Measurements and simulation with one subwoofer

[2234rew]

Hi @juicehifi

This question has nothing to do with the HDMI measurements thread so thought I would separate it here , even though the measurements below is from that.

This is a more general query about simulations.

I have one subwoofer handling bass below 80 Hz for left channel and 80.01 Hz for right channel.

AL measurement shows a left speaker dip around 65 Hz. This would be the left woofer.

But AL simulation also shows left channel dip here at 65 Hz, but at this frequency this should only be subwoofer (which has no dip at all anywhere below 80 Hz).

Should I ignore AL simulation for left speaker for anything below 80 Hz?

I'm mainly asking because I want to know if I should increase the max boost to correct for it, because boosting from 6dB max boost to 12dB max boost does indeed help this 65 Hz dip. But am I unnecessarily boosting here ? since the subwoofer below 80 Hz measures very accurately?

I thought a "left speaker" simulation would incorporate the subwoofer simulation too and the sub doesn't have any dips below 80Hz.

The measurement file is attached and lots of screenshots below too

Speaker setup:






Filtered measurement: the 65hz dip is only from the left speaker's woofer but this will be subwoofer domain anyway?:




Subwoofer simulation with max 6dB bass boost: (crossover is 80Hz) - looks clean up to crossover point




Left speaker with max 6dB boost shows the 65 Hz dip: but if this is subwoofer domain, should i correct it or leave it ?




Left speaker with max 12dB boost shows improvement in 65 Hz dip but is it irrelevant ?

 

[juicehifi]

You're crossing over at 80 Hz and the dip occurs at 65 Hz, so the dip occurs in the overlap region. I bet the two are in opposite phase dere.

 

[2234rew]

You're crossing over at 80 Hz and the dip occurs at 65 Hz, so the dip occurs in the overlap region. I bet the two are in opposite phase dere.

Thanks. I know phase and polarity are not exactly the same thing but I do know from the measurements it looks like they are in opposite polarity. Woofer is +ve polarity and tweeter and subwoofer are both in -ve polarity , looking at impulse response and step response measurements.

I had a feeling even before measuerments that internal wiring of tweeters was wrong. But was certain sub and woofers were correct.

There is a phase switch on back of sub i can try to change phase

I thought AL would automatically do phase correction though?

 

[2234rew]

You're crossing over at 80 Hz and the dip occurs at 65 Hz, so the dip occurs in the overlap region. I bet the two are in opposite phase dere.

Also, opposite phase only for left speaker but not right speaker?

Change phase on back of sub won't help there then would it

 

[2234rew]

You're crossing over at 80 Hz and the dip occurs at 65 Hz, so the dip occurs in the overlap region. I bet the two are in opposite phase dere.

I flipped the phase switch on back of the subwoofer (now it's negative phase) and it became easier to correct that left speaker 65 Hz hump ! I only need max boost to be 8dB now for 65Hz to look nicer, compared with 12dB max boost needed before.




I also swapped +ve and -ve terminals for tweeters, so now all impulse response measurements indicate all drivers are in positive polarity:

 

[2234rew]

You're crossing over at 80 Hz and the dip occurs at 65 Hz, so the dip occurs in the overlap region. I bet the two are in opposite phase dere.

Hi Bernt,

Also: which plot is best to use to see phase at 65 Hz for the left woofer and left sub?

Audiolense doesn't currently have a phase angle vs frequency plot does it?

Would be nice to see if phase is actually improved in measurement, when I flipped subwoofer phase switch

Measurements, before and after flipping subwoofer phase switch attached

 

[2234rew]

Hi @juicehifi ,

I found I could export measurements in .PCM format and import into REW - very cool feature of Audiolense! So I did some investigations

In REW I could clearly see that after flipping the subwoofer phase switch at back of sub, I can see now left woofer and subwoofer are approx IN PHASE at 65 Hz.

Whereas BEFORE, they were approx 90 deg out of phase at 65 Hz

Am I on the right path with my investigating here, as to why it seems it's easier now for AL to correct at 65 Hz freq resp dip?

Before subwoofer phase switch, approx 90 deg out of phase:




After subwoofer phase switch, now approx in phase:




And interestingly switching the tweeter + and - cable connections, the AMT tweeter is now in positive phase after XO point.

Before switching terminals, AMT tweeter was in negative phase after XO point





So using phase and impulse response and step response measurements, does everything now look wired correctly now?

 

[juicehifi]

Yes, in phase where it was out of phase before, but out of phase some 15-20 Hz lower down.

 

[2234rew]

Yes, in phase where it was out of phase before, but out of phase some 15-20 Hz lower down.

Thanks Bernt, if I reduce max boost to only 2dB to show simulation problem areas more clearly, I don't see any issues in the simulated response, below this 65 Hz dip.

So is that a good enough reason to not worry about the out of phase that's 20 Hz lower down?

Or is there something I can still try?

 

[juicehifi]

That is up to you to decide. I am just the tool provider.

 

[2234rew]

That is up to you to decide. I am just the tool provider.

Noted, I was moreso asking what other measurements you would recommend I look at, to see if this out of phase could be an issue.

The end result (sound quality) is of course my decision.

Your experience has helped me get this far :-)

 

[2234rew]

Hi @juicehifi

I run into an issue where "Generate correction filter" produces different simulation results, changing only the TTD subwindow values. Depending on if I generation filters on fresh startup of the app, or do it after a previous correction was done.

Here 3/3 is used on fresh startup of Audiolense 6.17:



Here 7.5/3 is used immediately after 3/3 used above (didn't close AL app):




Here 7.5/3 is used on fresh startup of Audiolense app:



And here 3/3 is used immediately after 7.5/3 was used above:





Different results, depending on sequence.

I assume the 'correct' is the correction done on startup of the app?

Quite important because I wouldn't want to miss the huge dip of the first

Measurement file attached

 

[juicehifi]

You will repeatable results, but you need to repeat the filtrering operation when you have change the correction procedure.

 

[jjazdk]

You will repeatable results, but you need to repeat the filtrering operation when you have change the correction procedure.

Interesting. It may be in the manual, but I haven't noticed. Perhaps an idea to make sure that is clearly written somewhere :-)

 

[2234rew]

You will repeatable results, but you need to repeat the filtrering operation when you have change the correction procedure.

When I launch AL fresh and i only hit the "generate correction filter" it automatically does the filtering operation.

But the next one (in same app session) it needs to be done manually?

Is there any reason why AL can't do filtering automatically like when first opening the app?

 

[juicehifi]

It will skip the filtering if the filtering has already been done. And if you have done the filtering, and then modify the correction procedure, you will get a correction that is blend of two different procedures.

 

[2234rew]

It will skip the filtering if the filtering has already been done. And if you have done the filtering, and then modify the correction procedure, you will get a correction that is blend of two different procedures.

Thanks Bernt, this is how I've observed things are currently happening. That's what I showed in screenshots above.

My question to you (or feature request) is can you change this behaviour, so that each time 'generate correction filter' button is hit, AL atomically runs filtering, regardless if correction procedure was changed or not ?

ie. change AL to behave the same way it behaves already on fresh app startup ?

to avoid any chance of this 'blend of two different procedures' ?

 

[juicehifi]

That's what the "filter measurement" button is for.