Alignment simulation on a passive PMC studio monitor of 2 2way sources using delay vs using All-pass filters   4 comments

 

So the idea for this post came to me during a short seminar in doing measurements with a FFT analyzer like in this case SMAART 7.4 that i give here in the Netherlands .

What if you have two 2 way loudspeakers with different x-over frequencies and different slopes .

A lot of 2 (or more) way systems i come across are aligned using delay only in order to match the timing on the high (mid in a 3 way cabinet) and low drivers in these systems .

But what if you have to combine them ? What will happen to the frequency response .

So i decided to compare 2 ways to align two 2 way systems and than combine them and see what happens .

The data in this post is a simulation on a single PMC passive studio monitor . I’m connecting the source signal to a APEX intelli-x processor . 4 of the outputs on this unit are routed back to a line mixer and the output is connected to the amp that drives the monitors . In this way i can simulate the 2 way loudspeakers on 1 monitor . The mic is a DPA 4007 and in this case is very close to the centre of the high and low speaker from the PMC monitor .

Again this is a simulation so not a real live situation !

(Colour is due to change on any trace you come across for visibility) .

First screenshot is the response of the PMC monitor with no processing applied what so ever . 

001

So this is the 1ste 2 way system .

It has a x-over point @ about 1200Hz with a 4rt order Linkwitz-Riley low-pass on the low driver and a 6th order Linkwitz-Riley high-pass on the high driver . You might take a look @ the x-over zone just past 1Khz . There is a rather large dip in the frequency response which would make me suspicious about the alignment @ the x-over zone .

002

By soloing both drivers and saving a trace from them i can compare both . Turns out a phase difference between both drivers exists . The high driver has a steeper slope @ the x-over area in the phase window compared to the low driver so the low driver has to be delayed to get the angles on the phase traces to match .

003

If i overlay the trace S1 start which is the SUM of them both you can see a bump in the phase trace around the  x-over area and along with it a dip in the frequency response . All are clear signs of a misalignment of both drivers . Keep in mind this is a simulation not a real live situation (in real live this might look a lot better but take it from me it will probably look a lot worse) .

004

So i add delay to the S1 low driver (0,2ms in this case) in order to get the down worth angle of the phase trace from the low driver to match with the angle of the phase trace from the high driver .

In the below screen shot the phase slopes on both drivers match but 1 of them needs polarity reversal  because a 180° phase difference exists from just past 500Hz up to about 2kHz .

005

So here’s a little side track for a moment . I sometimes get lost in discussions regarding polarity reversal . I my opinion polarity reversal has nothing to do with phase .

Phase in short is time over frequency . Polarity has nothing to do with time as you can see when you compare the IR window of the above screen shot to the IR window in the screen shot below . Both show the low driver @ a point in time just past the synchronization point (somewhere around 4,3ms because i delayed the low driver with 0,2ms to get the phase traces to match) .

S1 low 0,2ms delay polarity normal .

low 0.2ms

S1 low 0,2ms delay polarity reversed .

low pol rev 0.2ms

By reversing polarity on in this case the low driver the IR window shows the low driver @ the same point in time . So it doesn’t move in time . The only thing that changes is the peak of the screen shot above becomes a valley (? not sure if this is correct in English) in the screen shot below.

So by reversing polarity on the low driver the phase trace (pink) shows all frequencies shifted by 180° in order to match with the phase trace of the high driver (but it did not change the timing on the low driver as proved above) .

006

As you can see below here the low driver is now aligned to the high driver thus there will be a addition in the x-over area (max +6db) .

007

And indeed the SUM of them both shows addition in the x-over area .

008

So just to check if there is an improvement here are the S1 start vs the S1 aligned traces . The dip has gone away and a nice addition is clear around the x-over point .

009

That’s it for this moment on S1(system 1 just to be clear) .

Now lets move to the 2nd 2 way system S2 which has a different x-over point with different high & low pass filters compared to S1 .

So this is the system S2 strt no processing yet (except for the high & low pass filters). The x-over point is set @ 800Hz with a 4rt order LR high-pass on the high driver and a 2nd order LR low pass on the low driver .

Again a rather large dip is visible @ the x-over area from about 500Hz up to 2/3kHz indicating a misalignment between both “drivers”.

010

To get a better view of the problem I have to get the solo response of both sources . A 90° phase difference exists @ the x-over point . The phase trace of the low driver has a less steep angle down worth compared to the high driver indicating it needs to be delayed .

011

So if I overlay the SUM of both drivers you can see a bit of addition on the right side of the frequency scale indicating there is a phase difference between both drivers but still they are not aligned because the addition is about 3db instead of 6db which is to be expected because the phase difference between both drivers is about 90° .

 012

By adding 0,32ms delay to the low driver it aligns perfectly with the high driver .

013

Now the sum of both drivers shows a 6db addition around the x-over zone .

014

 For verification I have to check if the system has improved by adding the delay on the low driver . Below here are the traces from this system before and after the alignment . As you can see we have a large area with addition in the aligned system(brown trace) compared to the start of the measurement on this system (blue trace) .

015

And now i have 2 systems aligned properly which show about the same frequency response . They differ a bit but not to much . Now let’s see if i can combine both systems .

016

 Here you see both systems again with there phase data included . They do not match . If combined the low end will be cancelled between them provided they are matched in level because a phase difference exists between them ranging from 120° to 240° which will cause a cancelation anywhere from –infinity to 0db .

017

As said before the SUM of S1 and S2 causes the low end to cancel up to about 1K5 . From there on up the frequency scale both the high drivers still are not in phase but the difference doesn’t exceed the 120° border where both drivers would be cancelling again .

018

Now just for comparison the PMC no processing vs the S1 aligned .

As you can see there are minor differences in frequency response but a really large difference in the phase response due to the high and low-pass filters applied .

019

And again for comparison the PMC no processing vs the S2 aligned .

Again you can see there are minor differences in frequency response but a really large difference in the phase response due to the high and low-pass filters applied .

020

And the PMC no processing vs the SUM of S1 and S2 . Looks like this is a mismatch to me . I tried to find a solution for this problem without changing high and low-pass filters or using allpass filters but i could not find it whatever i tried .

021

Now i start from scratch again with both systems at there starting point so only the high and low-pass filters applied to S1 and S2 and still on the passive PMC monitor again .

Just as a reminder on the settings for S1 strt & S2 strt :

S1

Low-pass : LR4 @ 1200Hz on the low “driver”. High pass : LR6 @ 1200Hz on the high “driver”.

S2

Low-pass : LR2 @ 800Hz on the low “driver”. High pass : LR4 @ 800Hz on the high “driver”.

The challenge presented : combine both systems with as little change in frequency response as possible .

The delaying of drivers didn’t work in this configuration so now I’m going to try something different . I’m going to use my “secret weapon” being all-pass filters . First i have to align S1 not using delay but all-pass filters .

So below you see a comparison of S1 vs the original response of the PMC monitor . The dip in the frequency response of S1 due to a misalignment is back again . 

 022

Again you see a comparison of S2 vs the original response of the PMC monitor in this case. The dip in the frequency response of S2 due to a misalignment is also back again . 

 023

Again the solo traces of S1 so low and high traces no processing yet .

024

Again the SUM of both vs the solo traces of the high and low driver within S1 . The solo phase trace of the low driver has a less steep angle compared to the solo phase trace of the high driver so let’s see if i can match them using no delay but all-pass filters instead .

025

If i apply a 2nd order all-pass filter (ap2 from now on) on the low-driver @ 450Hz with a BW (bandwidth) of 2,543oct the low driver aligns perfectly with the high driver . So by adding the ap2 at this frequency i add group delay like any filter would do but THERE IS NO CHANGE IN THE FREQUENCY RESPONSE (the green-ish trace is underneath the brown trace in the magnitude window so you can not see it but you can see the change in the phase window) .

026

So if i SUM the low driver with a ap2 applied to it with the high driver i get a addition of +6db @ the x-over area .

027

And again i compare the start of S1 to the aligned system to see if S1 shows improvement via this way of aligning . And as you can see the frequency dip is gone .

028

 So now the real challenge starts .

I have to align S2 @ it’s x-over point in a similar way but i have to compare each trace to the solo traces of the aligned S1 system and keep the phase differences between them all between 0° to 120° to avoid cancelation .

So here we go .

First a comparison of the original trace no processing from the PMC monitor and S2 strt . Again the dip in the frequency response of S2 can be clearly seen @ the area from 250Hz up to about 2K5 due to misalignment of both drivers .

029

And again the low drivers phase trace indicates it needs to be altered to match the phase trace of the high driver . It needs more group delay to match with the trace from the high driver .

 030

So here’s the SUM again of both drivers with the solo traces of the high and low driver . Note that the SUMMED phase trace of the high and low driver takes “the middle road” between the high and low drivers traces . It’s the sum of level and phase differences between them .

031

So now there’s a extra challenge :

I have to align the S2 low driver with the S1 processed low driver . The LR4 low-pass filter @1200Hz and ap2 filter @450Hz bw. 2,543oct give the S1 low driver a certain amount of group delay and the S2 low driver has to get the same amount without changing the low-pas filter on it in order to keep the same frequency response .

032

So by applying a ap2 on S2’s low driver @433Hz bw. 2,543 oct it’s already matching up to about 500Hz but in the higher region of S2’s response it’s still not aligned properly .

033

Now it’s getting a bit more difficult to see so first an explanation of the traces you see in the screen shot below .

Brown S1 processed low driver (LR4 lp @1200Hz;ap2@450Hz bw.2,543oct ) .

Pink S2 processed low driver (LR2 lp @800Hz;ap2@433 bw.2,543oct) not there yet .

Green S2 processed low driver (LR2 lp @800Hz;ap2@433 bw.2,543oct) + a 2nd ap2 @ 3K bw.2oct .

034

So now the S2 low driver aligns with the S1 low driver (as you could see in the screenshot above) and the S1 high driver as you can see below here .

035

And now for the final challenge i have to align the S2 high driver to the S2 low driver and the S1 high driver without to much phase differences .

036

So here’s an explanation of the spaghetti in the screen shot below here .

Blue S1 high driver LR6 high-pass @ 1200Hz no further processing .

Dark green S2 high driver start LR4 @ 800Hz no further processing yet . 

Light green S2 low driver LR2 low-pass @800Hz;ap2@433Hz bw 2,543oct;ap2@3000Hz bw. 2oct.

The some kind of (wonderful) red trace is the S2 High driver LR4 high-pass;ap2@1200 bw. 0,573oct .

037

So now a comparison of the S1 high driver and the S2 high driver with processing (ap2 @ 1200Hz bw. 0,573oct) .

I could not get them to match perfectly but it’s close enough to not cause cancelation between both drivers and have a nice fat addition between them .

038

So now i have to check if the S2 high driver aligns with the processed S2 low driver .

There is a phase difference between them both around the x-over point where they are matched in level . The phase difference between both drivers increases going down the frequency scale but before causing cancelations the level from the S2 high driver is reduced enough to have a very little effect on the systems response .

039

So here’s the SUM of S2 vs the solo traces of S2 low and S2 high . There is an addition from about 600Hz going up the frequency scale and a very small cancelation just after the 500Hz point .

040

And now for the grand finale : lets see if both systems can be combined without the disastrous outcome when using delay .

So below you see the solo trace S1 system processed and the solo trace S2 system processed and S1 + S2 SUMMED together . Looks like we have a winner . Both systems give a addition of about +6db all over the frequency range in which both systems operate .

041

The same screen shot but with the SUM high lighted .

042

So now for the final comparison The PMC no processing trace @ +6db (blue) and the SUM of S1 and S2 .

There is a change in frequency response but you might put some eq on the the entire system which will work fine because the systems are matched as close as possible .

043

Equipment used in this simulation :

Apex intelli-x 4/8 io processor

SMAART 7.4

DPA 4007 mic

PMC studio monitor

Behringer line mixer for hooking up the processor outputs .

A load of cables and a shit load of time 😉

Greetz from the Netherlands .

4 responses to “Alignment simulation on a passive PMC studio monitor of 2 2way sources using delay vs using All-pass filters

Subscribe to comments with RSS.

  1. Very good job !!!

  2. https://timobeckmangeluid.files.wordpress.com/2012/12/017_thumb.jpg?w=1000&h=739
    S1 and S2 are 180° out of phase under 1.5k. You should change the polarity of S2 before summing. There would be some cancellation in HF, but not so disastrous as in https://timobeckmangeluid.files.wordpress.com/2012/12/018_thumb.jpg?w=1000&h=739

    • Hi primoz
      This blog post was to show that not every system can be combined with another system.
      If i reverse polarity on the low section of S2 s2 suffers from a mis-alignment between the highs and low’s so that’s not the way to go.
      In the 2nd part of the post i’m using 2nd order allpass filters and in the end the sum between both systems is ok. not pretty but ok so…..
      Thanx for reading my blog

Geef een reactie

Vul je gegevens in of klik op een icoon om in te loggen.

WordPress.com logo

Je reageert onder je WordPress.com account. Log uit / Bijwerken )

Twitter-afbeelding

Je reageert onder je Twitter account. Log uit / Bijwerken )

Facebook foto

Je reageert onder je Facebook account. Log uit / Bijwerken )

Google+ photo

Je reageert onder je Google+ account. Log uit / Bijwerken )

Verbinden met %s

%d bloggers liken dit: