Moar!!

On 8/17/2018 at 7:39 AM, SME said:

Subs make plenty of sound above the XO point, and that sound can have a surprisingly big impact on overall quality.  Problems there can make distortion sound a lot worse.  For example, even fairly high harmonic distortion may go unnoticed by listeners if it is fairly uniform vs. frequency, but if certain notes have spikes in harmonic distortion due to horn resonances and whatnot, you'll get notes that tend to stand out, which is a lot more likely to be audible.

This has actually been a part of my concern with the response of the Skhorn. When looking at the HD, there seems to be a correlation between the 2nd and 3rd HD, and the response peak at 220hz.

Quote

I looked at the sim and it is quite close below 80Hz.
This one should have less efficiency and less output above 40Hz on paper. It should have a bit more output near tuning. I'd say the original sims as being the slightly bit louder cab overall for music apps or live sound (kick drum, percussion fundamentals, most of the content being above 40Hz etc.).

You didn't compensate for "Lossy LE" in HR. When done the sims look like this:

So, with the added baffle gain added in, the response will be quite similar between about 40hz and 110hz, with the grey line having more output below 40hz and less above 110hz.

On 8/16/2018 at 11:54 PM, Ricci said:

I've found over the years I like smoother responding wider bandwidth systems vs systems optimized for more sensitivity over a narrower band of operation. This is part of the reason I decided not to do a straight slot. I wanted extra upper bass output and extension well beyond 100Hz as part of the design goal. Often times you'll see guys using subs for 30-70Hz and another set of kick bins for 70-140Hz or whatever. Usually due to sub bins with degraded upper range behavior or just not enough output. I'm not a fan of that approach. I'd much rather have one cab that can handle everything. I think the sim you proposed would be fine. I tried to get the top end as extended as possible on this one.

This is an interesting topic. Now, my sim has good output to about 120-130hz, which should cover the demands of a sub in about 99% of the cases. But what is interesting is the topic of extension above 100hz... Now, the more 'orders' that one adds to a solution, the more the impulse response and the SQ typically worsens. So, my thinking is: If one has a design that already has a 2nd order roll off above ~110hz, and then adds a 4th order lowpass over that, don't we have a 6th order lowpass, which typically should sound worse? So, couldn't the solution with the SQ 'issue' with bandpass subs with a narrow band be to use less steep filters? The upside of this approach would of course be that the HD gets rolled off in a BP design. That HD won't get filtered away with a regular low pass filter. 
The above could also be part of the reason why you've found that you like the sound of subs with wider bandwidths more.

Quote

Another reason I didn't choose a large straight slot was to more fully enclose the drivers inside the cab. Most of the time I notice non harmonic mechanical and operational noises before HD gets to be offensive. The acoustic roll off of the slot above 100Hz should help lower HD a bit, but with the driver cone edge only 10" from the end of a large slot with line of sight to the outside world I'd expect there to be more operational noise leakage, though HR doesn't show too much of a difference. I'd like to see that tested actually. Also the drivers would be a little more susceptible to rain or a drink spill, but either arrangement is safer than a direct radiator cab!

This is also interesting. I too would see tests of this. Having a 10" slot should naturally be worse, but since it rolls of HD earlier, it should roll off those mechanical and operational noises earlier as well. This all depends on where the noises occur of course. Above 350hz the different designs have similar sensitivity, but my sim has less sensitivity between 120hz and 350hz. Above 350hz the wavelengths start to get relatively short... could a slight bit of damping in the front 'chamber' be the answer to decreasing the noises?
The front of the drivers would be a bit more susceptible to rain or drinks, but in the regular SKhorn, standing up, one can pour a beer straight into the motor of one of the drivers, through the middle part of the port.

Quote

I didn't go with a push pull driver arrangement due to the minimum clearances needed for the drivers  (10.5" driver depth + a large dustcap and +40mm excursion worst case scenario minus some for the baffle thicknesses and depth to the dustcap below the frame plane) and the fact that most high excursion sub drivers make quite a racket from the motor and suspension when pushed hard. Even more of a reason was the uneven loading it would present on the drivers. It would be much more involved and difficult to design the cab to evenly load the drivers with one inverted. Not impossible but it would complicate things. I'd give a rough guess that there would be about a 25-30L difference in the vented air volumes seen by the drivers. Tuning would be slightly different unless compensated for and the throat area would be a concern. I know this has been done in plenty of cabs before but I don't know if I trust it with the kinds of pressures that can be developed inside something like these. I didn't think the potential lowered even order distortion outweighed the concerns with uneven loading and possible mechanical noise from the inverted driver motor. The forces inside these can be pretty ridiculous.

Of course the volume differences would be accounted for in a push-pull scenario, which would be easier to do in a design that only has 90 degree bends, but since the push-pull scenario has been left behind for other reasons it's a bit of a non topic.

Quote

The force cancelling works really well on it. It's not like the drivers are 90deg rotated from each other or even 45deg. It's only a 12deg offset and it's a 250+ lb sub. In use the cab has  no perceptible rocking from the driver operation. That small amount of offset from exact opposition plus the sheer size and mass just doesn't add up to any real rocking forces fore and aft. Technically I'd guess a perfect driver opposition would measure a bit lower with an accelerometer but in practice it worked like expected and it's a complete non issue. Regular old panel vibration and bracing is a much bigger concern. The Skhorn has been quite good as far as that is concerned when compared with most other large cabs, but this is always a battle on big subs with tons of output.

Panel vibrations would be much worse, yeah. It would just be nice to know that one has done what one can to reduce the vibrations from the drivers themselves. 0 degrees would be nicer than 12 degrees, if all else is equal, but it's not like it's a make or break aspect of it all.

Quote

About Edge and directivity. Keep in mind that the math is simplified and goes back to the point source mic placement method. If the radiating points are spread on one sub and focused on another you can never truly get the same mic distance from the two. In practice there is a very large area that one is trying to cover usually inside of a room with boundaries. It's complicated. One sub is more diffuse but that's not necessarily always a bad thing. Edge is a good tool but I don't get too caught up in an apparent half dB advantage here or there. Most of that directivity happens above 80Hz unless the baffle area starts getting truly large.

That's why I've set the mic in Edge to measure the sound at 100000m(100km) out - to mask the difference in distance from sources on the baffle itself. If the mic is put 2-3m off of the dead center line so far out we can be sure that the sims will still be very fine.
In the second part of the above quote I suppose that you weigh 2 sources spread out by 45" against 1 source in a single location, and question which would be better. Me, I wouldn't put a Skhorn in my living room - I'd get a AE TD18H+, put it in a simple design, place it well, integrate it well, and be done with it, but that's me. 

Quote

One way I like to think about this is with 2 different philosophies. The first would be maximize the baffle area for the cabinet (think DTS-10 shape or similar) and get as much out in front of the cab with the drivers and design as possible. However at what point does this become impractical? Let's say the cab has a 60x60" baffle and is very shallow and gets some extra forward gain over a very similar design with the same driver which has a baffle of  30x30" but is much deeper. At some point you run into limitations in available depth to use and/or available baffle area. How many 60x60" baffle cabs are going to be able to be arranged? In a lot of cases not too many and eventually you just need more cabs and drivers. You could fit 4 of the 30x30 subs into the same baffle area as the 60x60 that would outperform it. The best case as far as potential output density per baffle area goes is that one entire face of the sub is radiating surface area. That's not going to be the best for directivity control though. You also end up with deeper cabs to get your cab volume. Tradeoffs.

I'd bone this all down to a relation between the depth of the stack and the height of the stack. Drunk people fall over and push each other around. To prevent them from knocking a sub over in their fall I guess that a depth about half of the height of the stack would be enough. 48" tall by 24" deep, for example. If taller than 48" one most likely does sound at an event that has a fence anyways, so stuff wouldn't get knocked over, and stacking would be safe even though the relative depth would get lower.

Quote

Vent area is vent area I'll give you that as an advantage for sure. This is the #1 priority I would have when redesigning the Skhorn. Mostly to work better with the lowered tunings. I try to avoid turns in vents when possible though. The reduced area of the slot for the drivers would increase air velocity over the horn expansion type though. I have some new thoughts on designing vented subs that I haven't gone to far with yet but I'll share once I think it through a bit more.

The reduced area of the front part would mean that the particle velocity would be brought up to about 30m/s (with filters in place), which is fine, and which is far from 45m/s of the particle velocity of the port, and 50-55m/s in the Skhorn.

Something that we haven't brought up yet is where the sensitivity is most needed. According to the Equal-Loudness contours ( https://en.wikipedia.org/wiki/Equal-loudness_contour ), for something to be perceived as being 100db, you actually have to reproduce a 100hz tone at about a 107db level, a 60hz tone at about a 112db level, and a 30hz tone at about a 122db(!) level. Now, that means that we have to send 10 times more energy to a tone at 30hz for it to be perceived as as loud as the tone at 60hz.

Let's say that we try to reproduce a perceived 100db signal with a speaker that has 100db sensitivity throughout our hearing range, 10m from the source (no reflections included in this example). This means that we have to send 100w to our speaker at 1000hz, we have to send 500w at 100hz, we have to send 1600w at 60hz and we have to send 16000w to 30hz, for it all to be percieved as as loud. Where does it make the most sense to increase the sensitivity by, let's say, 3db? At 100hz, where it we could cut the needed power by half, which would correspond to a decrease in needed power by 250w, or at 30 hz, where it would correspond to a decrease in needed power by 8000w? Of course the increased sensitivity would make the most good down by 30hz. Even if we increased the sensitivity at 60hz indefinitely and impossibly much, this would only correspond to a cut in power by 1600w - not a lot compared to a decrease by 8000w.

The above is why my sim has more sensitivity below 40hz than above 110hz. It's also why I've chosen a larger port. 1-1,5db extra sensitivity below 40hz, plus 0,25db extra baffle gain, plus about another db from the decreased port compression, that should all make for about 2-3db more maximum output, where it's sorely needed the most. 

Ok, so I finally got some time to reply.

On 8/10/2018 at 5:53 AM, Ricci said:

You may find that what seems good in theory isn't practical to implement when it comes time to fit everything in the box. 

Ok, so I drew out a quick sketch by hand. This would be the layout. I don't see how anything doesn't fit in there.

In the above drawing, I took the same design volume as in the Skhorn sims and drew it into the same final design volume that the actual Skhorn is. Now, the sim volume and the actual volume within the actual Skhorn differ by some 65l, according to your numbers (18.35ft3(519,6L)vs584,3L) but for comparisons sake I've done as I have.

I've also skipped the one inverted driver in the above drawing, for fear of getting a drink accidentally tossed into the front chamber, sploshing about around the driver motor. *shivers down my spine*

Here are the inputs of the design that I've drawn:

Try to put these inputs in HR, put a 27hz 3rd order low pass and a 70hz 3rd order high pass on it, and compare it to the regular Skhorn sims with a 25hz 4th order low pass and a 68hz 3rd order highpass on it, with a 197hz 3q -10db EQ on it to pull that peak down.
One will find that the two designs are incredibly close as far as phase, group delay, excursion and driver power goes. The inputs above has a slight output edge near 35hz in the sims, otherwise the response is identical. The efficiency shows that the design above doesn't emphasize output around and above 100hz as much, and that it emphasizes 25-40hz more instead. 
As far as the points above goes, would it be worth it to redo the whole design because of them? Hell no. 
Where it gets interesting is in the points below though:
It has less output above 100hz, so it rolls of the response where the bulk of the distortion is , it's an easier build (no angled cuts), it's better boundary coupled (0,25-1db more output forward and hence also less backwards), it's a bit more force cancelled, and it's got bigger ports for less compression and less vent noise (20% less port particle velocity).
All in all, it should make for a design that fits well within the size of the cab, is a bit louder, a bit cleaner, and a bit easier to build. All in all, I think it's a good idea for a slight development of a design that people clearly have a VERY good time with! 

Now, again, I don't mean to crap all over your work, Josh. I am an extremely analytical mind with a huge passion for both live sound and hi-fi, and to design the holy graal of bass designs has been the goal ever since I wasn't satisfied with the FLH subs that I built almost 15 years ago. The Skhorn is incredibly potent in many ways for good reasons, and, if it weren't, you wouldn't have designed and built it. I know that I've introduced Edge, the baffle gain sim program, into your set of tools, so it's not like I haven't brought something useful to this forum, and I know that my mind can do more good around here, among you other analytical, geeky and passionate minds. 

1 hour ago, Ricci said:

There are reasons I didn't do many of the items you mention and I wouldn't make most of those changes if starting from scratch for various reasons...

I am bringing some fair points to the table, but I'm happy to hear what points are more important than my mentioned ones when it comes to going with one solution over the other.

With people building their own versions of this design, I thought I'd chime in with a few points.

The long, flared front horn gives a lot of sensitivity above 100hz, where it isn't needed or wanted. That means that we've spent cab volume for no good. In my simulations I've seen that by using a shorter straight horn, I can keep the sensitivity below 90hz unchanged while greatly reducing the sensitivity above 90hz, reducing the volume of the design by 10%, simplifying the build by removing the angled cuts, further force cancel the drivers by placing them absolutely parallel, and at the same time reduce the weight of the build, in part because of the fewer parts needed, and in part because of the actual lower design volume. Simpler, smaller, lighter, more force cancelled - why not?

While we're at it with force cancelled drivers in a large front chamber... why not invert the mounting of one driver to further reduce the distortion? Free decreased distortion - why not?

Why not place all the ports in the center of the front baffle instead of along the edges? The increased boundary coupling that this gives would give us another 0,25db down by 30hz, rising to 1db by 100hz, given that the baffle sizes are the same. Free increased sound pressure - why not? 

The exit for the front part of the design would be as a slot in the center of the baffle, with the ports exiting as slots above and below the front part's slot.

These few changes would make the Skhorn smaller, lighter, easier to build, louder, and it would have less distortion. What's not to love? 

Here are the inputs that I'm using: (These are meant to imitate the Skhorn as closely as possible below 90hz)

Although, I'd prefer these inputs even more: (Flatter response, bigger port for less compression)

On 2/27/2018 at 9:22 PM, Ricci said:

The Skhorn sim is the old development sim from before the cab was built and measured. It really doesn't match the real world performance of the cab.

Still, when simulating, if two systems are the same volume it is a great comparison. The pound for pound comparison is still there. Now that you've updated the HR inputs to match the real world measurement though, it's pretty evident that my design can't keep up - the lower tuning of the updated Skhorn HR sims calls for a level of excursion that only the 21ipal can reach.

On 2/27/2018 at 9:22 PM, Ricci said:

I noticed that the baffle size is listed as 1300x1900 in the Edge comparison above. That would be for a stack of 4 cabs if I am not mistaken. Is that being compared to a single Skhorn cab with a frontal area of 610x1372mm or a pair? 4 cabs is 3X the baffle area or stack size of a single Skhorn.

The 1300*1900 is for two cabs next to each other, in 2 pi. Edge is 4 pi by default, so one has to compensate for that. Those two are compared to a Skhorn laying vertically, in 2 pi. Try it out in Edge, and you'll find what I found - this:

On 2/13/2018 at 8:02 PM, Moar!! said:

Despite having the exact same total volume, my cabs have just shy of 1db more baffle gain down by 40hz, rising to just shy of 2db by 100hz.

10 hours ago, Ricci said:

You shouldn't use 1pi as an approximation for 2 cabs.

I know. The only thing that differs though is the impedance, which I knew would be in check and quite similar between the two designs, so I just saved myself some time instead. Half of the times that I want to go into the filter wizard or the multiple speaker tool I end up going into the other instead, which is mildly irritating at times, so I end up saving myself from some blood pressure rise too lol.

10 hours ago, Ricci said:

Also I noticed some of your inputs for the 21SW152-4 are off. Re should be 3.4.

True. I corrected that, increased the input by one volt and it was all identical again.

10 hours ago, Ricci said:

Also the original Skhorn sim isn't as close as it could be. The original inputs were as best I could guess at it feeding inputs from the cab design back into HR during the design phase, prior to building and measuring. I never went back and fit it better to the real result. I'm lazy!

The thing is though that I matched the volume of the two designs within half a liter of each other, so even if the Skhorn sim doesn't match real world to 100%, it should still be a very fair comparison in the simulations. Of course, correcting the sim will give and take a tad here and there in the response, shift the impedance somewhat, etc, but that 40% gap in power will still be there, although the peaks in the power demand will probably shift a little so that they will be a tad lower up top and higher down low, or vice versa.

10 hours ago, Ricci said:

That looks like a lot larger boost than I would guess off of such a minor difference in frontal area. 

It is. The contributor to that boost is not only the frontal area though, it's also where the source is located on the baffle, as I mentioned in the first section of my last post. Centering the source on the baffle or the stack gives you a good deal of extra baffle gain. Danley's BC218 is named like that because it's built around the baffle gain phenomenon - BC stands for 'boundary coupled'.


Again, I don't mean to be rude or anything, posting like this. I have an enormous amount of respect for what you've done and created here on data-bass. I simply wish to be of good use, and to exchange a thought or two

I'm glad that I could be of great help here!

Playing around with Edge, you'll find that you get more baffle gain if the source is located more towards the middle than the edges of the baffle (or more towards the middle and then the floor for a sub, because of the "mirrored sub below the floor" in a half space scenario).

I've been fiddling around with hornresp to come up with a cab that is smaller than the 21ipal loaded Skhorn (which is rather huge unless you're a rental company with big trucks and a range of smaller subs than the Skhorn for the times where it's size and power will be too much), that has more frontal area relative to the depth of the cab and the source located at the bottom of the cab (to be able to get more baffle gain both in singles and in multiples), that extends as deep as the Skhorn(for comparisons and good extensions sake), that is easier to build, is cheaper to build(the 21IPAL sure is pricey), that is preferably as powerful(because power), and that rolls off above 100-120hz(to roll off distortion).

I've kept the 6th order concept. It really is a power dense and good sounding design if done right. I've split the cab in half, ie, one 21" per cab, and therefore and for the prices sake I've gone with the 21sw152-4 instead. That is one great driver in price/performance, performance/weight ratios, in (Bl2/Re)/(Le/Re), and it's also 4ohm so they can be run in singles, and wired in parallel with two per amp channel if needed. For comparisons sake I've kept the volume per driver identical.
I've straightened out the 'horn' part of the Skhorn, because I often find in my simulations that keeping ducts straight is more space efficient than flaring them. They're also easier to build.

So, what I've come up with is a regular bandpass cab with a great driver lol. Pretty simple really, but with complex reasoning behind, as usual with us nerds.

Hornresp inputs: (I simply selected quarter space instead of selecting 2 speakers all the time in the tools menu.)

The form factor of the cab I have in mind would be 24" deep, 25,5" wide and about 34" tall. Slot port at the bottom, driver vertically mounted pointing forward in the cab, the hf sections slot port would be just above the lf sections slot port. Two cabs next to each other simulated in Edge would look like this, my cabs in red and a Skhorn, laying vertically, simulated in blue:
Despite having the exact same total volume, my cabs have just shy of 1db more baffle gain down by 40hz, rising to just shy of 2db by 100hz.

In the sims I've tried to subtract the baffle gain from the hornresp response both with the cab design and the filters used. If you add the baffle gain back you'll get a flat response with a high pass at ~100hz. And oh yeah, unless I want to see the raw response to look for warning signs, I always simulate with the filters in effect and try to go for a flat response(if the baffle gain would have been added back) with only the filters in use. The cabs will always be used with filters anyways so it makes most sense to sim it like that, I think.

21ipal Skhorn in black, 2 of my cabs with the 21sw152-4 in grey. Both have 12db/oct hpf and 24db/oct lpf, set at different points to adapt to the different responses, excursion behaviors, and so on.

So, yeah, about the same difference as in the Edge simulation, but inverted. The two responses should match up very well with the baffle gain added.

Delay:

Phase:

Efficiency:

Excursion:

All in check so far. By all accounts, everything so far is pretty much the same. Response, phase, delay and efficiency is pretty much exactly the same. The excursion differs, but they are both in check. What differs is the power per driver:

The Skhorn seems to need about 40% more power to get the same job done. Now, in my book, that's a huge difference. Especially considering that the drivers in the Skhorn are about 75% more expensive and the Skhorn has to be brought in in one piece whereas half of my option can be left at home if one would want to. My design has no angled pieces, so it's a bit easier to build. It rolls off fast above 120hz to roll off distortion, as shown below:

Now, could the 21ipal be used the same way and reap the same benefits as my cab loaded with the 21sw152? IMO, not practically so. One driver per cab loads the amp with too low of an impedance, so a pair is pretty much the only viable option. Would you like a double loaded cab that is 24" deep, 36" tall and 48" wide? I don't know. I wouldn't. I like to have a bit of scalability.

Areas where my design can't beat the Skhorn is the force cancelling mounting and how deep inside the cab the drivers are buried. I think less power needed, cheaper amps, more scalability, cheaper builds and easier builds outweighs the cons by a pretty good margin though.

Now, I don't mean to take a crap on Ricci's work. The importance of the work he/you have done can't be overstated. I just wish to contribute with my .02$ and to shed some light over this whole directivity deal, and show an example of how one can use it to one's advantage.

First post around here. *Yay*

Rad, I think you can attribute both the rising response and the high sensitivity up top to directivity. Under sane and normal circumstances, the highest sensitivity you can reach in half space is 105-106db. Of course, in quarter space, you can add another 6 to that. The bigger the baffle is in relation to the wavelength, the more the source will be seeing a quarter space scenario. (Then ofc you might be able to see another db of sensitivity here or there, caused by diffraction off of the edges of the baffle.)

This right here is a pretty neat little piece of software: http://www.tolvan.com/edge/help.htm
It calculates the gain caused by the baffle. (It also does calculate what kind of a passive crossover you'd need to compensate for the baffle gain.)

Toying around with that software(doubling the height of the cab to emulate a half space scenario), I find that the baffle adds about 1db at 30hz, rising to a 4,5db addition at 100hz. That should explain most of the phenomenon.