S&V Experiment: one 15, two 12, or four 8 inch subs

[terryj]

Ok, took me a minute to do just for illustration, takes longer to post it than to do it!

Here is the impulses from the subs and mains, the mains (woofers) being the two 'taller' ones, very close together (I don't sweat the difference, you can never put the mic back in the exact same place each time, so 'as long as it is in the ballpark' if you follow)

Of course, the mic was never moved.

Here they are 'perfectly' aligned in time (note, for MY ease of doing this I took the simplest way out, so if they don't line up as well as they could that is why. When I finally do it for myself I will take the extra time and effort to make the smaller adjustments required)

This one shows the subs ahead of the mains by about 1.3 ms..whatever that is in feet (or metres!) NOTE, this is in arrival time, nothing to do with actual physical distance)

In this next one, I set the arrival times to reflect the ACTUAL physical location of the subs in the room, the left sub is almost directly 1.3 m behind the left main, the right sub (not exactly behind) the right main, but a physical path difference of 2.3 m.

So the left sub arrival time and the right sub arrival time lag the mains by that amount

Obviously I could spend a day doing thousands of different combinations, but this should suffice to explain the point.

Now, here is an overlay of all of the three above situations, and we can see how the FR (which we will proceed to eq) changes.

The green is the subs leading by 1.3 ms (about a foot or so), the red when aligned, the purple when the arrival times reflect the actual path differences.

There was NO change to any of the individual response of the mains or subs, all that was changed was the arrival times.

As expected, the closer to the 'perfect' time the closer the FR is to the 'perfect' one, and vice versa. In any case, the graph with the best (and least amount of eq required) is the 'perfect' one.

I don't want to make huge claims about audibility etc, so I will limit it to the simple observation that it is best to apply the least amount of eq, and if we can do it why not do it?

 

[FrantzM]

Ok, took me a minute to do just for illustration, takes longer to post it than to do it!

Here is the impulses from the subs and mains, the mains (woofers) being the two 'taller' ones, very close together (I don't sweat the difference, you can never put the mic back in the exact same place each time, so 'as long as it is in the ballpark' if you follow)

Of course, the mic was never moved.

Here they are 'perfectly' aligned in time (note, for MY ease of doing this I took the simplest way out, so if they don't line up as well as they could that is why. When I finally do it for myself I will take the extra time and effort to make the smaller adjustments required)

This one shows the subs ahead of the mains by about 1.3 ms..whatever that is in feet (or metres!) NOTE, this is in arrival time, nothing to do with actual physical distance)

In this next one, I set the arrival times to reflect the ACTUAL physical location of the subs in the room, the left sub is almost directly 1.3 m behind the left main, the right sub (not exactly behind) the right main, but a physical path difference of 2.3 m.

So the left sub arrival time and the right sub arrival time lag the mains by that amount

Obviously I could spend a day doing thousands of different combinations, but this should suffice to explain the point.

Now, here is an overlay of all of the three above situations, and we can see how the FR (which we will proceed to eq) changes.

The green is the subs leading by 1.3 ms (about a foot or so), the red when aligned, the purple when the arrival times reflect the actual path differences.

There was NO change to any of the individual response of the mains or subs, all that was changed was the arrival times.

As expected, the closer to the 'perfect' time the closer the FR is to the 'perfect' one, and vice versa. In any case, the graph with the best (and least amount of eq required) is the 'perfect' one.

I don't want to make huge claims about audibility etc, so I will limit it to the simple observation that it is best to apply the least amount of eq, and if we can do it why not do it?

Would be very interested to have an idea of how you did it and how you measured .. Thanks in advance.. Very, very interesting interesting!

 

[FrantzM]

Ok, took me a minute to do just for illustration, takes longer to post it than to do it!

Here is the impulses from the subs and mains, the mains (woofers) being the two 'taller' ones, very close together (I don't sweat the difference, you can never put the mic back in the exact same place each time, so 'as long as it is in the ballpark' if you follow)

Of course, the mic was never moved.

Here they are 'perfectly' aligned in time (note, for MY ease of doing this I took the simplest way out, so if they don't line up as well as they could that is why. When I finally do it for myself I will take the extra time and effort to make the smaller adjustments required)

This one shows the subs ahead of the mains by about 1.3 ms..whatever that is in feet (or metres!) NOTE, this is in arrival time, nothing to do with actual physical distance)

In this next one, I set the arrival times to reflect the ACTUAL physical location of the subs in the room, the left sub is almost directly 1.3 m behind the left main, the right sub (not exactly behind) the right main, but a physical path difference of 2.3 m.

So the left sub arrival time and the right sub arrival time lag the mains by that amount

Obviously I could spend a day doing thousands of different combinations, but this should suffice to explain the point.

Now, here is an overlay of all of the three above situations, and we can see how the FR (which we will proceed to eq) changes.

The green is the subs leading by 1.3 ms (about a foot or so), the red when aligned, the purple when the arrival times reflect the actual path differences.

There was NO change to any of the individual response of the mains or subs, all that was changed was the arrival times.

As expected, the closer to the 'perfect' time the closer the FR is to the 'perfect' one, and vice versa. In any case, the graph with the best (and least amount of eq required) is the 'perfect' one.

I don't want to make huge claims about audibility etc, so I will limit it to the simple observation that it is best to apply the least amount of eq, and if we can do it why not do it?

Would be very interested to have an idea of how you did it and how you measured .. Thanks in advance.. Very, very interesting interesting!

 

[terryj]

sorry frantz, guessed people would automatically know. duh, dumb me

that is REW v5. When I want to look at time, I use that version, for straight eqing I go back to the previous version. You can have both open at the same time and switch between the two.

these impulse graphs are under the overlay function in V5.

Nothing special about the procedure, plonk the mic in place and away we go. As we are dealing with bass (in this example) repeatability of the mic position is not THAT important, even when I am 'on about' time...you can see that 1.3 ms did not make a huge difference, so half an inch with the mic etc is chicken feed.

Just by the by, it may be important to know the setup a little. I run my mains full range, and low pass the subs at 70 hz or so at something like 70 db/octave.

So at first glance that dip we see at 65 hz or so looks like it COULD be improper integration at the crossover point.

Decided to check for myself what that was about (as the subs are prob down 30 db or so by 65 hz) and ran the mains by themselves.

As I suspected, that dip actually corresponds to a dip in the mains at that point, of course the subs would hardly be influencing that given the slopes and frequency involved.

Here is the mains response superimposed on the previous full range response, we can see the dip comes from the mains.

An easy eq fix of course. Oh, not sure if I even said it in my previous post, those graphs were 'raw', ie no eq as yet. And no smoothing either.

Why not simply regard it as 'free gain'? I mean we all pay big bucks for the subs, the amps and whatever we control it with, so why not (as we have seen) make a few adjustments to the time as well as adjustments to the eq? Maximise what we have paid money for?

 

[terryj]

I should maybe add that I will probably be in 'fiddle mode' for a day or two, if anyone wants me to run any specific test I should be able to fit that in...depending on the request and my level of laziness at the time that is!

 

[DS-21]

As expected, the closer to the 'perfect' time the closer the FR is to the 'perfect' one, and vice versa. In any case, the graph with the best (and least amount of eq required) is the 'perfect' one.

I don't want to make huge claims about audibility etc, so I will limit it to the simple observation that it is best to apply the least amount of eq, and if we can do it why not do it?

I'm still not sure what you're arguing against. Hopefully nobody disagrees with you that one should try to optimize the response naturally by playing with placement, levels, and delay/"phase" before reaching for the EQ sliders! I certainly don't.

That's Geddes' approach. That's how Harman's SFM works. If you read my "modest multisub system" thread on AVS (DIY forum) you will see that equalization came dead last as well. (Of course, I claim no original methods, except perhaps in taking the spatial average in one go with the Velodyne SMS-1/MIC-5 combo rather than doing multiple measurements and averaging them in software. I basically follow Geddes' advice.)

Just by the by, it may be important to know the setup a little. I run my mains full range, and low pass the subs at 70 hz or so at something like 70 db/octave.

So at first glance that dip we see at 65 hz or so looks like it COULD be improper integration at the crossover point.

Decided to check for myself what that was about (as the subs are prob down 30 db or so by 65 hz) and ran the mains by themselves.

Why such a steep slope and low cutoff? Do your subwoofers have excessive inductance? Have you tried the system with more overlap? (Higher crossover and shallower slope.)

Super-steep slopes look great in theory, but in practice I've never heard them work as well as systems with greater overlap.

Also, is that slight rise from 50-40 Hz an intentional house curve? Depending on your listening levels, that could make a lot of sense. (I calibrate my bass flat now only because Audyssey DynamicEQ provides a house curve that gets more pronounced as overall levels drop)

Lastly, as for a "request," I suspect that 65Hz dip corresponds to a floor-ceiling mode. I bet you could fix it easily with a high-mounted filler sub. It doesn't need to be that big or powerful. In my previous condo, my floor subs were a 15 with 1kW and two 12's with 500W each. (All three of those subs have long-throw underhung motors.) A crappy "designer" sub with a 9" relatively short-throw woofer and 100W (Morel SoundSub 9) was sufficient for the high-mounted sub. It should barely be playing.

 

[RUR]

Dammit, Terry, those impulse graphs are cool and may push me over the edge to set up my REW. Neither XTZ nor Dayton will do this trick. The TacT sets acoustic distance automatically, but it would be nice to be able to double-check.

And no smoothing either.

How many PPO? There's definitely some smoothing in those graphs.

Ken

 

[terryj]

I'm still not sure what you're arguing against. Hopefully nobody disagrees with you that one should try to optimize the response naturally by playing with placement, levels, and delay/"phase" before reaching for the EQ sliders! I certainly don't.

Not sure! So I quickly scanned the thread to see what it was (if anything)

your first post pg 1 mentioned 'three prongs', none of which mentioned anything about time alignment. That is how I normally find any discussion of the geddes approach, tho I admit that placement can have a lot to do with time, obviously, yet never see 'and to ensure the same arrival time, you can use placement to achieve that if you have no other way'.

Your post 16 says To paraphrase, in bass in a small room, if you fix the FR you've fixed timing problems too. , I simply tried to show that the fr did NOT change or fix the time.

You then went further to say in small rooms time has no meaning, and cited waterfalls as proof of that. That is a 'different' time than I was talking about.

So I thought a few graphs (which was not a lot of skin off my nose to do..I was doing it anyway) could explain things a lot more quickly and more clearly.

That's Geddes' approach.

As I said, he may have since changed his position, but I do clearly remember discussing it with him four or five years ago. At that stage all he would say was 'there is no evidence for audibility'.

Why such a steep slope and low cutoff? Do your subwoofers have excessive inductance? Have you tried the system with more overlap? (Higher crossover and shallower slope.)

Have no idea of the inductance btw, the subs are maelstrom 18...I hazard you know FAR more about them than I!, the mains are PHL 18's, pro drivers. (7030 IIRC)

No, no rhyme nor reason for that, kinda just happened this time around. I used to run them UNDER the 18's, this time I have about an octave overlap.

Low? I recall you run yours up to 200?? I have crossed them at 95, which to my surprise I did not find too much wrong with (so much for the adamant 'You MUST cross below blah blah, else you will hear where they are' stuff. Rules...the sometimes apply and sometimes don't.)

Super-steep slopes look great in theory, but in practice I've never heard them work as well as systems with greater overlap.

Maybe.

Also, is that slight rise from 50-40 Hz an intentional house curve? Depending on your listening levels, that could make a lot of sense. (I calibrate my bass flat now only because Audyssey DynamicEQ provides a house curve that gets more pronounced as overall levels drop)

Funnily enough, I don't run ANY house curve, the bass is ample as it is! TBH, I find I like 'flat' to about 1k, then roll off after that. Once I have that baseline, I like to add a bit of the BBC dip, 2 or 3 db centred around 800 or so, very broad. Anyway, these graphs were 'just to show', as I mentioned there was NO eq in the graphs, so not a house curve. Something that small (hump wise) I'd probably ignore I'd say. Time will tell!

Lastly, as for a "request," I suspect that 65Hz dip corresponds to a floor-ceiling mode. I bet you could fix it easily with a high-mounted filler sub. It doesn't need to be that big or powerful. In my previous condo, my floor subs were a 15 with 1kW and two 12's with 500W each. (All three of those subs have long-throw underhung motors.) A crappy "designer" sub with a 9" relatively short-throw woofer and 100W (Morel SoundSub 9) was sufficient for the high-mounted sub. It should barely be playing.

Ok, no cheating mind! From this, tell me the height of my ceiling. Don't you go searching haha

Dammit, Terry, those impulse graphs are cool and may push me over the edge to set up my REW. Neither XTZ nor Dayton will do this trick. The TacT sets acoustic distance automatically, but it would be nice to be able to double-check.


How many PPO? There's definitely some smoothing in those graphs.

Ken

Hi ken, yeah I can get the deqx to do it automatically too, dunno if I prefer it or it has simply become habit but I tend to eq the bass manually. After all, as I just explained to DS I tailor the sound for myself anyway, so the mic yada yada is already set up.

As you say, always good to check 'assumptions', maybe a bit of fine tuning in both time and eq could help.

No, pretty sure I did not apply smoothing. Never do in the bass, that is how REW spits it out. So if there is smoothing somewhere, it is in the program before I see it on the screen.

 

[Robh3606]

It is easy to see (and trivial to show) that changes in timing between the subs and the mains produce (often wildly) differences in the FR.

So the upshot is that after EQ and sub placement, we have a nice looking FR that we want (that should allow for individual preference).

BUT, we can apply eq to a graph that is 'correct' (ie the result of the subs and mains being in time with each other) or one that is 'wrong', ie the subs and mains are just playing when they want so to speak.

Hello Terry

Yes but wouldn't you do any EQ as a last resort??. You can use the phase adjustment on a single or individual subs to help offset any phase/timing issues through the crossover point and then EQ from there as needed. I always go for the most output to make sure it's all in phase and go from there as far as levels and so on.

Rob

 

[terryj]

hi rob

yes, of course.

I thought I was being clear, but obviously not, long boring posts notwithstanding!

I might refrain from posting on this anymore, I'm not sure how I can be clearer so maybe it's best if I shut up now.

 

[DS-21]

your first post pg 1 mentioned 'three prongs', none of which mentioned anything about time alignment.

OK, then we were talking past each other, probably because I was too brief - me too brief, really? - in my comments. To elaborate on my phrase "set up sequentially" under the first prong, it means "adjust the levels, sub highpass, and delay/phase for smoothest natural response, starting with the first (usually corner) sub and mains, and then repeat the process with each additional sub."

"Time alignment" in and of itself is, as you've shown in your measurements, visible in the frequency response down low.

That is how I normally find any discussion of the geddes approach, tho I admit that placement can have a lot to do with time, obviously, yet never see 'and to ensure the same arrival time, you can use placement to achieve that if you have no other way'.

I think it's the other way around. Placement first, then everything else.

That does assume some flexibility in placement, which is often (usually?) an unwarranted assumption. In practice, usually subs go where they'll fit without being an eyesore in the room.

Your post 16 says To paraphrase, in bass in a small room, if you fix the FR you've fixed timing problems too. , I simply tried to show that the fr did NOT change or fix the time.

Your FR did change with changes in delay, though, so I'm confused as to why one would look at both, rather than just the one.

You then went further to say in small rooms time has no meaning, and cited waterfalls as proof of that. That is a 'different' time than I was talking about.

Let me ask you a question that I think will clarify where I'm coming from.

Assume for a minute you were only taking FR measurements. Which one looks best? While it's subtle, the one with equal arrival times is the smoothest, no?

I think we're converging on similar results, but going at it from different directions. I prefer my direction, because it requires looking at only one thing and not two.

Also, your measurements are all single-point, correct? Move the mic a foot in any direction, and what happens?

As I said, he may have since changed his position, but I do clearly remember discussing it with him four or five years ago. At that stage all he would say was 'there is no evidence for audibility'.

I think implied in that is "except inasmuch as it affects the frequency response." As you've shown, misalignment does affect the FR.

Have no idea of the inductance btw, the subs are maelstrom 18...I hazard you know FAR more about them than I!,

Which ones? There were, AFAIK, three separate runs of Maelstrom-X's. I only have experience with the very first ones, which had dual 8? coils and the lowest normalized inductance of the series by quite a bit (0.87mH Lep/ 3.1? Rep). I know that the Mk. I is clean out to past 200Hz, but the later ones as I mentioned did have higher inductance so I don't know if they are as good.

the mains are PHL 18's, pro drivers. (7030 IIRC)

Sealed or vented?

Low? I recall you run yours up to 200??

Yeah, I consider 70Hz pretty low. I usually start at 120Hz and adjust for smoothest response from there. I think the highest I have gone is a system with subs lowpassed in at 140Hz, 120Hz, 90Hz, and the fourth at something in between the two extremes. (Mains were run wide open.) Current system (which doesn't really count, as it's temporary and the good stuff is in storage pending a move) uses a 120Hz lowpass on all three subs, with a 120Hz highpass on the mains because of their limited cone area and volume displacement.

I have crossed them at 95, which to my surprise I did not find too much wrong with (so much for the adamant 'You MUST cross below blah blah, else you will hear where they are' stuff. Rules...the sometimes apply and sometimes don't.)

Rules that by and large were written for suboptimal setups (i.e. single-sub systems) don't necessarily apply to more advanced systems.

Ok, no cheating mind! From this, tell me the height of my ceiling. Don't you go searching haha

No idea. I don't know the math. I just know from experience setting up subwoofer systems that when I see a single notch in that general region that's consistent across multiple positions, I think floor-ceiling mode. And in practice I've always been able to eliminate or at least significantly reduce it with a high-mounted subwoofer.

 

[andy_c]

Terry, thanks very much for that set of graphs, very thought-provoking post! It seems to provide a deterministic way of setting up the delays - just line up the peaks of the individual impulse responses.

Also, I think some of the confusion in the discussion about "time" is from a potential ambiguity. It could relate to:

1) The individual delays assigned to each mains speaker system or sub, which, when done right as you have shown, affects the combined frequency response of mains and subs in a positive way.

2) The time-domain response of combined subs and mains at a given mic position. That is, when the frequency response of the combined subs and mains is optimized, is the impulse response of the combination optimized as well? Asked another way, can the effect of room modes on frequency response at any position in the room be described as a minimum-phase function? I have my doubts, but apparently Toole disagrees with me.

The fact that your impulse response technique for setting individual delays works so well suggests to me that Earl's approach is less unconventional than I previously thought.

 

[Mark Seaton]

I would advise some caution when getting too caught up with impulse response usefulness in mating subwoofers and main speakers. Impulses, and particularly the attractive, sharp peaks, are always dominated by the highest frequency content. By definition, the speaker with greater high frequency content will have a sharper pulse. You can look at both general starting points with an impulse and then the phase response to better see what is going on near the crossover. If you limit the content of an impulse to just the lower frequencies it becomes more useful, or some systems like my TEF will allow measurement of an energy time curve (ETC) which can be band limited. In the end it's another representation of phase with an absolute time reference.

I'm one who has been recommending to use the distance adjustments to improve the summed subwoofer - speaker integration for a long while now. My recommendation and practice is to be aware of which direction you are adjusting to make sure you are bringing things closer or in a preferred time offset (ie speaker arrival before subwoofer), while making adjustments that help the observed response at your listening position.

The assumption that a "flat" subwoofer pass band and "flat" extension of the main speaker = good combination is a long shot. Many aspects of the sub and speaker's behavior above and below the pass band of each greatly impact the phase & group delay which dictate how the two signals interact. Until you measure the combined response to confirm, you're at best making educated guesses, but you're still guessing.

 

[terryj]

Some of this will touch on marks comments etc.

Thanks DS, looks like we are closer than first appeared? Haha, you should know me by now, at times I am wont to wax philosophical! And here is another one

Can I make the observation that when a discussion is carried out in a non combative manner, it invariably moves along and is productive? None of which demands anyone capitulate their position BTW, so it is not a matter of being 'less manly'...that may be cryptic to some, if so then trust me, don't worry about it. Those that can guess where I am pointing would surely agree with me.

OK, then we were talking past each other, probably because I was too brief - me too brief, really? - in my comments. To elaborate on my phrase "set up sequentially" under the first prong, it means "adjust the levels, sub highpass, and delay/phase for smoothest natural response, starting with the first (usually corner) sub and mains, and then repeat the process with each additional sub."

"Time alignment" in and of itself is, as you've shown in your measurements, visible in the frequency response down low.

Yeah, we can see that, and twas the sole point I was trying to get across, that (as you observe soon) we can get 'free' gain and a smoother plot to work with if we address this issue before eq.

I think it's the other way around. Placement first, then everything else.

That does assume some flexibility in placement, which is often (usually?) an unwarranted assumption. In practice, usually subs go where they'll fit without being an eyesore in the room.

I take it you mean 'in the corner (for example) to maximise output' type stuff? Yeah, I can certainly see that. That it also kinda ensures that the sub and mains will NOT be in time and because of that is another reason to take a quick look at the time. I was only looking at the guy who has no other way to affect the time, he can ONLY do it by placement. In any case, we agree here I think.

Your FR did change with changes in delay, though, so I'm confused as to why one would look at both, rather than just the one.

Let me ask you a question that I think will clarify where I'm coming from.

Assume for a minute you were only taking FR measurements. Which one looks best? While it's subtle, the one with equal arrival times is the smoothest, no?

I think we're converging on similar results, but going at it from different directions. I prefer my direction, because it requires looking at only one thing and not two.

This is an interesting one, I'd like to spend a little bit of time on it if that is ok. If I have grabbed the wrong end of the stick then let me know yeah?

I see we essentially agree...that the 'best' response also coincides with the 'best' timing. But, as you never look at timing, here is how I picture the process going (and is where you might need to correct me)

If you only look at the FR...firstly HOW do you know it is not (magically) the 'best' FR? In other words, it is only by moving the subs, taking another sweep and comparing that you can start to tell you are converging on the 'best' timing?

Which way do you move them, left or right, front or back, a combination? Each time taking a sweep. (the starting point is that you have to get the best graph FR wise which coincides with the 'best' time)

Start to ad up those permutations and I fail to grasp how it is any easier...'because it requires looking at one thing not two'.

Anyway, it is possible that you don't quite know this? You might have to look at two graphs, BUT it is the one and the same measurement...once you have that measurement you simply switch back and forth to see it represented how you wish...phase, spl, impulse etc. They are just different presentations of the same data.

Again, a quick pic will show it faster than words.

It also addresses another point while we are at it

Also, your measurements are all single-point, correct? Move the mic a foot in any direction, and what happens?

Yeah, when I check the time I put the mic where (I think) the centre of my head is, so it represents the equidistant point from the speakers. Yes, single mic.

BTW, when I adjust the mains timing, I just set it up and don't worry about it after that on later measurements...as the mic WILL be in a different place, there is a time limit you reach after which you don't bother chasing anymore, that limit being what we reasonably expect mic variations to be.

The short answer to your question is 'no, in the range we are discussing here (bass), differences in position do not make much difference'. Of course, the higher frequencies will vary wildly with changing position, as I think will be clear.

(pretty sure you know all this stuff?? In any case, it is good to write it for a few reasons, others might not AND most importantly, it allows me to be corrected if I am wrong)

Again, took two minutes. I put the mic in five different places across the listening position, about six inch increments (ie, one foot to the left of the centre point, six inches to the left, centre point etc etc). I do this anyway as a matter of course. I mentioned above how I like the in room curve to be, I get the base position from this very method, and in the average tab you will see (hopefully it will be clear!) I average these positions across the LP and apply any full range eq to the average. Often there will be 'common' peaks and dips that appear across each graph, they are the obvious first points to address.

Anyway, here is a print screen (so it will show the full picture, not just the graphs) of those five measurements

First thing, we can see that tho anything above 200 (say) varies wildly, the bass is essentially the same. I have not bothered to show this, but if you remove the outer two measurements, one each side, then they agree very nicely. That is still at least a foot spread.

Just for the heck, here is the bass with no smoothing, the outer two removed (ie a foot spread) and with the same scale as the earlier graphs

BUT, I want to draw your attention to the tab at the top, labelled SPL naturally. Roughly in the middle is another tab called Impulse, click in that and this data will be presented as an impulse plot.

No extra measurements or work required, it is all there. The same applies of course if in the SPL plot we had, instead of full range measurements as this one is, seperate measurements of subs, mains, tweeters etc etc (that was how I did the earlier ones, you can see the seperate impulses from each, if I had shown the SPL tab of those graphs you would have seen the seperate SPL traces of each driver)

So, given that we seem to agree that we should converge on the 'best' timing to give us the 'best' FR, I think my method is way faster and more accurate.

Note again, this is just for everyone's 'information', I don't demand nor care if they agree or disagree, do it or not do it. Just data.

Which ones? There were, AFAIK, three separate runs of Maelstrom-X's. I only have experience with the very first ones, which had dual 8? coils and the lowest normalized inductance of the series by quite a bit (0.87mH Lep/ 3.1? Rep). I know that the Mk. I is clean out to past 200Hz, but the later ones as I mentioned did have higher inductance so I don't know if they are as good.

Dunno actually, both bought second hand. I know the first one was dual four ohm coils, I assumed the second one was too! Haha, never even checked and wired it up the same way as the first. Oh well, I just make it do what I want. When I rebuild them, I might check out of curiosity's sake.

Sealed or vented?

All sealed.

No idea. I don't know the math. I just know from experience setting up subwoofer systems that when I see a single notch in that general region that's consistent across multiple positions, I think floor-ceiling mode. And in practice I've always been able to eliminate or at least significantly reduce it with a high-mounted subwoofer.

Had a quick look at a calculator, don't think that particular one corresponds with ceiling height (seventeen feet btw)..what DID seem to come up that caught my eye was one around 110 or so, I see THAT one all the time! My width/depth depending on how you look at it is very close to the ceiling height, the last dimension is around 9m if memory serves (with a very big bay window to make it even more complicated!)

Thanks ever so much mark. Oh, that reminds me, you will see above in the impulse how sharp it is (certainly compared to the earlier ones!)..as mark pointed out the crispness of the impulse depends on the signal it was fed. I always thought it had to do with bandwidth, mark has said it is the high frequencies. Still learning.

Another point mark raised was 'where do you align it'. VERY true indeed. As we have seen it is not a crisp clean impulse. I usually use the peak of the first major 'rise', if that made sense.

why? dunno. Well, if you take a close mic reading of the sub there usually IS a very pronounced unmistakeable peak...look at the in room response and it might not be so clear! SoI just refer back to my 'knowledge' of how the raw sub produced the impulse (close mic) and apply that to the often scrambled in comparison in room response.

I agree with mark, you need to be aware of which way you are adjusting the time (if that is what you want to do)! Much quicker and easier to LOOK and see, rather than an iterative process.

Mark, is there a better 'signal' we can use to do this? Would you be able to tell us (if not a trade secret) how you go about it? Any variations on this outline, any complete stupid stuff I have said that needs correcting?

This is all about learning, not winning so to speak.

thanks.

 

[andy_c]

Thanks again for all the thought-provoking data, Terry.

I do have a question about your earlier impulse responses, the ones associated with the data with upper frequency limit 100 Hz. You mention the following:

Oh, that reminds me, you will see above in the impulse how sharp it is (certainly compared to the earlier ones!)..as mark pointed out the crispness of the impulse depends on the signal it was fed. I always thought it had to do with bandwidth, mark has said it is the high frequencies. Still learning.

I have not used REW yet. Are you saying that when you display a frequency sweep that only goes up to, say, 100 Hz, that the associated impulse response only makes use of data in that frequency range? If so, I guess the impulse responses previously shown would end up being low-pass-filtered versions of the true impulse responses. It seems to me that, if anything, this makes those plots more useful for setting the delay, not less, because then they only deal with subwoofer/main woofer integration, almost as if the mid and tweet were turned off.

Maybe I'll ask this at HTS also. I'm a member there but have not posted yet.

 

[terryj]

hi andy, if I get your question right then the answer is no. The measurement itself extended past 100 hz, I just limited the graph to 100 hz. Hence, the impulse response stays the same in both cases as it is generated from the measurement.

did that answer the question?

I think you then went on to ask 'how does it change when we bandlimit the measurement'??? As you probably know, it is easy to do that when you take the measurement, you can set the lower and upper limits to it.

The subs, naturally, set the limit they will measure by their characteristics. Not sure if I have helped tho...

 

[andy_c]

hi andy, if I get your question right then the answer is no. The measurement itself extended past 100 hz, I just limited the graph to 100 hz. Hence, the impulse response stays the same in both cases as it is generated from the measurement.

did that answer the question?

I think you then went on to ask 'how does it change when we bandlimit the measurement'??? As you probably know, it is easy to do that when you take the measurement, you can set the lower and upper limits to it.

The subs, naturally, set the limit they will measure by their characteristics. Not sure if I have helped tho...

Let me be more clear. Let's look at your post #21, where the upper limit of the graph is 100 Hz. The width of the impulse response of the mains in that plot is about 5-10 msec (between consecutive dips surrounding the peak), indicating low-pass filtering of the mains is happening in software. I don't know what the measurement bandwidth was, but the display bandwidth is limited to 100 Hz. Did you actually sweep the full bandwidth and then limit the display to 100 Hz? Or did you limit the measurement range to 100 Hz?

Then in post #34 where the maximum frequency of the display is much higher, the impulse response is, well, too narrow to tell the exact width, but it's many times smaller than in post #21.

So in order to get the "fat" low-pass filtered impulse response of the mains to display (which is what we want), must one limit the measurement range to 100 Hz max, or can one simply limit the display to 100 Hz max in the frequency domain, then switch over to the impulse response display?

 

[terryj]

Let me be more clear. Let's look at your post #21, where the upper limit of the graph is 100 Hz. The width of the impulse response of the mains in that plot is about 5-10 msec (between consecutive dips surrounding the peak), indicating low-pass filtering of the mains is happening in software. I don't know what the measurement bandwidth was, but the display bandwidth is limited to 100 Hz. Did you actually sweep the full bandwidth and then limit the display to 100 Hz? Or did you limit the measurement range to 100 Hz?

I am taking it step by step, so forgive the tortuous progress!

Yes, the graph is limited to 100 hz. Maybe a bit of data you are missing is that I ran a sweep of the subs (so they play roughly 20-70 hz say...ie bandlimited).

I ran a sweep of my WOOFERS, ie soloed the woofers, no other driver-lower and upper mid or tweeter-played anything. The woofers cover approx 40-180hz, ie bandlimited by the crossover.

AND, I only ran a sweep of that part of the frequency range each driver covered..no point sitting there whilst three quarters of the sweep frequency is not played because it is outside of the range of the driver in question.

I could have done all that with a 20-20k sweep, and still got the same result because that is the part of the spectrum they play. But, you get a clearer graph by playing a band limited sweep (of the appropriate frequency range) as it get's more data from the sweep. Same FR, same impulse etc, just cleaner.

Did that help?

I don't mind spending as much time as needed if it helps, but if it confuses I should shut up?

so yes, I played the full bandwidth of the driver in question, the graph was limited to 100 hz (as the subs don't go over that anyway, and we were talking 'what happens to the FR when we change nothing but the time for the subs')

I must say I ran a sweep of the mains too (180-20 000, but deliberately left it off to avoid confusion haha!, shows how easy it is to make an error in judgement)

Then in post #34 where the maximum frequency of the display is much higher, the impulse response is, well, too narrow to tell the exact width, but it's many times smaller than in post #21.

Ah, yes. Starting to get you now. That is back to the point mark made, and I tried to. THAT particular impulse was a full 20-20k one. Or maybe just the mains, forget now. The wider the bandwidth (or higher frequency as mark put it) the sharper the pulse. And vice versa. So it is the nature of the beast that when we run a sub sweep, well naturally it will be broad, leading to the earlier discussion of 'where do you take the peak?'

So in order to get the "fat" low-pass filtered impulse response of the mains to display (which is what we want), must one limit the measurement range to 100 Hz max, or can one simply limit the display to 100 Hz max in the frequency domain, then switch over to the impulse response display?

You know, I reckon it did help to take it bit by bit, I read it a lot easier now! I got the fat impulse because I can do it, I can play any particular driver I want at any time. So, I would imagine that if you had a full range speaker where you could not do what I can, and you happened to know the crossover point of the woofers (say), then yeah, limit your sweep...so in effect all you had playing was the woofer (or mid etc etc)

Not that it matter too much to others perhaps, but if you look at the impulse, you can see if a driver is inverted, yet another handy tool at times maybe.

 

[andy_c]

Okay, now I'm even more confused . Are you saying that you got the "fat impulse" for the mains by literally turning off the mid and tweet in an active system? Or were they enabled, but the frequency sweep limited in software? Or, repeating my previous question, can one simply limit the frequency span of the frequency domain display and have the software automatically bandlimit the impulse response to that range, thus "fattening it up"?

The answer to this is important, because if it's the former, only people using active mains for which individual drivers can be disabled can do it, while if either of the latter two, it's a more universal technique.

 

[terryj]

Okay, now I'm even more confused .

Oh dear..sorry 'bout that chief.

Less talk from me, more pics methinks?

Are you saying that you got the "fat impulse" for the mains by literally turning off the mid and tweet in an active system?

Yes, that is exactly what I do.

Or were they enabled, but the frequency sweep limited in software? Or, repeating my previous question, can one simply limit the frequency span of the frequency domain display and have the software automatically bandlimit the impulse response to that range, thus "fattening it up"?

Luckily the mic is still there, and lucky I am mucking about with it still!

(sounding good at the moment I am sure you all want to hear haha!) As you will see, not finished yet but can still explain with pics none-the-less.

The answer to this is important, because if it's the former, only people using active mains for which individual drivers can be disabled can do it, while if either of the latter two, it's a more universal technique.

Well, see if this helps.

(good, the tags at the bottom are there, that will help) I am presuming that people trying to integrate subs are able to play JUST the subs, and JUST the mains?

The fuzzy green is just the subs, but a full 20-20k sweep at 512k sample size.

The sharp red is just the subs, but a sweep of 20-100 (just covering the range of the subs) same sample size of 512k.

Obviously there is more data within the band of interest (512k in 20-100 vs 512k 20-20k), so we get a clearer picture of what we are after.

[the cryptic shorthand at the bottom kinda tells you what is what, you should be able to work it out from that too)

The purple graph is WOOFERS ONLY, again with a sweep only covering the range (a bit cleaner than 20-20k, but not a lot. Seems only the subs benefit greatly from that)

Lastly, the brown is FULL RANGE, no particular driver selected or deselected, but the sweep only covered the range of interest, the woofers. Ie, the sweep in my case was 20-200 hz. (damn, just noticed I forgot to label that sweep)

There is a slight change in the shapes, BUT the impulses essentially align in all the cases, which I think is what answers your questions.

Did that help?

Still have to make final adjustments between the woofers and subs, but for now good enough to evaluate the new set up.