REW Spectrogram interpretation

[TexDrone]

This measurement was taken from a tri-amped LaScala.  Do the tweeters need 2ms delay and horns 3.5ms? 

[Chris A]

Add 1.5 ms delay to your tweeter channel(a) 

 

Also, you should toggle the "Freq. Axis" button on the upper left side of the plot to return the vertical axis to full-scale so you can also see the woofer to midrange time alignment.  That time misalignment of the bass bin to the midrange needs to be added to the other two channels (midrange and tweeter delays are incremented by the same amount).

 

Chris

[Islander]

Not to be too picky, but all 3 drivers on the La Scala are horn-loaded:  the woofer, the squawker, and the tweeter.  The squawker and tweeter have shorter horns than the woofer, so they need to be delayed, in order for the 3 drivers to be time-aligned.  It’s a subtle improvement, but it’s worth doing, especially if you’re tri-amping your speakers.

[Chris A]

3 minutes ago, Islander said:

...It’s a subtle improvement...

Fairly major timbre shifts are created by all the time misalignments through passive crossovers in La Scala, Belle, Khorn and other fully horn-loaded loudspeakers, as well as soundstage issues--in my experience. 

 

It's not subtle when you finally get the time alignments dialed in while listening to truly hi-fi recordings.  Suddenly the timbre matches Jubilees...in my experience.

 

Chris

[Islander]

It’s entirely possible that my ear/brain combo is not as informed or as experienced as yours.  Also, I didn’t conduct precise before and after listening tests when I converted my La Scalas into their bi-amped time-aligned configuration.  This explains why the difference was only apparent to me with certain types of music, like with bass and treble lines that matched beat for beat.  Some of Leonard Cohen’s later music is like this.  It was not a day and night difference for me, but everyone’s experience will be different, of course.

 

Can you recommend any music that clearly shows how much better time-aligned La Scalas Sound?  And that sound great on Jubilees, too?  Thanks.

[Chris A]

36 minutes ago, Islander said:

It’s entirely possible that my ear/brain combo is not as informed or as experienced as yours.

What I'm talking about is not really close, i.e., you'll most surely hear it, too.  In a stock Belle that I had between two Jubilees, I couldn't tolerate the audible timbre shifts due to the passive crossover filters--and I'm confident you wouldn't, too.  Once I tri-amped and dialed in the Belle, the timbre problems magically disappeared. 

 

36 minutes ago, Islander said:

Some of Leonard Cohen’s music is like this.

I recommend listening to music that hasn't been "multitrack recorded", i.e., natural recordings made with acoustic instrumentation--all recorded in real time all together, with the musicians listening to each other in real time and adjusting their performances...and that were recorded in some venue other than a recording studio with their typically little "recording booths" using headphones, etc. to layer tracks together.  Those in-studio recording techniques apparently don't control for the inevitable phase shifts due to the recording and headphone playback chain, and other recording errors that make the original ambient sound field into corned beef hash.

 

In natural recordings, like those of orchestras, solo acoustic instrumentation (e.g., acoustic guitar recorded in a good auditorium, etc.) and unamplified voices (such as a capella choirs, etc.), these defects are largely avoided--and the effect on the presence or "you are there" subjective experience is dramatically heightened.  This is when the effects of time alignment really become extremely audible.

 

Chris

[Islander]

Thanks for the tips.  I mostly listen to rock, folk, blues, alternative, and pop, most of it recorded the way you described.  I'll have to look for some good live recordings, especially those without loud audience reactions.  I may have some like that in my collection already, maybe in the jazz section.

[Marvel]

Try some of the music by Chris Thile... he's done a bunch of recordings recorded with single mic's, or a pair. NPR's Tiny Desk concerts often have minimalist mic's and sound very natural. Thile does everything from Bach, jazz, bluegrass and more.

[Chris A]

1 hour ago, Marvel said:

Try some of the music by Chris Thile...

 

The two Nickel Creek multichannel SACDs are very, very good--especially in native DSD playback--which requires a DSD-capable disc player that supports DSD-over-HDMI (DOH) to a pre-pro that accepts native multichannel DSD files (i.e., USB or S/PDIF won't do this, nor will HDMI from a PC):

 

 

 

The "New Bluegrass" recordings represented by Chris Thile's group are quite engaging...the second album in particular is well done.  It sounds multitracked, but the multichannel (5.1) version generally puts one voice in each channel, and this is almost as good as a clean stage-recorded natural recording.

 

Chris

[TexDrone]

7 hours ago, Chris A said:

Add 1.5 ms delay to your tweeter channel(a) 

 

Also, you should toggle the "Freq. Axis" button on the upper left side of the plot to return the vertical axis to full-scale so you can also see the woofer to midrange time alignment.  That time misalignment of the bass bin to the midrange needs to be added to the other two channels (midrange and tweeter delays are incremented by the same amount).

 

Chris

 

This is what I got with 1.5ms on tweeter only.

 

 

 

 

 

This is 'zoomed' in to the lower range of the same image above, what type of delay do the horns need? 3.5-ish??

 

 

[Wirrunna]

Texdrone,

Looks like you have the Mid and Hi aligned, so I would start with 2ms for the bass.

 

Now, on the top right of Spectragraph REW screen there are a bunch of Icons, click on the Limits icon and then click the "Fit to data" field and that will adjust the Y axis to your data and fill the chart without the blue waste of space above 20k. From that you can better estimate the delay required for the Bass.

 

Another way is to click the GD field which hides in the row of clickable fields that starts with SPL & Phase across the top of the screen.

[TexDrone]

3 hours ago, Wirrunna said:

Texdrone,

Looks like you have the Mid and Hi aligned, so I would start with 2ms for the bass.

 

Now, on the top right of Spectragraph REW screen there are a bunch of Icons, click on the Limits icon and then click the "Fit to data" field and that will adjust the Y axis to your data and fill the chart without the blue waste of space above 20k. From that you can better estimate the delay required for the Bass.

 

Another way is to click the GD field which hides in the row of clickable fields that starts with SPL & Phase across the top of the screen.

 

So right now I have 1.5ms on the tweeters, 0ms on mids, and 0ms on woofers. Are you saying I should add 2ms to the woofers, and leave the tweeters and horns alone at 1.5ms and 0ms, respectively?

 

 

 

[Wirrunna]

Tex,

If you added 1.5ms to your tweeter channel to align the tweeter (hi) with the mid, then you add another 1.5ms  to the hi to give you 3ms and 1.5ms to the mid.

Then have a look at the plot, then try an extra 0.1ms. In other words step it out until it looks good.

[TexDrone]

21 hours ago, Wirrunna said:

Tex,

If you added 1.5ms to your tweeter channel to align the tweeter (hi) with the mid, then you add another 1.5ms  to the hi to give you 3ms and 1.5ms to the mid.

Then have a look at the plot, then try an extra 0.1ms. In other words step it out until it looks good.

 

Ok, this was taken with 3ms on tweeter and 1.5ms on mid. Including both the All SPL plot and the spectrogram. 

[Wirrunna]

Tex, it looks like you need to add another 1ms to get 4ms on the hi and 2.5ms on the mid.

I haven't used a Xilica, but it looks like you need some gain adjustment on 2 of the channels. Is there an output gain adjustment available? I would give the bass +2db, leave the mid at 0db and set the hi to -2.5db