Droogne

To fit my room and driver, I'm going to make a sub just like the skram, using a 21" Lavoce San 214.50. It will be 60cm(wide)x68cm(high)x62cm(deep).

I was however still wondering how much fronthorn I need. I tried to look at the freq. response and horn mouth velocity. Would it be reasonable to model my sub so the top velocity (in the 30-80hz region) of the Lavoce, excursion limited, is the same as the Skram (or Skhorn) using an IPAL21. The ipal produces twice the air (+3dB), so the horn also has to be twice as big. 

Anything else I need to look at when choosing the front loaded horn? 

I included the modelling from both a version with a sloped second part of the horn and without. I would 'not slope' it because of construction reasons. It does look like it helps tremendously with reducing velocity reduction. Just not sure if it is needed. 

UNSLOPED

SLOPED

3 hours ago, peniku8 said:

Ricci might have adjusted his inputs based on the measurements posted some time in the past, to match the measured response.

IIRC he said that the proximity to the back wall also pushed the port resonance down a bit, but I'm unsure how you ended up with an even longer port. I have not looked much into the Skram

I went with the rule of thumb stating that a slotted port ( using the wall of the cab) should have its (height/2) added to get an estimation of equivalent port length. 

Also @Ricci, how did you determine port length? If I go by the length through the center of the port (659,8mm+153mm =  812,8m) + 1/2 of the port height (135/2 = 67,5mm) I get 880,3mm, not the 857,1mm you listed in the parameters. 

Anyone with Skrams who has done some vent related tests? Compression and chuffing when pushing it. 

On 10/22/2019 at 2:45 PM, Ricci said:

Droogne I've done square ports before and they work well. The ideal is round and heavily flared or shaped which I've also done where it makes sense. Have you read the available AES and other papers on ports? There are many. Compression and chuffing noise do not necessarily correlate.

As far as I can tell compression doesnt relate to in/outflow patterns, but rather core turbulence aka the tube itself. Flaring wont change that. Only circumference and area. 

On 10/22/2019 at 2:45 PM, Ricci said:

There are also papers which point to some possible advantages with multiple smaller vents despite the increased skin effect.

I have been reading through https://drive.google.com/file/d/0B1gKnCRgylLcMnAya0dvc3NDZkE/view the last few days. Havent gotten through the most parts yet, but will try to see how much usefull info I can get out of it. I'm not technically educated, although I have had a lot of sciences (I'm a doctor), sohopefully I'll be able to plough through.

On 10/22/2019 at 2:45 PM, Ricci said:

In the case of the Skhorn and Skram variable tuning is an integral part of the design that I considered fundamental. That is why they have many smaller vents instead of a larger single. Also it is structurally very strong. When considering them as individual smaller vents they are not too far from square. 

Especilly in the Skram! The almost squares do help. But as you mentioned, primarly the area will be the determining factor. 

On 10/22/2019 at 2:45 PM, Ricci said:

Its all about the tradeoffs. I think you may be overestimating the effect a change to the shape may have on compression. A single square or even round vent the area of one of the Skhorn vents isn't going to gain much. They will still overload and compress. The area is not enough to cope with the demand once the sub is pushed at tuning. This is true of almost every sub tuned under 20Hz. In the Skhorn's native tuning with all vents open with 3X the vent area this isn't really an issue as the testing showed. At the end of the day it's all about vent area in my opinion. 

In the end I will probably end up building my adjusted Skram with the same port dimensions, as its proven to work well in the Skhorn, which should do it way worse than the Skram. Its just for future designs nice to know what is important and whats not. Looking at the Skram and Skhorn as a 'case-study' is important in that process. 

1 hour ago, peniku8 said:

You mentioned the port dividers and how they add more wall area to the port, which made it sound like you were thinking about removing them. If you did so, the panel vibrations of the outer (now unbraced) side panel would introduce more distortion that removing the braces would lower it. That's what I was refering to.

I tend to agree, but if you follow my comments earlier you can see that by using a square/less rectangular port and removing port dividers would allow to use a smaller area because there is less friction.. So yes, no port dividers and a more squarely shape would give less bracing to the outer wall BUT the smaller port would give more bracing (or would take in less wall space, so would allow for more bracing) . I think if you meet it somewhere in the middle there can something he gained. It would also allow for a smaller cabin, which is sturdier in itself. 

Quote

I still think vent noise can almost completely be ignored in 25Hz and above PA applications, while port compression needs to be minimized. 

Also agree. I just want to know it precisely because if you can achieve the same amount of compression with a smaller port you save up space. In other designs it could allow for longer ports (not so relevant in the Skram/SKhorn) 

Quote

If you think about your wall area and air friction so much, maybe try polishing the inside of the port or apply black paint -> clear coat -> polish. I'd be interested to see if theres any difference

I was actually planning on maybe using 'concrete plywood' (the dutch word, I don't know the English one.) its basicly plywood with a thin but very smooth layer on it. Specifically made to be smooth (and waterproof) actually. Wouldn't be much harder to work with than with regular ply, so I don't see why not. 

Also @Ricci, what is your opinion on using a rectangular port instead of a squared/rounded one? Design wise it's obvious why you decided on using a rectangular one, but by my calculations (based on info from the Flare-it app (https://www.subwoofer-builder.com/flare-testing.htm) using an elongated port like the one in the Skram or Skhorn decreases the required velocity where port compression starts by almost 50%!

I calculated velocity where compression start for the Skhorn (with IPAL), which resulted in around 40m/sec at 30hz. If youre interested I could PM/post my calculations here. In any case it seems to correlate with what people are noticing in their Skhorns. Using a square port with the same area I got a 80m/sec value. 

As port compression results from core turbulence (something that isnt influenced (much) by flares) due to friction/turbulence with the walls, the bigger (wall-circumference : area ratio) of a rectangular port results in compression at lower velocities. For the Skhorn the ratio is exactly 153% bigger (0,29) than for a square port (0,19) with the same area (418cm²), and increases further to 190% (0,37) when considering the dividers which add a lot of wall surface.  

Going back to the Skram we can ofcourse see that the bigger ports help reducing velocity, but as compression happens at a certain velocity at a specific port area the bigger ports of the Skram can also take on higher velocities before compressing. So both lower velocity, and better toleration of high velocities. Win win. Sadly, we do find the same 'problem' because the Skram is still has a very rectangular port which uses even more port dividers. The ratio is around 195% higher than a square version would, but the circumference/area ratio is still way better than the Skhorn (0,29 for the Skram vs the 0,37 of the Skhorn) so it will be able to take on higher velocities. 

Peniku did make me aware of the fact that panel vibration could be a problem when using square ports, but its clear that you dont need square ports to take advantage of this. Using less dividers (so less tuning opportunity and structural rigidity) or port which are less rectangular could help reduce compression OR help making ports smaller without changing compression. Pick which one you need  

3 minutes ago, Ricci said:

Don't miss the forest for the trees. It is not only air velocity at the mouth that if of concern, but also at the throat or entry inside of the cabinet. If the port is flared, bowtie or otherwise more complexly shaped the performance at the "choke" points should also be examined. Any of these points can trigger compression. It's more complex than just airspeed at the mouth. You may decrease audibility of air noise but greatly increase output compression. 

More involved programs such as ABEC or Akabak can model, velocities, pressures and accelerations at any point in a pipe, duct or horn. 

Yesss! Sorry I hadnt posted my conclussion, but here is what I realised and left in message to peniku: 

'It dawned on me that using sloped ports (so eg 250=>500=>750 instead of the regular straight, eg 500=>500) is wrong for the same reason[creating a bottleneck]. Port compression would occur sooner because you start with a throat that is smaller, so higher velocity at the throat in the sloped design than wherever in the straight design. I will not be going that route afterall for that reason.'

So yes, I realised exactly what you just said  I can post the modelling in hornresp here, but you already understand what I mean. 

Also, wouldnt noise created at the throat (if your  using a horn shaped port) be 'hornloaded'. So the noise would be created deeper in the cabin, but it would be increased. 

5 hours ago, SME said:

The problem with too much vent velocity is not just chuffing but also power compression, and substantial compression can set in well before the chuffing becomes audible.

Didnt think about that, thats true. Could you theoreticaly model this by the 'pressure' function? In any case, bit of a derailment here, so I'm just gonna start a side topic to explore this further. What I noticed was that 

the peak sound pressure in the mouth was around the same in at mouth (around 9-10 pascals, and a little bit lower for the sloped port) and exactly the same at the throat. So going from this I dont see a disadvantage of doing it like this. Lowering the velocity by at least 25%, without increassing pressure (if this is related to power compression ofcourse).

Could this same principle be applied to a port? No idea how this would effect tuning though.

In trying to see the influence a sloped port has on the velocity I modelled it as a rear loaded horn. Doing it like this requires some parameter juggling etc, but I did manage to create the almost exact same response for the 'rear loaded horn' as for the port in my Skram version. Modelled velocities only differed slightly.

Parameters

(my version has a 183L chamber, aka 183000 Vtc. Modelling done with the 21" LaVoce). 

Velocities

port modelled as portport modelled as rear loaded horn

To try see if a sloped approach could work I tried out a few different parameters. Goals:

1. Keep the port output the same

2. keep the port volume the same

3. use the bends to make up the horn (in my case, 35cm and 50cm)

Then we get the following input:

Which results in the following velocity at the mouth:

This is significantly lower than the normal unsloped port, from 33m/sec to 25m/sec. Granted, the velocity at S2 and port mouth is enormously high, but S2 is 50cm away from the mouth and and the horn throat 85cm, which might limit the audibility of the port?

Thinking about the size I thought of changing the dimensions while keeping the specs the same. This would end up being a 60 x 60 x 120cm cabin. Would this work? 

Also, could I make the horn part straight (so S1=S2=S3) as long as it models correctly? 

8 hours ago, Ricci said:

It looks ok based on Beyma's specs.

OK, as in equalish to the Lavoce and other cheaper drivers?

A little light on motor force but should be fine. The unknown is what the complex inductance specs look like but based on the published impedance curve it shouldn't change things too much.

 

I've asked this before, but I'd like to know if the Beyma 21SW1600ND would be a close match to the cheaper drivers. I can get a pair for 450, and my initial modelling (based on the manufacturers disclosed specs) are great. 

Anyone who can chime in? 

Me being in the EU makes the Eminence harder to get than the IPAL, sadly  I think I got the modelling right (checked vs the models listed above here), but still not sure to model with or without lossy Le. Thanks for running me through the process!

At this moment I have a pair of Sundown Audio ZV4 18D2 drivers, so I modelled them for reference. Keeping aside pragmatic problems (it being an 18" driver with a huge rubber hangup) I quite like this. A 5-10dB gain to the sealed I have right now, and only 2dB loss vs the 'cheaper' 21"ers mentioned above.

All ports open (for reference)

Compared to the sealed config I have them in now:

Sealed, 140L

6 hours ago, peniku8 said:

In the acoustical power window hit ctrl+M and enter values there.

Thanks! Awesome tool! I cant pick 'Lossy Le' when doing this, might explain some different values between us. What would be the most accurate modelling, lossy Le ON or OFF? In any case, this is what I get with the 2-port configuration (the one I'm most interested in for modelling) using their respective max voltage inputs and Xmax: 

SW152

La Voce

IPAL 21

Looks like 2-3dB might be closer. Considering the LaVoce is about half of the price of the IPAL, is readily available and is much more amplifier friendly I will probably go with that one. I was briefly considering using a 5000W 2 ohm car amplifier with an AC/DC transformator, but that seems somewhat complex to work out. 

PS: how do I put all these graphs over each other?

10 hours ago, peniku8 said:

@droogne that's a lot to reply to, I hope I manage to catch everything posting from my phone.

The 2nd graph I posted shows the maximum output of the cab/driver combination which is either Xmax limited or power limited (depending on frequencies).

How do you get these kind of graphs?

10 hours ago, peniku8 said:

I killed the colours for the B&C drivers, but the LaVoce graph line is two colours: green for power limited and red for excursion limited. You'll see that the graph is red from 0 to 29Hz and green from 29 to 38Hz. That correlates to the excursion minimum at the tuning frequency, so the driver hits its power limit before it hits Xmax at this frequency band in that cab.

If your system hits Xmax with 50% power at a certain frequency that's okay. Unless you're running heavy limiting, program material will not contain 0dbFS sine waves. When you're far away from excursion limits using the rated AES power, it is likely that your cab is too small.

As you can see in my comparison, the IPAL is about 3db ahead both in the power limited as in the excursion limited bands. I don't know how you're getting your 6db difference, but it doesn't seem correct to me, no matter the cab. How are you determining the headroom?

I see youre graph shows a 3dB, not 6 tot 8dB difference. I havent figured out how to compare several graphs like you. I modelled it based on the displacement graph in the wiazrd function:

This graph shows the displacement of the SW152 in the 2 port configuration at full 2000W 4ohm (89V, GRAY). Xmax is listed at 15mm, so you can see it maxes out at ~42Hz. When I reduce the volts to 62 you can see it stays below the 15mm Xmax. 

This results in the following output, which shows a 3-4dB loss due to reduced input.

When I do this for the IPAL 21 I get the following graphs, showing an Xmax that alway stays below the listed 22mm. 

Comparing those:

                                             IPAL 21                                                                                                                    SW152

=> I dont know if I made a mistake earlier, as the difference isnt that big anymore, but still. At 20hz there is a major advantage for the IPAL (at least in a 2 port config).

20hz: 125dB vs 118db aka 7dB

30hz: 125dB vs 123dB aka 2dB

40hz: 128dB vs  124dB aka 4dB

DISCLAIMER: i'm new to modelling, so dont take these results for what they are. 

58 minutes ago, peniku8 said:

@Droogne I don't know how you're getting to these numbers, but 3-4db difference between the SAN214.50 and the 21SW152 sounds like you're comparing them using the same voltages while using the 4Ohm version of the SW.

+1. This is why I keep posting my models here. I'm bound to make mistakes. Here is the SW152 in a 2 port config, with 2000W 4ohm (instead of the 2000W 8 ohm I used earlier..). On the other hand, the model I care the most about is the Xmax limited, which stays the same (SW152 already hits it Xmax at 60 volts, which is still lower than the 89V at 2000W 4ohm).

That said.. is 'Xmax limited' something that is a logical way to approach this? I made the same mistake on the ohm/voltage front, so for the IPAL you cannot input 53V (as 2500W 1ohm is 50V), but it doesnt make any real difference dB wise. This means the IPAL can be used to its full power in the Skram, while the SW152 and Lavoce clearly dont as they need to restrict their input by ~30-40V to stay below Xmax. This is clearly reflected in the output, which favors the IPAL even more. 6-8dB in favor of the IPAL (I see you modelling shows a slighter difference which is probably due to not being Xmax limited?) . Normally I'd say, 6-8dB makes an IPAL Skram x4-6 times more worth it, so the extra few euros (in my case the IPAL isnt even twice as expensive as the cheapest LaVoce I can find) would be saving me money by needing less subs (and space!), but sadly the 1 ohm load of the IPAL is the dealbraker. No way I can afford one of those ultra expensive amps that can take on a 1ohm/channel or 2om bridged load. Otherwise I would have placed my IPAL order this evening. 

Quote

Here is the max output graph (at their rated AES power, basically an excursion limited efficiency graph), LaVoce in green, SW in blue and 21IPAL in pink.

So.. Xmax limited you mean? Or just full 1700, 2000 and 2500W without looking at Xmax?

Quote

Also, 650€ for the 21IPAL is a very good price, could you paste a link to that supplier? That's like the regular street price for the 18IPAL here

It's up on Bluearan in the UK. Always the best price as far as I can see. Next one where I normally look is TLHP in France (but IPAL over there is still at the regular 900 euro).

Quote

 

Edit: all models used Josh's complex inductance parameters

I urgently need to use his parameters too, instead of manufacturer listed ones.. dont know why I didnt do that from the get go.

2 hours ago, Droogne said:

Lavoces here in Europe also go for a lot less. I was modelling them vs the SW152 drivers and I see around a 3-4dB difference (from 30hz and up) at full RMS input, so justifying the costs of the B&C. However, the Lavoce has an Xmax of 15,5mm (instead of 15mm) and a max power rating of 1700W (instead of 2000W), so I figured the SW152 might hit that 15mm Xmax sooner. 

When I model it to limit the output to their respective Xmax (at 40hz) it looks like the Lavoce can take more power (80V instead of 60V) which translated into only 3dB loss instead of 6dB, so a relative '3dB gain' for the Lavoce which effectively bridges the 3-4dB gap I modelled at their respective full power. So Xmax limited the Lavoce results in only slightly less output instead of a very relevant 3-4dB. 

Please correct me if I make wrong assumptions or modellings, I'm pretty new to this. 

Also, if I model the IPAL 21 I get similar results, though still massively favoring the IPAL (even when considering costs). First of all, consider the current marketprices over here (Belgium):

Lavoce 21 SAN: 358 euro

B&C SW152: 519

IPAL 21: 650 euro

When modelled to never exceed Xmax we get these results (2pi, 2 port configuration):

                                   20hz   30hz   40hz   50hz   60hz   70hz   80hz  amount of volts needed to reach Xmax   DB lost due to Xmax limiting   Price

Lavoce 21 SAN       117    123      125     125     125     127     128     80                                                                    3dB                                              358

B&C SW152             118    122      125     126     126     127     129     60                                                                    6dB                                              519

B&C IPAL 21            125    127      129     132     133     134     135     55                                                                    8dB                                              650

=> it looks like the IPAL suffers the most from limiting its output to Xmax, but it still kicks ass (especially in the low end), AND it needs LESS power to get there. 

Should I jump on the IPALs and get a pair before prices go up again?

Also sorry, to be someone else wondering about the CNC files, but I cant seem to figure out how to send this to my CNC operator in a way that he can effectively work with this. I downloaded the 3 files listed here above, but I cant seem to work out an 'easy' way to appoint the correct measurements, so my CNC guy can get to work. 

Greetings

2 hours ago, peniku8 said:

I've seen the LaVoce for 380$ on PE, so that's anyother even cheaper alternative. And the LaVoce models a bit differently, which can be great depending on the cab you're using. It doesn't show the bump in the 60-70Hz region which I sometimes see from the DS.

Lavoces here in Europe also go for a lot less. I was modelling them vs the SW152 drivers and I see around a 3-4dB difference (from 30hz and up) at full RMS input, so justifying the costs of the B&C. However, the Lavoce has an Xmax of 15,5mm (instead of 15mm) and a max power rating of 1700W (instead of 2000W), so I figured the SW152 might hit that 15mm Xmax sooner. 

When I model it to limit the output to their respective Xmax (at 40hz) it looks like the Lavoce can take more power (80V instead of 60V) which translated into only 3dB loss instead of 6dB, so a relative '3dB gain' for the Lavoce which effectively bridges the 3-4dB gap I modelled at their respective full power. So Xmax limited the Lavoce results in only slightly less output instead of a very relevant 3-4dB. 

Please correct me if I make wrong assumptions or modellings, I'm pretty new to this. 

On 9/14/2019 at 1:26 PM, Droogne said:

I did yes. Didnt use the "lossy Le" option.

Here is the difference:

2 port configuration:

@Ricci Does this look more like it? 

Also, what do you think of trying to make an 18" equivalent version? I very much like the performance of the Skram, but their size is scaring me a bit. Height and width aren't problematic, but the depth is. I would like to reduce it to 60-70cm, as it would make it more accessible for my living situation. 

If not possible to reduce the depth without changing the design completely, then I'll just leave the size as is. 

14 hours ago, Ricci said:

These don't look right to me. The response shape looks off. Are you using the generic LE (red Le) function? 

I did yes. Didnt use the "lossy Le" option.

Here is the difference:

2 port configuration:

1 hour ago, Ricci said:

There are 4 ports when all are open= Divide the Ap (port area) of 725.76cm2 / 4 = Each port has an area of 181.44cm. Multiply 181.44 by the number of ports that are open and input into Ap.

Also some volume from the vent is added back into the enclosure when blocked. Add 14.42L to the enclosure rear chamber volume (Vrc) for each port that is closed off.

3 port operation = an Ap of 544.32 and a Vrc of 216.15

2 port operation = Ap of 362.88 and a Vrc of 230.57

1 port operation = Ap of 181.44 and a Vrc of 245

Ok thanks. I modelled them all out for my own reference. Here the results for everyone interested. I Xmax limited the output (15mm for the 21Sw152)

1 port operation. Xmax at ~22hz, 60 volts

2 port operation. Xmax at 30hz, 70 volts

3 port operation. Xmax at ~22hz, 80 volts

On 7/16/2019 at 11:15 PM, Ricci said:

 

Great to know the hornresp parameters for the Skram with all vents, but how would I convert this to a configuration with only 1 vent open? 

Anyone who can help?