Got any practical tips for measuring drivers & crossover sims?

[JonW]

In designing my first speaker, I’ll soon be buying the drivers and putting them into a test cabinet. Then I need to take measurements in the cabinets and use the data to model the crossover. Any practical tips I could get for how best to go about that would be great.

I’ve got the ECM8000 microphone and MobilePre sound card. I’ve used them with Room EQ Wizard and things work fine. I’ve got the book “Testing Loudspeakers” but haven’t had a chance to read it yet.

The two programs that I see mentioned most often for this are Speaker Workshop and LspCAD. I’ve also read about people getting frustrated with Speaker Workshop. I’m happy to spring for LsPCAD if it would make things easier/quicker (it’s not that expensive).

What software program(s) are your favorites for collecting the measurements and then modeling the crossover?

And do I need anything else?

Thanks.

[Jed]

This is a good site about this topic. http://www.claudionegro.com/

[Dennis H]

JonMarsh uses LspCAD. 'Nuff said.

[cjd]

JonMarsh uses LspCAD. 'Nuff said.

Uh, I think he's moved on to Praxis...

[chasw98]

Dennis Murphy uses both Praxis and LSPcad.

This is a good site about this topic. http://www.claudionegro.com/

And here is Wallins page for building what looks to be a handy test jig.

http://mysite.verizon.net/tammie_eric/audio/jig2/jig2.html

Let me know if you need one built!

[Paul W]

I use LspCAD and Soundeasy. Soundeasy has a more complete measurement suite but LspCAD is far more intuitive.

My vote goes to LspCAD.

[Dennis H]

Uh, I think he's moved on to Praxis...

Errr, Praxis is the Porsche of measurements but it doesn't do crossovers. Still gotta have LspCAD to do something with those measurements.

[ThomasW]

Jon,

REQW is fine for amateurs dialing in their subs, it's not a tool for taking serious measurements.

A good very low cost software program is Audio Tester

Next up is ETF 5.0 and the free add-on/upgrade R+D Modal Analyzer. You can downlaod and use the demo version of ETF 5.0 to get your feet wet.

[cjd]

So, to get your feet wet Speaker Workshop with the basic cabling (as opposed to the jig) works beautifully.

[Jed]

So, to get your feet wet Speaker Workshop with the basic cabling (as opposed to the jig) works beautifully.

I've been waiting for someone to give me a reason to spend 200 on LSPCAD, when it looks like speakerworkshop can do just about everything LSPCAD can.

[JonMarsh]

Uh, I think he's moved on to Praxis...

I use Praxis and Clio for measurement, LspCAD for design development and modeling (including LF modeling).

I've never actually used Just MLS, which comes with LspCAD. (I'm a lazy git...

These days I leave CLIO setup for impedance measurements and LCR type measurements, and use Praxis for the mic setup. Only problem I have is that I never could seem to get my mic channel working right, and have been doing a work around for the mic input with the Probe 1 input. That's why I don't have reference levels- tried re-cal a few times, just don't seem to get any signal through on the mic line input, even though my ACCO mic puts out volts literally from the preamp.

[JonMarsh]

I've been waiting for someone to give me a reason to spend 200 on LSPCAD, when it looks like speakerworkshop can do just about everything LSPCAD can.

:rofl:

(no offense intended- I presume you're trolling to get a rise out of me?

Try modeling and optimizing the Natlie P or Modula MT in Speaker Workshop. Try creating a dipole bass model in Speaker Workshop. Try emulating a crossover design with music on your speaker. Uh, predicted polar plots based on driver diameters? Snapshots? Using optimizer to optimize impedance zobels? Selective frequency range optimization? range of acoustic target models built in? Import of acoustic target models from text files? Mixing active and passive filters? Saving reference sets for comparison?



Speaker Workshop is a GREAT program to start with, and if your'e limited in funds, it's obviously the way to go-

Going through their "Speaker Project" tutorial is a good overview of a lot of important stuff in the design process.

But though I don't recall all the details about it when I last played with it years ago, I didn't think it compared with LspCAD 5, much less 6.

JMO, of course, and YMMV...

Of course, I used to use Soundeasy, too, and switched from that to LspCAD half way through the X1 project, back in 97.

Of course, it's a poor workman who blames his tools, and in the end, I think it's really all about what you feel most comfortable working with. Some great designs are done with SoundEasy, it's had a lot of updates in the last several years, but I like how LspCAD works, especially the changes and steady updates in version 6, plus the organization it uses and it's overall flexibility- it's really well suited to a "production" environment if that's what you need, but I think it's also easier to use on a casual basis, too.

[Black300zx]

I just got done setting up Speaker Workshop the other week. Here's a list of things that got me hungup or frusterated:

1 - The Wallin II Jig. While it wasn't hard to build, it added a level of uncertainty when I was trying to troubleshoot. IE - If a test was coming out weird, I always had the possibility of a bad connection in the jig lingering in the back of my head. If you use the cable jigs initially to learn SW, it's much easier to look at the jig setup, follow a couple wires, and confirm the jig is hooked up correctly.

2 - Although my laptop's soundcard came up as being "compatible" it only has a headphone out and a mic in. The mic in must have a very high input impedence, as the soundcard impedence test was coming up around 900k ohm, when it should be in the 45-50k range. This frusterated me for a night or two until I hooked the jig up to my desktop and realized the problem was my laptop. I'll be getting a SB USB soundcard for it soon (~$50 from newegg).

3 - Latency. The graphs were defaulting to cover around 2 seconds along the x-axis. With a 1khz tone, the result is more or less a solid black block. The "Unofficial Manual" tells you to set the x-axis range to 0-4.0 ms for these graphs. I would do this and get nothing but a small voltage ripple that barely registered. Turned out that I had ~20ms of latency, and the reccomended x-axis range stopped before any signal was even recorded.

Those were probably the biggest hangups I can think of. Another thing is that the initial setup is pretty time consuming, and can get frusterating. Once you get past the setup, it's pretty easy. I think if you use the cable jigs, and only hit it for an hour or two at a time, the learning curve is much easier.

[augerpro]

I really like Soundeasy for an all in one package. ALOT of features, flexibility, and power. Unfortunately I've found the user forum less than helpful.

[capslock]

What kind of measurements can be done in Praxis that Soundeasy does not support?

I have been browsing the Arta manuals, which is anonther dedicated measurement suite, and I noticed that both response as well as absolute calibration issues are dealt with in more detail than SE, and that you are always given the choice of referenced (2 channel) and unreferenced measurements (1 channel). Actually, with a soundcard that has good channel to channel matching (both FR and sensitivity) and a good power amp, I would suspect single channel to be sufficient, and it is bound to have lower noise.

The two other points are that Arta supports Fourier in addition to MLS analysis, and that literally everything can be done with step sine excitation if the user chooses. Sines are also implemented in SE, but I don't know to what degree. I am pretty sure that Fourier analysis is patchier in SE.

[Jed]

no offense intended- I presume you're trolling to get a rise out of me?

Yup, a little bit. Sometimes I have to be absolutely sure that something offers a substantial performance upgrade over freeware. Some of the things you mention, like modeling a Modula crossover, is well within the means of speakerworkshop with very accurate results. Also, I can overlap FRD files, create target acoustic slopes, model zobels or optimize zobels, optimize networks based on target acoustic slopes with no issues in speakerworkshop. It doesn't do some of the features you mentioned, but I'm wondering how often they would be used anyway? From your response it sounds like you use all the features quite a bit. It would be interesting for me to see some examples of how each feature in LSPcad is used in some of your designs. That is how I learn best, that is to see an example and also see the supporting steps and features used as a practical solution to the problem.

[JonW]

Wow- thanks guys. Much more feedback than I expected. And some strong feelings for and against various software.

This speaker building is supposed to be 100% fun for me. And I spend so much time in front of a computer at work, I’m probably willing to spend some money for software that will take less fighting with. Maybe I’m lazy. Heck, I’m typing this from a Mac. (but have a PC at home for speaker work)

I’m hesitant to start down the Speaker Workshop road because I’ve read of many people being frustrated with it. For example:

I just got done setting up Speaker Workshop the other week. Here's a list of things that got me hungup or frusterated...

Another thing is that the initial setup is pretty time consuming, and can get frusterating.

And reading over this site:

This is a good site about this topic. http://www.claudionegro.com/

does not instill confidence that it will go smoothly. Then again, I was concerned I’d have trouble with the other programs but have found BDS, Edge, SPLtrace, WinISD, and Unibox all to be quite easy. But they are performing much simpler functions.

Thomas linked to some interesting programs I had not heard mentioned before.

So, to get your feet wet Speaker Workshop with the basic cabling (as opposed to the jig) works beautifully.

and then…

I use LspCAD and Soundeasy. Soundeasy has a more complete measurement suite but LspCAD is far more intuitive.

My vote goes to LspCAD.

...but I like how LspCAD works...

...but I think it's also easier to use on a casual basis, too.

Hmmm…

[JonW]

Dennis Murphy uses both Praxis and LSPcad.



And here is Wallins page for building what looks to be a handy test jig.

http://mysite.verizon.net/tammie_eric/audio/jig2/jig2.html

Let me know if you need one built!

Yeah, I came across that. So, presumably, any software like LspCAD will only get you measurements and crossover sims. But you need to build some sort of jig to get impedance data for any of the programs? And how important is it to get those data? Pretty important, I’d think- I don’t want to make a crossover that drops to 1 ohm and such.

[ThomasW]

Jon,

LspCAD doesn't do measurements. Included with LspCAD is JustMLS, it's a very basic measurement program. Arta, Audio tester, ETF 5.0, and the user hostile freeware Speaker Workshop, are better testing tools than JustMLS

To get started you need to measure the frequency response in your baffle. That data is then imported into a crossover modeling program.

[cjd]

You simply can't model without impedance data. You *can* do a lot of work without T/S parameters but you need them eventually. Pick the right components and you can cheat and use data others have published.

I use a smattering of tools, but I stand by Speaker Workshop + FRD Consortium tools being ideal if you're still not sure about the whole thing. LSPCad or Soundeasy or whatever, absolutely more polished - I would say professional. I think Speaker Workshop has lost any continued effort, which is its failing.

The FRD tools are absolutely superb but they have severe limitations due to the format. I know that Dan Cyr was working on a software crossover implementation that looked really promising but I haven't talked to him in a couple years (Ironically, I raced with him before I realized he was THE Dan from FRDC)

Speaker workshop is not, IMHO, any more mind-warping than good crossover design. In fact, I find it fairly intuitive for most things.

Jon, I *have* modeled the Natalie P in Speaker Workshop with success, and even spent some effort looking at a hybrid model for the 3-way towers. I NEVER use its built-in optimizer. Maybe my loss. I like to think it teaches me more to optimize manually. I've played with it to see what it does, just not used it. But it will model them accurately, regardless.

C

[JonW]

LspCAD doesn't do measurements.

I see. Didn’t realize that. I do understand I need take the measurement data and then import them into the crossover program.

Included with LspCAD is JustMLS, it's a very basic measurement program. Arta, Audio tester, ETF 5.0, and the user hostile freeware Speaker Workshop, are better testing tools than JustMLS

I chuckled at your use of “user hostile.” But that is exactly what I’m trying to avoid. I don’t want to spend days fighting with some software. Free time is scarce and this is supposed to be fun.

You simply can't model without impedance data. You *can* do a lot of work without T/S parameters but you need them eventually. Pick the right components and you can cheat and use data others have published.

That’s what I figured. I’d prefer to measure things myself, rather than rely upon other people’s data. Aside from the fact that it may be more accurate to my specific case, more importantly I’ll understand things better that way. So then how do people typically obtain the impedance and/or T/S parameters? Praxis is out of my newbie league and the Speaker Workshop jig looks like a headache from many accounts. That’s all that I’ve really seen people mention (not just here).

And thanks for the other thoughts.

[Paul H]

For clarity, while LSPCad doesn't take measurements directly, a sub-program Just MLS which is built into LSPCad does take impedance and frequency response measurements - the two basic measurement items you need to model the speaker.

I was in the same situation a couple of years ago as you JonW. I had the same concerns about wasting time on the learning curve for software, and was willing to spend some money to avoid that. I bought LSPCad and have been happy with that choice.

[JonW]

For clarity, while LSPCad doesn't take measurements directly, a sub-program Just MLS which is built into LSPCad does take impedance and frequency response measurements - the two basic measurement items you need to model the speaker.

I was in the same situation a couple of years ago as you JonW. I had the same concerns about wasting time on the learning curve for software, and was willing to spend some money to avoid that. I bought LSPCad and have been happy with that choice.

Hi Paul,

Yeah, exactly the same attitude I have. You’ve got LspCAD for the crossovers. I recall your mentioning that in another thread. What do you use for measurements of sound (Just MLS?) and then the impedance and T/S? You must have some hardware for the latter?

[Rick Craig]

There's a guy who recently posted on Madisound that he's selling a Clio system on E-Bay. I use Clio for measurements and LsPCad for crossovers. Both have worked very well for me.

What part of Indiana are you in? I'm originally from Fort Wayne.

[TacoD]

I once made a very long write-up about my findings.

[fjhuerta]

I use mainly LSPCad, and do my impedance measurements using the Woofer Tester 2.

There are things I still don't understand very well about measuring things with LSPCad, though.

If I follow the manual, and measure each driver on axis, and then use the x,y,z coordinates and the driver diameter to calculate on-tweeter axis response, the summed curve of all drivers active never matches the reference curve I measure with all drivers active.

So I always measure every driver on tweeter axis.

So I cannot model vertical on / off axis curves, or the power response curve.

I'm sure I'm doing something wrong, I just don't know what it is...

[JonW]

Hi Rick,
I never thought about trying to get such gear used. I’d think there are people out there who have moved on in the hobby. So there might be some available. I didn’t find any Clio measurement gear on ebay, but did see some very nice "clio" dresses.

I’m in the Lafayette area. Stop by if you ever come ‘home.’ And bring some speakers. I’ve read very good things about your designs.




Hi Taco-
Excellent post. I must have missed that earlier. You’re with the big boys running Praxis. If I were to buy LspCAD do you have a recommendation for how to collect the impedance data?




Javier,
One more vote for LspCAD. Thanks. And more stuff to get for impedance data. I haven’t seen many people mention Woofer Tester before (a few, but not many).

[ThomasW]

Jon,

Here's the cilo auction in ebay

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=220087192711

Note if you 'win' you'll need to find a PC that has a motherboard with an ISA slot

[JonW]

Thanks Thomas. Well, you're obviously better at searching between all the dress sales than I am.

My old PC at home used for speaker things is a laptop. So maybe I can't use Clio. Unless I were to get another computer. At which point I'm getting over my head. But if there happens to be an old, unused PC around work... Oh heck, I doubt I'll find one, but I can have a quick look.

[Dennis H]

JustMLS works about like the others for impedance measurements. Basically, you need a reference resistor and it compares the speaker to that. It's all in the pdf manual including how to build some probes with a couple of resistors so you don't accidentally put too much voltage into your sound card and make 'magic smoke'.

Edit: JustMLS looks as easy to use as any of them, and it does all the basic speaker stuff, so you could start with that and upgrade later if you felt the need.

[JonW]

Thanks Dennis. I just took a look at the resistor setup needed for JustMLS. Just some alligator clips (have those already), resistors and wire. Maybe this is all easier than I had feared.

[TacoD]

Impedance with JustMLS is indeed the same as with Praxis. And with JustMLS I got great results (SPL and impedance).

[hdspeakerman]

Taco - Long write up on measuring

I understand that Soundeasy has measuring capabilities beyond what you can get for LspCad but why do you considering it advanced compared to LspCad? I am in the same boat as John W and am definitely a type1 user as your speak about in your write up.

[ThomasW]

This thread contains various people's gear, software and recommendations. It's the thread that contains Taco's post.

https://www.htguide.com/forum/showthread.php?t=24008

[fjhuerta]

Hi Rick,




Javier,
One more vote for LspCAD. Thanks. And more stuff to get for impedance data. I haven’t seen many people mention Woofer Tester before (a few, but not many).

Jon, I got te Woofer Tester before I got LSPCad. The reason was, I didn't know if I got LSPCad I could do impedance plots without the WT2.

But - the Woofer Tester can also calculate all T/S parameters of a driver, test cables, help me unwind inductors, calculate VAS... it's a very handy tool to have around.

[Paul H]

Hi Paul,

Yeah, exactly the same attitude I have. You’ve got LspCAD for the crossovers. I recall your mentioning that in another thread. What do you use for measurements of sound (Just MLS?) and then the impedance and T/S? You must have some hardware for the latter?

I use JustMLS for MLS measurements ( ) and impedance. I rely on manufacturers' info for T/S parameters - the other info is less critical and used more for box sizing than crossover design.

I have a jig for impedance/freq measurements - it's a diy collection of toggle switches and wiring.

[JonP]

Wow- thanks guys. Much more feedback than I expected. And some strong feelings for and against various software.

This speaker building is supposed to be 100% fun for me. And I spend so much time in front of a computer at work, I’m probably willing to spend some money for software that will take less fighting with. Maybe I’m lazy. Heck, I’m typing this from a Mac. (but have a PC at home for speaker work)

I’m hesitant to start down the Speaker Workshop road because I’ve read of many people being frustrated with it. For example:

(snip)

Hmmm…

I'd started with Speaker Workshop, and yes it is quirky and frustrating. Once I figured out several things, an important one being how to set my soundcard up for clean, unclipped input and output, (get that program that Claudio suggests to save and reset your mixer settings!!) things weren't too bad and I could get things done. I was primarily using it for measurements, not as a full one end to the other design tool.

But I was frustrated enough to fork out for SoundEasy, thinking "professionally done, has more capabilities, should be better documented" etc. Well, I may have more frustration equity in SoundEasy than I ever logged in SW! To be fair, I was trying to use SE as a full design tool...

Things have gotten better, John K's example guide for SE got updated to match the current version's interface changes as well as expanded in scope, it goes a long way to helping you thru the winding paths... Jay Butterman's version 2 SW guide is a big help, as well as reading a few different' folk's takes on how to do things.. (Claudio, Roman Bednarak, etc)

Bottom line, I'm not sure if I can say I found SE any "easier" than SW!

Another note, you don't need a Wallin Jig to run SW, they all basically use the same method of series resistor measuring. So a few wires, that resistor and clips will do to get you started, but use a fairly good quality shielded cable, and keep the unshielded ends short.

I think it's conceptually easier to learn the process, less mistake prone, if you use simple cable jigs like Claudio illustrates. Many times I had problems where I had been successful in the past, only to find out I had the switches in the wrong positions. Later, when you're familiar with things, a nice "everything in one box" box is a good thing to have.

[fjhuerta]

I'm still deciding whether to "upgrade" from LSPCad 6 to SoundEasy.

On one hand, I can do distortion measurements. Something I don't think I'd be able to do with any other tool for as little money as SE.

On the other hand, I guess LSPCad is so easy even someone like me can use it.

I only have one question for all of you. Do you measure & design on-tweeter axis, or do you measure on -driver axis and then use LSPCad's driver positioning parameters to approximate an on-tweeter axis response? Reason I'm asking is, I have never been able to reconcile the measured response on-tweeter axis of the woofers with the predicted woofer response on-tweeter axis -especially with metal cone woofers.

[dlneubec]

FWIW,

I'm using SoundEasy and though a bit confusing in the interface, if you get John Kreskovsky's V13 design guide, it is really not that hard.
SoundEasy V13 Design Guide

I bought it in late November, learned to do measurments and crossover simulations, box modelling, etc. in just a few weeks.

John uses SE and so does Zaph. See what Zaph has to say about SE and why he chose it over other programs mentioned here.
Zaph on SE

I like it because it has all the tools you need in one program and once you get to know the interface and how it flows, it is very quick and easy to use, and very powerful, IMO. I have a pretty easy to build jig that I came up with after looking at various sources that you can use for various types of measurements (see below).

If you are going to use a laptop with an external USB sound card, I recommend this one for SE.
Edirol UA-1EX

It worked much better for me than the oft recommended M-Audio Transit.

[ThomasW]

I'm still deciding whether to "upgrade" from LSPCad 6 to SoundEasy.

It's really not an upgrade. The only thing SE offers is somewhat better measurement tools. And the downside is it's so hostle second parties need to write users manuals.

JonMarsh, who can have any software program he wants, started with SE and went to LspCAD because of it's superior modeling tools.

Since you already own and know how to use LspCAD, just buy a more sophisticated testing program. Doing that would be cheaper and less hassle than buying and learning SE.

[JonW]

This thread has been a great help to me. I hope it benefits others as well.

I placed my order for LspCAD yesterday.

[TacoD]

Money well spend. If you already have lspCAD, I see no reason the buy SE. Then you'll end up with two programs which overlap >50 %.