DEQX tutorial - time aligning a sub with main speakers

[almikel]

Hi guys,

edit 5/11/14 - Dave has pointed out that aligning peaks in the step response is a flawed approach - more edits to follow...

edit 7/11/14 - great discussion below so I won't edit the initial tutorial based on step response. I'll add at the bottom the official DEQX approach using impulse response - although as Dave has pointed out, with real world drivers and rooms this can also have issues identifying the apropriate peaks to time align....

 

 

Original post starts here - but don't use the Step Response:

Jiri showed me how to do this and I thought I'd share...

 

It would be better if I did it absolutely from scratch, but the kids are in bed and I don't want to start "woop woops" now - hopefully the existing room measurements I've done will sufficiently show the process....

If people are confused with where I've started, I'll edit and go back to scratch

 

In my case I'm aligning stereo subs (really mid bass TD18s) with my mains (PSE-144s), but the process is the same if aligning any main speaker with 1 or more subs.

 

Key points:

• Use "Room Measurement" not "Speaker Measurement"
• once the room measurement is finished you must open the room measurement in the Viewer, not the EQ tab - this is because only the Viewer has the Step response option
• DEQX mutes the gain between measurements - I find it useful to set the gain the same each time so I can compare different measurements - note the gain setting and set it the same for each measurement on the volume slider!

note in the EQ tab below all but the FR is greyed out

 

Starting point:

• tri-amp setup with 2 way mains + stereo subs
• speaker correction of mains implemented (which should time align tweeter and mid)
• crossovers implemented

Goal

• time align the main speaker woofers and the sub/subs

Steps

• do a listening position "Room Measurement" of the drivers you wish to time align - in my case left sub, left woofer and right sub, right woofer - note I find it helpful to keep the gain set between measurements the same (DEQX mutes each time) - take note and set it the same for each measurement - in this case -34dB

• uncheck open in EQ tab and hit "Finish"

 

• now hit the "Viewer" button

go and find the measurement you just made in the folder structure

• all the measurements will be displayed - note on the far right all the dB gains are the same as I've already gain set - that's another tutorial on its own!
• more importantly note you have access to the FR, IR, GD and Step Response buttons at the top - we'll use Step Response - IR would prob work also
• in my case I'm going to align left sub and left woofer, then right sub and right woofer

• hit the Step Response button

• de-select all but left sub in the bottom frame
• 
• hit normalise to 0dB - this isn't mandatory but helps bring the scale closer for easier comparison
• 
• now de-select that measurement, select the next and normalise to 0dB - only normalise individual measurements otherwise you get an average - not critical for time alignment - critical for gain setting but that's another tutorial
• now select 2 measurements you want to time align
• see how the peaks are mis-aligned
• 
• now zoom in using the + icon - click and drag over the peaks in the graph
• 
• when you click on one peak, you'll see the time measurement in the lower right (I can't screen grab as when I move the mouse off the curve it disappears
• 
• in my case - left sub and left woofer, the mis-alignment was 2ms

1. click on the peak of the sub woofer and look at the time measurement - write it down
2. click on the peak of the woofer and look at the time measurement - write it down
3. calculate the difference - write it down
4. work out which one needs to be delayed - write it down

• go into Configure
• 
• select the driver you want to delay (in my case the sub)
• in the "Time/Level" tab adjust the delay (in my case 2mS)
• 
• save all to DEQX
• 
• then do another room measurement to check following the steps above
• hopefully it should look like this
• 
• don't worry that the magnitudes are different - this tute is about time alignment!
• repeat for right sub and woofer
• remember to "Save All to DEQX"

If this was helpful I'll do the same for "Gain Setting" for a tri-amp setup.

 

cheers

 

Mike

Edited November 6, 2014 by almikel

[davewantsmoore]

I'd use phase for this .... but I assume there is no phase view which is useful ?!

[almikel]

Hi Dave - correct - the DEQX phase graph isn't useful in this case

 

Mike

Edited November 3, 2014 by almikel

[Satanica]

Hey there, I assume for correcting you can alternatively adjust the "Delay" or "Group" setting in the IO Manager?

Although I understand it can't be changed with as much precision (1.3 or 1.4 ms increments) like you can when entering a value you into the "Delay /Offset" of the Speaker Configuration, as you've shown.

I'm using version 2.70 of the DEQX software.

 

Cheers.

[almikel]

Hi Satanica

Correct, but as you say you don't get as small increments

The same with gain adjust

Mike

[Phil G]

Nice work Mike to explain this as well as you have.  Using REW impulse responce does/shows exactly the same thing, which is how I was taught.  It is good to see the DEQX method as I've not tried it before.

[almikel]

Cheers Phil

I must have another play with REW since I put new treatment in

Mike

[Mika75]

I notice in the FR graph your subs don't seem to be rolling off, does the DEQX have a high pass filter available or is it disabled when running a sweep?

[almikel]

Hi Mika,

the DEQX has a low pass and high pass "limit" filters that can be applied

In this case there is no high pass on the sub, I've actually applied EQ to push the response down (my "subs" are stereo TD18s in 60 litre sealed boxes - they don't play low on their own).

 

It's an interesting question you ask regarding whether filters are disabled during sweeps which has caught me out on numerous occasions with the DEQX.

 

For speaker measurements, all filters and EQ are disabled as you're measuring raw speaker response (it does use a high pass to protect the tweater).

For room measurements, all filters and optionally room EQ is enabled.

 

If you're making setting changes and want to see the effect - always use room measurements!

I made the same mistake when doing the time alignment - doing a speaker measurement and the step response peaks never moved even with gross changes to delay - doh!!

 

As a pure fluke, if you look at the very first picture in this thread you can see the EQ I've applied in room

 

Mike

Edited November 4, 2014 by almikel

[almikel]

Dave's point is that the peak in step response between sub and woofer is not a good reference for time alignment as the freq content is different and the woofer should have a faster rise time compared to the sub - I agree with his position.

 

I've mucked around with the DEQX for a few hours trying to work out if the DEQX Phase graph is a better measure and usable for time alignment.

I can never get the DEQX phase response close and the step response gets further off (the step response peaks should be useful as a "rough" guide)

 

I've PM'd and emailed Alan from DEQX to get guidance on time aligning subs and mains using DEQX.

 

Stay tuned

 

Mike

[Satanica]

I can't find an article on the DEQX website called "Where In Time Is Your Sub-woofer" anymore so perhaps they have removed it.

I actually kept the HTML webpage and screen shots of it, PM if your interested.

[davewantsmoore]

Hi guys.

 

One problem with looking at step/impulse measurements of a real driver like this, is that people don't immediately visualise what the response is supposed to look like.    It's hard to see in the jumbled response of a real driver what is actually going on.

 

.....  So to remove that issue, I'm going to demonstrate this using theoretical drivers .... This removes all the "wiggles", and will let you see what the time alignment should look like --- this will show you that lining the responses up like Mike did in his demo - doesn't necessarily time align the drivers.

 

 

Here's the 'simulated' response of Mike's drivers.   The bass (green) has a low pass at 300hz.... and the "mid" (blue) has a high pass at 300hz and a low pass at 1300hz.    The red response is the sum of the two drivers.    I've used 4th order crossover slopes - an arbitrary choice.

 

 

Because these are "perfect" responses, they are already aligned in time....   when we look to the step response(s) of the drivers....   we see how the bass and mid align in time .... and we see the result (sum) in red.   A "perfect" step.

 

 

(Note the step is not "perfect" because it does not include much above 1300Hz ....  If we had of omitted the high pass filter on our theoretical mid driver - then the step would have been perfect, sharp and steep)

 

 

We can see that for the step responses to be "time aligned", that the peak of the higher frequency driver, does not go in the same place as the peak of the lower frequency driver.

Edited November 6, 2014 by davewantsmoore

[Satanica]

We can see that for the step responses to be "time aligned", that the peak of the higher frequency driver, does not go in the same place as the peak of the lower frequency driver.

 

Interesting Dave, the way it's described in the DEQX article I mentioned in a previous post, it says that to time align then it does go in the same place and this is consistent with what Mike has done.

[davewantsmoore]

Interesting Dave, the way it's described in the DEQX article I mentioned in a previous post, it says that to time align then it does go in the same place and this is consistent with what Mike has done.

 

I wouldn't like to comment without reading the article..... perhaps there is some context in the article which means there advice is not "general".... or is special in some other way.        However I don't really know what that could be.     You cannot align the drivers by matching the peaks of the step response (when the drivers are producing different frequencies).

 

 

If you had the same frequency content (for example two subwoofer, or two full range speakers, or two identical drivers, or whatever)    .... then yes - the step response charts of each driver/speaker would be the same --- and you would align the peaks of the step responses to align the drivers in time.

 

... but for a "woofer" and "tweeter"... no.

 

-----------------------------------------------------------------------------------------------------------

 

This should all make sense when we think about the period of the signals we are talking about.... sometimes an extreme example helps to illustrate  (I hope)

 

Consider a subwoofer which plays < 100 hz... and a super tweeter which plays > 10 khz ....  They are aligned in time.

 

The drivers begin moving at the same time.

• The 10khz signal takes 0.05 milliseconds to reach it's peak
• The 100hz signal takes 5 milliseconds to reach it's peak

 

This is of course an extreme (-ly silly) example  ... a super tweeter and a subwoofer ......   but if you visualise this on a chart  (like the one posted above >   time vs amplitude)..... the tweeters peak as come and gone .... before the woofer has even started to barely register movement.

 

 

The take home message being that for two drivers to be in time alignment .... the step response peak for a higher frequency driver will come before the peak of a lower frequency driver  (because the higher frequency signal is shorter).

[Satanica]

Makes sense to me Dave, so I understand why you've made your point.

Hey, look what I've found: http://www.audiosense.nl/pdf/where%20in%20time%20is%20your%20sub-woofer.pdf

And you can tell me if this matters "Use the first positive peak as the point of reference".

I assume that the first peak is not necessarily the largest peak when doing a room correction?

Edited November 6, 2014 by Satanica

[Newman]

At those frequencies you are better off with a measuring tape. And, if using a normal subwoofer close to the front wall (not your case I realise), measure to the wall behind the sub (not the sub's driver) because that is the effective average source of the sub's output.

[aechmea]

Makes sense to me Dave, so I understand why you've made your point.

Hey, look what I've found: http://www.audiosense.nl/pdf/where%20in%20time%20is%20your%20sub-woofer.pdf

And you can tell me if this matters "Use the first positive peak as the point of reference".

I assume that the first peak is not necessarily the largest peak when doing a room correction?

That's the way that I do it, using the "time" button rather than the "step" button, but I think that it is the same result??

 

OT a bit.

The ear isn't all that good with bass frequencies and I read somewhere that it takes a couple of wavelengths to pass by before the ear latches on to bass.  Just wondering if a msecond or two will make any difference.  Are we digging too deep into detail that is inaudible?

 

Even more OT.

Just checked mine and my favorite profile has the mains not delayed by quite enough (by 3ms) - oops!

 

Some more OT.

Mike, you are crossing to subs at 300Hz.  I presume that these aren't normal subs or are crossed there for a particular reason.

[Satanica]

That's the way that I do it, using the "time" button rather than the "step" button, but I think that it is the same result??

 

Yes me too, I actually didn't pick that up, whoops!

[almikel]

Alan sent me the DEQX article - "where in time is your subwoofer" and the recommended method from DEQX is to use the Time (Impulse response) tab and the first peak.

 

I wonder if there's any difference between the Step Response and the Impulse Response from a time alignment perspective...we're only concerned with the front edge of the step which will give the same curves as the impulse.

 

I think the key is that when time aligning a sub and a woofer (not a tweeter) - using the peaks provides good reference points and Aechmea's point about the brain taking time to process bass means the innacuracy in this method produces a reasonable result.

 

@@aechmea's question

In my case I don't have subs at the moment, just my TD18s sitting under my PSE-144s, so I push them down as low as I can get, but they cross to the horn at 300Hz

 

where in time is your sub-woofer.pdf

 

mike

[almikel]

@@Satanica,

just noticed your attachment was the same document

[aechmea]

At those frequencies you are better off with a measuring tape.  snip

 

 

 

Yes but no.  That might be OK for an analogue sub set up, however since the mains and the subs are both subject to all sorts of digital filters (in the DEQX and in my case, the sub itself), the required delay can be quite large.  In my case I have to delay the mains by 20ms = 20 feet = 6 metres (roughly) even though they are only a couple of feet different as measured by a tape measure.  I'm thinking that ~20ms would be audible - dunno though.  Something to try one day.

[aechmea]

 

In my case I don't have subs at the moment, just my TD18s sitting under my PSE-144s, so I push them down as low as I can get, but they cross to the horn at 300Hz

 

 

Ahh, OK, thanks.

[davewantsmoore]

The ear isn't all that good with bass frequencies and I read somewhere that it takes a couple of wavelengths to pass by before the ear latches on to bass.  Just wondering if a msecond or two will make any difference.  Are we digging too deep into detail that is inaudible?

 

As far as "a couple of milliseconds" goes.....     It strongly depends on what frequency.

 

 

300hz takes 3.3ms ........ hence a couple of ms can put you so far out of alignment - that you can have the drivers totally cancel each other  (if they were half a cycle out, ie.  1.6)

 

... but at 80hz... a cycle takes 12.5ms .... and a couple of ms would put you about 1/6 cycle out ..... this still creates ripples in the frequency response --- which can be audible

 

 

 

But... Yes.   As the Hz goes down, the amount of time which is relevant changes  (a lot).

 

 

Alan sent me the DEQX article - "where in time is your subwoofer" and the recommended method from DEQX is to use the Time (Impulse response) tab and the first peak.

 

I wonder if there's any difference between the Step Response and the Impulse Response from a time alignment perspective...we're only concerned with the front edge of the step which will give the same curves as the impulse.

 

Yes - Impulse is different to step.     Impulse lines up the peaks.    (1000 words follows)  

 

 

[davewantsmoore]

That might be OK for an analogue sub set up, however since the mains and the subs are both subject to all sorts of digital filters (in the DEQX and in my case, the sub itself), the required delay can be quite large.

 

All filters have a time delay.....    analog, digital, passive crossover.... all delay the speaker.

 

Yes --- If the two drivers have the same crossover filters   (eg.  sub = 80hz 4th order low pass / woofer = 80hz 4th order high pass) .... then they will have the same delay.

 

However in practice they don't always have the same filter slope / frequencies   (even if the resulting acoustic responses are properly summing crossover slopes).

 

 

In my case I have to delay the mains by 20ms = 20 feet = 6 metres (roughly) even though they are only a couple of feet different as measured by a tape measure.  I'm thinking that ~20ms would be audible - dunno though.  Something to try one day.

 

20ms....   What is your crossover frequency?      If we assume it's 80hz .... then you are nearly 2 full cycles behind.     If you were out by exactly half a cycle (or a multiple of) then playing both your mains and sub at 80hz .... would results in no sound (total cancellation).

Edited November 6, 2014 by davewantsmoore

[almikel]

so using the impulse response instead of the step response to line up the peaks will gve te correct alignment?