Fastbike,
First off, I am not comparing MLTL benefits vs sealed boxes. I have yet to build a sealed speaker. I don't know about your builds, but I have measurements on mine showing that making it a MLTL has a HUGE impact. Like I said 15 Hz is not uncommon and in some cases 25 Hz extension. In one case the BR tuning was 75 Hz and with the MLTL it went down to 50 Hz. That is a big difference in the sound that comes from a speaker. In fact, it completely changes the character of a speaker when it can go from 75 to 50 Hz - certainly more than a few Hz as you suggest. If MLTL's were not big improvements in performance, we would all be doing BR designs on the Full-range forum. And it is not just the extension into the lower frequencies but the quality - it is very flat and tight.
Last edited:
Hi sir Xrk971!
Which speaker combination (box) you propose for the LM3886 audio amplifier!?
Offer your solution a lot because I appreciate your work!
thank you and cheers!
Gost,
I personally have not used the LM3886, but from what I hear, they are great amps from the Gainclone and Chipamp crowds. I think they are quite flexible and would work well with a number of drivers as they have lots of power so you are not limited to super high efficiency drivers. Have you built your Vistaton BG200 Weems pipe speaker yet? It should work well for that. If you want a recommendation, all I can suggest are the speakers that I have built which include:
1. Cornu spiral horn
2. BIB and Flat BIB or FIB
3. MLTL
4. Straight cross sectional area horn with offset driver
5. Folded tapered tapped horn with rear facing mouth (FH3-inspired)
It depends on the music you like and the room they will be used in. The LM3886 can drive all of these speakers.
In a smaller room, for near field I would go with MLTL, in bigger room I would go with Cornu, the BIB works in a bigger room as well and can work in medium room. In a medium room I would go with the FH3-inspired horn.
For jazz or vocals, classic rock, pop, with the best spatial imaging, the FH3-inspired is best. For all around every type of genre balanced, go with the MLTL. For bass heavy hip hop go with the BIB or MLTL. For folk, chamber music, jazz, vocals, big band, go with the Cornu.
Hope that helps. Personally I would go with a class D amp rather than the LM3886.
Regards,
X
I love this thread! I think your design method is excellent.
Obviously simulation does have additional advantages, but I think it's terrific that your method shows how to basically "improvise" a BR into an MLTL. MJK articulated the principles so clearly because he wanted people to understand the physics. And you can get most of the way there via your method.
(Of course as you already know, it is still well worth it to go the extra mile in MJK, and while it seems intimidating at first, it turns out to be astonishingly simple and lots of fun.)
Obviously simulation does have additional advantages, but I think it's terrific that your method shows how to basically "improvise" a BR into an MLTL. MJK articulated the principles so clearly because he wanted people to understand the physics. And you can get most of the way there via your method.
(Of course as you already know, it is still well worth it to go the extra mile in MJK, and while it seems intimidating at first, it turns out to be astonishingly simple and lots of fun.)
Thanks for the kind words Rjbond. I know that the distinction between a BR and MLTL is actually very thin, even MJK said so in his writings because many people will look at a MLTL and dismiss it as a BR. As the aspect ratio gets shorter and shorter, the MLTL will collapse to a BR in behavior and performance. I have been meaning to buy MJK's software but I don't have a Paypal account and do not want to set one up. If he simply took credit cards, he would get more sales I believe. The software is certainly worth a lot more than the $25 he is charging...
One thing that I am still trying to figure out is what rule of thumb to use to set the length? Right now, it is one of practical considerations as you usually don't want it too long. But at the same time, you want to keep it shorter than the quarter wave length of the fb. So give a driver with a certain fs, sometimes it is possible to push its tuning to a factor of 2x below its fs. Take the TC9FD for example with an fs of 120 Hz. Because it is a high Qts driver, I can get down to 50 Hz, which is crazy. I know there is probably quite a bit of harmonic distortion here and the purists will poo poo this. But, for me, just getting to hear some nice bass around there with my wimpy 3.5 in driver is cool enough. I have heard of rule of thumbs of not pushing your tuning below 75% of the fs. If I were to follow that for the TC9FD, I would have ended up at 90 Hz - not very spectacular or impressive at all. Anyhow, I think it has a lot to do with the Qts. I think this is where a program like WinISD is handy as it will let you see how ugly the BR curve looks like when you tune below the recommended frequency. The power of the MJK simulation is that you can actually predict how flat the final response in the MLTL will look, whereas with my AMLTL method - you are just hoping. So far the hope has panned out and produced several nice sounding designs.
I find the contrast of opinions about this thread quite entertaining. 
XRK, have you tried to work with the worksheets at all? I sat and stared at the free "test" worksheet sample for about 40 minutes one day. Made absolutely no progress and had no idea where to even begin.
I'd have to receive some sort of training -- at least a help file and/or a video walkthrough, before I would pay $25 for it. I have no doubt that it is worth it if you know how to use it, but if you don't, it is worthless.
Easy to use? Ha!
XRK, have you tried to work with the worksheets at all? I sat and stared at the free "test" worksheet sample for about 40 minutes one day. Made absolutely no progress and had no idea where to even begin.
I'd have to receive some sort of training -- at least a help file and/or a video walkthrough, before I would pay $25 for it. I have no doubt that it is worth it if you know how to use it, but if you don't, it is worthless.
Easy to use? Ha!
Cogitech,
Yes, I have played with the free test worksheet and it works great. I just can't change the driver parameters! Which is why you need to buy it. I used to use Mathcad as part of my job almost on a daily basis (way back when...). So I am not the best person to give an opinion of how easy it is to use. However, if you use it in conjunction with the white paper on designing the back loaded horn for example, it is easier to follow. I agree that it is not easy if you are not familiar with Mathcad. Someone should make video with a walk thru training with screenshots - that would be very helpful I agree.
I can give you this tip with the worksheets: it is all about the section where you describe the geometry. This is where you set how many points (segments) you have in your 1-d model. It goes from closed end to where the driver is, then from driver to open end. Each segment you specify the cross sectional area and the distance (and damping - stuffing if applicable). Look at the graph showing the cross sectional area vs distance that results and see if that matches your speaker design. Note that it is 1-dim so a folded horn just looks like a long series of expansions/contractions.
The easiest model to follow is the MLTL as there are only 3 segments: closed to driver, driver to terminus, and terminus to vent opening. If you look at Bjohanessen's TABAQ input files, you will see what I mean.
Don't give up, it will open a whole world of speaker building without guesswork! Or if you are happy with MLTL's you can use my AMLTL method and it will probably work OK. There is also Hornresp which is a little easier to use that MJK's sheets and is free. But it doesn't account for damping which makes the response curves smooth and flat. I always look at Hornresp freq resp outputs and say that looks terrible (full of peaks and valleys) and you have to mentally smooth them out to estimate what the response will probably look like in reality.
Yes, I have played with the free test worksheet and it works great. I just can't change the driver parameters! Which is why you need to buy it. I used to use Mathcad as part of my job almost on a daily basis (way back when...). So I am not the best person to give an opinion of how easy it is to use. However, if you use it in conjunction with the white paper on designing the back loaded horn for example, it is easier to follow. I agree that it is not easy if you are not familiar with Mathcad. Someone should make video with a walk thru training with screenshots - that would be very helpful I agree.
I can give you this tip with the worksheets: it is all about the section where you describe the geometry. This is where you set how many points (segments) you have in your 1-d model. It goes from closed end to where the driver is, then from driver to open end. Each segment you specify the cross sectional area and the distance (and damping - stuffing if applicable). Look at the graph showing the cross sectional area vs distance that results and see if that matches your speaker design. Note that it is 1-dim so a folded horn just looks like a long series of expansions/contractions.
The easiest model to follow is the MLTL as there are only 3 segments: closed to driver, driver to terminus, and terminus to vent opening. If you look at Bjohanessen's TABAQ input files, you will see what I mean.
Don't give up, it will open a whole world of speaker building without guesswork! Or if you are happy with MLTL's you can use my AMLTL method and it will probably work OK. There is also Hornresp which is a little easier to use that MJK's sheets and is free. But it doesn't account for damping which makes the response curves smooth and flat. I always look at Hornresp freq resp outputs and say that looks terrible (full of peaks and valleys) and you have to mentally smooth them out to estimate what the response will probably look like in reality.
Last edited:
It was a combination of not knowing the MathCad interface and not understanding most of the vocabulary/terminology being used. I am one of those people who needs to understand precisely what every word means before I can continue. When confronted with an entire interface of stuff I didn't understand, my eyes glazed over and that was the end of it.
I do appreciate your attempt to help, though.
How about I buy the software for you with my Paypal account and then you make a video tutorial of how to use it? Post it for the benefit of all us "others"?
But, then again, the point of the thread is "perhaps we don't need it."
I do appreciate your attempt to help, though.
How about I buy the software for you with my Paypal account and then you make a video tutorial of how to use it? Post it for the benefit of all us "others"?
But, then again, the point of the thread is "perhaps we don't need it."
Last edited:
Can you do any of that in the free "MathCAD Explorer". I thought it was more of a "read only" interface type of thing.
I have no plans to buy MathCAD proper.
MJK for Dummies
Hi
MJK´s models will educate you in how different T/S, shapes, volume, damping and driver offset influence the output.
First time i looked at the models, I gave up and made my first "Hide Away sub" from what i had read was good practise. However, the result was better than expected with a deep and clean bass. I then used several evenings to get hold of MJK´s software and found out that the tuning frequency of my sub cabinet was lower than I had expected. This was a result of the shape of the cabinet.
I then made another Hide Away sub, and finally changed the driver. Please look at the file attached.
Well, when i had got the hang of it, i wrote this articel as a MJK for dummies:
http://www.coolcat.dk/bjoern/QWforDummies.pdf
MJK´s models will predict any shape of box, I always use "sections corner".
Damping is a very important parameter, which is modelled by MJK. By having a gentle decrease of the bass response, you can avoid boomy bass caused by room gain.
Finally: The fee you pay to Martin J. King is a very small amount.
Have fun !
Hi
MJK´s models will educate you in how different T/S, shapes, volume, damping and driver offset influence the output.
First time i looked at the models, I gave up and made my first "Hide Away sub" from what i had read was good practise. However, the result was better than expected with a deep and clean bass. I then used several evenings to get hold of MJK´s software and found out that the tuning frequency of my sub cabinet was lower than I had expected. This was a result of the shape of the cabinet.
I then made another Hide Away sub, and finally changed the driver. Please look at the file attached.
Well, when i had got the hang of it, i wrote this articel as a MJK for dummies:
http://www.coolcat.dk/bjoern/QWforDummies.pdf
MJK´s models will predict any shape of box, I always use "sections corner".
Damping is a very important parameter, which is modelled by MJK. By having a gentle decrease of the bass response, you can avoid boomy bass caused by room gain.
Finally: The fee you pay to Martin J. King is a very small amount.
Have fun !
Amen. $25 is not a vast sum, nor are the sheets especially difficult to work with if you spend a while familiarising yourself with Martin's nomenclature and the basic layout. The critical points to remember are:
1/ Martin has done the mathematics so essentially you only need very basic (as in the ability to count) to use them as far as this side of things is concerned.
2/ They don't design a box for you. They model the behaviour of a driver / cabinet enclosure that you put into them.
1/ Martin has done the mathematics so essentially you only need very basic (as in the ability to count) to use them as far as this side of things is concerned.
2/ They don't design a box for you. They model the behaviour of a driver / cabinet enclosure that you put into them.
The dummy a.k.a. brute force method worked for me:
0. Get licensed and join the Yahoo group, but don't post yet.
1. Read all MJK's papers, and forum posts. Now freak out, because it's all over your head. But you press on because you have developed an unhealthy obsession.
2. Open the MLTL worksheet. Do. Not. Change. The. Values. Leave them as-is.
3. Scroll down -- this is what forces the graphs to re-calculate.
4. Focus on a single value, e.g. Fs, Qts, Vas, whatever, and increase / decrease it. Scroll down and observe the changes it makes in the upper-most frequency response graph (forget about the other graphs -- you're not ready).
5. Repeat for other values. You're not over-thinking, you're just screwing around with variables, la la la la la.
6. Keep playing for a month (or a year). Slowly, you get a feel for the values, just by shaping the upper-most frequency response curve.
7. Eventually you zero in on an ideal geometry, volume and tuning. Now scroll further down and see why your beautiful frequency response is now horrible, what?!
8. Play with boundaries, BSC, etc. and get a good in-room response.
9. Re-read the papers and voila, you are connecting the dots, if only very slowly. Not that it's even necessary, because MJK mastered the math and you just have to master a simple worksheet, c'mon!
P.S. It may go easier if you master WinISD (or similar box design program) before attempting MJK.
0. Get licensed and join the Yahoo group, but don't post yet.
1. Read all MJK's papers, and forum posts. Now freak out, because it's all over your head. But you press on because you have developed an unhealthy obsession.
2. Open the MLTL worksheet. Do. Not. Change. The. Values. Leave them as-is.
3. Scroll down -- this is what forces the graphs to re-calculate.
4. Focus on a single value, e.g. Fs, Qts, Vas, whatever, and increase / decrease it. Scroll down and observe the changes it makes in the upper-most frequency response graph (forget about the other graphs -- you're not ready).
5. Repeat for other values. You're not over-thinking, you're just screwing around with variables, la la la la la.
6. Keep playing for a month (or a year). Slowly, you get a feel for the values, just by shaping the upper-most frequency response curve.
7. Eventually you zero in on an ideal geometry, volume and tuning. Now scroll further down and see why your beautiful frequency response is now horrible, what?!
8. Play with boundaries, BSC, etc. and get a good in-room response.
9. Re-read the papers and voila, you are connecting the dots, if only very slowly. Not that it's even necessary, because MJK mastered the math and you just have to master a simple worksheet, c'mon!
P.S. It may go easier if you master WinISD (or similar box design program) before attempting MJK.
There are two ways to do things; the right way and any other way. If you absolutely refuse to pay for someone else's work, then use HornResp. Yes, it is free. Yes it will give you the right answer. Yes you have to do some visual extrapolating because HornResp does not include stuffing.
Yes, I always model a new driver in a BR box program. That provides a starting point for entry into the MJK worksheets. The final result may well be quite different because of the Q and Vas of any given driver. Your method will produce and OK answer in most cases and a correct answer by pure accident. Your choice of pipe length is extremely important to getting the right answer, and there is no ROT for that.
Bob
Yes, I always model a new driver in a BR box program. That provides a starting point for entry into the MJK worksheets. The final result may well be quite different because of the Q and Vas of any given driver. Your method will produce and OK answer in most cases and a correct answer by pure accident. Your choice of pipe length is extremely important to getting the right answer, and there is no ROT for that.
Bob
Bob Brines,
I think you are absolutely right on all points. Luck so far has played an important part. But what I agree with you most about is that it will produce an OK design. The pipe length is critical and perhaps sim in HR will let one narrow down on that. For folks who don't use HR yet, but are comfortable with a BR program like WinISD this is one way for them to imagine the possibilities of what the cabinet looks like as a MLTL. One way to cheat is to use a length similar to a known MLTL for a known driver size.
I think you are absolutely right on all points. Luck so far has played an important part. But what I agree with you most about is that it will produce an OK design. The pipe length is critical and perhaps sim in HR will let one narrow down on that. For folks who don't use HR yet, but are comfortable with a BR program like WinISD this is one way for them to imagine the possibilities of what the cabinet looks like as a MLTL. One way to cheat is to use a length similar to a known MLTL for a known driver size.
OK, yea, that will work sometime, most of the time if using the same driver. Several years ago, I gave a talk on quarter-wave generators at the Lone Star Audio Fest. I stressed the fact that once you get outside a golden ration BR, standing waves becomes a problem. One of the attendees was a builder of medium range ($3000-5000) boutique speakers. For him, the light came of as to why he had unexplained wiggles in his FR plots. Now he does all of his modeling with quarter-wave software.
Now, you can stuff the bejesus out of any speaker you can make the standing waves go away. What you end up with is an infinite baffle, which is exactly what a traditional TL is when stuffed to remove the second impedance peak.
Bob
Now, you can stuff the bejesus out of any speaker you can make the standing waves go away. What you end up with is an infinite baffle, which is exactly what a traditional TL is when stuffed to remove the second impedance peak.
Bob
Stuffing the bejeezus also kills all the efficiency you may have had to get the bass augmentation. There is also the nautilus spiral like B&W did without stuffing to get infinite baffle. I guess one could make the Cornu spiral start out large and taper down to get an infinite TL wall mount speaker with no stuffing.
I like the stealth sub woofer. Nice design that gets you down into 20 hz territory. Very cool.