REW and acoustics - help needed please

[almikel]

1 hour ago, andyr said:

 

 But given:

• the Maggie bass driver has its resonance at about 40hz and is flat to around 1Khz.  So as the -6dB point from the 24dB L-R HP filter is 100hz - surely Maggie "bass driver response" is taken out of the picture?

24dB down by 50 Hz (1 octave below xover) is likely OK - depends on the response of the 40Hz resonance - I'd prefer a bit steeper (say 48dB - but that would require de-doing time alignment etc

 

1 hour ago, andyr said:

 

• and the 15" Dayton Ultimax sub driver extends flat to at least an octave above 100hz - so, likewise, it has no "driver response" screwing up the L-R filter response?

 

 

same as above - 24dB down is not that much if the FR gets wiggly 1 octave above Xover.

 

regardless - if you determine your Maggies need a 48dB Xover, then best to keep hi pass and lo pass symmetrical unless using the speaker box as part of the Xover (you're not).

 

Given you run mini DSP - others may disagree - but I would implement 48dB LR Xovers on subs and Maggies, and accept you need to re-do the time alignment.

 

My recollection is that Mr Spencer did your sub setup? - I wouldn't do anything without discussing with him first.

 

I've never done time alignment with REW, only DEQX - but it shouldn't be hard if you head down that path.

Note that a 48dB low pass on your subs will mean a bigger time delay will be required on the Maggies.

 

cheers

Mike

Edited April 1, 2017 by almikel
clarification, typos

[aechmea]

OT - a bit.

 

The bass panel and the mid range panel that I had in my previous Maggie MG3.6R speakers were wired out of phase and with a huge overlap in freq. with a passive XO from the factory.  I didn't do that when I went active and let the DEQX do the XO the 'normal' way.

[almikel]

5 minutes ago, aechmea said:

OT - a bit.

 

The bass panel and the mid range panel that I had in my previous Maggie MG3.6R speakers were wired out of phase and with a huge overlap in freq. with a passive XO from the factory.  I didn't do that when I went active and let the DEQX do the XO the 'normal' way.

I think Andy is full active also.

What Xover do you currently run between bass and mid panel (freq and slope)?

 

Mike

[davewantsmoore]

1 hour ago, andyr said:

• the Maggie bass driver has its resonance at about 40hz and is flat to around 1Khz.  So as the -6dB point from the 24dB L-R HP filter is 100hz - surely Maggie "bass driver response" is taken out of the picture?
• and the 15" Dayton Ultimax sub driver extends flat to at least an octave above 100hz - so, likewise, it has no "driver response" screwing

 

In theory, maybe.   You'll actually have to measure it, and see...   a driver won't actually be flat to Khz in room, as the transition between 1/8, 1/4, and 1/2 space is happening.

 

The crux of the matter is, that you cannot make any judgements about your room by what you've posted, as you cannot narrow down the cause of the issue.    Posting the individual driver responses (and their sum) will help confirm or deny what is happening .... and then move on  (to the next thing to investigate).

[davewantsmoore]

22 minutes ago, almikel said:

What Xover do you currently run between bass and mid panel (freq and slope)?

 

 

We will see once the data is posted.       (stumbling around in the dark otherwise)

[aechmea]

Yeah, I know Andy is active.  That's why I prefaced my post with "OT".  I was just giving an example of two drivers which were deliberately out of phase like Dave was saying "fighting each other".

 

The speakers that I have now are passive 3-way.  The factory XO is linear or series or something, whatever that means, so doesn't allow access to the drivers without serious axe surgery.

 

Dave.  The question was directed to me I think.

[almikel]

11 minutes ago, aechmea said:

Dave.  The question was directed to me I think.

yes it was - what Xover did you implement actively between bass and mid panel in that setup?

 

13 minutes ago, aechmea said:

Yeah, I know Andy is active.  That's why I prefaced my post with "OT".  I was just giving an example of two drivers which were deliberately out of phase like Dave was saying "fighting each other".

Cheers Aechmea

lots of complicated stuff going on with Xovers - especially well designed passives!

 

Much easier IMHO to use tools like DEQX with steep linear phase filters.

 

You now run passives, but have kept the DEQX - my turn to be OT - how does it sound?

Do you run DEQX FIRs over the top of the passive Maggies, or only Xover to the subs and time alignment?

 

cheers

Mike

[aechmea]

Sorry people, we won't be long.

 

Hi Mike.  

 

1.  Current setup with passive mains.  In DEQX-talk it is a "single amp with stereo subs" configuration.  So I do a calibration run and correct amplitude and group delay, with L and R done together as a pair.  Getting a good quasi-anechoic measurement is hard with the physical size of them.  Then I do a measure of the subs with the mic right in their throat and then calibrate them.  Then do a LinkwitzR between the mains and the subs.  I experiment here with different XO points and slopes, but of course the lower the freq the flatter the slope has to be.  ATM I like running the mains full range with the subs doing 20 - 80 as well but at a reduced level; ie 4 bass drivers scattered around the room.  I don't believe that I am missing anything much by not being fully active.  I still have enough amps to go active  should the situation arise; but I'm not touching $20k speakers.  I reckon that the passive XO is well done by Magnepan because I can't see any evidence of it in my graphs.

 

2.  The previous MG3.6R speakers were dead easy to active bi-amp and it was even encouraged in the manual.  Disconnect the box with the XO from the back and there were the banana sockets to the drivers.  In that situation I Xo'd at 300Hz with the maximum linear XO that the software would allow.  I can't remember what it was exactly but something like 96dB.  The DEQX software inverted the polarity on the midrange panel.  I vertically bi-amped.

 

[edit:  yes, I left out time alignment of subs and mains.  My delay is 22ms. ]

Edited April 1, 2017 by aechmea

[andyr]

1 hour ago, aechmea said:

 

The speakers that I have now are passive 3-way.  The factory XO is linear or series or something, whatever that means, so doesn't allow access to the drivers without serious axe surgery.

 

 

My understanding, a, is that the 3-way MG-20.7 has an all-6dB series XO.  So all 3 drivers are connected in phase.

 

20 minutes ago, aechmea said:

 

yes, I left out time alignment of subs and mains.  My delay is 22ms. ]

 

 

How did you arrive at 22mS, a?  Was this suggested by your DEQX?  I ask because I am running 9ms atm - and this includes the need to compensate for my subs being about 1.3m further away from my ears than my Maggie bass panels.

 

 

Regards,

Andy

[aechmea]

31 minutes ago, andyr said:

 

My understanding, a, is that the 3-way MG-20.7 has an all-6dB series XO.  So all 3 drivers are connected in phase.

 

 

How did you arrive at 22mS, a?  Was this suggested by your DEQX?  I ask because I am running 9ms atm - and this includes the need to compensate for my subs being about 1.3m further away from my ears than my Maggie bass panels.

 

 

Regards,

Andy

Yes Andy.  By visual inspection of the DEQX graphs showing what I think is normally called the impulse response.  I measure the room at "the chair".  I look at the mains graph and see at what time the front arrives and then do the same for the subs.  By comparing the 4 times of arrival and then by subtraction I calculate the delay that is required for each.  Having applied the delay, I remeasure to see that the arrivals now coincide.  I can change the delays in real time so I can twiddle the setting and listen to what happens on either side of the measured setting.

 

So not really related to physical distance differences but rather the time it takes to get through the dig filters plus the physical distance.  I have a lot more delay than distances alone would suggest.  My subs are actually closer to the chair than the mains by 2 or 3 feet = 2 or 3ms.  Dunno why the delays are so long; complicated slow filters?

 

[almikel]

1 hour ago, aechmea said:

So not really related to physical distance differences but rather the time it takes to get through the dig filters plus the physical distance.  I have a lot more delay than distances alone would suggest.  My subs are actually closer to the chair than the mains by 2 or 3 feet = 2 or 3ms.  Dunno why the delays are so long; complicated slow filters?

 

something Dave reminded me of recently - filters cause delay.

Whether passive analog (caps/coils) or active analog (built around op-amps) or digital (DSP) - they all have delay - and the same filter (eg a 2nd order IIR Linkwitz Riley low pass) regardless of implementation (passive/active/DSP) will have the same delay (assuming "typical" computational power is available for the DSP (ie way more than is actually needed)).

 

The delay is inherent in the filter, and nothing to do with the computation time due to DSP (which is fast enough to not matter).

The steeper the filter the more delay and the lower the filter the more delay.

 

A 48dB/octave analog filter using 2 op-amps will have nearly identical delay to the same filter implemented in DSP (ignoring passives, as who builds passive 48dB filters?).

 

What you can't implement with op-amps are linear phase FIR filters (we'll ignore analog Bessel filters which do get close to linear phase).

Steep linear phase filters at low frequencies have inherent delay and computational challenges with the number of "taps" required.

Hardware solutions like DEQX won't allow steep linear phase FIR Xovers at low frequencies because the delays get so large - due to inherent delay and computational limitations.

I had a discussion with Alan Langford on this topic - all the DSP manufacturers are targeting achieving low computational latency for low freq FIRs - but the inherent delay in the filter will remain based on physics (in this universe at least).

 

In the specific case of DEQX filters - if DEQX lets you implement the filter, the computational delay will be miniscule compared to the inherent delay of the filter. 

 

cheers

Mike

 

 

 

 

 

 

[andyr]

10 hours ago, aechmea said:

Yes Andy.  By visual inspection of the DEQX graphs showing what I think is normally called the impulse response.  I measure the room at "the chair".  I look at the mains graph and see at what time the front arrives and then do the same for the subs.  By comparing the 4 times of arrival and then by subtraction I calculate the delay that is required for each.  Having applied the delay, I remeasure to see that the arrivals now coincide.  I can change the delays in real time so I can twiddle the setting and listen to what happens on either side of the measured setting.

 

So not really related to physical distance differences but rather the time it takes to get through the dig filters plus the physical distance.  I have a lot more delay than distances alone would suggest.  My subs are actually closer to the chair than the mains by 2 or 3 feet = 2 or 3ms.  Dunno why the delays are so long; complicated slow filters?

 

 

That is a cool feature of the DEQX! 

 

My subs are 1.3m further away - my calcs said this would be a delay (for the Maggies) of 4mS.  What I did was put delay in 0.5 sec increments into 4 stored miniDSP configs and then flick through them, whilst listening to a CD of the Sheffield Drum Record.  The best location I found (for hearing the result) was with my ears in the centre between the Maggies, and in the same plane.

 

When I got to 9mS, the drum strikes seemed to firm up - what I think is happening is that with the correct delay, the fundamental (coming from the subs) and the harmonics (mainly from the Maggie bass panels) are in phase (arriving at the same time), at 9mS.  Anything less than 9mS - or greater - causes the harmonics to arrive at a different time to the fundamental - smearing the drum strike.  There were 3 others in the room listening - they all heard the same thing.

 

8 hours ago, almikel said:

something Dave reminded me of recently - filters cause delay.

 

cheers

Mike

 

 

Do you know the maths behind this delay, Dave?  IE, is it something like 24dB cause the LP to be delayed by 'X' mS ... 48dB by '2X'?

 

Andy

 

[davewantsmoore]

1 hour ago, andyr said:

Do you know the maths behind this delay, Dave?  IE, is it something like 24dB cause the LP to be delayed by 'X' mS ... 48dB by '2X'?

 

It's frequency dependant.

 

1 hour ago, andyr said:

What I did was put delay in 0.5 sec increments into 4 stored miniDSP configs and then flick through them, whilst listening to a CD of the Sheffield Drum Record.

 

If you change the delay between two drivers the sum together, it will cause peaks and dips in the frequency response as they sum together more/less than they are supposed to.

 

In my experience when left to chose which one someone (eg, me.) likes the most on certain program material.....  the result is typically poor.

 

 

[andyr]

55 minutes ago, davewantsmoore said:

 

It's frequency dependant.

 

Sure.  But let's take my example of a 100hz, 24dB L-R HP/LP XO.

 

Quote

 

If you change the delay between two drivers the sum together, it will cause peaks and dips in the frequency response as they sum together more/less than they are supposed to.

 

Of course.

 

Quote

 

In my experience when left to chose which one someone (eg, me.) likes the most on certain program material.....  the result is typically poor.

 

 

 

But I don't consider the drum tracks to be "any program material"?  It's specifically a series of drum strikes and I would've thought that if you adjust delay and listen, and find one delay setting that makes the drums sound more coherent ... then that's the right delay between subs and bass drivers!

 

Anyway, I've just completed some more sweeps - pics attached:

1. L sub plus main.
2. L sub by itself.
3. L main by itself.

 

Even the sub by itself doesn't appear to have any flat areas ... so I'm unable to see what useful info "Excess Phase" can supply?  Maybe the large Maggie panel areas confuse REW?

 

Regards,

Andy

 

Edited April 2, 2017 by andyr

[almikel]

2 hours ago, andyr said:

 

That is a cool feature of the DEQX! 

 

you can do the same in REW using the Impulse function, but I've never done it - it may require the loopback to set the timing reference to compare between sub and mains.

With mic left in the same spot look for the first peak on each then dial in the required delay.

In DEQX I cheat a bit and just align the sub with one of my 18 mid bass.

 

 

37 minutes ago, andyr said:

 

Even the sub by itself doesn't appear to have any flat areas ... so I'm unable to see what useful info "Excess Phase" can supply?  

 

Regards,

Andy

Other than telling you that your room has no minimum phase areas - so don't use EQ to get a smooth response.

Use multiple subs, speaker/chair position and treatment instead.

 

Mike

[andyr]

28 minutes ago, almikel said:

you can do the same in REW using the Impulse function, but I've never done it - it may require the loopback to set the timing reference to compare between sub and mains.

With mic left in the same spot look for the first peak on each then dial in the required delay.

 

 

Thanks, Mike - I will try that next time I fiddle with REW. 

 

28 minutes ago, almikel said:

 

Other than telling you that your room has no minimum phase areas - so don't use EQ to get a smooth response.

Use multiple subs, speaker/chair position and treatment instead.

 

Mike

 

OK.  Thanks.

 

Andy

 

[Jventer]

I have moved my speakers forward as well as away from side walls , measured again. Still have same issues. 

A bit frustrating when you don't really know what you are doing. 

I have now armed myself with some "tools" being some plywood, doonas  and sponge mattresses and will see what that is doing.

 

 

 

[ghost4man]

1 hour ago, Jventer said:

I have moved my speakers forward as well as away from side walls , measured again. Still have same issues. 

A bit frustrating when you don't really know what you are doing. 

I have now armed myself with some "tools" being some plywood, doonas  and sponge mattresses and will see what that is doing.

 

 

 

 

Hi mate,

 

That is one of the biggest issues that abound.

I think that a very common mistake is assuming that going active will automatically result in better sound.

So far that is something that I have not experienced.

In my particular case, I  have identified an issue with the mains power as well as an amp issue which I am going to resolve by

going down the path of a Holton Precision Audio diy build. It will give me 800 wpc and more importantly unconditionally stability

into a 2 ohm load which is what maggies present.

 

I am in a luckier position than what Andy is in. In a way, given the nature of his speakers, he has no passive XO system to fall back on.

 

So making the best of his active setup is perhaps his only option.

 

With my Tympani 1Ds, I can always go back to passive and if I want to keep the true ribbon tweeter then I could possibly roll off existing tweeter

with another cap. In other words I am not limited to going active or nothing. 

 

At the moment I am about 50/50 about the benefits of an active setup. You do learn along the way that it is not necessarily the be all end all

that some people will promote.

[almikel]

On 3/18/2017 at 2:24 PM, Jventer said:

As recommended by @davewantsmoore I have done more measurements.

So, from the seating position at 3.9 m I have gone forward until I got stuck right in front of the TV screen at.35m.

 

 

Hi Jventer,

just loopng back on this

reviewing all your graphs, that consistent suckout could be speaker related - hopefully not.
I know your speakers are large and heavy, but an outdoor measurement would be useful to see if that dip is still there...

or drag 1 into the middle of the room and do some on axis measurements at different distances...

or borrow another speaker from someone (you may have 1 already) and measure it in your room - can be bookshelf sized as only needs to get down to 100Hz or so.

What I'm trying to determine is if the dip is room related or speaker related.

 

cheers

Mike

[Jventer]

@almikel

Thanks for last post.

I have done some playing around with moving the speakers as well as putting parts of a sponge mattress behind the speaker. The best position I have been able to get is the green in the graph.

(Just as a refresher:  moved into a neglected rental property end of December which is being fixed with my guidance/assistance, some things to be paid for by the owner and lots to be paid for by the previous tenant and the property agent as they have allowed several tenants to move out without reporting or fixing. Agent is doing their best not to pay so etc. Quite a struggle, but rentals here are are hard to get and the rent was discounted due to the current state. Not an ideal place for HiFi, but the house serves the needs of our family for the next five years.)

In preparation of carpets to be washed and painting to be done everything as in lounge dining and study had to be moved out. Quite funny, lived with speakers in the kitchen for a week as the double garage is full . )

Carpets have been washed, but painting not done and don't know when it will be done so we proceeded to the next stage. 

After a detailed investigation of every shop in town, every website you can think of, trying every possible place to put the furniture the couch and chairs are almost back in their original positions, but the big coffeetable is gone.(Dont know what is coming in its place yet, but the old one was just too big and a negative effect on the sound.)

We have also bought lots of curtains which I am hoping to have fully installed in about a week's time. The room is already a lot darker - kids love it and less live.

 

Last weekend in the process of moving three rooms into the already full garage I had to unpack and repack and cull what was possible. In the process I "found" 3 of my 4 pairs of other speakers. I have also installed my old AVR.  Valve amps and Cary is packed away, so I will start news tests and see if that makes a difference.

 

Good idea re trying other speakers, will do.

 

 

[Jventer]

Follow up. I have done 2 quick sweeps with a pair of JPW booshelves. (The graph is not 100% as the JPW's were done to get 83 dB whilst before I was trying to hit 85dB.)

The JPW look totally different at about 150hz but then expexperience a drop/suckout  depending on position at about 150 to 180hz

Also note the difference between left and right speakers at about 160hz whilst the left is low the right drops double as low.

There is something going on between 150 and 200hz and I am still stumped.

 

[davewantsmoore]

On 4/2/2017 at 10:55 AM, andyr said:

Sure.  But let's take my example of a 100hz, 24dB L-R HP/LP XO.

 

I'm not sure if you misunderstand.

 

For a "100hz, 24dB L-R HP/LP XO" ... what is the delay?   The delay varies with frequency.

http://i109.photobucket.com/albums/n70/jeffbagby/TextbbokLR4.gif

 

On 4/2/2017 at 10:55 AM, andyr said:

But I don't consider the drum tracks to be "any program material"?  It's specifically a series of drum strikes and I would've thought that if you adjust delay and listen, and find one delay setting that makes the drums sound more coherent ... then that's the right delay between subs and bass drivers!

 

In my experience, even extremely experienced people do not end up with the right delay....   You just end up with the one that you think sounds less bad than the other ones you tried.     The reality is that we don't know what the drums are supposed to sound like, and even if we did our ears are simply not accurate enough objective measurement tools to accurately assess what is going on with.

 

The correctly designed alignment will result in less distortion over all program material, and really will "sound better" generally.

 

On 4/2/2017 at 10:55 AM, andyr said:

Even the sub by itself doesn't appear to have any flat areas ... so I'm unable to see what useful info "Excess Phase" can supply?  Maybe the large Maggie panel areas confuse REW?

 

The excess phase chart tells you if you have more phase rotation in your data, them would be expected based on the (assumed to be minimum-phase) amplitude response.

 

You do (have excess phase).    This is very useful information, because it poses the question "why?".

 

The first thing I would investigate is.    Are you using a "loopback timing reference" for your measurement?  .... and if no (or I don't know), then did you click the button  "Estimate IR delay"  (under the gear icon) .... and the remove the 'time of flight' delay from the data?

 

If you have not, then that will be responsible for (some of the) phase rotation.

[davewantsmoore]

On 4/2/2017 at 11:37 AM, almikel said:

Other than telling you that your room has no minimum phase areas - so don't use EQ to get a smooth response.

 

This isn't a conclusion you have enough information to make.

[davewantsmoore]

11 hours ago, Jventer said:

I am still stumped.

 

 

There isn't a simple answer to this.

 

In order to understand what is causing the dip, you need to take one of two approaches:

 

Calculate.    Calculate the different [path length delays for reflections from the walls, floor, ceiling, etc. etc.... and see what matches the frequency.

 

Practical.     Measure the speaker, and get the dip.     Change the position of something drastically, and see if the dip goes away or moves.     Use what you changed, to infer what the cause of the dip is.

 

On 4/2/2017 at 4:31 PM, ghost4man said:

I think that a very common mistake is assuming that going active will automatically result in better sound.

 

Absolutely.    Both acoustics, and filter design are very very complicated. 

[andyr]

1 hour ago, davewantsmoore said:

 

I'm not sure if you misunderstand.

 

For a "100hz, 24dB L-R HP/LP XO" ... what is the delay?   The delay varies with frequency.

http://i109.photobucket.com/albums/n70/jeffbagby/TextbbokLR4.gif

 

 

I probably did misunderstand, Dave. 

 

I can't seem to access your link (I'll try again when I get home) but I can understand that the delay varies with frequency.  So surely that means if you decide to delay the mains, say, by 6.5mS compared to the subs ... then this will only be correct at one particular frequency?  If this delay is 'correct' at 100hz - then the delay will be wrong when a 120hz note is playing?

 

1 hour ago, davewantsmoore said:

 

In my experience, even extremely experienced people do not end up with the right delay....   You just end up with the one that you think sounds less bad than the other ones you tried.     The reality is that we don't know what the drums are supposed to sound like, and even if we did our ears are simply not accurate enough objective measurement tools to accurately assess what is going on with.

 

The correctly designed alignment will result in less distortion over all program material, and really will "sound better" generally.

 

 

That I can understand.  So the issue becomes ... how can the mic & REW help us get to this "right delay"?

 

1 hour ago, davewantsmoore said:

 

You do (have excess phase).    This is very useful information, because it poses the question "why?".

 

The first thing I would investigate is.    Are you using a "loopback timing reference" for your measurement?  .... and if no (or I don't know), then did you click the button  "Estimate IR delay"  (under the gear icon) .... and the remove the 'time of flight' delay from the data?

 

 

No, was not using the 'loopback timing reference'.  I'll have to read the manual more and then use this, next time I do more REW work. 

 

1 hour ago, davewantsmoore said:

 

Both acoustics, and filter design are very very complicated.

 

 

I'm afraid i beg to differ ... in the case of Maggies.  Attached is Magnepan's XO schematic for what was once their top model - the Tympani IVa.  You will note:

• it's designed in 2 sections, to make it easy to use it in a 2-way active config (simply replace the external box with a 2-way active XO providing bass LP and mid HP).
• there are no Zobels - only filters.

This is because the drivers are almost entirely resistive.  The schematic shows:

• an 18dB bass LP
• a 12dB mid HP
• a 12dB mid LP, and
• a 12dB ribbon HP.

If you feed the component values into an XO simulation program like lspCAD, you can get the individual filter curves and work out their roll-off points - in fact, someone on the Maggie Forum has created an Excel spreadsheet which does this!  You then feed these roll-off points into your miniDSP unit.

 

 

Andy

 

Tymp IVa XO Schematic.pdf

Edited April 19, 2017 by andyr