Keystone Sub Using 18, 15, & 12 Inch Speakers

1)In the case of the KS, wouldn't that result in the mouth facing upwards?
2)How different though is the loading on the driver between the KS and the TH18?
3) My POC3 uses the same fold at the TH18. I may be imagining things, but I've noticed a difference when it's placed on its side compared to when it's upright. I thought that might have something to do with the bottom panel not being against the ground when the sub is used on its side (of all the panels, the side panels flex / vibrate the least) . I've never tried to quantify it by measurements though. I'll try to do that this weekend.
Brian,

1) Yes.
2) The Keystone has a more rapid drop in frequency response below F3 than the TH18 or DSL TH118, indicating the loading is a bit different.
3) The differences you notice could be due to flex or the effect of different dispersion due to proximity of the exit to the ground plane. There was a fairly significant difference in the smoothness of the upper response of the Keystone when measured on it's side or upside down compared to the "normal" apex up configuration.
 
Brian,

1) Yes.
2) The Keystone has a more rapid drop in frequency response below F3 than the TH18 or DSL TH118, indicating the loading is a bit different.
3) The differences you notice could be due to flex or the effect of different dispersion due to proximity of the exit to the ground plane. There was a fairly significant difference in the smoothness of the upper response of the Keystone when measured on it's side or upside down compared to the "normal" apex up configuration.

I'd expect the results you saw with the KS in (3), as the mouth does not have rotational symmetry. The square(ish) mouths of the TH* subs do have this symmetry, so any changes in FR when rotated near to a planar surface (e.g. the ground) should not produce any significant changes in FR due to a change in loading at the mouth.

Panel flex however will introduce some changes, which should show up in the distortion plot as well as FR.

Here's a quick and dirty FR I did of my POC3 (which shares a similar layout to the TH* designs) in upright and sideways position. FR seems to be basically identical up to 150 Hz, but the distortion curve is different. Above 65 Hz, the distortion is actually HIGHER for the "sideways" position, but that may be due to the fact that I've done some changes to my PC to ensure that the mouth's bottom panel only touches the ground on one edge, which should minimize any additional distortion caused by the panel slapping against the ground.

Or who knows - maybe the change in distortion is due to the change in orientation of the driver :). I'll have to think up another experiment to eliminate that as a possibility...
 

Attachments

  • 20150512-POC3 orientation.png
    20150512-POC3 orientation.png
    20.7 KB · Views: 585
Panel flex however will introduce some changes, which should show up in the distortion plot as well as FR.

Here's a quick and dirty FR I did of my POC3. Above 65 Hz, the distortion is actually HIGHER for the "sideways" position, but that may be due to the fact that I've done some changes to my PC to ensure that the mouth's bottom panel only touches the ground on one edge, which should minimize any additional distortion caused by the panel slapping against the ground.
Brian,

All the subs I have tested have progressively less distortion in the upper passband, in the range of 1-2% above 100 Hz.
Distortion exceeding 100% above 130Hz indicates something is definitely wrong with the POC3 or the test gear.


Art
 
Last edited:
The only thing I changed during the testing was the orientation of the subwoofer.
With as much upper distortion as your subwoofer indicates, unless there is a problem with the test gear, I'd suspect a loose panel or dragging voice coil to be at fault, either of which could change with the orientation.

A slow sine wave sweep makes either of the above mentioned problems easy to identify.
 
With as much upper distortion as your subwoofer indicates, unless there is a problem with the test gear, I'd suspect a loose panel or dragging voice coil to be at fault, either of which could change with the orientation.

A slow sine wave sweep makes either of the above mentioned problems easy to identify.

I believe HolmImpulse's distortion graph shows distortion relative to the main signal. So I don't think you're seeing "100% distortion" there. The DSP in the amp is also providing something like a 48dB/oct LP signal @ 100 Hz or thereabouts. I'll have to check.
 
1)I believe HolmImpulse's distortion graph shows distortion relative to the main signal. So I don't think you're seeing "100% distortion" there.
2)The DSP in the amp is also providing something like a 48dB/oct LP signal @ 100 Hz or thereabouts. I'll have to check.
Brian,

1) Distortion relative to the fundamental signal means when they are equal level, distortion is 100%. Your graph displays THD crossing the fundamental signal both in the low range (expected) and in the upper range (unexpected).
-10 dB is 31.62% distortion, -20 is 10% and -40 =1% etc.
The Keystone (or bass reflex) loaded with the B&C or Lab 12s has distortion around -40 dB in the upper range.

2)Your graph does not appear anything like "48dB/oct LP signal @ 100 Hz or thereabouts." That said, panel flap or rubbing voice coils don't really care what the electrical crossover does.

Art
 
1) Distortion relative to the fundamental signal means when they are equal level, distortion is 100%. Your graph displays THD crossing the fundamental signal both in the low range (expected) and in the upper range (unexpected).
-10 dB is 31.62% distortion, -20 is 10% and -40 =1% etc.
The Keystone (or bass reflex) loaded with the B&C or Lab 12s has distortion around -40 dB in the upper range.

I don't think that when the curves cross that distortion is 100%. That level of distortion would be VERY audible and my ears aren't that bad yet :). Here's another measurement I took of the same sub, in upright position with all DSP turned off. Check the distortion curve..

2)Your graph does not appear anything like "48dB/oct LP signal @ 100 Hz or thereabouts." That said, panel flap or rubbing voice coils don't really care what the electrical crossover does.

The DSP section of the amp was applying a 100 Hz 24dB/oct LP filter to the signal in my previous measurements.

Note, both tests were done indoors, so there may be some in-room resonances that are adding their character to both the FR and distortion graphs.
 

Attachments

  • 20150512-POC3 distortion2.png
    20150512-POC3 distortion2.png
    18 KB · Views: 544
I don't think that when the curves cross that distortion is 100%. That level of distortion would be VERY audible and my ears aren't that bad yet :).

As a sanity check, here's a TrueRTA measurement at 138 Hz, just above where the FR and distortion curves cross. Notice the level of the distortion components next to the fundamental. I moved the mic a bit closer to the subwoofer to get the level up a bit. Ain't no way that could be "100% distortion".
 

Attachments

  • 20150512-POC3 distortion3.png
    20150512-POC3 distortion3.png
    62 KB · Views: 547
What was the orientation of the subwoofer during this test?

Good question! Turns out it was mounted upright instead of sideways for that measurement. However, at that frequency there's less than 3dB of difference in the FR graph and the distortion graph for the previous measurement that I did with the sub mounted upright. If Art's interpretation of what the graph is supposed to be showing is correct, there would be gross distortion components showing up in the RTA measurement done at that frequency (-3dB THD =~ 71% THD). The distortion components actually observed at that frequency in the RTA measurement will add up to nowhere near that. The highest component is just over 50dB lower than the fundamental.

I will repeat the test sideways when I get the opportunity. I think I have an idea of what may have contributed to the increase in distortion for the sideway measurement however (large flat vibrating panel + uneven floor). Maybe we should take a page from the HiFi people and use 3 spiked feet per sub... :)
 
I repeated the upright / sideways test, this time correcting one major thing - the gain on the mic was set too low! The output has DSP applied to it to address the out of band peaks and smooth the TH's passband a bit, but the HP and LP filters are turned off. And the FR is smoothed by HolmImpulse's default smoothing algorithm.

In red is the FR/distortion curves with the POC3 upright
In blue is the FR/distortion curves with the POC3 on its side

The blip in the distortion around 100 Hz is due to panel flex. It decreases when I add about 185 lbs of weight to the TH (by sitting on it, LOL). I haven't figured out which panel is the main culprit yet, but my bet is that it's the bottom one. At frequencies nearer to the lower end of the TH's passband, the POC3 shows less distortion when mounted sideways, though I don't think the difference would be audibly significant.
 

Attachments

  • 20150514-POC3 orientation.png
    20150514-POC3 orientation.png
    14.4 KB · Views: 576
5) Having only looked at measurements of Th18 or an Othorn, I'd expect them to be similar, with the Keystone having a bit (2 dB, IIRC) more output than the TH18 according to PASC. The Othorn goes lower, which could translate to sounding less "punchy". Art

Art, any idea how the pal12 would sound if widened to around 22 inches and fitted with an 18" ? Do you know if anyone did a sim for it with an 18? I am wandering how it would compare with a TH18 or a KS if it had an 18" driver.

I am also wandering if the keystone shape would increase SPL to a pal12, considering you settled on that shape after much experimentation. The exit on the pal12 seems pretty small. That KS shape seems to have the magic.
 
Art, any idea how the pal12 would sound if widened to around 22 inches and fitted with an 18" ?

You might be able to get close in Hornresp by taking the areas of the Pal12's S1, S2, etc. and adding the percentage of area you get by widening the enclosure to your desired dimension. If, for instance, you're adding 15% to the internal areas of the Pal12, internally - you may be able to add 15% to the Pal12's S1, S2, etc., then enter some 18" driver parameters to see what comes of it.

Someone might chime in to tell me how accurate/inaccurate this method might be.
 
You might be able to get close in Hornresp by taking the areas of the Pal12's S1, S2, etc. and adding the percentage of area you get by widening the enclosure to your desired dimension. If, for instance, you're adding 15% to the internal areas of the Pal12, internally - you may be able to add 15% to the Pal12's S1, S2, etc., then enter some 18" driver parameters to see what comes of it.

Someone might chime in to tell me how accurate/inaccurate this method might be.

It should work out fine, once the only dimension you are changing is the width and you're using "par" segments for the HornResp model.

Trouble is, once you load the driver in the model, you may find that you need to increase or decrease the ratio between S1 and S4 to improve the response, and that would require more substantial changes to the box.
 
1)Art, any idea how the pal12 would sound if widened to around 22 inches and fitted with an 18" ?
2)Do you know if anyone did a sim for it with an 18?
3)I am also wandering if the keystone shape would increase SPL to a pal12, considering you settled on that shape after much experimentation.
1) Depends on the 18" used, some will sound like crap, others would work OK.
2) I doubt it, since the pal 12 thread was about a Lab 12 TH.
3) The pal12 uses a fold that would not benefit from a keystone shaped exit.
 
Art, any idea how the pal12 would sound if widened to around 22 inches and fitted with an 18" ? Do you know if anyone did a sim for it with an 18? I am wandering how it would compare with a TH18 or a KS if it had an 18" driver.

I am also wandering if the keystone shape would increase SPL to a pal12, considering you settled on that shape after much experimentation. The exit on the pal12 seems pretty small. That KS shape seems to have the magic.

I already had that idea. Here...

http://www.diyaudio.com/forums/subwoofers/242420-p-18-th-18tbw100-30x38x22-5-a.html
 
MRGSR, How many PAL12s would you say give the same SPL as a Lab horn?

I looked at the sim for the 18. As an 18 has twice the surface are of a 12, shouldn't that give at least a 3 db increase in SPL over the 12? It is pushing more air, twice the amount of air. Shouldn't that show up in the sim?