Ashley XR1001 limit frequencies to the passive crossovers.

 

fools loud.png

fools low.png

Link to comment
Share on other sites

Wow...

 

Now I'm wondering if the Indy guys have even tried to test for this.  It shouldn't be happening, IMHO, but there it is.  That's probably audible. 

 

Modulation distortion levels are strongly level (amplitude) dependent: that's backed up by PWK's writings and Klippel's white papers and test procedures.

 

Perhaps the source of that particular type of modulation distortion within the driver/passive network/amplifier electrical circuit plus the drivers themselves (AM and FM distortion) has been identified by Klippel.  From what I understand, the harmonic and amplitude modulation (AM) distortion shouldn't go away simply because you've separated the passive circuitry channels.  That's extremely interesting.  It says that everything that you're seeing is FM distortion, and that it is a function of the electrical circuitry.

 

Chris

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...