The art of speaker design?

[Wheels]

Its been a long time since I've posted here... I've been lurking in the shadows when I could. So Hello again!

So I now have some spare time again and was wanting to design a set of speakers for HT and music purposes. I know its a very large task, but I have this complex where I just like to design stuff from scratch. I found some very useful info from Here but when I was reading their equations it appears it was the database for their calculators and some of the inputs looked like variables for their stuff and I couldn't figure out what they were actually using.

I figure I'll have to read a lot, but I don't know really where to start to make sure I'm learning the right stuff. It doesn't matter how much you tweak a design if your modeling it wrong to begin with and I want to make sure I'm going down the right path.

For the crossover and box calculations, I'm an electrical engineer, specializing in signal processing. I'm not afraid of the math and would personally like to figure it out on my own and have someone review it to make sure I'm not doing something stupid. I'm planning on making a few smaller test runs to pick drivers and see how they respond before I try to build something like mains.

so basically what I think I'm looking for for starters is:
1. How do you model a speaker in a circuit? Is it as simple as a resistor and Inductor in series? If I get a circuit drawn I can crunch the numbers and play with stuff to tweak crossovers.

2. Qts, Qms, Qtot? I know its used to determine box size, but what exactly does it mean? I guess I'm really looking for what specs do you look at when picking speakers and why do you look at them, and what are you looking for?


Thank you in advance. I'm not asking for your trade secrets necessarily, but any tips you have for someone wanting to learn how to design speakers would be greatly appreciated. But stuff like how the mini statements have the tweeter and mid in reverse polarity. How did Curt and Jim know to do that? And their design with the open back mid, how do you calculate the size and length for those things?

I apologize if I missed a thread already covering this. I'm kinda in a search overload right now...

[Undefinition]

There's information all over the internet, but it's very scattered. Speaker design is still a black art.

The very best intro to speaker design that I know is Ray Alden's book Speakerbuilding 201. Sure, some of the design methodology in there is a bit antiquated in the age of computer simulation. But it's very solid fundamentals. Especially great is the first chapter, where he very carefully and clearly explains all the mechanics of a driver, and explains what the Thiele/Small parameters are all about.

[---k---]

Are you saying your not wanting to use available crossover and enclosure modeling software?

[Wheels]

Thanks, I'll have to check that book out.

depends on the price. There is a lot of free stuff out there that I assume would "work" but the few I have played with a few box size programs a couple years ago you don't have to input all the values they ask for to get a number out of it. So I am wondering why the extra values are needed (its been too long for me to remember which ones I could omit), and how do they effect the outcome. And I have read a few articles where people didn't set the model up properly and the results when they built it were significantly different.

As far as crossovers, I assumed the software just calc'd a freq response of the circuit given the correct impedance of the speaker was used. I don't particularly want to pay someone a couple hundred bucks if that is the case, because I have done that way more than I wanted to. Now if the crossover design ends up using the box parameters and more than just speaker impedance (ie Vas, Q...) sure I wouldn't mind looking into some software, but I still would like to know how to calculate what it is doing for a sanity check when I'm done to make sure I am setting things up right.

[---k---]

Yeah, computer modeling software has progressed.

Unibox

http://audio.claub.net/software/kougaard/ubmodel.html

for enclosure design is very accurate. It is what most of the people here use. If it is off, it is usually because what you inputed for stuffing or volume is incorrect.

PCD
for crossover design is a FREE program. It is also extremely accurate. It can use both the driver impedance and a measured frequency response to design the crossover. Many people (including Paul above, make sure you visit his site and read his FAQ's) use it to design crossovers and the measurements after match the predicted response. Jeff Bagby who wrote the software, I'm would assume uses it for some of the crossovers he does for very expensive, well acclaimed commercial speakers. Very worth checking out.

http://audio.claub.net/software/jbabgy/PCD.html

SpeakerWorkshop is another FREE crossover design program. CJD uses it to design many of his speakers. Again, very accurate and can use real measured frequency response of speakers. So, you're not just throwing a text book filter at the speaker, you're combine the real response of the driver with the crossover to get a real filter.

http://www.speakerworkshop.com/index.htm

You can see CJD's process with my Khanspires or lots of other threads in Mission Accomplished.

[Wheels]

hey, thanks for the links. The only catch right now is I primarily use linux and Mac and I don't know that I still have the student edition of ms office I got while in school. I guess I'll have to look for that tomorrow (or later today as I just realized what time it was... woops)

Does anyone know if the above programs are compatible with excel '07 or '10? Or should I look for an '03 version? The reason I ask is I have had VBA compatibility problems before because Microsoft doesn't understand the concept of a standardized programming language!! I'll save the rest of that rant for another day...

[maynardg]

Start here:

http://sites.google.com/site/undefinition/diy-gettingstarted

Jeff Bagby's software is available for Excel 2003 and 2007.

[JonMarsh]

The newest version of Office on the Mac, Office 11, FINALLY has decent compatibility with the current PC version, Office 10, including most of the macro language and I believe VBA, too.

For measurement on the Mac, FuzzMeasure 3 Pro is the way to go.

FuzzMeasure Pro



Getting Started with FuzzMeasure Pro

Has good options for gate windowing, filtering, data export and manipulation, extracting minimum phase response, etc.

I use low cost Firewire audio interfaces with it, like the TC Konneckt 8 or the M-Audio Firewire Solo. Easy to setup, works like a charm for the most part.

Only caveat I can think of is that if you do wide range measurements (say, 10 Hz to 40 kHz), the resulting data export file can overwhelm some simulation programs. Also, not all simulation programs are equally adept at parsing text delimited files from different sources. If your data looks funny when plotted in the modeling program (like LspCAD or Sound Easy), then that's what to look for.

I have Praxis on the PC, and a dedicated PC for running it, but it's been years since I've used it- can do pretty much everything I need with Fuzzmeasure, and no copy protection to deal with (via a hardware USB pod). Both use swept sine chirps that are convoluted to an impulse response, manipulated there for gating and windowing filters, then displayed as a frequency response or distortion plot. One measurement can do both in FuzzMeasure.

[---k---]

CJD has gotten Speakerworkshop to run under Linux. I believe using Wine. I'm not sure if it worked flawlessly or not. I'm sure he'd chime in if the thread was titled 'Speakerworkshop and Linux'.

[cjd]

Yeah, the modeling features of SW worked fine in Linux (under WINE). I never tried the measuring side of it, but then I've stopped using that even on Windows at this point.

I haven't tried any of the FRDC Excel stuff in the latest version of Office on the Mac (come to think of it, I think I don't have it, I got my latest replacement a few weeks too early...) but it definitely does not work on anything 2008 or older.

Measuring is by far the worst part, but getting past that is worthwhile - especially if you understand the math.

[---k---]

Just saw this posted at PE. A web base box modeling program. Looks simple, but really box modeling is.

http://udv.softica.dk/sbp/

[Justaguy]

I am also a noob to speaker design. At first I tried researching on the web and found a lot of useful information. The problem is that the information is scattered and it can be difficult to form a really good picture of what to do and why. I ended up buying "Speaker Building 201" and "The Loudspeaker Design Cookbook 7th Edition". I found that I had to read "Speaker Building 201" first. Gave me a really solid understanding of some fundamental concepts. I then read the "Loudspeaker Design Cookbook". That book does cover fundamentals but at an accelerated pace. It is also VERY heavy on graphs and charts. It also helped me to understand different crossover types and how and why they are employed.
I liked both books because they gave me a good place to start. All the information is in one location that is easily referenced later, something I do often.
The time spent reading a couple of speaker design books is time well spent.

[Wheels]

Wow. thanks for all the responses. Its been a crazy weekend and I just now had time to get back here. I do have an M-audio 2 channel interface box I can use. I'll give it a try in linux, but I found out my folks still have their version of '03 office they aren't using anymore. I have an XP machine... but i'd still rather give it a go in Mac or linux

Do you guys have any measuring mic recommendations?

[Bear]

Wow. thanks for all the responses. Its been a crazy weekend and I just now had time to get back here. I do have an M-audio 2 channel interface box I can use. I'll give it a try in linux, but I found out my folks still have their version of '03 office they aren't using anymore. I have an XP machine... but i'd still rather give it a go in Mac or linux

Do you guys have any measuring mic recommendations?

A Behringer ECM8000 that has been calibrated (<$50 for the mic, ~$100-ish for the calibration) will carry you along for a long, long time. Parts Express also sells the EMM-6, which is basically the same thing, and they offer a calibration file for ~$50 all-in, IIRC. I'm not sure I'd trust a mic and a calibration for $46 all-in, though (labor is expensive and calibration should be done under NIST conditions). If you want better, Earthworks makes some good mics as do others. Eventually, all roads lead to B&K (not the amp/processor people).

http://www.bksv.com/

Parts Express is selling a variation on the ECM8000 with a built-in DAC and a USB port (Omni Mic). There have been very positive reviews, but I haven't found anyone with specific knowledge doing high-end work with it yet.

http://www.parts-express.com/pe/showdetl.cfm?Partnumber=390-790

[BOBinGA]

Herb Singleton at Cross Spectrum Labs sells the PE EMM-6 calibrated for about $70 which is a very good deal. I got one from him and was very pleased with his promptness and service.

http://www.cross-spectrum.com/measurement/calibrated_dayton.html

Along with the mic, you will need to purchase a mic cable, mic stand, a mic preamp with a phantom voltage source (the Beringer XENYX 502 will do just fine - http://www.parts-express.com/pe/show...number=248-574 ) and an RCA to 1/8" adapter cable to get from the Beringer to your computer's aux or mic input on the sound card.

There are many freeware measurement programs for Windows (I use ARTA - http://www.fesb.hr/~mateljan/arta/index.htm ), but not so many for Linux and Mac. Jon's suggestion to use FuzzMeasure is undoubtedly good, since our dear moderator has measured more speakers with it than I'll measure in my lifetime.

-Bob

On second thought, you might need the Beringer 802 since the 502 doesn't appear to have the phantom power source for the mic. The PE catalog says the 502 has phantom power, but the web page doesn't. You might call them to see which is correct.

[Bear]

Herb Singleton at Cross Spectrum Labs sells the PE EMM-6 calibrated for about $70 which is a very good deal. I got one from him and was very pleased with his promptness and service.
http://www.cross-spectrum.com/measur...ed_dayton.html

That's a good deal! They came way down on the price of what they are charging for the calibration. They calibrated my ECM8k, and the calibration alone was more than $70, IIRC.

[Wheels]

Thanks guys for the recommendations. I'll check those out.

[NateTTU]

I can't remember where I read it, but the guy from cross spectrum said generally the Dayton EMM-6 microphones were overall better quality. He had more bad units out of the Behringer ECM8000 mics over the Dayton ones. However, he did say the Behringer mic comes with better extras.

[JonP]

Along with the mic, you will need to purchase a mic cable, mic stand, a mic preamp with a phantom voltage source (the Beringer XENYX 502 will do just fine - http://www.parts-express.com/pe/show...number=248-574 ) and an RCA to 1/8" adapter cable to get from the Beringer to your computer's aux or mic input on the sound card.

There are many freeware measurement programs for Windows (I use ARTA - http://www.fesb.hr/~mateljan/arta/index.htm ), but not so many for Linux and Mac. Jon's suggestion to use FuzzMeasure is undoubtedly good, since our dear moderator has measured more speakers with it than I'll measure in my lifetime.

-Bob

On second thought, you might need the Beringer 802 since the 502 doesn't appear to have the phantom power source for the mic. The PE catalog says the 502 has phantom power, but the web page doesn't. You might call them to see which is correct.

I went thru the buy 502, find no phantom power, :M return and get 802 shuffle... BUT, as I understand it, that was true for the older models. In the new "Xenyx" model line I'm pretty sure the 502 now includes the missing phantom +48V. Check first before ordering...

And, another :T for Herb at Cross Spectrum... I got one of the early batches of EMM6's that he purchaced and calibrated. Good service, good product.

Final, unrelated thing on Jon's Fuzzmeasure pic. Bose? BOSE??? Tell me its not true!!! :E

:rofl:

[Wheels]

I went thru the buy 502, find no phantom power, :M return and get 802 shuffle... BUT, as I understand it, that was true for the older models. In the new "Xenyx" model line I'm pretty sure the 502 now includes the missing phantom +48V. Check first before ordering...

Ive got a M-audio pre-amp interface unit. it has phantom power, plus its got the dedicated ADC that sends the signal out via USB directly to the computer. Its setup as my default sound device in windows so I think it would work correctly with the measuring software


I think I'll look for the Dayton then