Distortion measurements

Nicholas_23

Whats good and whats bad when it comes to distortion measurements ,visually when looking at the graph?
How do you guys do it ? I saw Craigs post earlier on the SB ceramic divers that he recently purchased . So crank that shit up and let it go nuts , or... how do you guys do it ?

NavyGuy

Well, first I'd change the scale to 5 dB increments so that you can actually read the data. 

I usually test at 2 different levels. The biggest thing is you have to make some kind of standard for yourself so that your measurements are comparable. So you need to test at the same levels (output or voltage - I use output) and distance every time or you are just getting a bunch of random information that you can't compare.

Wolf

I tend to test at 90/95dB to swamp out the noise-floor, and I actually do test at a meter distance because I do it just like the FR measurements I take, and in the same setup. Of course- I do where ear plugs while I do this, and make sure the woofers aren't bottoming. This is a comparison of the Cecropia project with the CAW634 and the TSCW636. The xovers/woofers are different, and therefore so is the HD.

jr@mac

I shoot for around 96db at 1/2m for raw drivers. Can let the smoke out of shitty little woofers and tweeters, though - so be careful.

On finished systems I generally target 90db at 1M, which is loud enough to let you know if the woofer is going to dump on you below tuning - they generally do. 

My measurements are generally useful to me primarily, since I do not follow any real established "method'. I set level at the distance I plan on measuring using my SPL meter and pink noise. Then I let her rip with a 128K sine wave. I like to torture during distortion testing. 

The things I look for are: odd peaks here and there. They can indicate suspension/cabinet issues. Where the woofer starts shooting up in the 3rd harmonic, ditto for the tweeter. 

One thing I look for in a design is a clean transfer from driver to driver on distortion levels. Not to pick on Ben, but his first iteration of the Cecropia shows a significant cliff in distortion at the hand-off from woofer to tweeter. The second iteration shows a markedly better hand-off. Those cliffs are due to bread-and-butter motors on woofers being mated to much nicer tweeters. That being said, the first version is not exactly a DS-175 level of shitty (it is fairly respectable, to be honest) - so don't read too far into my comments about "bread-and-butter" motors. Just pointing out one of those little "secrets" about "synergy". 

Another thing to do is to keep cranking the sweeps in 2db increments and overlay. You have to subtract 2db off each subsequent sweep to get a good overlay. It can tell you how loud you can play before compression kicks in. My experience has been that the typical woofer starts taking a shit by jumping the gap long before compression is an issue, so IMHO, the value of compression testing is of concern only if you have already taken steps to protect the woofer from being the weak spot. so throw a high pass on it to protect the woofer, and have fun with ear plugs trying to see where the tweeter starts failing. It is usually way louder than you would ever listen to - again, IMHO, making the value of compression testing somewhat marginal since the majority of us are not designing towards outdoor rock show levels. Additionally, if you perform distortion testing loud enough with a long enough sweep - non-linearities due to compression will probably show up anyways. 

Basically, crank it up and be prepared for losing a driver or two as you learn it. Yes, also put things in a little higher resolution. I use 5db in ARTA, and try to minimize the min/max db as well. Avoid the "oh my god my speaker is so smooth and flat" default display settings of Omnimic if you can. 

Wolf

The reason for the spike in that version was actually the cone diameter dip and resultant energy storage. The Supreme just does not have that problem. Obviously, it was one reason I was unhappy with it and why I upped the anty(sp). The bump on the tweeter HD is higher in the lower end 2nd order because of the added Xpad to flatten out the baffle response on the revision. The first didn't really need flattened out there as it was pretty flat, but the dip at 1.2k really left something missing. Much better now!

Nicholas_23

Thanks for the input guys, much appreciated! I will head back in with the info given and give it another go. 

Kornbread

Do you test tweeters full range at that 90+ level?

jr@mac

I do, yes. Let 'er rip!

One of those hidden costs of DIY, actually, is the chance of letting the smoke out of a driver or two. The people who ride the wave of kits/established designs benefit greatly from people who are willing to spend the time and money, running the risk of blowing an expensive driver while rooting around for the truth.

isaeagle4031

Except for a ribbon/amt, I test every tweet at 90db at 1m distance, full range. Ribbon/amt get a cap f or protection. 

dcibel

AMTs are much more robust than ribbons, maybe even more so than most domes. The only AMTs that I've seen fail were used improperly and the mylar diaphragm started to melt. Under this condition the "coil" was still in tact and they still made noise.

Keep in mind that 90dB into most tweeters is still 1W or less, and you are playing them for only a few seconds at a time. It will probably push the suspension to the limit, but I think its very unlikely to burn the coil over that duration.

Kornbread

Dang! 90+db full range @1 meter seems awfully hard on a tiny little tweeter.  Accidentally tested the freq of an xt25 full range at somewhere around that level (maybe a bit higher) and it made squawking noises that didn't sound so nice.  Thought I killed it. 

   ... but that's the way to do it?  I've been taking it too easy on them. 

rjj45

good discussion!
I'm really enjoying the different methods.