Jump to content

Deep End DIY - My first speaker project


acg

Recommended Posts

25 minutes ago, acg said:

It's pretty good....

Im super happy for you listening to the sound impression on the youtube clip. Sure its only recorded on a phone and the stack is incomplete but the tonality of the upper mids seems excellent.  Good times ahead.  Its still a huge project to finish and dial in but I hope that will be more fun when you can listen to the changes.

 

I wonder if you could load up more youtube videos for us?  Do you have a USB mic to plug into your phone? Stereo or binaural recordings would be great. Then we can hear an impression of the evolution.

 

 

  • Like 1
Link to comment
Share on other sites



@Nada, Hugh, I have here a calibrated microphone and an excellent ADC as part of my RTX6001 Audio Analyser.  There is also a Focusrite Scarlet 2i2 that would be OK.  Four issues:  I am only to turn-on stage with the RTX6001 so will need some time to get across its use;  the mic has been used extensively by the kids to sing (badly) and is likely chock full of nasties that may render its calibration moot;  pretty sure that Flac files are not a file type that can be uploaded to SNA;  I am not really sure there is a lot to gain by uploading in-progress clips of system progress.

 

So the answer is possibly, maybe.  In some ways it would be nice to be able to look back and listen to changes as the room progresses, but there are going to be so many incremental updates that it will hardly be conclusive.  Plus no video, unless I can sort out a USB mic in the meantime.

  • Like 1
Link to comment
Share on other sites

32 minutes ago, acg said:

pretty sure that Flac files are not a file type that can be uploaded to SNA

...just change the file extension to .jpg or .pdf ... @Marc will never know :) ...but seriously, no recording is going to do this system justice!

Having followed this thread for many years I'm hoping @acg will have a GTG or 2 or 3 when completed - Brisbane is just down the road, and I'd love to hear it in the flesh (following on with the Pink Floyd references)...

...getting very close on a very long journey - fantastic work!

 

cheers,

Mike

 

  • Like 1
Link to comment
Share on other sites

ML1's have been my speakers for years now and if you push a pair of them past say 88dB-90dB average they start to have a few issues in my room.  Finally, I decided to crank up the mono incomplete Macondo/Melquiades system to a  normal comfortable listening level and the damn thing is averaging 100dB (peaks to 110db) at the listening chair and sounds as pristine and clear as it does at 70dB average.  No doubt it can go quite a bit louder and still be comfortable to listen to so I am going to have to be really careful about hearing loss...the speaker upper listening limit has been shifted way out of the safe zone and into dangerous territory.

 

Fun days ahead.

 

@almikel, I am sure I will end up with plenty of visitors that come not only for my sparkling company but to listen to the my horns.  You would be most welcome when the time comes.

Edited by acg
  • Like 1
Link to comment
Share on other sites

On 21/02/2019 at 10:51 AM, acg said:

pristine and clear

I could hear that even on the recording.   :) 

 

People often misunderstand what's mostly responsible for this.  They think it's to do with harmonic distortion....  that high sensitivity (low power required), and the band-limiting of the drivers by the air loads of the horn .... reduce the distortion to super low levels.

 

It's due to something bigger and more fundamental (pun not intended Ant).....  In simple terms, what you are hear is that (compared to a typical speaker) is that the "direct sound" is much louder than the "room sound".

Edited by davewantsmoore
  • Like 2
Link to comment
Share on other sites



31 minutes ago, davewantsmoore said:

I could hear that even on the recording.   :) 

 

It's very interesting that "audiophiles in general"  (I know that's a big blanket statement, but .... and it's not intended to make me "sound superior")  misunderstand what causes this.

 

They think it's to do with harmonic distortion....  that high sensitivity (low power required), and the band-limiting of the drivers by the air loads of the horn .... reduce the distortion to super low levels.

 

It's due to something much bigger and more fundamental (pun not intended Ant).....  In simple terms, what you are hear is that (compared to a typical speaker) is that the "direct sound" is much louder than the "room sound".

 

I have zero doubt that by far the major effect is the reduction of the room sound Dave.  Also 8 x 10" woofers handling bass versus 1 x 6" on the ML1's has a major effect in how forced things end up sounding.  Not much above 90dB the ML1's mud-out and things get harsh.

 

At the moment I am playing some Tame Impala to listen to some of the larger bass notes.  At quite a bit of volume I put my hand on the side of the Bass Cannons and I am unable to detect any box vibration with my hands, however it is difficult to concentrate because the noise is causing my shirt and skin to flutter while it appears the individual Cannons are quite inert.  If this is really so, one of the major design objectives has been achieved namely inert bass boxes that do not add themselves too much to the sound.  At a later date I plan to pull out the accelerometer to see if I can measure my success.

  • Like 1
Link to comment
Share on other sites

35 minutes ago, acg said:

At a later date I plan to pull out the accelerometer to see if I can measure my success.

I know you will (I would too, 'cos "interestment") ....  but I wouldn't bother.

 

I'd be pulling out the impedance curve, or the microphone ... to see if I could see any distortion caused by the internal cabinet dimensions.    Although a super rough guess from me says 80Hz  (assuming the cannon is 1m long) ... is where you will see an acoustic cancellation.

Link to comment
Share on other sites

37 minutes ago, davewantsmoore said:

I know you will (I would too, 'cos "interestment") ....  but I wouldn't bother.

 

I'd be pulling out the impedance curve, or the microphone ... to see if I could see any distortion caused by the internal cabinet dimensions.    Although a super rough guess from me says 80Hz  (assuming the cannon is 1m long) ... is where you will see an acoustic cancellation.

 

I did measure things at the prototype stage and I am pretty sure I got rid of the major cancellations and a couple of minor impedance glitches, but 80Hz is about where the Cannons will crossover to the upperbass horn.  It will be interesting to see how that time spent on a 4 stack converts to the 8 stack.  Now that I actually have an amp that can drive the Cannons I will measure in-room response and start playing a little with room treatment.  The plan is to put a 400mm wide bench all the way under the windows at the back of the room and fill underneath entirely with soft/fluffy insulation to try to knock down a bit of bass reverberation time.  The big driver here for me is not necessarily to get started on room treatment but to get all my records out of the spare room, and on top of the aforementioned benches I will place some vinyl flips.

  • Like 1
Link to comment
Share on other sites

22 minutes ago, acg said:

The plan is to put a 400mm wide bench all the way under the windows at the back of the room and fill underneath entirely with soft/fluffy insulation to try to knock down a bit of bass reverberation time.

This is a good idea.   Put as much (and as thick) fluffy in the room, but not in positions where they will impact lateral reflections  (unlike what you see in typical treated rooms, where they do that on purpose)

Link to comment
Share on other sites

4 minutes ago, davewantsmoore said:

This is a good idea.   Put as much (and as thick) fluffy in the room, but not in positions where they will impact lateral reflections  (unlike what you see in typical treated rooms, where they do that on purpose)

 

Is your reasoning for not putting soft/fluffy at lateral reflection points because of how it soaks up everything, not just the bass?  I have a lateral modal cancellation circa 30Hz that is going to be a doozy to treat and is a big reason that I am so interested in the VPR absorbers:  firstly to get that low; secondly because I should be able to tune it not to go too high.

  • Like 1
Link to comment
Share on other sites



1 hour ago, acg said:

Is your reasoning for not putting soft/fluffy at lateral reflection points because of how it soaks up everything, not just the bass?

Yes....  and, if you put less than "lots fluffy" ....eg. a typical acoustic treatment panel, in those spots ..... then the change the spectral balance of the reflections.

 

In a typical system .... The spectral balance of the reflections is already wrong, and they are already too loud .... so anything is an improvement.      If your reflections are much quieter, and already of the correct spectral balance .... then mucking with the reflections is not a good idea.

 

1 hour ago, acg said:

I have a lateral modal cancellation circa 30Hz

You must have a larger room than I thought.    I assume you worked this out from simulations (?!)

Assuming yes, then it will be a different story in practise .... the simulator does not take into consideration that your house absorbs 30Hz.

If you do see the mode in practise .... then you will likely need to just "ignore" it.

 

1 hour ago, acg said:

VPR absorbers:  firstly to get that low; secondly because I should be able to tune it not to go too high.

How far up it reaches won't be a big concern, I wouldn't think.    Anything you can "absorb" below 200Hz or so, is a basically always a good thing.

Link to comment
Share on other sites

On 19/02/2019 at 5:03 PM, acg said:

It's missing a couple of channels, nothing had been measured or tuned, but frometh comes sound.  Yay!  Kind of an appropriate song considering how long this has taken me thus far.

 

Awesome Anthony. Really cool! Lots of promise there. And, James sounds as incredibly good, as he should. A huge fan here. Reckless.......maybe, but then again its a matter of debatable perspective. I think not

 

Please post more AV clips as you see fit. You have a captive  audience of nutters :)

Edited by aertex
Link to comment
Share on other sites

On 21/02/2019 at 12:36 PM, davewantsmoore said:

I'd be pulling out the impedance curve, or the microphone ... to see if I could see any distortion caused by the internal cabinet dimensions.    Although a super rough guess from me says 80Hz  (assuming the cannon is 1m long) ... is where you will see an acoustic cancellation.

 

Here 'tis Dave.  This is the first Bass Cannon of 8 paralleled drivers including the speaker cable, so is exactly what the amplifier is driving.  1.1R to 3.1R impedance range.  At the moment this is low passed first order at about 60Hz (but won't show in these measurements of course).  I doubt that I will have to change the crossover to any higher than 100Hz.  That blip at 600Hz is something I knew was there but did not really think I had to do anything about.  Datasheet or the drivers is here.  Your comments are welcome, but as a relative layman, I think it looks excellent, particularly below resonant frequency when compared to the datasheet.  I'll be very interested to measure the in-room SPL response, but that may not happen for a few days (am working this weekend).

 

 

1934310584_DATSFirstCannon(Left).thumb.jpg.14b9eab827854ba8fa82cdfdeaf986a1.jpg

Link to comment
Share on other sites

19 hours ago, aertex said:

Awesome Anthony. Really cool! Lots of promise there. And, James sounds as incredibly good, as he should. A huge fan here. Reckless.......maybe, but then again its a matter of debatable perspective. I think not

 

Please post more AV clips as you see fit. You have a captive  audience of nutters :)

 

Thanks Alan, I do like James Reyne as well.  His solo debut occurred at a formative time of my youth and I still listen to several of his particularly early albums regularly.

 

Link to comment
Share on other sites

13 hours ago, acg said:

Here 'tis Dave.

Zoom in :) 

Also... If it's not too much hassle, grab the impedance of just one cannon, instead of the whole the battalion.

 

Not that there is much you can do at this stage.... but, it might help you make sense of you in room measurements, which are going to be hard to interpret below a few hundred Hz.

Link to comment
Share on other sites



1 hour ago, davewantsmoore said:

Zoom in :) 

 

 

Done, and I've included the individual data points as bars.

 

70218480_DATSFirstCannon(Left)Zoomed.thumb.jpg.e9ed57ff05083253f3682109205f6e01.jpg

 

 

1 hour ago, davewantsmoore said:

Also... If it's not too much hassle, grab the impedance of just one cannon, instead of the whole the battalion.

 

I have that info from the prototyping stage.  It is a hassle to do it now for a single cannon because they are all wired up in parallel and it is a fiddly job to unwire it then wire it up again later.

 

Here is the empty prototype Cannon, without any damping or stuffing...

 

1830938093_BC0Sheets.thumb.jpg.e05a9d384df61c3bcbae9b7554cb4609.jpg

 

...problems are clearly visible.  I did a lot of experimenting with stuffing and damping to sort out the resonant behaviour of the individual Cannons and to get Fs lower.  This is what I ended up with for the damped and stuffed half-stack, which is four Cannons...

 

1614663838_HalfStack.thumb.jpg.3dec38794f1c756547cd7359752e49b3.jpg

 

As you can see the result is quite similar to the full stack although double in magnitude.

 

1 hour ago, davewantsmoore said:

Not that there is much you can do at this stage.... but, it might help you make sense of you in room measurements, which are going to be hard to interpret below a few hundred Hz.

 

At the moment the electronic filter is set at about 60Hz first order low pass so when I do run SPL sweeps it will be inclusive of this filter as well, unless I remove the filter which I hope not to have to do.

 

 

Link to comment
Share on other sites

52 minutes ago, acg said:

Done, and I've included the individual data points as bars.

Further (and no bars, just a line) .... although I think it's navel gazing from here, perhaps.     What we're looking for is between 80 and 500, where the impedance might wiggle up and down by 10, or 20%.   The absolute value in ohms is not important, it's the relative % change.

 

All this is working towards the problem where you will take in room measurements, and you will wonder "what causes that wiggle", and it will be hard to tell because you won't have reflection free data, and you will have low resolution data.

Link to comment
Share on other sites

1 hour ago, acg said:

At the moment the electronic filter is set at about 60Hz first order low pass so when I do run SPL sweeps it will be inclusive of this filter as well, unless I remove the filter which I hope not to have to do.

There shouldn't be any need to remove the filter.

 

The peak at 700hz is interesting. I wonder is it the round surface v. close to the driver.      Normally it would be out of the pass band, but you are only going to be less than 18dB down.

Edited by davewantsmoore
Link to comment
Share on other sites

45 minutes ago, davewantsmoore said:

Further (and no bars, just a line) .... although I think it's navel gazing from here, perhaps.     What we're looking for is between 80 and 500, where the impedance might wiggle up and down by 10, or 20%.   The absolute value in ohms is not important, it's the relative % change.

 

All this is working towards the problem where you will take in room measurements, and you will wonder "what causes that wiggle", and it will be hard to tell because you won't have reflection free data, and you will have low resolution data.

 

Here is a further zoom...I can't zoom any further in the software.

 

717390094_DATSFirstCannon(Left)Zoomed2.thumb.jpg.f9c25b7c4288f9661a39ef1f9ac012e2.jpg

 

100Hz to 500Hz is +/- 0.04r or +/- 3.3% max.  80Hz is 0.08r higher or 6.7%.  10% higher at 70Hz, 20% higher at 62Hz.  

 

50 minutes ago, davewantsmoore said:

The peak at 700hz is interesting. I wonder is it the round surface v. close to the driver.      Normally it would be out of the pass band, but you are only going to be less than 18dB down.

 

If I remember correctly, the stuffing and damping did not change that blip too much and I had deduced that it is something either inherent in the drivers or something to do with the shape of the Cannon...perhaps you are right about the cylindrical surface adjacent to the basket.  It may even be a magnetic interaction with the steel of the Cannon.  Looking at the datasheet there is a wiggle in that area of the graph but not a significant as it is in my circumstances.  The magnitude is 0.15r so about 12% deviation above the mean of the 100Hz-500Hz range, and the phase angle is in the 6-10 degree range across the blip so it should be reasonably easy for the amplifier to negotiate.

 

The wiggle at about 3k is shown both on the datasheet and in my measurements at approximately the same scale.

 

Link to comment
Share on other sites

3 hours ago, acg said:

Here is a further zoom.

Ace

Quote

something to do with the shape of the Cannon

They're about 25cm wide?!  ;) 

Quote

...perhaps you are right about the cylindrical surface adjacent to the basket.

On second thoughts, I doubt it.

Quote

It may even be a magnetic interaction with the steel of the Cannon.

Wouldn't be at such a specific Hz.

Quote

Looking at the datasheet there is a wiggle in that area of the graph

Could be related

Quote

so it should be reasonably easy for the amplifier to negotiate.

This isn't the concern.   We're looking for (potential) issues in the motion of the driver / acoustic output....   This is reflected in the impedance curve of the driver  (like if you touched your finger gently on the cone, that would show up in the impedance data as a blip).

 

... not that these things are necessarily a huge concern, or anything you can do a lot about now.    Just interesting.... and in some ways all you have now.   It's very difficult to get accurate/useful acoustic data in room.... where as the impedance data is more straight forward.

Edited by davewantsmoore
Link to comment
Share on other sites



2 hours ago, davewantsmoore said:

 It's very difficult to get accurate/useful acoustic data in room.... where as the impedance data is more straight forward.

 

Well I'm certainly not taking them outside to measure...that is half a day to take one stack apart, cart them individually down the stairs and outside...then a day to build them again...measure for a little while...half a day to pull them down and cart them back upstairs and a day to put them back together.  That is 3 or 4 days work that I am not up for....haha.  I could take one Cannon off and waste a day doing measurements for something that is unlikely to be changed.  I am confident that I have the impedance behaviour as benign as I can make the stack in this particular situation.   On the datasheet the drivers show good directivity to about 1kHz  which is 4 octaves into the low pass filter so the tonally poorer stuff will I guess start at about -24dB which I figure is a lot better than anyone could ever get with a 15" (-18dB) or 18" (-12dB) driver combining with a first order filter.  It is the size of these Scanspeak drivers that give a first order filter some hope of success in this situation, but if it does end up to be a problem I can always increase the slope.

 

 

2 hours ago, davewantsmoore said:

They're about 25cm wide?!  ;) 

 

10" steel pipe so about 263mm internal diameter from memory.

 

 

I guess I'll have to go measure something.

 

 

  • Like 1
Link to comment
Share on other sites

1 hour ago, acg said:

On the datasheet the drivers show good directivity to about 1kHz  which is 4 octaves into the low pass filter so the tonally poorer stuff will I guess start at about -24dB which I figure is a lot better than anyone could ever get with a 15" (-18dB) or 18" (-12dB) driver combining with a first order filter.  It is the size of these Scanspeak drivers that give a first order filter some hope of success in this situation

I don't think I understand what you're getting at here....

1 hour ago, acg said:

10" steel pipe so about 263mm internal diameter from memory.

Your blip is almost certainly due to the width of the pipe.

Link to comment
Share on other sites

Some room SPL measurements with the trapdoor down and the windows closed.  The left Bass Cannon driven by its dedicated channel in the DSET amplifier, with a first order low pass filter applied from about 60Hz.  The correct operation of the amplifier and the filter is yet to be measured and confirmed...another day. 

 

 

1135613763_TrapdoorDownwindowsallclosed..thumb.jpg.484eaf6266e21b4daaa60368c1f5f9a1.jpg

 

So in-room I have good response down to 18Hz where it drops off at -12dB/octave (not shown) pretty much to DC. 

 

Looking at the REW Room Simulation below I have an axial (width) mode just below 30Hz that is completely alleviated by listening exactly in the centre of the two side walls.  Then there is a big longitudinal reinforcement mode at about 40Hz between the front and back walls, a tangential suckout circa 50Hz, axial reinforcement at 57Hz, tangential mode at 70Hz, suckout at 75Hz, reinforcement at 79Hz et cetera.  I am quite amazed at how closely the room sim has predicted this basic SPL behaviour and it looks like it gives me very good information regarding the causes of those particular humps and hollows, especially considering my ceiling is not flat at 2.4m but is peaked at 3.3m.

 

To be honest I was not expecting the effects of the longitudinal nodes to be so strong in my measurements because of the windows behind the listening position, but they most certainly are quite strong. 

 

1442271571_REWRoomSimulation.thumb.jpg.ebf3d0a82c3d3e25f9bd22826715d07f.jpg

 

To test a few things I brought up a couple of still closed bags of Greenstuf R2.5 polyester insulation which I intend to use as soft/fluffy inside the bench traps that go under the windows along the entire northern wall.  They are still compressed in their plastic packing but are good enough just to give me an idea of potential effects.  Have a look at the next post.

 

Edited by acg
  • Like 1
Link to comment
Share on other sites

So, I traipsed two closed bags of Greenstuf R2.5 batts into the room and experimented with them in different parts of the room.

 

Green Line = control, no Greenstuff bags in room, trapdoor down, windows closed

Red = one bag of Greenstuff in both room corners behind the bass cannons.  At 1.2m high, the bags do not go from floor to ceiling.

Blue = both Greenstuff bags behind the active Bass Cannon stacked floor to ceiling

 

As you can see, the Greenstuff in the corners behind the speakers is having a big effect on the horizontal tangential modes at 70Hz and 95/98Hz in particular (50Hz is not improved) and has a tremendous effect on the 65Hz dip which may be a tangential/height mode with my cathedral ceiling.  This is very encouraging!

 

 

292370571_SPL2GSbags.thumb.jpg.c9386a6b906e431184fbd41a18ad1cb0.jpg

 

So, from what I understand, my most difficult bass problems are likely to be the 41Hz longitudinal reinforcement, the 50Hz tangential/horizontal suckout (although I do have good hopes for this with more Greenstuf in the corners), the 57Hz axial/horizontal reinforcement and the group of axial, longitudinal, tangential/horizontal modes that accumulate at 114Hz-119Hz.  To control axial modes I will build some VPR's.  Likewise it might be best to use VPR's on the wall between the speakers to alleviate the longitudinal modes.  I'll see how I go, but I had always planned for VPR's in the corners behind the Cannons, but the Greenstuff was just so effective and I might end up going that way.

 

One thing was clear:  just having those two bags of Greenstuf in the room shortened the decay times in the bass frequencies remarkably, and it almost did not matter where they were put.  Below is the waterfall plot of the standard room with the trapdoor down, windows closed and no Greenstuf...

 

 

1772401563_ControlRoomWaterfall.thumb.jpg.0c4de768462c9eedf530506da24d0175.jpg

 

Then add two bags of Greenstuf stacked floor to ceiling behind the active Bass Cannon...

 

716449209_Waterfall2bagsGreenstufbehindactiveCannon.thumb.jpg.5451558924aa7d316934b2b0368512e6.jpg

 

Edited by acg
  • Like 1
Link to comment
Share on other sites

1 hour ago, acg said:

So in-room I have good response down to 18Hz where it drops off at -12dB/octave (not shown) pretty much to DC. 

Your room is quite well sealed.

 

Below 40Hz we expect the response to decrease by 12dB/octave

Below 30Hz we expect the response to increase by 12dB/octave

 

... and we see exactly that... for half octave below 40, it drops ~12dB/octave .... then it is flat, down to ~16hz, below here you walls are absorbing/leaking energy.... and to continue to get a flat response, you would need to put in more energy from the speaker.

 

1 hour ago, acg said:

I am quite amazed at how closely the room sim has predicted this basic SPL behaviour

Heh.... people only have to play with this stuff a short time, to flip ... and be amazed that people are amazed.  ;) 

 

Because the wavelengths are so huge, then statistically the response reduces down to something relatively simple.

 

1 hour ago, acg said:

To be honest I was not expecting the effects of the longitudinal nodes to be so strong in my measurements because of the windows behind the listening position, but they most certainly are quite strong. 

Things are conspiring to make the peak look bigger than it is.     ie.    How much of the "peak" at 40hz is the reinforcement itself, and how much is due to the strong dips either side (the cancellation at 50, and the roll off below Fc).    I guess your peak is ~5dB .... which means it about half the strength (-3dB) predicted..... so in that sense you could say "your windows are absorbing half the energy" (which is, like you expected, significant).

 

2 hours ago, acg said:

To control axial modes I will build some VPR's

EQ is a very valid way to solve these issues.... especially seeing as you are likely to have less position dependance than most.

 

In many respects, there's not even a reason to think that traps will solve the issue better than EQ.     After you've built what traps you can accommodate and tune ..... then I would strongly recommend that you prototype some simple PEQ filters on your computer, and then translate these into your line-level passive filter.   

 

Remember.... correcting these peaks with EQ, will also correct the ringing.   ?     If you are unsure about whether that is true (which is common) .... then EQ a similar peak (to a room reinforcement) into the response ... and see what the waterfall looks like  ;) 

 

2 hours ago, acg said:

One thing was clear:  just having those two bags of Greenstuf in the room shortened the decay times in the bass frequencies remarkably

It may see 'pedantic' .... but it reduced the magnitude of the modal peaks ..... and the "ringing reduction" was simply a consequence of that.

 

If you meditate on this, you may see the difference in 'how to think about it' is enlightening.

1 hour ago, acg said:

To test a few things I brought up a couple of still closed bags of Greenstuf R2.5 polyester insulation

?   The  'under bench' trap is a very good idea.     Is the 'northern wall' behind you? (sorry, I'm not paying attention).   If so, even better.

 

  • Like 2
Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top