Zaph|Audio

Posted May 9, 2005. Will be updated again when I get time.

About SoundEasy
When it came time to choose a design/measuring package, I looked at Soundeasy, LSPcad and Speaker Workshop. Being free, I was already running Speaker Workshop but decided I could do better. I chose SoundEasy over the other options for several reasons.

Speaker Workshop was the most user unfriendly program I've ever used. It's also the most difficult to set up and calibrate. However, once you get past those issues, the core modeling ability works as well as any other package. Most people don't make it past the learning curve however, and I have a lot of respect for those that do.

LSPcad is probably the most popular choice among designers. It's easy to learn and use, at least relative to other packages. However, it falls seriously short in the measurement department. It comes with a simple measurement app called JustMLS, and the name is suitable because that's all you get. I can't see owning LSPcad without also having to purchase a measurement app such as Praxis or CLIO. That brings the price well up into the 4 digit range.

SoundEasy's Easylab is a large suite of measurement applications. This is the reason I chose SE over the other options. It's pretty much everything I need in one package. While LSPcad's interface may be slightly more intuitive, it is my opinion that it is difficult to design a good sounding system using LSPcad alone. Without sufficient distortion testing apps, You're really just taking a guess at where the best crossover point is, or even if drivers are suitable to be used with each other. Modeling a flat response is only one portion of the design process.

Support is equally bad for all of these packages. In most cases, you'd be lucky just to get a response from an email sent to the program author. It is somewhat understandable however, as their mailboxes are filled with questions that they should not be asked. These programs are power tools to aid a designer. They are not training tools, and there are way too many people buying these packages without the proper background and experience. Leave the authors alone until you know what you're doing.

Most help can be obtained from the software's respective forum. In the case of SE, it's the SoundEasyUsersList on Yahoo. Be sure you read the manual before you get there, or it will only be a matter of time before Kreskovsky smacks you down with a "RTFM". Heheh, sorry John. But seriously, questions like "How do I reverse driver polarity" should probably not be asked there. All that and more is in the manual.

I started with SE version 8. I am currently running version 14. Bohdan, SE's developer, releases frequent updates, usually with major new features in addition to fixes and tweaks. At this point, SE has grown far, far beyond any other software package in this price range.

Hardware
My current hardware, as of 05/09/2005: P4/3Ghz, 1 gig ram, Creative Labs Audigy 2 ZS Platinum. This is the version that has the drive-bay mounted panel. This panel has a pair of RCA input jacks for Auxiliary 2, which is what I use for my testing probes.

Audigy 2 ZS front panel

The Audigy 2 ZS is a very nice card with some limitations. First off, this is one of the quietest cards you can buy. Creative's claims of a 108dB SNR are real and are actually somewhat conservative. I've tested noise floors as low as 115dB. It's my other equipment that brings my useful noise floor limit up to about 100dB. This sound card is only 1/2dB down at 44Khz. The limitations of this Audigy are that it can't do more than 2 channels. 4 channel emulation is out of the question. That may be an issue for some people.

Audigy response curve

I started out with a DIY Panasonic capsule mic and a DIY preamp. One day my preamp died so I just bought new Behringer equipment. It was cheap and worked perfectly. I came to the conclusion that I should have just got that equipment in the first place. This is one situation where the few bucks saved is clearly not worth the DIY effort.

My mic is now the popular Behringer ECM8000, and my pre-amp is a Behringer EURORACK UB1202. Note the UB802 will work fine, but the UB502 will not since it doesn't have phantom power for the mic. These items were purchased at 8th Street Music. (8thstreet.com) I rounded out the package with a boom mic stand from Rat Shack and a good but inexpensive low Z XLR mic cable. The ECM8000 is a very flat mic. You can probably get by without even using a calibration file, as it's only about 1.5 dB down at 20kHz. On the other hand, the common calibration files for this mic are very consistent and will probably work fine for you. My mic is calibrated and the cal file includes the mic cable and the speaker cable.

My integrated amp is a Teac Reference series. It's a well built small amp with very flat response out to 50kHz and a low noise floor.

SE - Creative Labs setup
Getting any Creative Labs card to work can be a complete pain in the ass. By default, several processing options are turned on and troublesome drivers are installed. Here's some tips:

My SE setup is based on the hookups shown in the manual. There are three distinct hookup modes: calibrate, T/S measure and SPL measure.

No need for a Wallin Jig
All my connections are desk mounted. There's really no need for a wallin Jig. As shown in these pictures, from the left to right I have L and R rca inputs, 2 pairs of banana plug jacks and a switch. The rca inputs run to the Auxiliary input on my sound card. The next two banana plug jacks connect only to a 10 ohm resistor. The final two banana plug jacks connect to my amplifier. The switch is a double pole double throw, normally closed. It has 6 tabs on the back. The middle two come from the amp, the two lower tabs run to the banana plug jacks, and the top two run to my speakers. This could probably be done with a SPDT and a common ground, but I had this switch laying around. It's a switch that allows me to go from desktop music to speaker testing, and may not be needed for some. Then of course, there is the 1/8" stereo to rca jack that connects my sound card line out to my amplifier input.

Plugs, SPL testing

Plugs, T/S testing

This whole setup is based on the concept of using piggyback banana plugs. These are banana plugs that allow another banana plug to be plugged into the back of it. The assemblies needed are a pair of probes rca with a 47k resistor on the middle contact and banana plugs on the other end, a single jumper with banana plugs on both ends, and the speaker connectors with banana plugs on one end and alligator clips on the other. All wires shown in the image below are 18 gauge. If your SPL testing takes place in another room, like me, you may elect to have heavier gauge wires for that usage to minimize the effect of the wire. But for impedance testing, I'll just use my smaller and shorter set of speaker connectors shown below since the run is very short.

Probes, jumper and speaker connectors

Infinite baffle
For initial driver testing, I use an infinite baffle. Of course it's not really infinite, as I live in a normal size house, not an anechoic chamber. It's a 4x8 foot multiple layered and braced MDF with an 11" square cut out of the middle of it. I cut the top 3 feet off so I could move it around without problems. I have a couple of fold-out legs so I can stand it up in the middle of a room. When I'm not using this baffle, it sits in the back of my listening room, covered with a hanging rug. When I want to test a new driver that doesn't fit one of my existing cutouts, I just make a new insert. This insert just screws in. The insert is held in place on the back side with a couple of boards, and the edge of these boards are lined with felt to minimize the early reflection they cause.

I usually countersink drivers for accurate response curves. I always countersink tweeters, but I only countersink woofers when I intend to do so in the final box mounted design.

Mic position varies with the driver tested. With tweeters, I can get as close as 4" and still have an accurate response curve. With woofers, I have to back the mic up more and more as the driver gets bigger for an accurate representation of the top end response. Tweeters can be done in one step with a small amount of gating taking care of the room reflections. Woofers however have to be done in two steps with close mic (1/4") and no gating, and spaced mic with gating. The gating varies depending on how close I can get the mic.

Tweeter on infinite baffle w/nearfield mic

This infinite baffle is actually just a very large dipole. There are several factors influencing this type of measurement. Floor reflections and dipole cancellation are the primary issues. Close mic measurements do not have these issues, but as the mic is moved away from the driver, they begin to show up. Reflections are always an issue, but for close mic measurements they don't show up because the initial signal is much higher than the reflected signal.

Woofer on back side of baffle insert

It's easy to think that an infinite baffle is just a big board with a cutout for mounting drivers, but that's not the case. It required a lot of experimentation to get it right. Just propping up a big board with a driver in the middle of it results in some pretty bad board resonances that can propagate back into some of the measurements. The baffle has to be both stiffened and dampened. In the below picture, you can see the multiple layers of boards comprising the constrained layer damping, and the stiffening braces. Notice the weird cutouts on the braces - those had to be put in there to remove the reflection from the back of the baffle which will go back through the cone and into your measurements. Notice the thick felt layer around the insert back boards - yes, another potential place for reflections. Unfortunately, this testing baffle weighs about 150 pounds after all this, so it's not exactly fun to get out all the time for one or two driver tests.

Backside bracing

Vas testing box
I always use high level delta compliance testing because it is much more accurate compared to delta mass testing. (not to mention the questionable practice of sticking stuff to your cone) The nifty thing about my baffle insert is that the same insert also bolts on to a Vas testing enclosure. I have one main enclosure of roughly an 11" cube that is .57 cubic feet in volume. Then I have a couple different size inserts for testing smaller woofers that need less volume for a better delta.

Silence!
My testing computer is not in the same room as the infinite baffle. This is done intentionally so CPU fan noise does not become a factor. 30 feet of mic and speaker cable is fully compensated in my calibration files. I take pride in the fact that the noise floor in my home is incredibly low. The house was built with 6" fully insulated walls and is on located a quiet street. I take full control of all things in the house that generate noise, including the refrigerator and the HVAC. And last but not least, my spouse knows when to keep quiet. That is to say, she knows when I'm testing and will avoid polluting my impulses and sweeps by slamming cupboard doors or putting dishes away.

Bottom line, I've got an 80dB noise floor on the low end, and almost 100dB in the midrange.

Of course, I'm listening with every test, and when the surprise noise does happen, I'll restart.

Tips
After initial infinite baffle testing of response curves and distortion, and the T/S parameters are obtained, I am then in a position to decide on the remaining design issues such as driver compatibility, crossover type/frequency and enclosure design. If I'm still not comfortable making those decisions, I can use the infinite baffle curves saved in my driver files and add the diffraction response to them and begin modeling to see what works the best. Of course I have to note differences in level if required and I have to adjust for phase differences if I use different a mic distance. If I wanted to, I could completely design from infinite baffle measurements. At some point however, I build the box to increase accuracy and begin the serious tweaking.

Before I go any further with the tips, I need to point to John Kreskovsky's SE design guide. (well worth $11) I started with his original walkthrough for version 8 and it had me up and running within a couple hours. Some of you with less experience may require more time than that, but then realize that you would be completely lost without this design guide.

Versions 10 and above have 5 post processing slots, located on a separate tab. Check boxes allow you to add or subtract responses and place the result in the buffer. The 5 slots are separate from buffer and driver. The buffer slot is a "working area" and the driver slot is where the actual measurements go when you first take them. Overall, this post processing setup makes working with measurements easier and less time consuming, particularly if you screw up at some point. You won't have to start completely over, you'll just have to redo that step.

In this example, I use all 5 slots to get the proper final curve of a vented woofer enclosure. These steps are not the definitive way of doing it, nor are they detailed. They may help steer you in the right direction however.

This is the woofer measurement of a tall, slimline floor standing vented system.

1. Initial measurement, gated, 4ms window.

2. Nearfield measurement, 200ms window, with diffraction added in. HBT out a small bit of garbage below 15hz, and adjust delay to get the phase lined up at the merge point.

3. Merged at 600hz. The best merge frequency varies.

4. Nearfield port, with the level adjusted so the curve comes close to converging at 0hz with curve 3.

5. 3 and 4 added together.

D. One more HBT to tail the ends and slightly adjust delay through the midrange, then copy it to the driver file. Save the driver file.

I'm sorry, I'm not accepting questions in support of SoundEasy. This document is all you get from me. If you have more questions after reading this, you can ask for help at the SoundEasyUsersList or ask Bohdan himself. Before you do that however, be sure to read the manual.

Page done by John "Zaph" Krutke © 2005
Also visit -Zaph|Audio-