Jump to content

Active Bi-Amping/Tri-Amping FAQ


Chris A

Recommended Posts

10 hours ago, etc6849 said:

If you click on the link at the bottom of my post, the REW file is there.

Okay, looking at the REW file that you posted...some of the differences could be related to the exact levels of the dual tones.  The front P-39F looks to be about 7 dB lower SPL than the rear P-39F at the microphone position for the 2.4kHz and 8kHz tones, which shows perhaps the greatest differences in side band levels.  That could be significant.

Link to comment
Share on other sites

How closely do I need to match the EQ the passives are adding if I were to fully tri-amp?

 

Is this an OK process since my active crossover is analog (can only adjust gain, freq and damping near -3dB point, slope is fixed at 24dB/octave): 

http://ashly.com/wp-content/uploads/2015/08/xr-1001-2001-4001-crossovers-r01.pdf

 

1. Measure driver + passive network one at a time in REW (e.g. measure LF while muting MF and HF) with the mic pointed at the center of each driver (versus at the MLP like I've been doing)

2. Note slope and -3dB point of the filter from the REW plot (although I can't change slope in the Ashley XR1001)

3. Take my speakers apart and remove the crossovers, then make some jumpers to reach each of the connectors

4. Put settings into the Ashley crossover with gain very low, then raise gain while taking REW measurements until I reach a level close to what it was in 1.

5. Repeat for the other two drivers.

 

I was going to ignore any other EQ the passive XO's are doing as Dirac can easily handle +/-10dB dips and peaks and fix automatically.

 

My 600W per channel amp will be very large for the MF and HF drivers, but I don't think this should hurt anything?  However, since the horns are much more efficient than the LF drivers, I will have to adjust the MF and HF gain dials on the Ashley to lowest setting.

 

Also how should I be adjusting gain?  Using the overall freq response plot or by using the signal generator and sending a single sine wave? 

 

You note that my gain is off when looking at the dual test tone plots.  I ask this because clearly I mucked up level before as the room modes were causing some peaks for LF, I set LF to be lower than what it should be.  I also typically don't use any smoothing in REW.  Maybe I should use 1/48 smoothing when setting gains this time around?

 

10 hours 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.

 

Link to comment
Share on other sites

54 minutes ago, etc6849 said:

How closely do I need to match the EQ the passives are adding if I were to fully tri-amp?

 

You can re-EQ upstream of the crossover, assuming that you set the gains for the three channels using the active crossover and that there aren't any rough transitions in the interference bands between drivers, i.e., there aren't any SPL "bumps" in response in one or the other driver.  Those bumps in the crossover bands can be mitigated by moving the crossover frequencies slightly up or down.

 

The flatness of response can be +/- 3 or 4 dB for the measurements (assuming that you're just measuring, then returning the passives back to the mix).  If you intend to keep the active without the passives, then you'd want to EQ flatter, in my opinion.  The flatter, the better--for listening.

 

54 minutes ago, etc6849 said:

Is this an OK process since my active crossover is analog

Yes, but you don't have to remove the passives - only disconnect them.  I'd adjust the gains to be neutral, initially, for the woofer channel, then add the midrange, then the tweeter.  You'll be attenuating the midrange and tweeter channels, and not boosting gain on any channel if you do it that way.

 

Then you can transfer the settings to the other loudspeaker channel.

 

Chris

Link to comment
Share on other sites

1 hour ago, etc6849 said:

My 600W per channel amp will be very large for the MF and HF drivers, but I don't think this should hurt anything?

Just be careful with the tweeter channel.  They usually can't take more than a few watts. 

 

If you get a digital crossover at some point in the future, you can set its integrated fast limiter to protect the midrange and tweeter channels--and it works well.

Link to comment
Share on other sites

The manual for the Ashley has this interesting feature to fine tune the crossover bands if there are dips.  So should I just go off the speaker specs at first and adjust for crossover frequency for peaks only (probably also going to look at THD also) and use the response/damping factor control for dips?

 

"Response - adjusts the damping of the filter affecting the response shape at the crossover point. The dial calibrations refer to the amount of attenuation effected by the filter at the crossover frequency, i.e., a setting of 3dB means that the filter’s high-pass and low-pass outputs are each “rolled off 3dB at the crossover point”.  This describes Butterworth filter response, or a gentle 3dB peak at the crossover point where the two filter output signals overlap. To obtain a flat signal, or “Linkwitz-Riley” response, set the Response control to “6”.  To obtain a notch at the crossover point, turn down the response control past “6” to best suit your needs. The purpose of this control is to help offset the inaccuracies inherent in typical loudspeakers, helping you to achieve a flat system response.  NOTE: The Response control is not a “slope” control.  The Response control only affects filter response shape in the immediate vicinity of the crossover frequency; the ultimate crossover slope is a fixed parameter."

 

Also, forgot to ask this, but should I check polarity of the midrange and tweeter and what's the best way to do this without removing them (will only have access to the bottom of the speaker)?  Is it true that sometimes drivers are installed with polarity swapped (when compared to that of the woofer) to get the best possible  passive crossover region?

 

3 hours ago, Chris A said:

Those bumps in the crossover bands can be mitigated by moving the crossover frequencies slightly up or down.

 

Link to comment
Share on other sites

When you run your first sweep with all channels on, you'll see whether or not a driver needs its polarity reversed: you'll see a notch at the crossover frequency. 

 

As far as how to do it: I always start with the controls at neutral and go from there.  You can use Dirac or REW to set EQs.  It's difficult to know where to set the channel gains for the midrange and tweeter, but you'll see the relative gain mismatches once you run a sweep with all channels on.  Follow your instincts.  You can use lower amplifier gains (in the 70s dBSPL or perhaps the low 80s) at first, then later check at levels closer to 100 dBSPL once you feel that everything is falling into place.

 

Chris

Link to comment
Share on other sites

I don't.  Never had an issue.  If you were using tube amplifiers, you might want to do that because the likelihood of a tube failure resulting in DC on the output.

Link to comment
Share on other sites

Do you have a way to change the electrical signal without moving the speakers, and the mic? And do you have a way to guarantee exactly matched gains? 

 

Distortion is tricky because it's never flat versus amplitude. There's a lot of setup involved to exactly match things.

 

There's also going to be a lot of variation between speakers.

 

Does REW have a way to sweep distortion versus input signal? Seeing the curve helps account for minor gain differences.

Link to comment
Share on other sites

Lazy approach:

Current Setup2.jpg

 

Ashley Settings:

Used in Mono 3-way mode; on XR1001 serves the left, another serves the right.  All levels set to U on Ashely (XMC-1 sets overall channel level and Dirac on the PC is able to fine tune level is very small increments).  Response affects damping at the crossover region only and is not slope, all Response settings set to 6.   LF XO set to 500Hz, MF XO set to a little below 4kHz.

 

Ashley Part 1 cropped.jpg

Ashley Part 2 cropped.jpg

 

Dirac on my PC acts as a virtual soundcard that uses filters I measure via 9 measurements at the MLP.  It is setup to use a slightly downward tilted/sloping house curve.  I play all my media through a server in another room using a long Redmere HDMI cable from Monoprice.

 

2 hours ago, DrWho said:

What are the active xover frequencies you're using right now for the lazy approach? 

 

 

 

Link to comment
Share on other sites

I only removed the bars that would normally allow only one speaker wire to connect to LF, MF and HF at the same time.  Internal passive XO's are still in circuit though, they just aren't tied together with the external bars that would normally go under the binding posts.

 

LF passive XO (board #1 in speaker)

has hookups for individual woofers (top, mid and bottom) and an input.

 

MF/HF passive XO (board #2 in speaker)

has separate hookups for midrange and high frequency drivers.

 

Can't attach anymore pics right now due to 2MB limit.  Will post pics on next post.  May have to wait for someone else to post here though.

 

2 hours ago, DrWho said:

Oh, and how are the drivers wired inside? Are you just removing the binding posts? 

 

Link to comment
Share on other sites

Yes.  I do a frequency sweep using an Earthworks M30 mic and a Stienberg UR22 mkii USB audio interface as input to REW.  REW's output can be over HDMI to my processor, or out the UR22 mkii or even an Oppo HA-1 using it's USB input.

 

REW can do THD and individual harmonic distortion curves (thinking mine is set to show up through 10th harmonic) that are displayed after doing the above sweep measurement.

 

The Earthworks M30 mic I have should be +/-1dB or less, but I'm lazy so haven't sent it to them to calibrate it.  I plan to point the mic one meter at the center of the associated driver for the level matching using the Ashley dials (unless you have other advice).  I plan to tie the woofers together, although the P39f might be a 3.5 way design (LF passive XO has separate connections for each woofer).


Dirac will fine tune/EQ things at the MLP using the same mic, but I normally point it at the ceiling for the 9 Dirac measurements that are averaged at the main seating position (one seat calibration).

Image result for dirac one seat

 

 

3 hours ago, DrWho said:

Do you have a way to change the electrical signal without moving the speakers, and the mic? And do you have a way to guarantee exactly matched gains? 

 

Distortion is tricky because it's never flat versus amplitude. There's a lot of setup involved to exactly match things.

 

There's also going to be a lot of variation between speakers.

 

Does REW have a way to sweep distortion versus input signal? Seeing the curve helps account for minor gain differences.

 

Link to comment
Share on other sites

A couple thoughts:

  1. Different speakers may have different distortion characteristics.
  2. The microphone will be contributing some of its own distortion.
  3. The preamp will contribute some of its own distortion.
  4. Microphones respond to particle velocity and pressure - not just pressure. Room modes can play weird tricks in that regard. Notice how test tones sound very different as you move your head around? Even if the mic stays put - the speakers are in different positions so the location of peaks and nulls from the room modes will be different. This will affect distortion numbers greatly because the distortion products are calculated relative to the fundamental - and the room mode behavior also applies to the distortion frequencies, but it's going to be different because the wavelengths are different.

If you're trying to quantify the distortion benefit of the lazy triamping, then you'll really want to be using the same speaker with the same mic and everything at the exact same acoustic SPL. It's more work to do the A/B, but it looks like you're already really diving into this.

 

All that to say, I think there are too many moving variables to draw firm conclusions from what you've presented thus far. That's not to say it isn't real, but it's within the margin of the variables that have been moving around.

 

Now that said - amplifiers definitely create IMD products and they will go away when using a single tone. The IMD products only occur when there is a non-linearity in the system - and the magnitude of the distortion is a function of the relative impedances. A perfectly low output impedance amplifier will drive a perfect sine-wave into a diode. However, if you put a perfect resistor in series with that perfect amplifier, then the diode creates distortion. I can create a quick schematic and simulation to demonstrate the idea if that's not clear.

 

The back EMF from the speakers is working against the output impedance of the amp. When you have two tones going, the "bias point" for the higher tone is modulated by the lower tone. If the output impedance changes with the shift in bias (which it does ever so slightly for transistors), then that is a non-linearity that causes the back EMF to get rectified. I prefer to think of it as the output current modulates, but it's the same principal. This is why I don't like to have any impedance between my amplifiers and drivers. Any non-linearity in the system increases when the output impedance increases. It's just ohm's law at that point.

 

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). That's why I don't like passives.

Link to comment
Share on other sites

Thanks for taking a look.  What you say makes a lot of sense as I can hear bass differences in certain areas of my room (even though my bass performance is very very good).

 

The only plots where things are very close to the same are at the bottom of the last page where I compare lazy tri-amping using the Oppo HA-1 with the Ashley versus using a wye XLR adapter to feed the MF and HF signal directly from the Oppo HA-1 preout.  However, even those have a little less than 1-2dB loudness difference since I removed the Ashley from the circuit!

 

Both have no crossover to subs, use the Oppo HA-1 lefts channel via USB DAC input and I didn't move mic from MLP.

 

I have already took out the passive XO's for the left channel (no turning back now).  Hopefully I can get some measurements very close to the levels found below for the same speaker with/without sub.

 

Fool's tri-amping low volume (using XLR Wye adapter to feed MF and HF amps):

fools low.png

 

Versus the lazy tri-amping low volume (using Ashley XR1001):

Oppo lazy tri-amping.png

 

Fool's tri-amping loud volume:

fools loud.png

Oppo lazy tri-amping loud.png

 

1 hour ago, DrWho said:

 

If you're trying to quantify the distortion benefit of the lazy triamping, then you'll really want to be using the same speaker with the same mic and everything at the exact same acoustic SPL. It's more work to do the A/B, but it looks like you're already really diving into this....

...

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). That's why I don't like passives.

 

Link to comment
Share on other sites

@DrWho @Chris A

Have the left hooked up with no passive xo.

 

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.   Probably do it at the back of the amp and solder it on the speaker cable.

 

Was looking at these: http://www.parts-express.com/mills-45-ohm-12w-non-inductive-resistor--005-4.5

 

So would you live with a hiss or add a resistor?

 

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

phase shift with no passives.png

 

phase shifts.png

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...