diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Full Range (http://www.diyaudio.com/forums/full-range/)
-   -   New version of Martin King's MathCad Worksheets is coming soon! (http://www.diyaudio.com/forums/full-range/69760-new-version-martin-kings-mathcad-worksheets-coming-soon.html)

Scottmoose 14th December 2005 09:52 PM

New version of Martin King's MathCad Worksheets is coming soon!
 
2 Attachment(s)
OK guys

It’s time to be getting very excited. Along with Martin, and Bob Brines (not sure about anyone else) I’ve been testing a new version of Martin King’s MathCad worksheets, and it’s a stunner. Before I go any further (and Martin gets inundated with emails requesting a copy of the new sheet!), I must stress the word ‘testing.’ Martin understandably wants to ensure that there are no bugs in the sheet before it is released into the public domain, so please remain patient whilst we trial it to make sure everything is optimised.

For your interest, at present, the changes have only been made to the Ported Box worksheet (‘only!’) but they will be implemented to all of the sheets –TLs, Horns, Sealed Box, Isobaric etc early in the New Year. I believe Martin intends to implement the changes, then release them all on his site, rather than one sheet at a time, to save confusion.

What am I doing starting this thread then, if the sheets aren’t available yet? Well, it’s partially to let you all know what is afoot, and also to allow Martin to gauge the level of interest in the new worksheets. Over these next few posts, I’m going to show you the new facilities, together with some comments about what each of them is and does. Over the next few days and weeks, I will post further examples of plots, along with notes on how they were made, comments on what I have seen, what I will try next etc. By doing this, I am going to make quite a lot of very public mistakes. When I find that I have, I will tell you, together with what needed to be done to correct it, or I will ask for advice and / or ideas from the rest of you. If you see something wrong that you think I have missed, please tell me. That’s good; that’s what we want. Knowing what mistakes I have made during this learning process will help you, when the time comes, to avoid the errors I made. Please join in –if you have anything to contribute, any questions, suggestions, ideas; even requests for additional facilities you might like, please say so. You never know –it might well be possible to include them. Martin wants a good debate you see. And so do I. That's how we find out new things.

All I am going to do today is show you the new sheets, with a (very) rough sample input into them. So, without further waffle, we’ll get on with it. What is included in the new sheet? Sitting down? Good. Got a drink? No? Get one. Make it a double.

Welcome back. This is the part where I have a big grin. You’re going to like this.

The top part of the sheet is unchanged, apart from one thing that I for one find a Godsend. A little extra facility for calculating any extra series resistance you might wish to add to bring down a slightly rising response. As many people use SS amps with full-range drivers (Martin and myself included), I think this is extremely useful. If you use series resistance provided by a thin-gauge wire, then I would advise calculating this resistance in one of the several calculators available for the purpose (I use the one from TNT audio. I’m not a fan of their site particularly, but this is a useful little tool) and then enter it. It’s worth doing –you’ll get a more accurate response calculation. I won’t bother positing a picture of this part –you can imagine what it’s like.

Below, I present the anechoic frequency response for what I term my ‘Small Thor.’ Like many others I’ve found the well-known Thor TL designed by Dr Joseph D’Apollito to be sub optimal, and have been messing around, coming up with some better cabinets for these drivers. (Ok, Ok, it’s an MTM 2 way, but that doesn’t matter, this is just an example. This design is also an MLTL rather than a vented box if it comes to that, but again, that isn’t an really an issue here. The sheet doesn’t care –it just models what you put into it.)

Enclosure size is as follows: 7.5” x 10.5” x 45” (WxDxH) internally. 0.5lbs ft^3 of stuffing. 0.5ohms series resistance applied. Driver centre (see Martin’s article on modelling 2 drivers in 1 enclosure for more on this) 9” below the top of the cabinet. 3”x4” port (WxL) 4” from the base of the cabinet.

Scottmoose 14th December 2005 09:53 PM

2 Attachment(s)
So far, so normal. You could do all of this, bar adding the series resistance, in the current versions of the sheets. What then, I hear you say does the new sheet offer. OK, the facility for adding series resistance is useful and all that, but is that it?

Err, not exactly. You see, you know where the current sheet ends –you have a set of graphs presenting the calculated anechoic response, of which the last is the System Time Response for an Impulse Input? Keep scrolling on down guys. We have arrived at Part 2 of the new worksheet, and as you will see, Martin has only just begun. Because this new version of the sheet isn’t limited to ‘just’ anechoic anymore. Oh no. This time, we can calculate the in-room frequency response too. Yep. The in-room response. You can tailor make your new speaker specifically to suit your room and listening position. Anyone interested in being able to do that? Thought so.

Right, Part 2. Here we go. The calculation includes the exact horizontal and vertical location of the driver and port on the baffle (which you can adjust, and you can select between a front or rear port this time too); the baffle step diffraction of the driver and port, and the floor and rear-wall reflection of the driver and port. You can see the values that I have adjusted for the Small Thor in the attached image below. Some I have not needed or bothered to change from the default settings; those are the values in black. Those I have altered are in green, as usual with Martin’s MathCad sheets.

As you can see in this image, we enter the exact external width, height and depth of the cabinet we are planning to build, and also, the distance of the front baffle (not the rear, the front! Thought I’d stress that as it’s easy to make a mistake) from the rear wall. All self explanatory, and very easy so far. Nothing to it; anyone who’s used the current sheets will find this a breeze, and even those of you who haven’t –well, rocket science it isn’t.

The next value you won’t be familiar with however. Here we can enter the number of points per quadrant of the baffle edge. The default value is 9, as you can see. Basically, when performing its calculations MathCad looks at a number of specifically defined points along the baffle edge to calculate the in-room response. The higher the number you enter, the more points MathCad will look at, and the more accurate the end prediction will be. I wouldn’t advise going for a number lower than the default settings –if you have the potential for greater accuracy, and you’re going to the trouble to use the sheets, don’t sacrifice it for the sake of a few more seconds calculation time. Those of you with processor speeds faster than the combined wit of the cast of Monty Python’s Flying Circus and enough RAM to cause a mainframe to blush can enter some higher value if you wish to get even greater accuracy. One thing to be aware of though guys, and I might as well let you know here as anywhere: these are heavy-weight calculations being made, and it takes time to run. You know it takes perhaps 5 seconds to calculate the response in the existing sheets (or Part 1 of the new one)? Well, this takes over a minute on my machine, which has a 2Ghz processor and 1Gb of RAM. OK, not cutting edge, but not too slow either. Average so far is around 75 seconds. It’s worth it though. The best things come to those who wait, so pour yourself a pint of Guinness while waiting for your PC to finish doing its thing.)

The driver geometry input allows you to position the centre of your driver exactly where you want it on the baffle, in both the horizontal and vertical planes. Note that there is another of those ‘number of points’, in this case across the diameter, that MathCad will use to calculate the in-room response. The default is, as you can see, 5, which I have left it at in this case.

We haven’t finished with Part 2 though yet. We’ve got 3 more sets of variables to enter yet, and these will be shown and described in the next post.

Scottmoose 14th December 2005 09:55 PM

2 Attachment(s)
Welcome back. As you can see from the attached image, the three sets of variables we enter here are the port location on the front or rear baffle in the horizontal and vertical planes (and another of those ‘number of points’ values, which I again in this case have left at the default value of 4, but you can adjust to whatever you want).

We also can enter the number of degrees off-axis we listen to our speakers from our favourite armchair. Again, I have left it at the default value of 0, because I tend to listen to multi-way speakers like Small Thor on axis, but you, obviously, can adjust, and also experiment with, to find the best response and toe-in angle. I would suggest adjusting the angle in increments of 5 degrees until you find the best approximate response, then adjust in single degrees and refine that way.

The final thing you can select is whether you have a carpeted, or a hard floor. All you need to do here is enter a single number, 0 if you have a hard-floor, 1, if you have carpets. (Cork flooring –who said that?!? ;-)

Scottmoose 14th December 2005 09:56 PM

2 Attachment(s)
A very short post this. Below you will see the worksheet’s graphic depiction of the enclosure geometry you entered: useful for seeing at a glance if you have made a mistake, like putting the driver at the bottom of the enclosure, as I nearly did the first time I fired up the worksheet. Go on –have a laugh at my expense… (mutter, growl etc.) Again, this is for my Small Thor enclosure.

Completed the inputs? Good. Now scroll down, have a rest and a Guinness while waiting for the calculations to be completed.

Scottmoose 14th December 2005 09:57 PM

2 Attachment(s)
Aha. We’ve finished, and we have 6 new graphs. It’s the last two which are probably of greatest interest to you at the present time: this is the in-room response, as it stands, of Small Thor, on axis, in a carpeted room with the front baffle 32” from the rear wall.

Scottmoose 14th December 2005 09:59 PM

2 Attachment(s)
Hmm. How gratifying. It’s not too bad for an in-room plot at all. (I've seen a whole-lot worse) There is, however, a bit of a baffle-step issue.

Now, at times like these, we hit the textbooks and internet searching for appropriate values for a BSC circuit to correct this response, right? And we have to make a best guess at where the centre of this baffle-step is, right?

Not any more you don’t.

Welcome to Part 3 of the new MathCad worksheets. This is the BSC circuit designer. And guess what? Martin is a very nice man indeed, because the sheet has already worked out where the centre of the baffle-step problem occurs, and what it considers to be the necessary level of attenuation. It also suggests the values required for the inductor and resistor the circuit needs, and displays on a final graph (depicted below) the in-room response with BSC applied. You can adjust these yourself if they need modification. I won’t post a picture here of the input section –it’s all pretty obvious, and shall instead show the in-room response with the suggested BSC circuit applied.

It looks like the circuit has gone too far doesn’t it? Possibly. My interpretation however is that the circuit has flattened the response out, and extremely effectively, but the bass is overly prominent.

Compare the anechoic response, which is nominally completely flat, even in the bass, to the dramatic boost in these frequencies shown in the in-room plot, which I assume to be the result of room-gain. We can play around with the values we entered in Part 2 to try to rid ourselves of this unwanted, huge peak (a small peak can be quite effective, adding some weight or warmth to the sound, but 12db is beyond the pale in my view. Great for those home-theatre enthusiasts, but not so good for music) What is needed then, is a redesign of the enclosure which will result in a bass-response which will gradually fall away on an anechoic simulation, but when we factor room gain into the equation, will result in a nominally flat response. A modification to the basic enclosure geometry could also be desireable, as the 15ohm resistor requred to attenuate the predicted response in my room at least is going to kill the efficency stone-dead. Alternatively, I could try listening about 30 degress or so off-axis.

I’m not going into the different methods of achieving this –there are numerous ways, but that’s for another time. I’ll probably look toward adjusting the port dimensions to see if that can help out first, and also loose the 0.5ohms of series resistance I added in the first place to flatten the anechoic response slightly. If that doesn’t work, I’ll start playing with the cabinet somewhat. Suffice it to say however that I don’t subscribe to the ‘increase stuffing to damage the fundamental and thereby reduce the bass’ approach. A properly engineered cabinet shouldn’t need that.

And that’s it for now guys! I look forward to reading any comments, suggestions, advice and questions, as I’m sure Martin will too. Sorry about the sheer length of these, and the number of posts, but I couldn't really do it in a shorter way whilst still giving you the same information.

Best
Scott

soongsc 14th December 2005 11:28 PM

1. Does it take into acount driver size and Xmax effect on low frequency drive capability? I beleive the original worksheets just told you how much excursion would occur regardless of the Xmax limits.

2. Does it take into acount Bl variation?

3. Does it take into acount listening distance?

4. Which resonance directions in box resonancd are calculated?

5. Can you just input non-standard design shape data fron CAD programs like the Alibre? Smooth horns are understood to react differently from piecewise straight curves.

Dumbass 14th December 2005 11:49 PM

This looks like very cool, and useful, stuff. The BSC modelling, especially, could be very handy.

Kudos.

rjb 15th December 2005 12:40 AM

Brilliant!

This is certainly a major improvement. The old set were good, this is much better.

However a question re in-room response. I'm currently building a three-way where the bass is really a sub- ie 100 Hz or so- but the driver faces the wall. Can it do this? ( the other drivers face front)

I guess it would be similar to where the driver faces the floor.

The same question also would apply to side mounted drivers.

Maybe something has to be left for the next upgrade:-))

LineSource 15th December 2005 12:54 AM

Will this program cover line source and line array speakers?

Will this program cover putting speakers up against the wall? Putting speakers into the wall?

Flat panels hanging on the wall are changing many room layouts, and putting the speakers against the wall is demanded by many interior designers.


All times are GMT. The time now is 04:04 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2