Jump to content

Deep End DIY - My first speaker project


acg

Recommended Posts

On 31/08/2017 at 8:05 PM, acg said:

How to correct it though?  Because the wiggles happen when there is a certain threshold of high density stuffing in the tube all that I can think of is to try different types of stuffing....

 

 

 

Yeah, sounds like go lower in density.   Something around 10kg/m^3

 

The shape of your enclosure becomes relevant as you go up to non-subwoofer frequencies .... and the enclosure is a bad shape.   So it will be a battle.

 

On 02/09/2017 at 11:22 AM, acg said:

I've added REW "dB Offsets" to make the three lines meet roughly at their maximum volume circa 150Hz.

 

Align them at the lowest frequencies.   (Hint:  It didn't give you more bass output)

 

On 02/09/2017 at 2:24 PM, georgepapa said:

Hello bass extension, and it is likely that some of the impedance issues could be minimised

 

 

I don't really see how a port will solve any of the wiggles being discussed here  (like you say, it's the shape of the enclosure) .....   I see the merits of ports being only where you absolutely need the extra efficiency gain.    This system absolutely doesn't with 16 drivers planned.

 

On 02/09/2017 at 2:38 PM, 125dBmonster said:

open back would do

 

Nah.

 

If you were going to move to a BR design, then it would be important to actually design a properly working port.

 

Simply opening the back would make it a dipole, which would mean less bass .... and a dipole peak to worry about - the two way that is deal with are, either EQ (incompatible with Ant goal), and not using the enclosure above that dipole peak frequency (also incompatible with Ant goal).

 

Link to comment
Share on other sites



10 hours ago, davewantsmoore said:

 

 

 

Yeah, sounds like go lower in density.   Something around 10kg/m^3

 

The shape of your enclosure becomes relevant as you go up to non-subwoofer frequencies .... and the enclosure is a bad shape.   So it will be a battle.

 

 

Yes, it is proving to be a little battle, but it seems as though I am winning.  I just climbed up into the attic and checked the label of the Greenstuf R1.5 wall batts that are doing a good job in the Cannon and I calculate the density to be 7.8kg/m3 which is about one-third of the Bradford Soundscreen batts I initially used (25.71kg/m3).  I have found some glassfibre batts with a lower density (6kg/m3) and do wonder if I should give them a try.  The Greenstuff are polyester batts but have the advantage of being able to really stand on their own inside the tube and I doubt they will deform over time and leave a gap at the top, unlike low density glassfibre which I am not so sure will stand on their own.  I have also trialled a gap at the top of the tube and it does not work as well as coverage of the entire cross-section of the tube.

 

 

10 hours ago, davewantsmoore said:

 

Align them at the lowest frequencies.   (Hint:  It didn't give you more bass output)

 

 

No, I am under no illusions that I magically increased the bass response with that high density stuffing.  Almost certainly I have absorbed the higher frequencies in order to get that flatter SPL response line, which in itself could be a good thing I think for this project, but it comes at the expense of the non-linear SPL response at 150Hz+ in the crossover filter range.  Still, those HD batts are a weapon in my arsenal should the need arise, especially if I end up with a higher order low pass filter for the channel and those SPL wiggles are not so important anymore.

 

Link to comment
Share on other sites

:) Can't help thinking about this but feel a little more can be done about the length the sound waves must travel inside the canon. 

The little drawing of the baffles in the enclosure to lengthen and diffuse the path will most likely have a better effect than varying densities of stuffing, or even both used together.

The suggestion of even a spiral in the cannon would be of note as well. Is all about making it further and harder for any resonance  in the enclosure to effect the electrical impedance of the driver and/or get back to the back of the driver

Just thoughts.

Link to comment
Share on other sites

How about some half circle baffles, odd spacing and 180deg to each other?

So if you looked at it from an open end you wouldn't see past the second baffle. Half the tube would be covered by the first baffle then the other half would be covered by the second baffle rotated 180deg.

 

This should help stop any standing waves in the tube. 

Link to comment
Share on other sites

So here are the measurements with various amounts of Greenstuff Polyester Batts stuffed inside the tube:

 

First off, the impedance measurements:  Orange = no stuffing; Yellow = 5 Sheets; Orange (lower) = 4 sheets; Cyan = 6 sheets; blue = 7 sheets.

59ac79c9f0e37_GSComparison.thumb.jpg.3fc4394e1b00577c74d97d70a3eca44f.jpg

 

There is no real difference in the impedance response  between 5, 6 or 7 sheets which indicates that 5 sheets is enough to sort out the pipe resonance.

 

The SPL graph has been adjusted so that each of the measurements has been roughly equalised at 100Hz which is the upper end of the expected crossover frequency range.  

 

59ac7c87de926_GSMoremeasurements.jpg.1cebd0c0a9023748e485d3ba54d7934e.jpg

 

 

There is not much in it, but any of 4 sheets to 7 sheets will give an acceptable frequency response (the low pass filter start at  about 80-100Hz), but 5 or 6 sheets seems the most optimal when using low output impedance SS amplification.

 

 

Link to comment
Share on other sites



26 minutes ago, 125dBmonster said:

:) Can't help thinking about this but feel a little more can be done about the length the sound waves must travel inside the canon. 

The little drawing of the baffles in the enclosure to lengthen and diffuse the path will most likely have a better effect than varying densities of stuffing, or even both used together.

The suggestion of even a spiral in the cannon would be of note as well. Is all about making it further and harder for any resonance  in the enclosure to effect the electrical impedance of the driver and/or get back to the back of the driver

Just thoughts.

 

1 minute ago, nzlowie said:

How about some half circle baffles, odd spacing and 180deg to each other?

So if you looked at it from an open end you wouldn't see past the second baffle. Half the tube would be covered by the first baffle then the other half would be covered by the second baffle rotated 180deg.

 

This should help stop any standing waves in the tube. 

 

They are both good thoughts, but I would much prefer to sort this out easily without much more modification of the tubes.  Cutting batts is easy...installing baffles is not so easy.  The pipe resonance is at about 190Hz so I should be able to control it without baffles...hopefully...at least it seems so right now unless I am missing something important.

Link to comment
Share on other sites

I actually have an enclosure designed to a similar principal, it's a stand mount full range enclosure (off set baffles for rear of enclosure) may be easier if after work today take a photo or 2, all for the better good of learning something slightly just slightly out side of the square.. Can understand the desire to not sink even more time into building the stereo, just ideas

 

EDIT, enclosure has a good usable response to 150Hz by the measurements to my eyes, what is the high pass set for  at the transition in design ?

Edited by Guest
Link to comment
Share on other sites



Hi Anthony, I'm pretty sure someone mentioned pillows earlier but I stuffed my enclosures with poly fill from some Kmart pillows (the cheapest ones they do - $5 for two!) and it seems to work really well. I used about a pillow per side from memory.

 

Like yours the enclosures are 12" cylinders, but only about 45cm deep so the problem frequencies will be higher up than yours but I can't detect any issues from near field measurements or the impedance sweep with DATs. Worth a try maybe given that it's so cheap and easy?

 

Here's the impedance curve and the frequency response with a slight (16dB) low shelf boost. 

 

59ad3ada759e0_ScreenShot2017-09-04at9_13_41pm.thumb.png.4590ed7a4ea1d849a8403f3bb61c62d4.png

 

59ad3ade00466_MorelBoosted.jpg.40df192be3b1017e9081d4bb481e911c.jpg

  • Like 1
Link to comment
Share on other sites

Thank's for chipping in @NADMAD.  I'm a bit jealous at your lack of problems with a tube subwoofer.  Yes, I am sure that the 1m deep tubes allow a longer wavelength with more energy to cause the resonance or feedback.  Also, the drivers I am using have quite a soft surround and are probably a little easier for that standing wave to influence...sort of a perfect storm.  Anyway, I think I am though it.  Hopefully soon I will find some time to stack more cannons on top and then commence with some more experiments.

Link to comment
Share on other sites

On 9/4/2017 at 8:07 AM, acg said:

The SPL graph has been adjusted so that each of the measurements has been roughly equalised at 100Hz which is the upper end of the expected crossover frequency range.

 

You should choose a frequency where you know that each have the same output  (low frequency).

 

Doing it at 100Hz makes the chart v.difficult to interpret.

 

Link to comment
Share on other sites

1 hour ago, acg said:

Thank's for chipping in @NADMAD.  I'm a bit jealous at your lack of problems with a tube subwoofer.

 

Check the ohm scale on his chart and your chart ;)

 

1 hour ago, acg said:

I am sure that the 1m deep tubes allow a longer wavelength with more energy

 

There is no more energy.    It is just the dimensions wrt to the wavelengths.

 

On 9/4/2017 at 8:54 AM, acg said:

Low pass will be 80 - 100 Hz.

 

Looks to be like a filter with ~6dB/octave attenuation in the passband will do the trick well to EQ the driver

 

... as for the "crossover", you'll need to see your next driver up before you can think about that.

Link to comment
Share on other sites

13 minutes ago, davewantsmoore said:

 

Check the ohm scale on his chart and your chart ;)

 

Yes, I did notice that, but the two impedance lumps that stand out like dogs bollocks on my initial impedance sweep would even show up at the scale of NADMAD's graph.

 

16 minutes ago, davewantsmoore said:

 

Looks to be like a filter with ~6dB/octave attenuation in the passband will do the trick well to EQ the driver

 

... as for the "crossover", you'll need to see your next driver up before you can think about that.

 

Still gotta finish that horn...seems like I am putting it off but I really am not.

 

30 minutes ago, davewantsmoore said:

 

You should choose a frequency where you know that each have the same output  (low frequency).

 

Doing it at 100Hz makes the chart v.difficult to interpret.

 

 

Ok, I can align them at 20Hz or so where the mic calibration seems to drop off...see next post.  I've not been too worried about actual SPL levels thus far, more about the slope and shape of the rolloff either side of 150Hz or so.

 

 

 

Link to comment
Share on other sites



@davewantsmoore, here is the graph with a new alignment at 20Hz.  In addition, I have included the "gold" line which is 4 sheets of 25kg/m3 high density fibreglass stuffed at the far end of the tube.  You can see the dramatic difference in SPL to the other lines as the frequency rises but it is a pity about that second peak at about 330Hz.  I wonder if I should go back to the 4 sheets of HD at the end of the tube and put some low density stuff in front of it to see what happens with that second SPL peak?  It almost looks as though that peak is being reflected by the high density stuffing.

 

59ae07764e9c3_GSMoremeasurements2.jpg.f54c960b7a38b9ab2057c20f5f43b10c.jpg

Link to comment
Share on other sites

Paul @Silent Screamerwas in town yesterday and called in with some of his soft acoustic foam.  What do two audiophiles do when they get together like this?  Apparently listening to music is not the done thing...we stuffed and re-stuffed the Bass Cannon that I have under test.  Pauls acoustic foam is very soft, more soft than the Greenstuff I am using, but also less rigid and more difficult to pack into the tube so that its cross section is filled.  In the end its performance for this application was very similar to the Greenstuff, and the GS is much easier to use but we did learn some potentially useful things about tube stuffing.  

 

Most pertinent was the realisation that if the stuffing comes right up close to the back of the driver then we get more SPL at resonance and a noticeably more linear decay slope in the 150Hz+ region.  This is all in the filtered band but could be useful when tuning the system.  Also, placing an amount of stuffing near the driver is quite effective at reducing Fs...we got it down to 43Hz a couple of times...so I now have the means of tuning the resonance from 43Hz to probably 70Hz just by using different stuffing techniques.  When I get the tube amplifier into the testing and also start thinking about to circa 40Hz room node that I have at my listening position, I may have a useful tool to EQ things a little with the stuffing type and placement within the tubes.

 

Anyway, thanks for dropping by Paul, it was good to see you again. 

  • Like 2
Link to comment
Share on other sites

On 9/5/2017 at 0:19 PM, acg said:

... it is a pity about that second peak at about 330Hz.

 

Some may be tempted to dismiss this as semantics, but these are not peaks.    They're cancallations (at ~210Hz and its harmonics).    If you had a higher frequency resolution,  you could see they extend much lower (in theory, to zero - representing a complete interference cancellation).

 

I would surmise they are larger in the gold line becuase this stuffing is less absorbent than the other type of stuff (?!).

 

 

In a super-high performance system they are definitely worth considering .... although the dB scale belies their size ;)

 

 

Link to comment
Share on other sites

15 hours ago, davewantsmoore said:

 

Some may be tempted to dismiss this as semantics, but these are not peaks.    They're cancallations (at ~210Hz and its harmonics).    If you had a higher frequency resolution,  you could see they extend much lower (in theory, to zero - representing a complete interference cancellation).

 

 

I don't think so Dave.  There is no smoothing applied to any of my graphs yet and there is plenty of resolution in the signal tone to display complete cancellations much more accurately than that rolling gold line depicts.

 

15 hours ago, davewantsmoore said:

I would surmise they are larger in the gold line becuase this stuffing is less absorbent than the other type of stuff (?!).

 

 

The high density stuffing is this stuff...the Greenstuf is low density polyester and has no acoustic properties that I can find.  I strongly doubt that the GS is more absorbent than the HD stuff.

 

To throw a little further light on what is happening have a look at these two graphs (summary below):

 

59b1b5f21ef29_MoreHDandGSBatts.jpg.e389e1dc22b9a9f3288ffef8ee25b6d9.jpg

 

Orange (above & below):  empty tube with zero stuffing

 

Magenta (above & below): 4 discs of HD at far end of tube

 

Blue (above & below): 4 discs of HD at far end of tube + 5 discs of GS behind driver (a little too close judging by the impedance wiggle)

 

Cyan (above & below): 4 discs of GS at middle of tube and 4 discs of GS behind driver

 

Discussion:  The HD batts definitely absorb more sound:  the SPL is lower at all frequencies especially around peak SPL and that first "cancellation".  1.5dB down at the SPL peak is a relatively large reduction But adding GS behind the driver raises the peak SPL (it lowers Fs) and flattens out the response through the waves above 150Hz.  I think that the HD batts encourage some sort of non-linear absorption but have no evidence other than the SPL graph:  perhaps the phase of the cancellations is diffused by the HD batts at the far end which is causing the more sinusoidal shape of the SPL response rather that the sharp cancellation (but we don't see any sign of the cancellation in the raw SPL response).  The all GS scenario further lowers Fs and gives more SPL at peak and a smoother response 150Hz+.  Below you can see the impedance graphs each with their own wiggles, but the Blue line is probably the best. 

 

59b0f28b02f15_MoreHDandGSBatts.thumb.jpg.a523328cabeb272da795631261d7654a.jpg

 

 

Link to comment
Share on other sites



Now that I have seen the real mccoy I am so looking forward to seeing them finished, even half done they look like a real professional job. They look so good that if you weren't following this thread and you saw them at an audio show you would easily mistake them for mega dollar speakers and amps.

 

Despite having seen pictures of the amp parts I was still suitably gobsmacked seeing the parts in real life. These amps are on some serious steroids. Top job Anthony :thumb:

  • Like 2
Link to comment
Share on other sites

  • 2 weeks later...
On 08/09/2017 at 7:13 AM, acg said:

I don't think so Dave.  There is no smoothing applied

 

Understood - I did not say, or mean to infer that there was any smoothing involved....  I am talking about the frequency resolution  (the number of data points per Hz).    Higher would show the dips more accurately.

 

The SPL measurement error is potentially a consideration.  <0.5db can be measurement error  (it's worth investigating in your setup, anyways).

 

The higher damping of the driver looks good, although it's good to begin without these issues (and not having mechanical damping added to remove them).

 

 

Point being.   I wouldn't stress too much about this.   More importantly is the implementation of the filters to achieve the crossover, and how the driver and summed responses look in the listening position, and all that. :)

Link to comment
Share on other sites

  • 2 weeks later...
On 19/09/2017 at 10:07 PM, davewantsmoore said:

 

Understood - I did not say, or mean to infer that there was any smoothing involved....  I am talking about the frequency resolution  (the number of data points per Hz).    Higher would show the dips more accurately.

 

The SPL measurement error is potentially a consideration.  <0.5db can be measurement error  (it's worth investigating in your setup, anyways).

 

The higher damping of the driver looks good, although it's good to begin without these issues (and not having mechanical damping added to remove them).

 

 

Point being.   I wouldn't stress too much about this.   More importantly is the implementation of the filters to achieve the crossover, and how the driver and summed responses look in the listening position, and all that. :)

 

I have been measuring using a 256k log sweep from DC to 20kHz.  I've not stopped to figure out if that is enough to show the cancellations, but it certainly does show them at higher frequencies.

Link to comment
Share on other sites

I've been so busy at work, charging 80-90 hours each week for the few weeks and I am completely knackered and of course have no time for my audio hobby at the moment even though I am itching to get in and build these amps and speakers.  Yesterday a parcel (finally) arrived that contained the freshly hard chrome plated rollerblocks for the horizontal vibration isolation.  This morning I could not resist and took an hour off work to install a few into the base of the amplifiers and onto their allocated positions on the horn stack and then loaded the amplifier base onto the stack.  Wow...they work well...give the base a tap and it will rock away there for at least a minute.  I put one of the ball bearings in a dome on my desk and managed to eat all my Weet-bix before the ball stopped jiggling.  If I wave my hand past the shift in air set the ball off rolling on the rollerblock so it is an excellent result.

 

 

Rollerblock all chromed up...

59cc1c7a2e929_RollersChromed.jpg.e9dc1f35da5b3ba48c279644ff1fea36.jpg

 

In-situ on top of the pneumatic isolator...

 

59cc1c78b4b2b_Insitu.jpg.e96a4dda88bfd36c9256fc928975cf45.jpg

 

 

...and a view from a little way back showing the whole amplifier base sitting on the horn stack...

 

59cc1c7745a0b_Insitu2.jpg.d1270630cddffc95ad1344bbc8f041fe.jpg

 

Late one night I also succumbed to my desire to build and installed a lot of the bias pots and l-pads into their positions...

 

Knobs.jpg.74e536a49377d9867c06a7f7e2dc6c35.jpg

 

Knobs2.jpg.af572d6a0cfb0a1c92e77da65ff205d1.jpg

 

An executive decision was also made regarding the input to the amplifier with both BNC and RCA inputs installed.  There were a great number of options for the inputs swirling through my mind with thoughts of buffers and multiple interconnects but in the end I decided to just keep it simple.

 

59cc1c735a5e6_AmplifierInput.jpg.7681bf2e646d52eaa855d61b48bee215.jpg

 

Back to work now!

 

 

  • Like 3
Link to comment
Share on other sites

On 28/09/2017 at 7:23 AM, acg said:

I have been measuring using a 256k log sweep from DC to 20kHz.  I've not stopped to figure out if that is enough to show the cancellations, but it certainly does show them at higher frequencies.

The length of the log sweep doesn't have any effect on the frequency resolution.

 

Take a look at the IR window settings.    The other things which will affect them (fill them up) is reflected sound with different phase.

Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.




×
×
  • Create New...
To Top