New FAQ - How to design without measurement gear

[E Biddy]

Re: New FAQ - How to design without measurement gear

I'm following the guide and I get to the "auto extract phase" part and I press the button and nothing happens. I'm pretty sure that I didn't miss a step and all of the other steps worked perfectly.

Any help is appreciated.

Eric

[dlr]

Re: New FAQ - How to design without measurement gear

Would the cavity of the adjacent driver (as well as it's surround) cause a diffraction ripple for the tweeter? If so, is this taken into account in the modeling? Or can it be?

This can at times be more influential than the baffle edge. In between tweeter and adjacent driver is the first place I look to add felt. Hard coned drivers tend to diffract more, possibly due to there being little cone damping and/or the cone depth is greater. Modeling seldom shows this as it's so varied. Measurements are needed, but it's unlikely to be useful to try to fix it in the crossover as it's very sensitive to axis. Use of felt fixes the issue for pretty much all axes.

I'm also not aware of any studies that indicate whether or not there's any real audibility. Where it may be more likely to come into play might be trying to use the crossover to fix it that only does so on some particular axis.

dlr

[Ryan_M]

Re: New FAQ - How to design without measurement gear

Sounds like you have the hang of it!
Keep it up!
Wolf

Correct on both counts. (I find most small speaker experience the full step though, but what you described is the way to add some back into the simulation if you want to).

Excellent.

I've come up with a few more questions...

1.) Would the cavity of the adjacent driver (as well as it's surround) cause a diffraction ripple for the tweeter? If so, is this taken into account in the modeling? Or can it be?

2.) What about driver spacing? I've read the XO frequency should be less then one wavelength. Is this something that is automatically included in the modeling or something I just need to be aware of.

3.) How do you model an MTM? Again I've read the mid spacing should be less than the wavelength at the XO frequency. Are there any other pitfalls or does it go similar to an MT?

4.) I notice a lot of boxes are designed to be just wide enough to give clarance for the largest driver. Is there a reason other than just cosmetics that the baffle should be kept narrow? I've played with the diffraction sim and I can see differences widening the baffle by an inch or so but I don't know what to consider better or worse.

5.) Is there a standard convention used for speaker orientation when modeling? i.e. straight on at tweeter level. AFAIK the rule of thumb is to set up an equilateral triangle between the speakers and listener for a standard setup, and most people like the speakers to be facing straight ahead - for cosmetic reasons. If the speaker height was setup the the tweeter was at ear height then that would put the woofer slightly below a few degrees vertically and both tweet and woofer would be off horizontally. Should I be including this in the modeling? Do others? Am I getting too nit-picky?

[WayneN]

Re: New FAQ - How to design without measurement gear

Excellent thread Paul!

It’s nice to see steps in layman’s terms, its people like yourself, Jeff B, Wolf, dlr, and the many other seasoned veterans taking the time to share the knowledge you’ve gained over the years.

Ten years ago and a couple of back surgeries later I changed jobs from Head Custodian to IT, I was asked to write three tutorials on how to use the 3 different Energy Management Systems in one School, Barber Coleman, Honeywell and Johnson Controls, I know firsthand how much time is involved in taking information and converting it into layman’s terms, I cant thank you gentlemen enough for doing this!

It might be just me? But I find after sitting in front of a computer for 6-8 hours a day I prefer to read my information or articles in paper form, it would be great if you gentlemen could have your information available in a PDF this way we could download it, print it and keep it in a binder for reference.

Jeff, I do have your 35 page guide printed and in my binder with a tab under your name, just need to get all the renovations done then I can sit down and do some serious reading

Thanks again gentlemen for all your efforts and knowledge sharing!

WayneN

[maynardg]

Re: Warning Possibly Heretical Controversy

I still fire up the PCD when I get jammed by a design but I would rather spend my time listening to and adjusting a new project. For me that is where the fun really exists. I'll measure FR and Z as I go along but I don't let that data be my one and only rudder.

Ditto. In the final voicing stage (to which the most overall time is devoted, by far), I've been misled by trying to get too perfectionist and overworking XO's (earlier, LSPCad, more recently PCD). It may be that I find it easier to spend time in front of the computer. Easier, but but not better in the later stages of XO work.

The "art" part of the process is a discipline as well - to be informed, but not driven, by the model.

[Jeff B.]

Re: New FAQ - How to design without measurement gear

I don't know how I missed this thread for so long but it was a nice find! Thanks again Paul (et al) for taking the time to share your knowledge!

I've done this type of stuff before but for whatever reason wasn't nearly as slick as your tutorial. Maybe a different program here or there or possible just because these are newer versions. I followed the instructions and before I knew it I was loading up the crossover design simulator... I kept having this nagging feeling like "that's it?". Of course following someones lead vs. doing completely on my own is a whole other animal, but it's still a testament to your authoring skills!

I have a couple questions though:

1.)when splicing in the modeled box response below 200Hz, do you adjust the SPL scale so it matches up the best with the existing curve? I assumed it should and that's what I did but I want to double check.

2.) From what I understand modeling the baffle step gives you the FULL baffle loss as if the speaker was way up in the air somewhere giving the theoretical 6db loss. If one wanted to only add say 4db of baffle loss then would I - after that is modeled - apply an "inverse" 2db in that section? again that's what I did because it looked right but I'm not sure.

I'm going to start a thread of my results because it went so well (using your method and then slightly modifying an existing design) that i'm actually a little suspicious that I did everything right.

Thanks again!

Correct on both counts. (I find most small speaker experience the full step though, but what you described is the way to add some back into the simulation if you want to).

[shawn_a]

Re: Warning Possibly Heretical Controversy

Not heresy from my chair only because I agree with you. I evolved from being a modeling junky to using my instinct and ears with a smattering of measurements. Once a person has a goodly amount of projects under their belt and starts to understand different XO topologies and what happens when you do action X to piece Y in a filter I think that person really should be using their ears more and more. As I've said before, they will not lead you wrong as long as you are honest with yourself and know how an instrument or a person's voice should really sound. Listen for the details, the grit and minutiae in a recording. There's a wealth of info in the little spurious sounds too after all.

Case in point; I have one particularly clean recording where you can hear a music stand squeaking right of center of the orchestra as if someone had turned it to the side and then back to face them. That's a noise I know pretty well after spending most of my school life in music classes of one kind or another. I've listened to that particular part of the track more times than I care to admit because for me it is a reference bit of crud that I happened to know what sounded like in the flesh, or metal in this case.

I'm not dissing modeling at all because it's too valuable of a tool to use whether you're just getting started in the hobby or you are a seasoned veteran, so to speak. I still fire up the PCD when I get jammed by a design but I would rather spend my time listening to and adjusting a new project. For me that is where the fun really exists. I'll measure FR and Z as I go along but I don't let that data be my one and only rudder.

And I've ran headlong into what you said about the room doing some really interesting things to live instruments or voices. The stairwell at my mom's place is a super cool place to sing. Go figure it's right under the shower, another wonderful place to belt out some neat vocals. At some point I'm going to sit down on the bottom step and record some tracks.

shawn

[Wolf]

Re: New FAQ - How to design without measurement gear

Sounds like you have the hang of it!
Keep it up!
Wolf

[Ryan_M]

Re: New FAQ - How to design without measurement gear

I don't know how I missed this thread for so long but it was a nice find! Thanks again Paul (et al) for taking the time to share your knowledge!

I've done this type of stuff before but for whatever reason wasn't nearly as slick as your tutorial. Maybe a different program here or there or possible just because these are newer versions. I followed the instructions and before I knew it I was loading up the crossover design simulator... I kept having this nagging feeling like "that's it?". Of course following someones lead vs. doing completely on my own is a whole other animal, but it's still a testament to your authoring skills!

I have a couple questions though:

1.)when splicing in the modeled box response below 200Hz, do you adjust the SPL scale so it matches up the best with the existing curve? I assumed it should and that's what I did but I want to double check.

2.) From what I understand modeling the baffle step gives you the FULL baffle loss as if the speaker was way up in the air somewhere giving the theoretical 6db loss. If one wanted to only add say 4db of baffle loss then would I - after that is modeled - apply an "inverse" 2db in that section? again that's what I did because it looked right but I'm not sure.

I'm going to start a thread of my results because it went so well (using your method and then slightly modifying an existing design) that i'm actually a little suspicious that I did everything right.

Thanks again!

[dano]

Go

Thanks guys now were getting somewhere. I remember years ago I designed an auto system for IASCA competition. I designed and build the crossovers and we put everything in. When we finished and started listening to it everyone was pumped.

The owner and my installers were all high five-in, exclaiming how good it sounded. I said it aint right! I heard a faint raspy sound so I put a zoble on the mid woofers. That small change made the difference. The mids rising impedance defeated the crossover letting the mids play too high. I was hearing their cones breakup.

I don't know a lot about designing home speakers. I appreciate you guys helping me out. I'm not saying I was great at car stereo but I got by.
________
Fifth-Generation Ford Taurus History

[LouC]

Warning Possibly Heretical Controversy

I agree totally with Paul, particularly with metal cones. I know some tweeters (of any material) just grate on me when pushed too high.

Flat is probably a good goal, but if you do get there, most people find boxes voiced that way to be on the bright and edgy side. I recall Dennis Murphy visited Tucson once, and we chatted about that topic. I think we both agreed that a 1-2 db downward tilt from about 2K and up usually sounds better to most people in real rooms. (The BBC dip) (With your music, & your hearing...)

Personally, I design flat, then voice by ear. Usually that winds up with more tweeter padding. Sometimes I go back and start the whole crossover over again. Once it sounds right, I put the measurement gear away.

I'm a rock/pop/blues guy. Flat response often collapses the stage, and puts the vocalist right in my face. If the master/recording is poorly done, it sounds like someone is spitting in the microphone. I'd rather sit in the audience, than inside the band, but that's my preference.

Xavier came by my house one day, and played a couple of short pieces on his viola for my wife and I. In our carpeted great room, he sounded dead and tiny, when we moved into the kitchen, the difference in the life of the music was astounding. I'll bet they measured the same from 1 meter.

[Paul Carmody]

Re: New FAQ - How to design without measurement gear

"If you don't know what you are doing designing speakers, you'll get flat measuring designs that don't sound good"
Could you explain a little? Do you mean obvious things like crossing tweeters to low or woofers too high, or something else less obvious? I'm trying to learn all I can.

That is an excellent question.

When I was first designing my own speakers, I figured "If the overall FR is flat, then it's all good. It all comes out in the wash." And really, since I started out using drivers with pretty flat FR, that theory pretty much worked for me.

But then after working with metal cone drivers, I started to notice that a speaker would have extra sibilance or a sort of "glassy" sound in the upper midrange/lower treble, even though its overall FR was flat. I realized that this was the breakup mode of the woofer; it didn't show up in the system FR, but it was clearly audible and annoying. So I learned how to use notch filters, and the extra sibilance went away.

"So why doesn't it all come out in the wash?" I'd wonder. Then I thought of an analogy that helped me. For example, say I were listening to an orchestra, and say the oboe was playing a counter-melody to the rest of the orchestra. If I were simply watching a VU meter of the performance I'd probably only see the peaks created by the orchestra as a whole. However, most people listening could probably pick out the oboe part with its counter-melody. For a non-classical example, think of a rock song with an acoustic guitar or tambourine added; again, it doesn't show up on the VU meter, but you can hear it.

[robert]

Re: New FAQ - How to design without measurement gear

When I built my 8" 2-way, I measured farfield/nearfield/port outputs and spliced them together. I also measured the 27TDFC. When I started designing the system, I shot for flat frequency response with about 4-5dB of baffle step compensation. I brought these to Iowa and they sounded heavy and dull compared to other's work.

2 reasons this happened:

1.) The cabinet and driver were capable of 28Hz or so and hit room modes with a vengeance, so bass response was heavy even though it measured as having considerable rolloff. I should have used a smaller cab, 1.4-1.5cf.

2.) With that bass response, keeping the driver's response near flat through the midrange made the system sound even heavier. Note that with a smaller driver, this may make the system sound more full, but with an 8, especially the HiVi, it wasn't a good idea.

When I reduced the series coil on the woofer and tweaked everything else out, it sounded much better, although the strong bass (lots would like it in their living rooms) is still a touch heavy, though much improved from Iowa. I may actually further reduce the woofer coil to see what happens. Easy enough.

Another issue: the Seas 27TDFC requires a series RC circuit to roll off the very top end a bit. When I measured, this looked pretty benign. When I listened, OUCH! You could hear scratchiness, something I hate from a tweeter. I looked at Zaph's SR-71 and used the same circuit tweaked a bit, and scratchiness away.

At Fort Wayne, I got Wolf feedback and ended up increasing the series resistor on the tweeter 1/2 Ohm to further sweeten the tweeter. Really good final touch.

So lots of things that looked good on measurement didn't sound good. Another argument for a proven design to compare.

Also, if the driver has high harmonic distortion, you need to dip it a bit where it peaks. This is because to the ear, it'll sound brighter than it measures, since the harmonics add to the output. I don't have a quiet enough house/neighborhood to do distortion measurements but hope to someday.

Cheers / Robert

[dano]

Go

"If you don't know what you are doing designing speakers, you'll get flat measuring designs that don't sound good"
Could you explain a little? Do you mean obvious things like crossing tweeters to low or woofers too high, or something else less obvious? I'm trying to learn all I can.
________
SHINTO FORUM

[robert]

Re: New FAQ - How to design without measurement gear

Unfortunately, some of the measurement freeware out there is very tough to use. I won't touch ARTA with a stick, no documentation. As an electronics tech, measuring things and setting up test equipment came pretty naturally, getting the PC to do what it was supposed to was a whole different animal, and can be pretty much at the whim of the pc mfr. In other words, an individual experience, especially with laptops.

SoundEasy's measurement suite is non-obvious but by blindly following JohnK's CD and figuring out what was going on, I was pretty successful, and pretty much understand the process now. I read the manual too, a big help. You also have to practice it lots so setup and getting things to work is automatic. JohnK goes into farfield/nearfield splicing, how to ensure your splice was correct, etc. With sims, your bass response shows flat to F3, and it ain't so. Did I lose time when I could have made 6-7 speakers? Yes.

Measurement capability doesn't necessarily mean a good design, either. If you don't know what you are doing designing speakers, you'll get flat measuring designs that don't sound good. This can also happen with sims. In the end, I am happy that I took the time to learn it, but it would have been nice to produce, too. I agree with Paul - do a couple in sim, then break out and get some documented measuring software and equipment, and set up just like someone else that has been successful. Don't customize, don't buy that freakin' cool sound card you just have to have, get one that has worked for someone else without complex issues. I followed Dan Neubecker's rig, and used the same sound card. TA DA! It worked and I have never had a lag or delay issue with 2 different pc's. In fact, Ed and I shot some perfect minimum phase measurements a few weeks ago. Amazing, following a proven formula :D

Again, it took some time out of my building and buying drivers, but I bought myself a 12-pack when I had measurement, modelling, and sim all up in SE.

Cheers / Robert