Using REW to Determine Time Delays Between Drivers

[Rudy81]

This is the Left Main spectrogram with NO EQ on the Oris. Don't have a reading at 3 feet with EQ on the Oris.

 

 

[Chris A]

Thanks Rudy. 

 

Chris

[Rudy81]

23 minutes ago, Chris A said:

Thanks Rudy. 

 

Chris

Hah....thank you sir for all the help.

[Rudy81]

Not to take away from this thread, but this is an observation after all this work.  After tuning the speakers, I felt that the sound was pretty darn good.  Very much like what I heard with Audyssey running BEFORE I changed things.  After all the tuning, I re-ran Audyssey Pro and it promised good results, which in fact, did sound very good.  2-ch and multi-channel was excellent.  After the new settings, when I turned Audyssey off, it sounded dull and lifeless.  I can't help but wonder if Audyssey does something to the sound when Audyssey is off to make it rather 'unappealing'.  Strange, but that is what I experienced today. 

 

EQ by itself sounded very good.  Audyssey after everything was tuned was also very good.  But then, when turning Audyssey off it sounded worse than before I ever ran Audyssey....interesting.

 

I am currently contemplating upgrading to an Anthem AVM 60 for one of the new Emotiva processors. I know Anthem uses ARC processing to set EQ and folks rave about the results.  Integra, what I currently use, has gotten away from Audyssey. Emotiva uses the newer Dirac equalization.  Although there seem to be lots of problems with the latest offerings from Emotiva.

[Chris A]

11 hours ago, Rudy81 said:

Not to take away from this thread, but this is an observation after all this work.  After tuning the speakers, I felt that the sound was pretty darn good.  Very much like what I heard with Audyssey running BEFORE I changed things.  After all the tuning, I re-ran Audyssey Pro and it promised good results, which in fact, did sound very good.  2-ch and multi-channel was excellent.  After the new settings, when I turned Audyssey off, it sounded dull and lifeless.  I can't help but wonder if Audyssey does something to the sound when Audyssey is off to make it rather 'unappealing'.  Strange, but that is what I experienced today. 

Could it be that the choices that you made with a "house curve" to make it more pleasant sounding...was the source of the difference?  In other words, you put in a house curve before running Audyssey, and Audyssey mostly took out the house curve.  If instead you put flatter response back into the setup before running Audyssey, perhaps you won't need Audyssey at all.  (At least that's been my experience.)

 

Toole's writings on "listening window" and house curves, etc.  are often misrepresented it seems.  When you read carefully, Toole recommends setting the loudspeakers to flat response to within 1.5 dB (using a smoothed response).  The downward curve that he most often refers to is the natural off-axis response of the loudspeaker, not putting in a downward curve to compensate for music recordings and movies mastered poorly (i.e., too hot).  This downward SPL off-axis is not something that is controllable through EQ (like through using DSP crossovers or Audyssey).  Rather, it's a characteristic of the loudspeakers themselves.  I don't recommend "house curves".  Those should be in the recordings themselves assuming that you've got full-range directivity (a BIG assumption--that's usually not true). 

 

What I have found is that the perception of loudspeakers being too "bright" is largely a function of their phase response.  If the HF drivers lead the LF drivers in phase by as little as 90 degrees (locally in the frequency spectrum) and 360 degrees (globally), you've got sharp sounding loudspeakers.  Other factors are confounded with this (such as full-range directivity and low modulation distortion), but in general, otherwise well-performing loudspeakers will sound sharp and harsh when their phase response is wrapping up.  The difference is heard in the transients. That topic is really the subject of this thread:

 

 

Chris

[Rudy81]

Well, the only issue is that I did not allow the REW suggestions to create a house curve type of response.  I always set the target to flat to get a baseline.  In my initial 'pre-Audyssey' audition the EQ settings sounded very good. I did not EQ above 10kHz because I only have 4 EQ positions per output channel on the Ashly xover (this has me looking at better xover options).  I began to think at that point that Audyssey is really not needed if you have an active xover with EQ capability as we do. 

 

I did, however, add a bass house curve to Audyssey. I have always added the house curve to the low end.   Audyssey attenuates bass too much in my experience.  I have huge amounts of bass available (6-15" main woofers and 6-18" sub drivers) and Audyssey always attenuates it way too much, so at the end of the Audyssey process I add a 3dB bass house curve from 20-500hz.

 

I guess the lack of a house curve when I turn Audyssey off is what gives that sensation of a lifeless sound. 

 

Concerning the very high range, Audyssey always allows 10khz and above to roll off.  Years ago while messing with Audyssey I used the house curve to raise the area above 10khz to intentionally make it flat. That was simply horrible!  It only took a few minutes to reach HF sound fatigue.  I have never tried to make the very high end flat again.

 

My take away from all this is that with a little care, effort and active EQ, you don't need Audyssey at all.  Using your measurement techniques and then adding speaker distances to a pre/pro should provide pretty good,  if not better, sound than what Audyssey does. The last few weeks I have been looking at the Anthem pre/pro since my ancient Integra is showing signs of video board problems.  If my understanding is correct, the Anthem uses ARC to 'fix' room issues but it only does so to the frequency range below 5khz.  Folks rave about the ARC results. 

 

Anyway, back to speaker alignment...

[Chris A]

11 minutes ago, Rudy81 said:

My take away from all this is that with a little care, effort and active EQ, you don't need Audyssey at all.  Using your measurement techniques and then adding speaker distances to a pre/pro should provide pretty good,  if not better, sound than what Audyssey does.

That's fortunate. I only use Audyssey to tell me what the 5.1 array channel delays should be (i.e., the inputs to the DSP crossover) since I use more than one DSP crossover (each crossover type having a little different insertion delay).  Then I transcribe those suggested channel delays to the PRE/PRO without using Audyssey, a method that I've used for many years now.

 

12 minutes ago, Rudy81 said:

The last few weeks I have been looking at the Anthem pre/pro since my ancient Integra is showing signs of video board problems.  If my understanding is correct, the Anthem uses ARC to 'fix' room issues but it only does so to the frequency range below 5khz.  Folks rave about the ARC results. 

You can measure what Audyssey and any other "room correction software" does using REW (except for any compression added by the application, if any).  I'm pretty sure that there isn't any more "magic" than EQ (including loudness-based PEQ) and adjusting channel delays.

 

Chris

[Rudy81]

Chis, how many EQ points can you set on each Xilica output channel?  That was one limitation I ran into yesterday.  The Ashly I use only allows 4 EQ points to be modified on each output channel. Rather limiting when REW EQ comes up with up to 7 or 8 at times. 

[Chris A]

22 minutes ago, Rudy81 said:

Chis, how many EQ points can you set on each Xilica output channel? 

Eight (8).

 

22 minutes ago, Rudy81 said:

The Ashly I use only allows 4 EQ points to be modified on each output channel.

That's the same for the EV Dx38.  I found that to be such a limitation that I don't use the Dx38 anymore.  I shifted to the Xilica about three years ago...and never looked back. 

 

The miniDSP 2x4 HD has 10 output PEQ filters/channel available--which is the crossover I use to bi-amp the two surround channels. I'm bi-amping the Jubs/TADs in the front corners, and tri-amping the center K-402-MEH/BMS 4592ND using the Xilica XP8080, leaving only one spare output channel on the Xilica.  The two subwoofers are EQed by the DSP front ends in the Crown amplifiers). 

 

Chris

[Rudy81]

looked at the Youtube on the Xilica XP4080.  Very nice!  If I get more into this self EQ thing I will pick one up for sure. I also like the fact that it does not seem to have a fan on it.  My Ashly has a very small fan that I can hear.

[rplace]

19 hours ago, rplace said:

21 hours ago, Chris A said:

Look at the broadening of the spectrogram horizontally around 300 Hz, and the smoothness of the decay (width of the spectrogram) with increasing frequency above that frequency.  Below that frequency, there is something occurring in the 100 Hz and 60 Hz regions, but that's typical for woofers to show that kind of behavior (i.e., resonances internally, in the box, or simply the 1/4 wavelength depth of horn point has been reached.  If he had crossed over at a higher frequency, you would have seen it to the right of the main impulse area as a broadening where the characteristic internal damping of the HF driver hands over to the LF driver (woofer).

 

 

Pointy part getting wider and shifting to blue?

If not too much trouble could you mark up that Spectrogram with paint or similar and post back? I'm guessing you are talking about the colors and curves. All I've really determined so far is I want a straight line vertically. I often see on my multiple sweeps with minor changes that the line will stay the same but the intensity or the colors or the shapes will change. I'm sure it means something but I don't know what. I'm a pretty visual learner. Show me once and I'll learn it....tell me 10 times and you might need an 11th.

 

As always thanks so much for being helpful and quick to respond.

 

This might have been overlooked yesterday. @Chris A any help/input? Curious what the colors and intensities tell us.

[Rudy81]

Oh oh.  My common sense side is raising its ugly head when it comes to what I did yesterday.  In my mains,  I delayed the Oris driver a slight bit based on the phase plots and the reference timing reported by REW. 

 

Problem is that the Oris Fostex driver voice coils are physically about 7" BEHIND the Eminence woofer voice coils!  Did I delay the wrong drivers?  I suspect my delay added to the Oris simply moved the phase back 180 degrees to match the woofer phase.

 

Any help would be appreciated.....Chris, this means you....

[Chris A]

37 minutes ago, rplace said:

Pointy part getting wider and shifting to blue?

If not too much trouble could you mark up that Spectrogram with paint or similar and post back? I'm guessing you are talking about the colors and curves. All I've really determined so far is I want a straight line vertically. I often see on my multiple sweeps with minor changes that the line will stay the same but the intensity or the colors or the shapes will change. I'm sure it means something but I don't know what.

I've made some annotations below in the spectrogram:

 

 

In general, you're looking for non-impulse shapes (the impulse is a column of colors toward the red end of the spectrum that is broadening as frequency decreases--with outward curving sides) and irregular areas to the left or right of the spectrogram impulse. 

 

1) An irregular left side usually means that you've got time misalignments in the loudspeaker--usually between ways, but also within a driver that is undergoing vibrational mode transformations. 

2) Irregular right-hand side (without corresponding left-hand side irregularities at the same frequency) usually means driver/box/horn resonances, or even perhaps room coupling (at low frequencies).  Areas on the right-hand side that look like they are repeating means that you've got bouncing energy or more than one part of the driver/loudspeaker emitting.  This includes mouth bounce of sound off of internal discontinuities (hard edges) and mouth bounce as sound reaches the end of the horn re-radiates from that point.

3) Widening of the impulse spike at certain frequencies without drop-outs in acoustic energy intervening--this almost always means resonances or couplings with horn/baffle/room, or in the case of woofers--internal resonances of the driver or within the box.

4) The peak energy curve (the dotted line) isn't smoothly moving down and to the right with largely vertical slope until it reaches about 200 Hz, below which the coupling to the nearfield and room boundary moves the peak energy time curve to the right due to the increased path length to those boundaries.  Any jumps to the left or right indicate some dynamic is occurring within the loudspeaker or driver/horn. 

 

Some more discussion, this time of a 1979 Cornwall spectrogram, below.

 

 

Chris

[Rudy81]

I'm guessing that rectangular area around 6k might be where the whizzer cone gets involved....yes?

[Chris A]

25 minutes ago, Rudy81 said:

Problem is that the Oris Fostex driver voice coils are physically about 7" BEHIND the Eminence woofer voice coils!  Did I delay the wrong drivers? 

What kind of crossover filter are you using between the Fostex and woofer?  The order of the crossover filter used will delay the lower frequency driver by 90 degrees for every order of the filter pair used--i.e., first order is 90 degrees, second order is 180 degrees, 4th order is 360 degrees (one full wavelength delay on the lower frequency driver), etc.  

 

Additionally, the acoustic center of the driver moves around vs. frequency.  That's why the phase is moving around vs. frequency within one driver assembly: the acoustic center is moving toward the listener or away from the listener.

 

29 minutes ago, Rudy81 said:

Any help would be appreciated.....Chris, this means you....

 

Chris

[Rudy81]

8 minutes ago, Chris A said:

What kind of crossover filter are you using between the Fostex and woofer?  The order of the crossover filter used will delay the lower frequency driver by 90 degrees for every order of the filter pair used--i.e., first order is 90 degrees, second order is 180 degrees, 4th order is 360 degrees (one full wavelength delay on the lower frequency driver), etc.  

 

Additionally, the acoustic center of the driver moves around vs. frequency.  That's why the phase is moving around vs. frequency within one driver assembly: the acoustic center is moving toward the listener or away from the listener.

 

 

Chris

Current crossover in use is L/R 24dB filter at 300hz.

REW reports bass bin farther away than Oris Fostex, but physically, that is not the case.  I measured this morning.

 

When I originally measured the timing, I took the xovers out and ran a sweep from 100hz-500hz on both drivers. I figured this would give me the actual distance measured by REW compared to the reference signal.  I really need to figure this out.

[Chris A]

24 minutes ago, Rudy81 said:

I'm guessing that rectangular area around 6k might be where the whizzer cone gets involved....yes?

Likely, but I'm not sure unless measuring the driver itself and looking at the diaphragm(s) and how they deform under external load using the fingers, etc., and looking where the voice coils are driving the diaphragm radially from the center.

 

My guess is that you're right guessing its either the whizzer cone surrounded by the voice coil driving force, or the whizzer cone is continuing to put out acoustic energy at higher frequencies due to the placement of the voice coil directly behind the whizzer cone, but the surrounding portion of the diaphragm is lagging the whizzer cone in phase.

 

21 minutes ago, Rudy81 said:

Current crossover in use is L/R 24dB filter at 300hz...REW reports bass bin farther away than Oris Fostex, but physically, that is not the case.  I measured this morning.

That's due to the delay of the crossover filters themselves, either passive filters, or the "IIR" filters in the Ashly. That delay is what you'd get for a two wavelength delay of the lower frequency drivers at 300 Hz--using 8th order crossover filters, assuming they were physically time aligned at their acoustic centers at the crossover frequency.

 

You have to use FIR filtering to avoid this phenomenon. That means using a digital input to the crossovers, eliminating the pre/pro, then running something like JRiver as a digital front end on a PC to implement the FIR filters.  That gets a little more complicated, but there are many people over on diyAudio using FIR filters. I haven't found the need for them yet due to the filters/drivers/horns that I'm using, so I haven't yet gone down that path.

 

Chris

[Rudy81]

So, in trying to set the proper delay, do I measure each driver independently with or without the xover set at 300hz? I had done the calculation with no xover set on the drivers and then added the delay and xover.  Did I do that incorrectly? One thing I do know is that the phase plots are on top of each other when I finished what I did.

[Chris A]

Calculations will get you close, but you have to measure both drivers/ways in the same sweep to really know what your delays are for each driver/horn with crossover filters in the circuit, and look at the resulting spectrogram and group delay plots to set the delay(s).

[Rudy81]

1 minute ago, Chris A said:

Calculations will get you close, but you have to measure both drivers/ways in the same sweep to really know what your delays are for each driver/horn, and look at the spectrogram and group delay plots.

Is that measurement with or without xover set, in this case to 300hz? Sorry for the confusion.